Search results for: transformer fine-tuning

Authors: Ma. Gracia Corazon Cayanan, Kai Yuen Cheong, Li Sha

Abstract:

Training a language model for a minority language has been a challenging task. The lack of available corpora to train and fine-tune state-of-the-art language models is still a challenge in the area of Natural Language Processing (NLP). Moreover, the need for high computational resources and bulk data limit the attainment of this task. In this paper, we presented the following contributions: (1) we introduce and used a translation pair set of Tagalog and English (TL-EN) in pre-training a language model to a minority language resource; (2) we fine-tuned and evaluated top-ranking and pre-trained semantic textual similarity binary task (STSB) models, to both TL-EN and STS dataset pairs. (3) then, we reduced the size of the model to offset the need for high computational resources. Based on our results, the models that were pre-trained to translation pairs and STS pairs can perform well for STSB task. Also, having it reduced to a smaller dimension has no negative effect on the performance but rather has a notable increase on the similarity scores. Moreover, models that were pre-trained to a similar dataset have a tremendous effect on the model’s performance scores.

Keywords: semantic matching, semantic textual similarity binary task, low resource minority language, fine-tuning, dimension reduction, transformer models

Procedia PDF Downloads 211

Authors: Babak Forouraghi

Abstract:

A genetic circuit is a collection of interacting genes and proteins that enable individual cells to implement and perform vital biological functions such as cell division, growth, death, and signaling. In cell engineering, synthetic gene circuits are engineered networks of genes specifically designed to implement functionalities that are not evolved by nature. These engineered networks enable scientists to tackle complex problems such as engineering cells to produce therapeutics within the patient's body, altering T cells to target cancer-related antigens for treatment, improving antibody production using engineered cells, tissue engineering, and production of genetically modified plants and livestock. Construction of computational models to realize genetic circuits is an especially challenging task since it requires the discovery of flow of genetic information in complex biological systems. Building synthetic biological models is also a time-consuming process with relatively low prediction accuracy for highly complex genetic circuits. The primary goal of this study was to investigate the utility of a pre-trained bidirectional encoder transformer that can accurately predict gene expressions in genetic circuit designs. The main reason behind using transformers is their innate ability (attention mechanism) to take account of the semantic context present in long DNA chains that are heavily dependent on spatial representation of their constituent genes. Previous approaches to gene circuit design, such as CNN and RNN architectures, are unable to capture semantic dependencies in long contexts as required in most real-world applications of synthetic biology. For instance, RNN models (LSTM, GRU), although able to learn long-term dependencies, greatly suffer from vanishing gradient and low-efficiency problem when they sequentially process past states and compresses contextual information into a bottleneck with long input sequences. In other words, these architectures are not equipped with the necessary attention mechanisms to follow a long chain of genes with thousands of tokens. To address the above-mentioned limitations of previous approaches, a transformer model was built in this work as a variation to the existing DNA Bidirectional Encoder Representations from Transformers (DNABERT) model. It is shown that the proposed transformer is capable of capturing contextual information from long input sequences with attention mechanism. In a previous work on genetic circuit design, the traditional approaches to classification and regression, such as Random Forrest, Support Vector Machine, and Artificial Neural Networks, were able to achieve reasonably high R2 accuracy levels of 0.95 to 0.97. However, the transformer model utilized in this work with its attention-based mechanism, was able to achieve a perfect accuracy level of 100%. Further, it is demonstrated that the efficiency of the transformer-based gene expression classifier is not dependent on presence of large amounts of training examples, which may be difficult to compile in many real-world gene circuit designs.

Keywords: transformers, generative ai, gene expression design, classification

Procedia PDF Downloads 59

Authors: Babak Forouraghi

Abstract:

A genetic circuit is a collection of interacting genes and proteins that enable individual cells to implement and perform vital biological functions such as cell division, growth, death, and signaling. In cell engineering, synthetic gene circuits are engineered networks of genes specifically designed to implement functionalities that are not evolved by nature. These engineered networks enable scientists to tackle complex problems such as engineering cells to produce therapeutics within the patient's body, altering T cells to target cancer-related antigens for treatment, improving antibody production using engineered cells, tissue engineering, and production of genetically modified plants and livestock. Construction of computational models to realize genetic circuits is an especially challenging task since it requires the discovery of the flow of genetic information in complex biological systems. Building synthetic biological models is also a time-consuming process with relatively low prediction accuracy for highly complex genetic circuits. The primary goal of this study was to investigate the utility of a pre-trained bidirectional encoder transformer that can accurately predict gene expressions in genetic circuit designs. The main reason behind using transformers is their innate ability (attention mechanism) to take account of the semantic context present in long DNA chains that are heavily dependent on the spatial representation of their constituent genes. Previous approaches to gene circuit design, such as CNN and RNN architectures, are unable to capture semantic dependencies in long contexts, as required in most real-world applications of synthetic biology. For instance, RNN models (LSTM, GRU), although able to learn long-term dependencies, greatly suffer from vanishing gradient and low-efficiency problem when they sequentially process past states and compresses contextual information into a bottleneck with long input sequences. In other words, these architectures are not equipped with the necessary attention mechanisms to follow a long chain of genes with thousands of tokens. To address the above-mentioned limitations, a transformer model was built in this work as a variation to the existing DNA Bidirectional Encoder Representations from Transformers (DNABERT) model. It is shown that the proposed transformer is capable of capturing contextual information from long input sequences with an attention mechanism. In previous works on genetic circuit design, the traditional approaches to classification and regression, such as Random Forrest, Support Vector Machine, and Artificial Neural Networks, were able to achieve reasonably high R2 accuracy levels of 0.95 to 0.97. However, the transformer model utilized in this work, with its attention-based mechanism, was able to achieve a perfect accuracy level of 100%. Further, it is demonstrated that the efficiency of the transformer-based gene expression classifier is not dependent on the presence of large amounts of training examples, which may be difficult to compile in many real-world gene circuit designs.

Keywords: machine learning, classification and regression, gene circuit design, bidirectional transformers

Procedia PDF Downloads 60

Authors: Qisheng Hu, Jianglei Han, Yue Yang, My Hoa Ha

Abstract:

When a company is considering purchasing software for commercial use, contract risk assessment is critical to identify risks to mitigate the potential adverse business impact, e.g., security, financial and regulatory risks. Contract risk assessment requires reviewers with specialized knowledge and time to evaluate the legal documents manually. Specifically, validating contracts for a software vendor requires the following steps: manual screening, interpreting legal documents, and extracting risk-prone segments. To automate the process, we proposed a framework to assist legal contract document risk identification, leveraging pre-trained deep learning models and natural language processing techniques. Given a set of pre-defined risk evaluation problems, our framework utilizes the pre-trained transformer-based models for question-answering to identify risk-prone sections in a contract. Furthermore, the question-answering model encodes the concatenated question-contract text and predicts the start and end position for clause extraction. Due to the limited labelled dataset for training, we leveraged transfer learning by fine-tuning the models with the CUAD dataset to enhance the model. On a dataset comprising 287 contract documents and 2000 labelled samples, our best model achieved an F1 score of 0.687.

Keywords: contract risk assessment, NLP, transfer learning, question answering

Procedia PDF Downloads 129

Authors: Brandon Foggo, Nanpeng Yu

Abstract:

Power distribution circuits undergo frequent network topology changes that are often left undocumented. As a result, the documentation of a circuit’s connectivity becomes inaccurate with time. The lack of reliable circuit connectivity information is one of the biggest obstacles to model, monitor, and control modern distribution systems. To enhance the reliability and efficiency of electric power distribution systems, the circuit’s connectivity information must be updated periodically. This paper focuses on one critical component of a distribution circuit’s topology - the secondary transformer to phase association. This topology component describes the set of phase lines that feed power to a given secondary transformer (and therefore a given group of power consumers). Finding the documentation of this component is call Phase Identification, and is typically performed with physical measurements. These measurements can take time lengths on the order of several months, but with supervised learning, the time length can be reduced significantly. This paper compares several such methods applied to Phase Identification for a large range of real distribution circuits, describes a method of training data selection, describes preprocessing steps unique to the Phase Identification problem, and ultimately describes a method which obtains high accuracy (> 96% in most cases, > 92% in the worst case) using only 5% of the measurements typically used for Phase Identification.

Keywords: distribution network, machine learning, network topology, phase identification, smart grid

Procedia PDF Downloads 299

Authors: P. Bhavya, P. R. Jayasree

Abstract:

This paper presents a low cost transformer isolated boost half bridge micro-inverter for single phase grid connected PV system. Since the output voltage of a single PV panel is as low as 20~50V, a high voltage gain inverter is required for the PV panel to connect to the single-phase grid. The micro-inverter has two stages, an isolated dc-dc converter stage and an inverter stage with a dc link. To achieve MPPT and to step up the PV voltage to the dc link voltage, a transformer isolated boost half bridge dc-dc converter is used. To output the synchronised sinusoidal current with unity power factor to the grid, a pulse width modulated full bridge inverter with LCL filter is used. Variable step size Maximum Power Point Tracking (MPPT) method is adopted such that fast tracking and high MPPT efficiency are both obtained. AC voltage as per grid requirement is obtained at the output of the inverter. High power factor (>0.99) is obtained at both heavy and light loads. This paper gives the results of computer simulation program of a grid connected solar PV system using MATLAB/Simulink and SIM Power System tool.

Keywords: boost-half-bridge, micro-inverter, maximum power point tracking, grid connection, MATLAB/Simulink

Procedia PDF Downloads 341

Authors: Fatima Khurshid, Noor Ul Ain, Syed Abdul Rehman Kashif, Zainab Riaz, Abdullah Usman Khan, Muhammad Imran

Abstract:

As the world is moving towards extra-high voltage (EHV) and ultra-high voltage (UHV) power systems, the performance requirements of power transformers are becoming crucial to the system reliability and security. With the transformers being an essential component of a power system, low health index of transformers poses greater risks for safe and reliable operation. Therefore, to meet the rising demands of the power system and transformer performance, researchers are being prompted to provide solutions for enhanced thermal and electrical properties of transformers. This paper proposes an approach to improve the health index of a transformer by using nano-technology in conjunction with bio-degradable oils. Vegetable oils can serve as potential dielectric fluid alternatives to the conventional mineral oils, owing to their numerous inherent benefits; namely, higher fire and flashpoints, and being environment-friendly in nature. Moreover, the addition of nanoparticles in the dielectric fluid further serves to improve the dielectric strength of the insulation medium. In this research, using the finite element method (FEM) in COMSOL Multiphysics environment, and a 2D space dimension, three different oil samples have been modelled, and the electric field distribution is computed for each sample at various electric potentials, i.e., 90 kV, 100 kV, 150 kV, and 200 kV. Furthermore, each sample has been modified with the addition of nanoparticles of different radii (50 nm and 100 nm) and at different interparticle distance (5 mm and 10 mm), considering an instant of time. The nanoparticles used are non-conductive and have been modelled as alumina (Al₂O₃). The geometry has been modelled according to IEC standard 60897, with a standard electrode gap distance of 25 mm. For an input supply voltage of 100 kV, the maximum electric field stresses obtained for the samples of synthetic vegetable oil, olive oil, and mineral oil are 5.08 ×10⁶ V/m, 5.11×10⁶ V/m and 5.62×10⁶ V/m, respectively. It is observed that for the unmodified samples, vegetable oils have a greater dielectric strength as compared to the conventionally used mineral oils because of their higher flash points and higher values of relative permittivity. Also, for the modified samples, the addition of nanoparticles inhibits the streamer propagation inside the dielectric medium and hence, serves to improve the dielectric properties of the medium.

Keywords: dielectric strength, finite element method, health index, nanotechnology, streamer propagation

Procedia PDF Downloads 141

Authors: Mohamed Ahmed Khali

Abstract:

Numbers of deep well anode ground beds (GBs) have been retrieved due to un operated anode chains. New identical magnetite anode chains(MAC) have been installed at Raslanuf complex impressed current Cathodic protection(ICCP) system, distributed at different plants(Utility, ethylene and polyethylene). All problems associated with retrieving and installation of MACs have been discussed, rectified and presented. All GB associated severely corroded wellhead casings were well maintained and/ or replaced by new fabricated and modified ones. The main cause of wellhead casings internal corrosion was discussed, and the conducted remedy action to overcome future corrosion problem is presented. All GB connected anode junction boxes (AJBs) and shunts were closely inspected, maintained, and necessary replacement/and or modification were carried out on shunts. All damaged GB concrete foundations (CF) have been inspected and completely replaced. All GB associated Transformer-Rectifiers units (TRUs) were subjected to through inspection, and necessary maintenance has been performed on each individual TRU. After completion of all MACs and TRU maintenance activities, each cathodic protection station (CPS) has been re-operated. An alternative current (AC), direct current (DC), voltage and structure to soil potential (S/P) measurements have been conducted, recorded, and all obtained test results are presented. DC current outputs has been adjusted, and DC current outputs of each MAC has been recorded for each GB AJB.

Keywords: magnatite anode, deep well, ground bed, cathodic protection, transformer rectifies, impreced current, junction box

Procedia PDF Downloads 112

Authors: Ioan Rusu

Abstract:

In the operation of power transformers of 110 kV/MV from substations, these are traveled by fault current resulting from MV line damage. Defect electrical contacts are heated when they are travelled from fault currents. In the case of high temperatures when 135 °C is reached, the electrical insulating oil in the vicinity of the electrical faults comes into contact with these contacts releases gases, and activates the electrical protection. To avoid auto-flammability of electro-insulating oil, we designed a security system thermal of electrical contact defects by pouring fire-resistant polyurethane foam, mastic or mortar fire inside a cardboard electro-insulating cylinder. From practical experience, in the exploitation of power transformers of 110 kV/MT in oil electro-insulating were recorded some passing disconnecting commanded by the gas protection at internal defects. In normal operation and in the optimal load, nominal currents do not require thermal secure contacts inside electrical transformers, contacts are made at the fabrication according to the projects or to repair by solder. In the case of external short circuits close to the substation, the contacts inside electrical transformers, even if they are well made in sizes of Rcontact = 10^‑6 Ω, are subjected to short-circuit currents of the order of 10 kA-20 kA which lead to the dissipation of some significant second-order electric powers, 100 W-400 W, on contact. At some internal or external factors which action on electrical contacts, including electrodynamic efforts at short-circuits, these factors could be degraded over time to values in the range of 10^-4 Ω to 10^-5 Ω and if the action time of protection is great, on the order of seconds, power dissipation on electrical contacts achieve high values of 1,0 kW to 40,0 kW. This power leads to strong local heating, hundreds of degrees Celsius and can initiate self-ignition and burning oil in the vicinity of electro-insulating contacts with action the gas relay. Degradation of electrical contacts inside power transformers may not be limited for the duration of their operation. In order to avoid oil burn with gas release near electrical contacts, at short-circuit currents 10 kA-20 kA, we have outlined the following solutions: covering electrical contacts in fireproof materials that would avoid direct burn oil at short circuit and transmission of heat from electrical contact along the conductors with heat dissipation gradually over time, in a large volume of cooling. Flame retardant materials are: polyurethane foam, mastic, cement (concrete). In the normal condition of operation of transformer, insulating of conductors coils is with paper and insulating oil. Ignition points of its two components respectively are approximated: 135 °C heat for oil and 200 ⁰C for paper. In the case of a faulty electrical contact, about 10^-3 Ω, at short-circuit; the temperature can reach for a short time, a value of 300 °C-400 °C, which ignite the paper and also the oil. By burning oil, there are local gases that disconnect the power transformer. Securing thermal electrical contacts inside the transformer, in cardboard tube with polyurethane foams, mastik or cement, ensures avoiding gas release and also gas protection working.

Keywords: power transformer, oil insulatation, electric contacts, Bucholtz relay

Procedia PDF Downloads 158

Authors: Mohamed Ahmed Khalil

Abstract:

The number of deep well anode ground beds (GBs) have been retrieved due to unoperated anode chains. New identical magnetite anode chains (MAC) have been installed at Raslanuf complex impressed current Cathodic protection (ICCP) system, distributed at different plants (Utility, ethylene and polyethylene). All problems associated with retrieving and installation of MACs have been discussed, rectified and presented. All GB-associated severely corroded wellhead casings were well maintained and/or replaced by new fabricated and modified ones. The main cause of the wellhead casing's severe internal corrosion was discussed and the conducted remedy action to overcome future corrosion problems is presented. All GB-connected anode junction boxes (AJBs) and shunts were closely inspected, maintained and necessary replacement and/or modifications were carried out on shunts. All damaged GB concrete foundations (CF) have been inspected and completely replaced. All GB-associated Transformer-Rectifiers Units (TRU) were subjected to thorough inspection and necessary maintenance was performed on each individual TRU. After completion of all MACs and TRU maintenance activities, each cathodic protection station (CPS) has been re-operated, alternative current (AC), direct current (DC), voltage and structure to soil potential (S/P) measurements have been conducted, recorded and all obtained test results are presented. DC current outputs have been adjusted and DC current outputs of each MAC have been recorded for each GB AJB.

Keywords: magnetite anodes, deep well, ground beds, cathodic protection, transformer rectifier, impressed current, junction boxes

Procedia PDF Downloads 119

Authors: Ke Zhang, Yongli Zhu

Abstract:

Due to rapid load growths in today’s highly electrified societies and the requirement for green energy sources, large-scale wind farm power transmission system is constantly developing. This system is a typical multi-terminal power supply system, whose structure of the network topology of transmission lines is complex. What’s more, it locates in the complex terrain of mountains and grasslands, thus increasing the possibility of transmission line faults and finding the fault location with difficulty after the faults and resulting in an extremely serious phenomenon of abandoning the wind. In order to solve these problems, a fault location method for multi-terminal transmission line based on wind farm characteristics and improved single-ended traveling wave positioning method is proposed. Through studying the zero sequence current characteristics by using the characteristics of the grounding transformer(GT) in the existing large-scale wind farms, it is obtained that the criterion for judging the fault interval of the multi-terminal transmission line. When a ground short-circuit fault occurs, there is only zero sequence current on the path between GT and the fault point. Therefore, the interval where the fault point exists is obtained by determining the path of the zero sequence current. After determining the fault interval, The location of the short-circuit fault point is calculated by the traveling wave method. However, this article uses an improved traveling wave method. It makes the positioning accuracy more accurate by combining the single-ended traveling wave method with double-ended electrical data. What’s more, a method of calculating the traveling wave velocity is deduced according to the above improvements (it is the actual wave velocity in theory). The improvement of the traveling wave velocity calculation method further improves the positioning accuracy. Compared with the traditional positioning method, the average positioning error of this method is reduced by 30%.This method overcomes the shortcomings of the traditional method in poor fault location of wind farm transmission lines. In addition, it is more accurate than the traditional fixed wave velocity method in the calculation of the traveling wave velocity. It can calculate the wave velocity in real time according to the scene and solve the traveling wave velocity can’t be updated with the environment and real-time update. The method is verified in PSCAD/EMTDC.

Keywords: grounding transformer, multi-terminal transmission line, short circuit fault location, traveling wave velocity, wind farm

Procedia PDF Downloads 263

Authors: Jingwei Du

Abstract:

The need of optimizing air conditioning systems for existing buildings calls for control methods designed with energy-efficiency as a primary goal. The majority of current control methods boil down to two categories: empirical and model-based. To be effective, the former heavily relies on engineering expertise and the latter requires extensive historical data. Reinforcement Learning (RL), on the other hand, is a model-free approach that explores the environment to obtain an optimal control strategy often referred to as “policy”. This research adopts Proximal Policy Optimization (PPO) to improve chiller plant control, and enable the RL agent to collaborate with experienced engineers. It exploits the fact that while the industry lacks historical data, abundant operational data is available and allows the agent to learn and evolve safely under human supervision. Thanks to the development of language models, renewed interest in RL has led to modern, online, policy-based RL algorithms such as the PPO. This research took inspiration from “alignment”, a process that utilizes human feedback to finetune the pretrained model in case of unsafe content. The methodology can be summarized into three steps. First, an initial policy model is generated based on minimal prior knowledge. Next, the prepared PPO agent is deployed so feedback from both critic model and human experts can be collected for future finetuning. Finally, the agent learns and adapts itself to the specific chiller plant, updates the policy model and is ready for the next iteration. Besides the proposed approach, this study also used traditional RL methods to optimize the same simulated chiller plants for comparison, and it turns out that the proposed method is safe and effective at the same time and needs less to no historical data to start up.

Keywords: chiller plant, control methods, energy efficiency, proximal policy optimization, reinforcement learning

Procedia PDF Downloads 28

Authors: Chun-Wei Huang, Jyh-Cherng Gu, Ming-Ta Yang

Abstract:

Non-synchronous breakage or line failure in power systems with light or no loads can lead to core saturation in transformers or potential transformers. This can cause component and capacitance matching resulting in the formation of resonant circuits, which trigger ferroresonance. This study employed a wavelet transform for the detection of ferroresonance. Simulation results demonstrate the efficacy of the proposed method.

Keywords: ferroresonance, wavelet transform, intelligent electronic device, transformer

Procedia PDF Downloads 496

Authors: Ma Cecilia Siva

Abstract:

This study introduces a Bidirectional Encoder Representations from Transformers (BERT)-based machine learning model aimed at improving educational counseling by automating the process of recommending academic strands for students. The framework is designed to streamline and enhance the strand selection process by analyzing students' profiles and suggesting suitable academic paths based on their interests, strengths, and goals. Data was gathered from a sample of 200 grade 10 students, which included personal essays and survey responses relevant to strand alignment. After thorough preprocessing, the text data was tokenized, label-encoded, and input into a fine-tuned BERT model set up for multi-label classification. The model was optimized for balanced accuracy and computational efficiency, featuring a multi-category classification layer with sigmoid activation for independent strand predictions. Performance metrics showed an F1 score of 88%, indicating a well-balanced model with precision at 80% and recall at 100%, demonstrating its effectiveness in providing reliable recommendations while reducing irrelevant strand suggestions. To facilitate practical use, the final deployment phase created a recommendation framework that processes new student data through the trained model and generates personalized academic strand suggestions. This automated recommendation system presents a scalable solution for academic guidance, potentially enhancing student satisfaction and alignment with educational objectives. The study's findings indicate that expanding the data set, integrating additional features, and refining the model iteratively could improve the framework's accuracy and broaden its applicability in various educational contexts.

Keywords: tokenized, sigmoid activation, transformer, multi category classification

Procedia PDF Downloads 8

Authors: Prashanth Pillai, Purnaprajna Mangsuli

Abstract:

Oil exploration and production companies invest considerable time and efforts to extract essential well attributes (like well status, surface, and target coordinates, wellbore depths, event timelines, etc.) from unstructured data sources like technical reports, which are often non-standardized, multimodal, and highly domain-specific by nature. It is also important to consider the context when extracting attribute values from reports that contain information on multiple wells/wellbores. Moreover, semantically similar information may often be depicted in different data syntax representations across multiple pages and document sources. We propose a hierarchical multi-source fact extraction workflow based on a deep learning framework to extract essential well attributes at scale. An information retrieval module based on the transformer architecture was used to rank relevant pages in a document source utilizing the page image embeddings and semantic text embeddings. A question answering framework utilizingLayoutLM transformer was used to extract attribute-value pairs incorporating the text semantics and layout information from top relevant pages in a document. To better handle context while dealing with multi-well reports, we incorporate a dynamic query generation module to resolve ambiguities. The extracted attribute information from various pages and documents are standardized to a common representation using a parser module to facilitate information comparison and aggregation. Finally, we use a probabilistic approach to fuse information extracted from multiple sources into a coherent well record. The applicability of the proposed approach and related performance was studied on several real-life well technical reports.

Keywords: natural language processing, deep learning, transformers, information retrieval

Procedia PDF Downloads 193

Authors: Y. Jeon, A. Bissessur, J. Lin, P. Ndungu

Abstract:

Most transformers employ the usage of cellulose paper, which has been chemically modified through the Kraft process that acts as an effective insulator. Cellulose ageing and oil degradation are directly linked to fouling of the transformer and accumulation of large quantities of waste insulating paper. In addition to technical difficulties, this proves costly for power utilities to deal with. Currently there are no cost effective method for the rejuvenation of cellulose paper that has been documented nor proposed, since renewal of used insulating paper is implemented as the best option. This study proposes and contrasts different rejuvenation methods of accelerated aged cellulose insulating paper by chemical and bio-bleaching processes. Of the three bleaching methods investigated, two are, conventional chlorine-based sodium hypochlorite (m/v), and chlorine-free hydrogen peroxide (v/v), whilst the third is a bio-bleaching technique that uses a bacterium isolate, Acinetobacter strain V2. Through chemical bleaching, varying the strengths of the bleaching reagents at 0.3 %, 0.6 %, 0.9 %, 1.2 %, 1.5 % and 1.8 % over 4 hrs. were analyzed. Bio-bleaching implemented a bacterium isolate, Acinetobacter strain V2, to bleach the aged Kraft paper over 4 hrs. The determination of the amount of alpha cellulose, degree of polymerization and viscosity carried out on Kraft-cellulose insulating paper before and after bleaching. Overall the investigated techniques of chemical and bio-bleaching were successful and effective in treating degraded and accelerated aged Kraft-cellulose insulating paper, however, to varying extents. Optimum conditions for chemical bleaching were attained at bleaching strengths of 1.2 % (m/v) NaOCl and 1.5 % (v/v) H2O2 yielding alpha cellulose contents of 82.4 % and 80.7 % and degree of polymerizations of 613 and 616 respectively. Bio-bleaching using Acinetobacter strain V2 proved to be the superior technique with alpha cellulose levels of 89.0 % and a degree of polymerization of 620. Chemical bleaching techniques require careful and controlled clean-up treatments as it is chlorine and hydrogen peroxide based while bio-bleaching is an extremely eco-friendly technique.

Keywords: alpha cellulose, bio-bleaching, degree of polymerization, Kraft-cellulose insulating paper, transformer, viscosity

Procedia PDF Downloads 270

Authors: Dylan Elliott, Katya Pertsova

Abstract:

Preposition errors are some of the most common errors created by L2 speakers. In addition, improving error correction and detection methods remains an open issue in the realm of Natural Language Processing (NLP). This research investigates whether the bidirectional encoder representations from the transformers model (BERT) have the potential to correct preposition errors accurately enough to be useful in error correction software. This research finds that BERT performs strongly when the scope of its error correction is limited to preposition choice. The researchers used an open-source BERT model and over three hundred thousand edited sentences from Wikipedia, tagged for part of speech, where only a preposition edit had occurred. To test BERT’s ability to detect errors, a technique known as multi-level masking was used to generate suggestions based on sentence context for every prepositional environment in the test data. These suggestions were compared with the original errors in the data and their known corrections to evaluate BERT’s performance. The suggestions were further analyzed to determine if BERT more often agreed with the judgements of the Wikipedia editors. Both the untrained and fined-tuned models were compared. Finetuning led to a greater rate of error-detection which significantly improved recall, but lowered precision due to an increase in false positives or falsely flagged errors. However, in most cases, these false positives were not errors in preposition usage but merely cases where more than one preposition was possible. Furthermore, when BERT correctly identified an error, the model largely agreed with the Wikipedia editors, suggesting that BERT’s ability to detect misused prepositions is better than previously believed. To evaluate to what extent BERT’s false positives were grammatical suggestions, we plan to do a further crowd-sourcing study to test the grammaticality of BERT’s suggested sentence corrections against native speakers’ judgments.

Keywords: BERT, grammatical error correction, preposition error detection, prepositions

Procedia PDF Downloads 147

Authors: Rita Chattopadhyay, Vivek Anand Thoutam

Abstract:

Speech is the most acceptable means of communication where we can quickly exchange our feelings and thoughts. Quite often, people can communicate orally but cannot interact or work with computers or devices. It’s easy and quick to give speech commands than typing commands to computers. In the same way, it’s easy listening to audio played from a device than extract output from computers or devices. Especially with Robotics being an emerging market with applications in warehouses, the hospitality industry, consumer electronics, assistive technology, etc., speech-based human-machine interaction is emerging as a lucrative feature for robot manufacturers. Considering this factor, the objective of this paper is to design the “Audio-Visual Co-Data Processing Pipeline.” This pipeline is an integrated version of Automatic speech recognition, a Natural language model for text understanding, object detection, and text-to-speech modules. There are many Deep Learning models for each type of the modules mentioned above, but OpenVINO Model Zoo models are used because the OpenVINO toolkit covers both computer vision and non-computer vision workloads across Intel hardware and maximizes performance, and accelerates application development. A speech command is given as input that has information about target objects to be detected and start and end times to extract the required interval from the video. Speech is converted to text using the Automatic speech recognition QuartzNet model. The summary is extracted from text using a natural language model Generative Pre-Trained Transformer-3 (GPT-3). Based on the summary, essential frames from the video are extracted, and the You Only Look Once (YOLO) object detection model detects You Only Look Once (YOLO) objects on these extracted frames. Frame numbers that have target objects (specified objects in the speech command) are saved as text. Finally, this text (frame numbers) is converted to speech using text to speech model and will be played from the device. This project is developed for 80 You Only Look Once (YOLO) labels, and the user can extract frames based on only one or two target labels. This pipeline can be extended for more than two target labels easily by making appropriate changes in the object detection module. This project is developed for four different speech command formats by including sample examples in the prompt used by Generative Pre-Trained Transformer-3 (GPT-3) model. Based on user preference, one can come up with a new speech command format by including some examples of the respective format in the prompt used by the Generative Pre-Trained Transformer-3 (GPT-3) model. This pipeline can be used in many projects like human-machine interface, human-robot interaction, and surveillance through speech commands. All object detection projects can be upgraded using this pipeline so that one can give speech commands and output is played from the device.

Keywords: OpenVINO, automatic speech recognition, natural language processing, object detection, text to speech

Procedia PDF Downloads 80

Authors: H. M. Wilhelm, P. O. Fernandes, L. P. Dill, C. Steffens, K. G. Moscon, S. M. Peres, V. Bender, T. Marchesan, J. B. Ferreira Neto

Abstract:

Deterioration of insulating oil is a natural process that occurs during transformers operation. However, this process can be accelerated by some factors, such as oxygen, high temperatures, metals and, moisture, which rapidly reduce oil insulating capacity and favor transformer faults. Parts of building materials of a transformer can be degraded and yield soluble compounds and insoluble particles that shorten the equipment life. Physicochemical tests, dissolved gas analysis (including propane, propylene and, butane), volatile and furanic compounds determination, besides quantitative and morphological analyses of particulate are proposed in this study in order to correlate transformers building materials degradation with insulating oil characteristics. The present investigation involves tests of medium temperature overheating simulation by means of an electric resistance wrapped with the following materials immersed in mineral insulating oil: test I) copper, tin, lead and, paper (heated at 350-400 °C for 8 h); test II) only copper (at 250 °C for 11 h); and test III) only paper (at 250 °C for 8 h and at 350 °C for 8 h). A different experiment is the simulation of electric arc involving copper, using an electric welding machine at two distinct energy sets (low and high). Analysis results showed that dielectric loss was higher in the sample of test I, higher neutralization index and higher values of hydrogen and hydrocarbons, including propane and butane, were also observed. Test III oil presented higher particle count, in addition, ferrographic analysis revealed contamination with fibers and carbonized paper. However, these particles had little influence on the oil physicochemical parameters (dielectric loss and neutralization index) and on the gas production, which was very low. Test II oil showed high levels of methane, ethane, and propylene, indicating the effect of metal on oil degradation. CO₂ and CO gases were formed in the highest concentration in test III, as expected. Regarding volatile compounds, in test I acetone, benzene and toluene were detected, which are oil oxidation products. Regarding test III, methanol was identified due to cellulose degradation, as expected. Electric arc simulation test showed the highest oil oxidation in presence of copper and at high temperature, since these samples had huge concentration of hydrogen, ethylene, and acetylene. Particle count was also very high, showing the highest release of copper in such conditions. When comparing high and low energy, the first presented more hydrogen, ethylene, and acetylene. This sample had more similar results to test I, pointing out that the generation of different particles can be the cause for faults such as electric arc. Ferrography showed more evident copper and exfoliation particles than in other samples. Therefore, in this study, by using different combined analytical techniques, it was possible to correlate insulating oil characteristics with possible contaminants, which can lead to transformers failure.

Keywords: Ferrography, gas analysis, insulating mineral oil, particle contamination, transformer failures

Procedia PDF Downloads 225

Authors: Leila Baghaarabani, Parvin Razzaghi, Mennatolla Magdy Mostafa, Ahmad Albaqsami, Al Warith Al Rushaidi, Masoud Al Rawahi

Abstract:

Predicting the interaction between drugs and their molecular targets is pivotal for advancing drug development processes. Due to the time and cost limitations, computational approaches have emerged as an effective approach to drug-target interaction (DTI) prediction. Most of the introduced computational based approaches utilize the drug molecule and protein sequence as input. This study does not only utilize these inputs, it also introduces a protein representation developed using a masked protein language model. In this representation, for every individual amino acid residue within the protein sequence, there exists a corresponding probability distribution that indicates the likelihood of each amino acid being present at that particular position. Then, the similarity between each pair of amino acids is computed to create a similarity matrix. To encode the knowledge of the similarity matrix, Bi-Level Routing Attention (BiFormer) is utilized, which combines aspects of transformer-based models with protein sequence analysis and represents a significant advancement in the field of drug-protein interaction prediction. BiFormer has the ability to pinpoint the most effective regions of the protein sequence that are responsible for facilitating interactions between the protein and drugs, thereby enhancing the understanding of these critical interactions. Thus, it appears promising in its ability to capture the local structural relationship of the proteins by enhancing the understanding of how it contributes to drugprotein interactions, thereby facilitating more accurate predictions. To evaluate the proposed method, it was tested on two widely recognized datasets: Davis and KIBA. A comprehensive series of experiments was conducted to illustrate its effectiveness in comparison to cutting edge techniques.

Keywords: BiFormer, transformer, protein language processing, self-attention mechanism, binding affinity, drug target interaction, similarity matrix, protein masked representation, protein language model

Procedia PDF Downloads 7

Authors: Jae Ni Jang, Young Uk Kim

Abstract:

Background: Chat Generative Pre-training Transformer-3 (ChatGPT; San Francisco, California, Open Artificial Intelligence) is an artificial intelligence chatbot based on a large language model designed to generate human-like text. As the usage of ChatGPT is increasing among less knowledgeable patients, medical students, and anesthesia and pain medicine residents or trainees, we aimed to evaluate the accuracy of ChatGPT-3 responses to questions about the American Society of Anesthesiologists (ASA) classification based on patients’ underlying diseases and assess the quality of the generated responses. Methods: A total of 47 questions were submitted to ChatGPT using textual prompts. The questions were designed for ChatGPT-3 to provide answers regarding ASA classification in response to common underlying diseases frequently observed in adult patients. In addition, we created 18 questions regarding the ASA classification for pediatric patients and pregnant women. The accuracy of ChatGPT’s responses was evaluated by cross-referencing with Miller’s Anesthesia, Morgan & Mikhail’s Clinical Anesthesiology, and the American Society of Anesthesiologists’ ASA Physical Status Classification System (2020). Results: Out of the 47 questions pertaining to adults, ChatGPT -3 provided correct answers for only 23, resulting in an accuracy rate of 48.9%. Furthermore, the responses provided by ChatGPT-3 regarding children and pregnant women were mostly inaccurate, as indicated by a 28% accuracy rate (5 out of 18). Conclusions: ChatGPT provided correct responses to questions relevant to the daily clinical routine of anesthesiologists in approximately half of the cases, while the remaining responses contained errors. Therefore, caution is advised when using ChatGPT to retrieve anesthesia-related information. Although ChatGPT may not yet be suitable for clinical settings, we anticipate significant improvements in ChatGPT and other large language models in the near future. Regular assessments of ChatGPT's ASA classification accuracy are essential due to the evolving nature of ChatGPT as an artificial intelligence entity. This is especially important because ChatGPT has a clinically unacceptable rate of error and hallucination, particularly in pediatric patients and pregnant women. The methodology established in this study may be used to continue evaluating ChatGPT.

Keywords: American Society of Anesthesiologists, artificial intelligence, Chat Generative Pre-training Transformer-3, ChatGPT

Procedia PDF Downloads 47

Authors: Brennan Lodge

Abstract:

Amidst a complex regulatory landscape, Retrieval Augmented Generation (RAG) emerges as a transformative tool for Governance Risk and Compliance (GRC) officers. This paper details the application of RAG in synthesizing Large Language Models (LLMs) with external knowledge bases, offering GRC professionals an advanced means to adapt to rapid changes in compliance requirements. While the development for standalone LLM’s (Large Language Models) is exciting, such models do have their downsides. LLM’s cannot easily expand or revise their memory, and they can’t straightforwardly provide insight into their predictions, and may produce “hallucinations.” Leveraging a pre-trained seq2seq transformer and a dense vector index of domain-specific data, this approach integrates real-time data retrieval into the generative process, enabling gap analysis and the dynamic generation of compliance and risk management content. We delve into the mechanics of RAG, focusing on its dual structure that pairs parametric knowledge contained within the transformer model with non-parametric data extracted from an updatable corpus. This hybrid model enhances decision-making through context-rich insights, drawing from the most current and relevant information, thereby enabling GRC officers to maintain a proactive compliance stance. Our methodology aligns with the latest advances in neural network fine-tuning, providing a granular, token-level application of retrieved information to inform and generate compliance narratives. By employing RAG, we exhibit a scalable solution that can adapt to novel regulatory challenges and cybersecurity threats, offering GRC officers a robust, predictive tool that augments their expertise. The granular application of RAG’s dual structure not only improves compliance and risk management protocols but also informs the development of compliance narratives with pinpoint accuracy. It underscores AI’s emerging role in strategic risk mitigation and proactive policy formation, positioning GRC officers to anticipate and navigate the complexities of regulatory evolution confidently.

Keywords: cybersecurity, gen AI, retrieval augmented generation, cybersecurity defense strategies

Procedia PDF Downloads 95

Authors: John K. Avor, Choong-Koo Chang

Abstract:

The transfer of electrical loads at power generation stations from Standby Auxiliary Transformer (SAT) to Unit Auxiliary Transformer (UAT) and vice versa is through a fast bus transfer scheme. Fast bus transfer is a time-critical application where the transfer process depends on various parameters, thus transfer schemes apply advance algorithms to ensure power supply reliability and continuity. In a nuclear power generation station, supply continuity is essential, especially for critical class 1E electrical loads. Bus transfers must, therefore, be executed accurately within 4 to 10 cycles in order to achieve safety system requirements. However, the main problem is that there are instances where transfer schemes scrambled due to inaccurate interpretation of key parameters; and consequently, have failed to transfer several critical loads from UAT to the SAT during main generator trip event. Although several techniques have been adopted to develop robust transfer schemes, a combination of Artificial Neural Network and Fuzzy Systems (Neuro-Fuzzy) has not been extensively used. In this paper, we apply the concept of Neuro-Fuzzy to determine plant operating mode and dynamic prediction of the appropriate bus transfer algorithm to be selected based on the first cycle of voltage information. The performance of Sequential Fast Transfer and Residual Bus Transfer schemes was evaluated through simulation and integration of the Neuro-Fuzzy system. The objective for adopting Neuro-Fuzzy approach in the bus transfer scheme is to utilize the signal validation capabilities of artificial neural network, specifically the back-propagation algorithm which is very accurate in learning completely new systems. This research presents a combined effect of artificial neural network and fuzzy systems to accurately interpret key bus transfer parameters such as magnitude of the residual voltage, decay time, and the associated phase angle of the residual voltage in order to determine the possibility of high speed bus transfer for a particular bus and the corresponding transfer algorithm. This demonstrates potential for general applicability to improve reliability of the auxiliary power distribution system. The performance of the scheme is implemented on APR1400 nuclear power plant auxiliary system.

Keywords: auxiliary power system, bus transfer scheme, fuzzy logic, neural networks, reliability

Procedia PDF Downloads 171

Authors: Kayne Uriel K. Rodrigo, Jerriane Hillary Heart S. Marcial, Samuel C. Brillo

Abstract:

Rice leaf diseases significantly impact yield production in rice-dependent countries, affecting their agricultural sectors. As part of precision agriculture, early and accurate detection of these diseases is crucial for effective mitigation practices and minimizing crop losses. Hence, this study proposes an enhancement to the Convolutional Neural Network (CNN), a widely-used method for Rice Leaf Disease Image Classification, by incorporating MobileViTV2—a recently advanced architecture that combines CNN and Vision Transformer models while maintaining fewer parameters, making it suitable for broader deployment on edge devices. Our methodology utilizes a publicly available rice disease image dataset from Kaggle, which was validated by a university structural biologist following the guidelines provided by the Philippine Rice Institute (PhilRice). Modifications to the dataset include renaming certain disease categories and augmenting the rice leaf image data through rotation, scaling, and flipping. The enhanced dataset was then used to train the MobileViTV2 model using the Timm library. The results of our approach are as follows: the model achieved notable performance, with 98% accuracy in both training and validation, 6% training and validation loss, and a Receiver Operating Characteristic (ROC) curve ranging from 95% to 100% for each label. Additionally, the F1 score was 97%. These metrics demonstrate a significant improvement compared to a conventional CNN-based approach, which, in a previous 2022 study, achieved only 78% accuracy after using 5 convolutional layers and 2 dense layers. Thus, it can be concluded that MobileViTV2, with its fewer parameters, outperforms traditional CNN models, particularly when applied to Rice Leaf Disease Image Identification. For future work, we recommend extending this model to include datasets validated by international rice experts and broadening the scope to accommodate biotic factors such as rice pest classification, as well as abiotic stressors such as climate, soil quality, and geographic information, which could improve the accuracy of disease prediction.

Keywords: convolutional neural network, MobileViTV2, rice leaf disease, precision agriculture, image classification, vision transformer

Procedia PDF Downloads 22

Authors: Amira Zouaoui, Hamed Belloumi, Ferid Kourda

Abstract:

This paper presents the analysis and design of a DC–AC resonant converter applied to induction heating. The proposed topology based on the series-parallel half-bridge resonant inverter is described. It can operate with Zero-Voltage Switching (ZVS). At the resonant frequency, the secondary current is amplified over the heating coil with small switching angle, which keeps the reactive power low and permits heating with small current through the resonant inductor and the transformer. The operation and control principle of the proposed high frequency inverter is described and verified through simulated and experimental results.

Keywords: induction heating, inverter, high frequency, resonant

Procedia PDF Downloads 464

Authors: R. S. Al Abri, M. H. Albadi, M. H. Al Abri, U. K. Al Rasbi, M. H. Al Hasni, S. M. Al Shidi

Abstract:

This paper presents an analysis study on the impacts of the changes of the capacitor banks, the loss of a transformer, and the installation of distributed generation on the voltage total harmonic distortion and harmonic resonance. The study is applied in a real system in Oman, Sohar Industrial Port–C Substation Network. Frequency scan method and Fourier series analysis method are used with the help of EDSA software. Moreover, the results are compared with limits specified by national Oman distribution code.

Keywords: power quality, capacitor bank, voltage total harmonics distortion, harmonic resonance, frequency scan

Procedia PDF Downloads 617

Authors: Nagesh Vangala, Rayudu Mannam

Abstract:

Soft switching has attracted the interest of various researchers constantly. Many techniques are in vogue to achieve soft switching (ZVS and/or ZCS) in Boost converters. These techniques utilize an auxiliary switch to incorporate the ZCS/ZVS. Such schemes require additional control circuit and induce complexity in design. This paper proposes an elegant fly back approach which guarantees zero current switching of the main Switch without the need for any additional active device. A simple flyback transformer scheme is implemented which absorbs the initial turn ON energy (or the Reverse recovery energy of Boost diode) and delivers to the output.

Keywords: boost converter, complete energy transfer, flyback, zero current switching

Procedia PDF Downloads 397

Authors: Tahar Alimi, Rahma Boujebane, Wiem Derouich, Lamia Hadrich Belguith

Abstract:

Machine translation task of low-resourced languages such as Arabic is a challenging task. Despite the appearance of sophisticated models based on the latest deep learning techniques, namely the transfer learning and transformers, all models prove incapable of carrying out an acceptable translation, which includes Arabic Dialects (AD), because they do not have official status. In this paper, we present a machine translation model designed to translate Arabic multidialectal content into Modern Standard Arabic (MSA), leveraging both new and existing parallel resources. The latter achieved the best results for both Levantine and Maghrebi dialects with a BLEU score of 64.99.

Keywords: Arabic translation, dialect translation, fine-tune, MSA translation, transformer, translation

Procedia PDF Downloads 61

Authors: Rade M. Ciric, Sasa Mandic

Abstract:

Students of Higher Education Technical School of Professional Studies, in Novi Sad follow the subject Maintenance of electric power equipment at the Electrotechnical Department. This paper presents educational plan and program of the subject Maintenance of electric power equipment. The course deals with the problems of preventive and investing maintenance of transformer stations (TS), performing and maintenance of grounding of TS and pillars, as well as tracing and detection the location of the cables failure. There is a special elaborated subject concerning the safe work conditions for the electrician during network maintenance, as well as the basics of making and keeping technical documentation of the equipment.

Keywords: educational plan and program, electric power equipment, maintenance, technical documentation, safe work

Procedia PDF Downloads 467

Authors: Md.Noman Habib Khan, M. S. Tajul Islam, T. S. Gunawan, M. Hasanuzzaman

Abstract:

The photovoltaic (PV) panel with no galvanic isolation system is well-known technique in the world which is effective and deliver power with enhanced efficiency. The PV generation presented here is for stand-alone system installed in remote areas when as the resulting power gets connected to electronic load installation instead of being tied to the grid. Though very small, even then transformer-less topology is shown to be with leakage in pico-ampere range. By using PWM technique PWM, leakage current in different situations is shown. The results that are demonstrated in this paper show how the pico-ampere current is reduced to femto-ampere through use of inductors and capacitors of suitable values of inductor and capacitors with the load.

Keywords: photovoltaic (PV) panel, duty cycle, pulse duration modulation (PDM), leakage current

Procedia PDF Downloads 534