Search results for: Kaiming initialization

Authors: Ben Soltane Cheima, Ittansa Yonas Kelbesa

Abstract:

Speaker Identification (SI) is the task of establishing identity of an individual based on his/her voice characteristics. The SI task is typically achieved by two-stage signal processing: training and testing. The training process calculates speaker specific feature parameters from the speech and generates speaker models accordingly. In the testing phase, speech samples from unknown speakers are compared with the models and classified. Even though performance of speaker identification systems has improved due to recent advances in speech processing techniques, there is still need of improvement. In this paper, a Closed-Set Tex-Independent Speaker Identification System (CISI) based on a Multiple Classifier System (MCS) is proposed, using Mel Frequency Cepstrum Coefficient (MFCC) as feature extraction and suitable combination of vector quantization (VQ) and Gaussian Mixture Model (GMM) together with Expectation Maximization algorithm (EM) for speaker modeling. The use of Voice Activity Detector (VAD) with a hybrid approach based on Short Time Energy (STE) and Statistical Modeling of Background Noise in the pre-processing step of the feature extraction yields a better and more robust automatic speaker identification system. Also investigation of Linde-Buzo-Gray (LBG) clustering algorithm for initialization of GMM, for estimating the underlying parameters, in the EM step improved the convergence rate and systems performance. It also uses relative index as confidence measures in case of contradiction in identification process by GMM and VQ as well. Simulation results carried out on voxforge.org speech database using MATLAB highlight the efficacy of the proposed method compared to earlier work.

Keywords: feature extraction, speaker modeling, feature matching, Mel frequency cepstrum coefficient (MFCC), Gaussian mixture model (GMM), vector quantization (VQ), Linde-Buzo-Gray (LBG), expectation maximization (EM), pre-processing, voice activity detection (VAD), short time energy (STE), background noise statistical modeling, closed-set tex-independent speaker identification system (CISI)

Procedia PDF Downloads 280

Authors: Alexander I. Dovgyallo, Evgeniy A. Zinoviev, Svetlana O. Nekrasova

Abstract:

The most of works applied for the pulse tube converters contain the workflow description implemented through the use of mathematical models on stationary modes. However, the study of the thermoacoustic systems unsteady behavior in the start, stop, and acoustic load changes modes is in the particular interest. The aim of the present study was to develop a mathematical thermal excitation model of acoustic oscillations in pulse tube engine (PTE) as a small-scale scheme of pulse tube engine operating at atmospheric air. Unlike some previous works this standing wave configuration is a fully closed system. The improvements over previous mathematical models are the following: the model allows specifying any values of porosity for regenerator, takes into account the piston weight and the friction in the cylinder and piston unit, and determines the operating frequency. The numerical method is based on the relation equations between the pressure and volume velocity variables at the ends of each element of PTE which is recorded through the appropriate transformation matrix. A solution demonstrates that the PTE operation frequency is the complex value, and it depends on the piston mass and the dynamic friction due to its movement in the cylinder. On the basis of the determined frequency thermoacoustically induced heat transport and generation of acoustic power equations were solved for channel with temperature gradient on its ends. The results of numerical simulation demonstrate the features of the initialization process of oscillation and show that that generated acoustic power more than power on the steady mode in a factor of 3…4. But doesn`t mean the possibility of its further continuous utilizing due to its existence only in transient mode which lasts only for a 30-40 sec. The experiments were carried out on small-scale PTE. The results shows that the value of acoustic power is in the range of 0.7..1.05 W for the defined frequency range f = 13..18 Hz and pressure amplitudes 11..12 kPa. These experimental data are satisfactorily correlated with the numerical modeling results. The mathematical model can be straightforwardly applied for the thermoacoustic devices with variable temperatures of thermal reservoirs and variable transduction loads which are expected to occur in practical implementations of portable thermoacoustic engines.

Keywords: nonlinear processes, pulse tube engine, thermal excitation, standing wave

Procedia PDF Downloads 350

Authors: Hamza Benzerrouk, Alexander Nebylov

Abstract:

In classical GNSS/INS integration designs, the loosely coupled approach uses the GNSS derived position and the velocity as the measurements vector. This design is suboptimal from the standpoint of preventing GNSSoutliers/outages. The tightly coupled GPS/INS navigation filter mixes the GNSS pseudo range and inertial measurements and obtains the vehicle navigation state as the final navigation solution. The ultra‐tightly coupled GNSS/INS design combines the I (inphase) and Q(quadrature) accumulator outputs in the GNSS receiver signal tracking loops and the INS navigation filter function intoa single Kalman filter variant (EKF, UKF, SPKF, CKF and HCKF). As mentioned, EKF and UKF are the most used nonlinear filters in the literature and are well adapted to inertial navigation state estimation when integrated with GNSS signal outputs. In this paper, it is proposed to move a step forward with more accurate filters and modern approaches called Cubature and High Degree cubature Kalman Filtering methods, on the basis of previous results solving the state estimation based on INS/GNSS integration, Cubature Kalman Filter (CKF) and High Degree Cubature Kalman Filter with (HCKF) are the references for the recent developed generalized Cubature rule based Kalman Filter (GCKF). High degree cubature rules are the kernel of the new solution for more accurate estimation with less computational complexity compared with the Gauss-Hermite Quadrature (GHQKF). Gauss-Hermite Kalman Filter GHKF which is not selected in this work because of its limited real-time implementation in high-dimensional state-spaces. In ultra tightly or a deeply coupled GNSS/INS system is dynamics EKF is used with transition matrix factorization together with GNSS block processing which is well described in the paper and assumes available the intermediary frequency IF by using a correlator samples with a rate of 500 Hz in the presented approach. GNSS (GPS+GLONASS) measurements are assumed available and modern SPKF with Cubature Kalman Filter (CKF) are compared with new versions of CKF called high order CKF based on Spherical-radial cubature rules developed at the fifth order in this work. Estimation accuracy of the high degree CKF is supposed to be comparative to GHKF, results of state estimation are then observed and discussed for different initialization parameters. Results show more accurate navigation state estimation and more robust GNSS receiver when Ultra Tightly Coupled approach applied based on High Degree Cubature Kalman Filter.

Keywords: GNSS, INS, Kalman filtering, ultra tight integration

Procedia PDF Downloads 259

Authors: Irina P. Starodubtseva, Aleksandr N. Pavlenko

Abstract:

Quenching is a term generally accepted for the process of rapid cooling of a solid that is overheated above the thermodynamic limit of the liquid superheat. The main objective of many previous studies on quenching is to find a way to reduce the total time of the transient process. Computational experiments were performed to simulate quenching by a falling liquid nitrogen film of an extremely overheated vertical copper plate with a structured capillary-porous coating. The coating was produced by directed plasma spraying. Due to the complexities in physical pattern of quenching from chaotic processes to phase transition, the mechanism of heat transfer during quenching is still not sufficiently understood. To our best knowledge, no information exists on when and how the first stable liquid-solid contact occurs and how the local contact area begins to expand. Here we have more models and hypotheses than authentically established facts. The peculiarities of the quench front dynamics and heat transfer in the transient process are studied. The created numerical model determines the quench front velocity and the temperature fields in the heater, varying in space and time. The dynamic pattern of the running quench front obtained numerically satisfactorily correlates with the pattern observed in experiments. Capillary-porous coatings with straight and reverse orientation of crests are investigated. The results show that the cooling rate is influenced by thermal properties of the coating as well as the structure and geometry of the protrusions. The presence of capillary-porous coating significantly affects the dynamics of quenching and reduces the total quenching time more than threefold. This effect is due to the fact that the initialization of a quench front on a plate with a capillary-porous coating occurs at a temperature significantly higher than the thermodynamic limit of the liquid superheat, when a stable solid-liquid contact is thermodynamically impossible. Waves present on the liquid-vapor interface and protrusions on the complex micro-structured surface cause destabilization of the vapor film and the appearance of local liquid-solid micro-contacts even though the average integral surface temperature is much higher than the liquid superheat limit. The reliability of the results is confirmed by direct comparison with experimental data on the quench front velocity, the quench front geometry, and the surface temperature change over time. Knowledge of the quench front velocity and total time of transition process is required for solving practically important problems of nuclear reactors safety.

Keywords: capillary-porous coating, heat transfer, Leidenfrost phenomenon, numerical simulation, quenching

Procedia PDF Downloads 107

Authors: George Zhou, Yunchan Chen, Candace Chien

Abstract:

Kidney replacement therapy is the current standard of care for end-stage renal diseases. In-center or home hemodialysis remains an integral component of the therapeutic regimen. Arteriovenous fistulas (AVF) make up the vascular circuit through which blood is filtered and returned. Naturally, AVF patency determines whether adequate clearance and filtration can be achieved and directly influences clinical outcomes. Our aim was to build a deep learning model for automated AVF stenosis screening based on the sound of blood flow through the AVF. A total of 311 patients with AVF were enrolled in this study. Blood flow sounds were collected using a digital stethoscope. For each patient, blood flow sounds were collected at 6 different locations along the patient’s AVF. The 6 locations are artery, anastomosis, distal vein, middle vein, proximal vein, and venous arch. A total of 1866 sounds were collected. The blood flow sounds are labeled as “patent” (normal) or “stenotic” (abnormal). The labels are validated from concurrent ultrasound. Our dataset included 1527 “patent” and 339 “stenotic” sounds. We show that blood flow sounds vary significantly along the AVF. For example, the blood flow sound is loudest at the anastomosis site and softest at the cephalic arch. Contextualizing the sound with location metadata significantly improves classification performance. How to encode and incorporate categorical metadata is an active area of research1. Herein, we study ordinal (i.e., integer) encoding schemes. The numerical representation is concatenated to the flattened feature vector. We train a vision transformer (ViT) on spectrogram image representations of the sound and demonstrate that using scalar multiples of our integer encodings improves classification performance. Models are evaluated using a 10-fold cross-validation procedure. The baseline performance of our ViT without any location metadata achieves an AuROC and AuPRC of 0.68 ± 0.05 and 0.28 ± 0.09, respectively. Using the following encodings of Artery:0; Arch: 1; Proximal: 2; Middle: 3; Distal 4: Anastomosis: 5, the ViT achieves an AuROC and AuPRC of 0.69 ± 0.06 and 0.30 ± 0.10, respectively. Using the following encodings of Artery:0; Arch: 10; Proximal: 20; Middle: 30; Distal 40: Anastomosis: 50, the ViT achieves an AuROC and AuPRC of 0.74 ± 0.06 and 0.38 ± 0.10, respectively. Using the following encodings of Artery:0; Arch: 100; Proximal: 200; Middle: 300; Distal 400: Anastomosis: 500, the ViT achieves an AuROC and AuPRC of 0.78 ± 0.06 and 0.43 ± 0.11. respectively. Interestingly, we see that using increasing scalar multiples of our integer encoding scheme (i.e., encoding “venous arch” as 1,10,100) results in progressively improved performance. In theory, the integer values do not matter since we are optimizing the same loss function; the model can learn to increase or decrease the weights associated with location encodings and converge on the same solution. However, in the setting of limited data and computation resources, increasing the importance at initialization either leads to faster convergence or helps the model escape a local minimum.

Keywords: arteriovenous fistula, blood flow sounds, metadata encoding, deep learning

Procedia PDF Downloads 52

Authors: Arjun Kumar Rath, Titus Dhanasingh

Abstract:

Embedded product development requires software to test hardware functionality during development and finding issues during manufacturing in larger quantities. As the components are getting integrated, the devices are tested for their full functionality using advanced software tools. Benchmarking tools are used to measure and compare the performance of product features. At present, these tests are based on a variety of methods involving varying hardware and software platforms. Typically, these tests are custom built for every product and remain unusable for other variants. A majority of the tests goes undocumented, not updated, unusable when the product is released. To bridge this gap, a solution accelerator in the form of a framework can address these issues for running all these tests from one place, using an off-the-shelf tests library in a continuous integration environment. There are many open-source test frameworks or tools (fuego. LAVA, AutoTest, KernelCI, etc.) designed for testing embedded system devices, with each one having several unique good features, but one single tool and framework may not satisfy all of the testing needs for embedded systems, thus an extensible framework with the multitude of tools. Embedded product testing includes board bring-up testing, test during manufacturing, firmware testing, application testing, and assembly testing. Traditional test methods include developing test libraries and support components for every new hardware platform that belongs to the same domain with identical hardware architecture. This approach will have drawbacks like non-reusability where platform-specific libraries cannot be reused, need to maintain source infrastructure for individual hardware platforms, and most importantly, time is taken to re-develop test cases for new hardware platforms. These limitations create challenges like environment set up for testing, scalability, and maintenance. A desirable strategy is certainly one that is focused on maximizing reusability, continuous integration, and leveraging artifacts across the complete development cycle during phases of testing and across family of products. To get over the stated challenges with the conventional method and offers benefits of embedded testing, an embedded test framework (ETF), a solution accelerator, is designed, which can be deployed in embedded system-related products with minimal customizations and maintenance to accelerate the hardware testing. Embedded test framework supports testing different hardwares including microprocessor and microcontroller. It offers benefits such as (1) Time-to-Market: Accelerates board brings up time with prepacked test suites supporting all necessary peripherals which can speed up the design and development stage(board bring up, manufacturing and device driver) (2) Reusability-framework components isolated from the platform-specific HW initialization and configuration makes the adaptability of test cases across various platform quick and simple (3) Effective build and test infrastructure with multiple test interface options and preintegrated with FUEGO framework (4) Continuos integration - pre-integrated with Jenkins which enabled continuous testing and automated software update feature. Applying the embedded test framework accelerator throughout the design and development phase enables to development of the well-tested systems before functional verification and improves time to market to a large extent.

Keywords: board diagnostics software, embedded system, hardware testing, test frameworks

Procedia PDF Downloads 116

Authors: M. Mueller, M. Kuehn, M. Voelker

Abstract:

In small and medium-sized enterprises (SMEs), the challenge is to create a well-grounded and reliable basis for process analysis, optimization and planning due to a lack of data. SMEs have limited access to methods with which they can effectively and efficiently analyse processes and identify cause-and-effect relationships in order to generate the necessary database and derive optimization potential from it. The implementation of digitalization within the framework of Industry 4.0 thus becomes a particular necessity for SMEs. For these reasons, the abstract presents an analysis methodology that is subject to the objective of developing an SME-appropriate methodology for efficient, temporarily feasible data collection and evaluation in flexible production and logistics systems as a basis for process analysis and optimization. The overall methodology focuses on retrospective, event-based tracing and analysis of material flow objects. The technological basis consists of Bluetooth low energy (BLE)-based transmitters, so-called beacons, and smart mobile devices (SMD), e.g. smartphones as receivers, between which distance data can be measured and derived motion profiles. The distance is determined using the Received Signal Strength Indicator (RSSI), which is a measure of signal field strength between transmitter and receiver. The focus is the development of a software-based methodology for interpretation of relative movements of transmitters and receivers based on distance data. The main research is on selection and implementation of pattern recognition methods for automatic process recognition as well as methods for the visualization of relative distance data. Due to an existing categorization of the database regarding process types, classification methods (e.g. Support Vector Machine) from the field of supervised learning are used. The necessary data quality requires selection of suitable methods as well as filters for smoothing occurring signal variations of the RSSI, the integration of methods for determination of correction factors depending on possible signal interference sources (columns, pallets) as well as the configuration of the used technology. The parameter settings on which respective algorithms are based have a further significant influence on result quality of the classification methods, correction models and methods for visualizing the position profiles used. The accuracy of classification algorithms can be improved up to 30% by selected parameter variation; this has already been proven in studies. Similar potentials can be observed with parameter variation of methods and filters for signal smoothing. Thus, there is increased interest in obtaining detailed results on the influence of parameter and factor combinations on data quality in this area. The overall methodology is realized with a modular software architecture consisting of independently modules for data acquisition, data preparation and data storage. The demonstrator for initialization and data acquisition is available as mobile Java-based application. The data preparation, including methods for signal smoothing, are Python-based with the possibility to vary parameter settings and to store them in the database (SQLite). The evaluation is divided into two separate software modules with database connection: the achievement of an automated assignment of defined process classes to distance data using selected classification algorithms and the visualization as well as reporting in terms of a graphical user interface (GUI).

Keywords: event-based tracing, machine learning, process classification, parameter settings, RSSI, signal smoothing

Procedia PDF Downloads 100

Authors: Ho-Ting Wu, Kai-Wei Ke, Bo-Yu Huang, Liang-Lin Yan, Chun-Ting Lin

Abstract:

With the advances in advanced technology, wireless sensor network (WSN) has become one of the most promising candidates to implement the wireless industrial internet of things (IIOT) architecture. However, the legacy IEEE 802.15.4 based WSN technology such as Zigbee system cannot meet the stringent QoS requirement of low powered, real-time, and highly reliable transmission imposed by the IIOT environment. Recently, the IEEE society developed IEEE 802.15.4e Time Slotted Channel Hopping (TSCH) access mode to serve this purpose. Furthermore, the IETF 6TiSCH working group has proposed standards to integrate IEEE 802.15.4e with IPv6 protocol smoothly to form a complete protocol stack for IIOT. In this work, we develop key network technologies for IEEE 802.15.4e based wireless IIoT architecture, focusing on practical design and system implementation. We realize the OpenWSN-based wireless IIOT system. The system architecture is divided into three main parts: web server, network manager, and sensor nodes. The web server provides user interface, allowing the user to view the status of sensor nodes and instruct sensor nodes to follow commands via user-friendly browser. The network manager is responsible for the establishment, maintenance, and management of scheduling and topology information. It executes centralized scheduling algorithm, sends the scheduling table to each node, as well as manages the sensing tasks of each device. Sensor nodes complete the assigned tasks and sends the sensed data. Furthermore, to prevent scheduling error due to packet loss, a schedule inspection mechanism is implemented to verify the correctness of the schedule table. In addition, when network topology changes, the system will act to generate a new schedule table based on the changed topology for ensuring the proper operation of the system. To enhance the system performance of such system, we further propose dynamic bandwidth allocation and distributed scheduling mechanisms. The developed distributed scheduling mechanism enables each individual sensor node to build, maintain and manage the dedicated link bandwidth with its parent and children nodes based on locally observed information by exchanging the Add/Delete commands via two processes. The first process, termed as the schedule initialization process, allows each sensor node pair to identify the available idle slots to allocate the basic dedicated transmission bandwidth. The second process, termed as the schedule adjustment process, enables each sensor node pair to adjust their allocated bandwidth dynamically according to the measured traffic loading. Such technology can sufficiently satisfy the dynamic bandwidth requirement in the frequently changing environments. Last but not least, we propose a packet retransmission scheme to enhance the system performance of the centralized scheduling algorithm when the packet delivery rate (PDR) is low. We propose a multi-frame retransmission mechanism to allow every single network node to resend each packet for at least the predefined number of times. The multi frame architecture is built according to the number of layers of the network topology. Performance results via simulation reveal that such retransmission scheme is able to provide sufficient high transmission reliability while maintaining low packet transmission latency. Therefore, the QoS requirement of IIoT can be achieved.

Keywords: IEEE 802.15.4e, industrial internet of things (IIOT), scheduling mechanisms, wireless sensor networks (WSN)

Procedia PDF Downloads 132