For speaker independent emotion recognition, scholars have recently proposed Fourier parameter and mean Fourier parameter models. This study aims to propose fast Fourier coefficient features such as minimum, maximum, mean, and standard deviation. In addition to source features, i.e., glottal volume velocity, this study aims to separate speech segments into voiced speech segments, unvoiced speech segments, and silent segments so that the effect of each part of speech corpus on emotion recognition can be observed. Experimental results indicate that the proposed method improves speech emotion recognition rate to 80.85% for EmoDB. For Ryerson audio-visual database of emotional speech and song (RAVDESS) for eight emotions, a highest emotion recognition rate of 70.19% was achieved.
Most machine learning (ML) algorithms reported for ultrasonic guided wave structural health monitoring (GW-SHM) applications, particularly for damage assessment in the presence of time-varying environmental and operating variables, use networks with a lot of parameters, necessitating cloud computing or expensive computing and storage infrastructure. This raises the cost of the solution and may prevent the widespread use of GW-SHM. In this paper, we describe a viable alternative based on TinyML framework, to create lightweight machine learning models that can be instantly deployed on embedded edge devices. We present a practical implementation of this scheme on a custom-designed FPGA-based embedded system for GW-SHM of a honeycomb composite sandwich structure (HCSS) in presence of large thermal variations.
Fueled by the rapid development of machine learning (ML) and greater access to cloud computing and graphics processing units (GPUs), various deep learning based models have been proposed for improving performance of ultrasonic guided wave structural health monitoring (GW-SHM) systems, especially to counter complexity and heterogeneity in data due to varying environmental factors (e.g., temperature) and types of damages. Such models typically comprise of millions of trainable parameters, and therefore add to cost of deployment due to requirements of cloud connectivity and processing, thus limiting the scale of deployment of GW-SHM. In this work, we propose an alternative solution that leverages TinyML framework for development of light-weight ML models that could be directly deployed on embedded edge devices. The utility of our solution is illustrated by presenting an unsupervised learning framework for damage detection in honeycomb composite sandwich structure (HCSS) with disbond and delamination type of damages, validated using data generated by finite element (FE) simulations and experiments performed at various temperatures in the range 0{\deg}C to 90{\deg}C. We demonstrate a fully-integrated solution using a Xilinx Artix-7 FPGA for data acquisition and control, and edge-inference of damage.
Fueled by the rapid development of machine learning (ML) and greater access to cloud computing and graphics processing units, various deep learning based models have been proposed for improving performance of ultrasonic guided wave structural health monitoring (GW-SHM) systems, especially to counter complexity and heterogeneity in data due to varying environmental factors (e.g., temperature) and types of damages. Such models typically comprise of millions of trainable parameters, and therefore add to cost of deployment due to requirements of cloud connectivity and processing, thus limiting the scale of deployment of GW-SHM. In this work, we propose an alternative solution that leverages TinyML framework for development of light-weight ML models that could be directly deployed on embedded edge devices. The utility of our solution is illustrated by presenting an unsupervised learning framework for damage detection in honeycomb composite sandwich structure with disbond and delamination type of damages, validated using data generated by finite element simulations and experiments performed at various temperatures in the range 0-90 °C. We demonstrate a fully-integrated solution using a Xilinx Artix-7 FPGA for data acquisition and control, and edge-inference of damage. Despite the limited number of features, the lightweight model shows reasonably high accuracy, thereby enabling detection of small size defects with improved sensitivity on an edge device for online GW-SHM.
Conventional damage localization algorithms used in ultrasonic guided wave-based structural health monitoring (GW-SHM) rely on physics-defined features of GW signals. In addition to requiring domain knowledge of the interaction of various GW modes with various types of damages, they also suffer from errors due to variations in environmental and operating conditions (EOCs) in practical use cases. While several machine learning tools have been reported for EOC compensation, they need to be custom-designed for each combination of damage and structure due to their dependence on physics-defined feature extraction. In this work, we propose a CNN-based automated feature extraction framework coupled with Gaussian mixture model (GMM) based EOC compensation and damage classification and localization method. Features learnt by the CNNs are used for damage classification and localization of damage by modeling the probability distribution of the features using GMMs. The Kullback-Leibler (KL) divergence of these GMMs with respect to corresponding baseline GMMs are used as signal difference coefficients (SDCs) to compute damage indices (DIs) along various GW sensor paths, and thus for damage localization. The efficacy of the proposed method is demonstrated using FE generated GW-data for an aluminum plate with a network of six lead zirconate titanate (PZT) sensors, for three different types of damages (rivet hole, added mass, notch) at various temperatures, with added white noise and pink noise to incorporate errors due to EOCs. We also present experimental validation of the method through characterization of notch damage in an aluminum panel under varying and non-uniform temperature profiles, using a portable custom-designed field programmable gate array (FPGA) based signal transduction and data acquisition system.
Most reported research for monitoring health of pipelines using ultrasonic guided waves (GW) typically utilize bulky piezoelectric transducer rings and laboratory-grade ultrasonic non-destructive testing (NDT) equipment. Consequently, the translation of these approaches from laboratory settings to field-deployable systems for real-time structural health monitoring (SHM) becomes challenging. In this work, we present an innovative algorithm for damage identification and localization in pipes, implemented on a compact FPGA-based smart GW-SHM system. The custom-designed board, featuring a Xilinx Artix-7 FPGA and front-end electronics, is capable of actuating the PZT thickness shear mode transducers, data acquisition and recording from PZT sensors and generating a damage index (DI) map for localizing the damage on the structure. The algorithm is a variation of the common source method adapted for cylindrical geometry. The utility of the algorithm is demonstrated for detection and localization of defects such as notch and mass loading on a steel pipe, through extensive finite element (FE) method simulations. Experimental results obtained using a C-clamp for applying mass loading on the pipe show good agreement with the FE simulations. The localization error values for experimental data analysed using C code on a processor implemented on the FPGA are consistent with algorithm results generated on a computer running Python code. The system presented in this study is suitable for a wide range of GW-SHM applications, especially in cost-sensitive scenarios that benefit from on-node signal processing over cloud-based solutions.
Abstract Guided wave (GW) based structural health monitoring (SHM) techniques being developed by researchers frequently use amplitude and group velocity variations between healthy and damage-affected GW modes to detect and localise damage. Nonetheless, external variables such as temperature and moisture influence these features, which were not considered in previous studies, particularly in the presence of damage in honeycomb composite sandwich structures (HCSSs). Therefore, a coordinated numerical and experimental study was carried out in an effort to examine the characteristics of GW propagation in an HCSS for two damages: a disbond between the face sheet and the core, and delamination between the face sheet layers for a temperature range of 0 ∘ C–90 ∘ C. Computationally efficient two-dimensional numerical models were developed using COMSOL Multiphysics that takes into account a variety of temperature-related phenomena, such as thermal stresses and changes in the material properties of honeycomb sandwich and piezoelectric wafer transducers (PZTs). The amplitude and group velocity of the fundamental anti-symmetric (A0) mode are found to increase in the presence of a disbond and decrease in the presence of face sheet delamination. However, it is observed that there is a linear decrease in the amplitude of A0 mode for both the healthy and damaged cases with an increase in temperature. Since the A0 mode is widely employed for interrogation due to its defect sensitivity, an amplitude and group velocity adjustment equation with temperature change is proposed. Finally, considering the amplitude difference of normalised A0 mode, the two damages are localised within a network of PZTs by using a probability-based signal difference coefficient method, which is found to be efficient and reliable for SHM of HCSS under variable temperature conditions.
Most reported research for monitoring health of pipelines using ultrasonic guided waves (GW) typically utilize bulky piezoelectric transducer rings and laboratory-grade ultrasonic non-destructive testing (NDT) equipment. Consequently, the translation of these approaches from laboratory settings to field-deployable systems for real-time structural health monitoring (SHM) becomes challenging. In this work, we present an innovative algorithm for damage identification and localization in pipes, implemented on a compact FPGA-based smart GW-SHM system. The custom-designed board, featuring a Xilinx Artix-7 FPGA and front-end electronics, is capable of actuating the PZT thickness shear mode transducers, data acquisition and recording from PZT sensors and generating a damage index (DI) map for localizing the damage on the structure. The algorithm is a variation of the common source method adapted for cylindrical geometry. The utility of the algorithm is demonstrated for detection and localization of defects such as notch and mass loading on a steel pipe, through extensive finite element (FE) method simulations. Experimental results obtained using a C-clamp for applying mass loading on the pipe show good agreement with the FE simulations. The localization error values for experimental data analyzed using C code on a processor implemented on the FPGA are consistent with algorithm results generated on a computer running MATLAB code. The system presented in this study is suitable for a wide range of GW-SHM applications, especially in cost-sensitive scenarios that benefit from on-node signal processing over cloud-based solutions.
Structural health monitoring (SHM) of pipelines using ultrasonic guided waves (GW) is highly suitable for long-range and real-time monitoring. Most such demonstrations utilize bulky piezoelectric transducer rings and laboratory-grade ultrasonic NDT equipment to monitor pipelines and are therefore difficult to translate from laboratory to on-field deployable systems for real-time monitoring of state-of-health of pipelines. In this work, we have demonstrated a novel damage identification and localization algorithm for pipes, realized with a compact FPGA-based smart GW-SHM system. The custom-developed board with Xilinx Artix®-7 FPGA (Digilent Cmod A7) and frontend electronics is capable of actuating the PZT thickness shear transducers, data acquisition and recording from PZT sensors. The algorithm is a variation of the common source algorithm, adapted for cylindrical geometry, and is applied to detect location of mass loading on a steel pipe. The experimental results were verified with finite element method (FEM) simulations.
'%3e%3ccircle r='20' fill='%23008'/%3e%3ccircle r='17.5' fill='%23fff'/%3e%3ccircle r='3.5' fill='%23008'/%3e%3cg id='d'%3e%3cg id='c'%3e%3cg id='b'%3e%3cg id='a' fill='%23008'%3e%3ccircle r='.875' transform='rotate(7.5 -8.75 133.5)'/%3e%3cpath d='M0 17.5.6 7 0 2l-.6 5L0 17.5z'/%3e%3c/g%3e%3cuse xlink:href='%23a' transform='rotate(15)'/%3e%3c/g%3e%3cuse xlink:href='%23b' transform='rotate(30)'/%3e%3c/g%3e%3cuse xlink:href='%23c' transform='rotate(60)'/%3e%3c/g%3e%3cuse xlink:href='%23d' transform='rotate(120)'/%3e%3cuse xlink:href='%23d' transform='rotate(-120)'/%3e%3c/g%3e%3c/svg%3e)
