Wavelet Packet Decomposition Algorithm Research Articles

Research on the identification method based on energy distribution of adaptive WPD for railway tunnel lining void

ABSTRACT As the main disease of railway tunnels, void disease affects the safe operation of railways. The traditional detection technology recognition void relies on the intuitive characteristics of the signal, there are still problems such as poor void edge recognition capability and low intelligence level. To solve this problem, this study used Comsol software to establish a sound-solid coupling finite element model of the tunnel, extracted the acoustic signals of voids under various conditions, analysed the acoustic signal characteristics under different conditions and proposed the acoustic sensitive frequency band of tunnel lining void. The test model of Partial tunnel lining is established by the method of layering pouring. The wavelet packet algorithm is improved by the maximum cross correlation coefficient, the optimal wavelet basis decomposition algorithm is proposed. The layer number of the wavelet packet decomposition (WPD) algorithm was calculated based on the sensitive frequency band and the energy distribution of the sub-signal is obtained. Then, the energy distribution of the sensitive frequency band is used to estimate the void boundary. The experimental results show that the acoustic frequency band of the tunnel void mainly distributes in the range of 0–10,000 Hz and the sensitive frequency band of the void concentrated around 0–3000 Hz. According to the sampling frequency and the sensitive frequency band of the void, the sub-signal energy distribution of each layer is obtained. The energy distribution of sub-signal 1 has strong void boundary recognition ability. The research results will be beneficial to the damage detection and state assessment of tunnel lining.

Detection of partial discharge patterns in hybrid high voltage power transmission lines based on parallel recognition method

Due to their specially designed structures, the partial discharge detection of hybrid high-voltage power transmission lines (HHVPTL) composed of overhead lines and power cables has made it difficult to monitor the conditions of power transmission lines. A parallel recognition method for partial discharge patterns of HHVPTLs is proposed by implementing wavelet analysis and improved backpropagation neural network (BPNN) to address the shortcomings of low efficiency, poor accuracy, and inability to parallel analysis of current partial discharge (PD) detection algorithms for HHVPTLs. Firstly, considering the non-smoothness of the partial discharge of the HHVPTLs, the wavelet packet decomposition algorithm is implemented to decompose the PD of the HHVPTL and resolve the relevant signal indicators to form the attribute vectors. Then, BPNN is implemented as a classification model. A beetle optimization (DBO) algorithm based on orthogonal contrastive learning improvement is introduced to optimize the BPNN parameters since BPNN has a slow convergence problem and fails easily into a local optimum. The proposed IDBO-BPNN is employed as the model that recognizes and analyzes the parallel partial discharge patterns of HHVPTLs. Finally, the suggested model is implemented to investigate the local discharge data of an HHVPTL in the Kaggle Featured Prediction Competition and is compared with other algorithms. The experimental results indicate that the proposed model can more accurately identify whether PDs occur in an HHVPTL and detect phases where PDs occur, with higher overall accuracy and efficiency. An excellent practical performance is achieved. The proposed model can achieve the recognition accuracy of 95.5%, which is 5.3333% higher than that of the DBO-BPNN and far more than other recognition algorithms.

Electric vehicle motor fault diagnosis using improved wavelet packet decomposition and particle swarm optimization algorithm

This study addresses the issue of diagnosing faults in electric vehicle motors and presents a method utilizing Improved Wavelet Packet Decomposition (IWPD) combined with particle swarm optimization (PSO). Initially, the analysis focuses on common demagnetization faults, inter turn short circuit faults, and eccentricity faults of permanent magnet synchronous motors. The proposed approach involves the application of IWPD for extracting signal feature vectors, incorporating the energy spectrum scale, and extracting the feature vectors of the signal using the energy spectrum scale. Subsequently, a binary particle swarm optimization algorithm is employed to formulate strategies for updating particle velocity and position. Further optimization of the binary particle swarm algorithm using chaos theory and the simulated annealing algorithm results in the development of a motor fault diagnosis model based on the enhanced particle swarm optimization algorithm. The results demonstrate that the chaotic simulated annealing algorithm achieves the highest accuracy and recall rates, at 0.96 and 0.92, respectively. The model exhibits the highest fault accuracy rates on both the test and training sets, exceeding 98.2%, with a minimal loss function of 0.0035. Following extraction of fault signal feature vectors, the optimal fitness reaches 97.4%. In summary, the model constructed in this study demonstrates effective application in detecting faults in electric vehicle motors, holding significant implications for the advancement of the electric vehicle industry.

Short-term traffic flow prediction based on secondary hybrid decomposition and deep echo state networks

Short-term traffic flow prediction is a significant and challenging research topic as it is closely related to the application of intelligent transportation systems. Due to the variable and random characteristics of the transportation system, raw traffic flow data often contain noise, and predicting the raw data directly may reduce the accuracy and effectiveness of the prediction models. Therefore, a hybrid method is established in this research which combines denoising schemes and deep learning models to improve the prediction accuracy. The time series denoising schemes include two parts: the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), and wavelet packet decomposition (WPD). Firstly, the raw traffic flow data are decomposed by CEEMDAN to obtain intrinsic mode functions (IMFs) and a residual. Then the IMFs are divided into anti-persistent and persistent components through the Hurst Exponent index. The anti-persistent components are re-decomposed by the WPD algorithm, and persistent components are aggregated into one component. Finally, these components and residual are forecasted by the deep echo state network (DeepESN) model. In the experiment, to investigate the prediction performance of the proposed CEEMDAN-WPD123456–7a11-DeepESN model, the LSTM, CEEMDAN-LSTM, CEEMDAN-WPD-LSTM, DeepESN, CEEMDAN-DeepESN, CEEMDAN-WPD1-DeepESN, CEEMDAN-WPD123456-DeepESN and CEEMDAN-WPD1a6–7a11-DeepESN models are considered to be comparison models. The experimental results demonstrate that the proposed model has superior performance on both efficiency and accuracy.

Flywheel energy storage controlled by model predictive control to achieve smooth short-term high-frequency wind power

The use of energy storage systems to improve the fluctuation of wind power generation has garnered significant in the development of wind power. However, the fluctuation of the signals in the high-frequency part of the wind turbine output is particularly drastic and short in duration. As a kind of physical energy storage device, the flywheel energy storage device has a fast response speed but higher requirements on the control system. In order to improve the control effect of the flywheel energy storage device, the model predictive control algorithm is improved in this paper. First, the high-frequency components of the wind farm output power data are extracted by the wavelet packet decomposition algorithm, and the high-frequency components are optimized by mathematical interpolation as the data basis. Secondly, a mathematical model of the flywheel energy storage system applied in the model predictive control algorithm is proposed, and the model predictive control algorithm is used to configure the flywheel energy storage device to achieve a smooth output power of the wind farm. Finally, the simulation is performed in MATLAB and the experimental parameters are adjusted. The experimental results show that the configuration of the flywheel energy storage system based on the model predictive control algorithm can effectively smooth the fluctuation of the high-frequency component of the output power data of the wind farm. The experimental results take the wind power data of different time periods for energy storage configuration, and the comparison verifies the reliability of the system designed in this paper.

A Two-DOF Active-Passive Hybrid Vibration Isolator Based on Multi-Line Spectrum Adaptive Control

In order to effectively control the low-frequency vibration of ship machinery, based on the improved multi-line spectrum adaptive control algorithm, a two-degree-of-freedom (two-DOF) active-passive hybrid vibration isolator composed of an electromagnetic actuator, rubber spring, and the hydraulic device is proposed. The dynamic model of the two-DOF vibration isolation system is established and the main control force demand of the vibration isolation system at different damping forces is analyzed. By introducing the improved wavelet packet decomposition algorithm with the Hartley block least mean square algorithm to the filter-x least mean square (FxLMS) algorithm, an improved wavelet packet Hartley block filter-x least mean square (IWPHB-FxLMS) algorithm is established. The experimental results show that the IWPHB-FxLMS algorithm has better control performance. Compared with the traditional FxLMS algorithm, the IWPHB-FxLMS control algorithm improves the convergence speed by 91.7% and the line spectrum power spectrum attenuation by 58.1%. The active-passive hybrid vibration isolator is based on multi-line spectrum adaptive control and can achieve good control effects under the excitation of multi-frequency line spectrum and constant frequency line spectrum.

An Improved Fault Diagnosis Approach Using LSSVM for Complex Industrial Systems

Fault diagnosis is a challenging topic for complex industrial systems due to the varying environments such systems find themselves in. In order to improve the performance of fault diagnosis, this study designs a novel approach by using particle swarm optimization (PSO) with wavelet mutation and least square support (LSSVM). The implementation entails the following three steps. Firstly, the original signals are decomposed through an orthogonal wavelet packet decomposition algorithm. Secondly, the decomposed signals are reconstructed to obtain the fault features. Finally, the extracted features are used as the inputs of the fault diagnosis model established in this research to improve classification accuracy. This joint optimization method not only solves the problem of PSO falling easily into the local extremum, but also improves the classification performance of fault diagnosis effectively. Through experimental verification, the wavelet mutation particle swarm optimazation and least sqaure support vector machine ( WMPSO-LSSVM) fault diagnosis model has a maximum fault recognition efficiency that is 12% higher than LSSVM and 9% higher than extreme learning machine (ELM). The error of the corresponding regression model under the WMPSO-LSSVM algorithm is 0.365 less than that of the traditional linear regression model. Therefore, the proposed fault scheme can effectively identify faults that occur in complex industrial systems.

Hybrid Deep Learning Model for Short-Term Wind Speed Forecasting Based on Time Series Decomposition and Gated Recurrent Unit

Accurate wind speed prediction has been becoming an indispensable technology in system security, wind energy utilization, and power grid dispatching in recent years. However, it is an arduous task to predict wind speed due to its variable and random characteristics. For the objective to enhance the performance of forecasting short-term wind speed, this work puts forward a hybrid deep learning model mixing time series decomposition algorithm and gated recurrent unit (GRU). The time series decomposition algorithm combines the following two parts: (1) the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), and (2) wavelet packet decomposition (WPD). Firstly, the normalized wind speed time series (WSTS) are handled by CEEMDAN to gain pure fixed-frequency components and a residual signal. The WPD algorithm conducts the second-order decomposition to the first component that contains complex and high frequency signal of raw WSTS. Finally, GRU networks are established for all the relevant components of the signals, and the predicted wind speeds are obtained by superimposing the prediction of each component. Results from two case studies, adopting wind data from laboratory and wind farm, respectively, suggest that the related trend of the WSTS can be separated effectively by the proposed time series decomposition algorithm, and the accuracy of short-time wind speed prediction can be heightened significantly mixing the time series decomposition algorithm and GRU networks.

Electroencephalogram-Based Motor Imagery Classification Using Deep Residual Convolutional Networks.

The classification of electroencephalogram (EEG) signals is of significant importance in brain-computer interface (BCI) systems. Aiming to achieve intelligent classification of motor imagery EEG types with high accuracy, a classification methodology using the wavelet packet decomposition (WPD) and the proposed deep residual convolutional networks (DRes-CNN) is proposed. Firstly, EEG waveforms are segmented into sub-signals. Then the EEG signal features are obtained through the WPD algorithm, and some selected wavelet coefficients are retained and reconstructed into EEG signals in their respective frequency bands. Subsequently, the reconstructed EEG signals were utilized as input of the proposed deep residual convolutional networks to classify EEG signals. Finally, EEG types of motor imagination are classified by the DRes-CNN classifier intelligently. The datasets from BCI Competition were used to test the performance of the proposed deep learning classifier. Classification experiments show that the average recognition accuracy of this method reaches 98.76%. The proposed method can be further applied to the BCI system of motor imagination control.

Electrical Harmonic Energy Measurement Based on Wavelet Packet Decomposition and Reconstruction Algorithm

In order to study the accurate measurement of electric energy in complexindustrial field, a method of harmonic electric energy measurement basedon wavelet packet decomposition and reconstruction algorithm, as well asthe calculation formula of harmonic power and the principle of harmonicelectric energy measurement are proposed. Using db42 wavelet function tocarry out harmonic energy metering simulation analysis, the results showthat: The fundamental frequency of the simulation signal is 50 Hz, two-layerwavelet packet transform is adopted, the simulation input signals within 40fundamental wave cycles are taken, and the sampling frequency fs is 800 Hz.Conclusion: The three-phase harmonic energy metering device based onvirtual instrument technology has realized the measurement of each harmonicactive power and reactive power, and the accuracy reaches 0.2 s.

Anti-Interference Test Technology of Transformer Noise

In order to enhance the transformer noise test accuracy, an anti- interference method based on wavelet packet decomposition and spectral subtraction algorithm is proposed in this paper. The tested original noise signal is decomposed with the wavelet packet method. The occurring time and frequency bands of the noisy signals are observed. In these bands, the noisy signals are further processed by spectral subtraction method. Finally, the pure transformer noise is obtained with wavelet packet reconstruction. The anti-interference effects of transformer noises in the conditions of various interference sources and signal to noise ratios are simulated. The result shows that the proposed method is effective in restraining the influence of diverse burst intermittent environmental noises on the transformer noise test result.

A Novel Method to Restrain Ambient Noise Interference in Transformer Noise Test

A preprocessing method based on wavelet packet decomposition and spectral subtraction algorithm is proposed in this paper. The tested original noise signal is decomposed with the wavelet packet analysis. It is divided into several bands with different frequency range. In these bands, the noisy signals are further processed by spectral subtraction method. The theoretical analysis result shows that the proposed method is effective in restraining the influence of surrounding noises on the result of transformer noise measurement. The proposed method can be referenced for the noise measurement of power transmission and transformation projects.

Fault Diagnosis for Variable Frequency Drive-Fed Induction Motors Using Wavelet Packet Decomposition and Greedy-Gradient Max-Cut Learning

In this paper, a novel fault diagnosis method for variable frequency drive (VFD)-fed induction motors is proposed using Wavelet Packet Decomposition (WPD) and greedy-gradient max-cut (GGMC) learning algorithm. The proposed method is developed using experimental stator current data in the lab for two 0.25 HP induction motors fed by a VFD, subjected to healthy and faulty cases under various operating frequencies and motor loadings. The features are extracted from stator current signals using WPD by evaluating energy eigenvalues and feature coefficients at decomposition levels. The proposed method is validated by comparing with other graph-based semi-supervised learning (GSSL) algorithms, local and global consistency (LGC) and Gaussian field and harmonic function (GFHF). To enable fault diagnosis for untested motor operating conditions, mathematical equations to calculate features for untested cases are developed through surface fitting using features extracted from tested cases.

An audio-based intelligent fault diagnosis method for belt conveyor rollers in sand carrier

The roller is an important part of the belt conveyor in sand carrier at sea. A good fault diagnosis method of the rollers provides an effective guarantee for the system’s optimal operation. In this paper, a novel intelligent fault diagnosis method for rollers is proposed by using audio wavelet packet decomposition and Convolutional Neural Networks (CNN). Firstly, the wavelet packet decomposition algorithm is used to decompose the audio data of the rollers into several frequency bands. Secondly, the lowest frequency data are adjusted under consideration of the excessive energy proportion of the low frequency data. Then, CNN is used to classify the features of each frequency band to diagnose rollers’ faults. The experiment shows that the diagnosis method has high accuracy, high speed and strong robustness, which greatly improves the efficiency of fault diagnosis of rollers of sand carrier.

Application analysis of machine learning in fault diagnosis: A bibliometric analysis

This article to analyze the application of machine learning in fault diagnosis by using bibliometrics, co-citation network analysis and cluster analysis methods. The analysis found that in the application research of machine learning in fault diagnosis, China has the largest number of published papers and cited papers, among which IEEE access is the most published journal, MECHANICAL SYSTEMS AND SIGNAL PROCESSING is the most cited journal,followed by the United States and India. Using document co-citation relationship analysis, a set of key documents in this field was identified., Currently, the important algorithms fault diagnosis include support vector data description method, transfer learning algorithm, convolutional neural network algorithm, natural inspired optimization algorithm, bayesian network, wavelet packet decomposition algorithm, fuzzy logic algorithm and so on. All of them are clustered according to keywords, and their application fields include acoustic emission and fault diagnosis of doubly-fed induction generators.

Practical application of a safe human-robot interaction software

PurposeBecause of the increased use of robots in the industry, it has become inevitable for humans and robots to be able to work together. Therefore, human security has become the primary noncompromising factor of joint human and robot operations. For this reason, the purpose of this study was to develop a safe human-robot interaction software based on vision and touch.Design/methodology/approachThe software consists of three modules. Firstly, the vision module has two tasks: to determine whether there is a human presence and to measure the distance between the robot and the human within the robot’s working space using convolutional neural networks (CNNs) and depth sensors. Secondly, the touch detection module perceives whether or not a human physically touches the robot within the same work environment using robot axis torques, wavelet packet decomposition algorithm and CNN. Lastly, the robot’s operating speed is adjusted according to hazard levels came from vision and touch module using the robot’s control module.FindingsThe developed software was tested with an industrial robot manipulator and successful results were obtained with minimal error.Practical implicationsThe success of the developed algorithm was demonstrated in the current study and the algorithm can be used in other industrial robots for safety.Originality/valueIn this study, a new and practical safety algorithm is proposed and the health of people working with industrial robots is guaranteed.

Lifelong Condition Monitoring Based on NB-IoT for Anomaly Detection of Machinery Equipment

The condition monitoring for machinery equipment is vital for safe and economical production in the industry. In modern manufacturing, data-driven methods for machine prognosis and health management (PHM) have been paid greater importance due to the development of machine learning. For another, the emerging Internet of Things (IoT) technique makes it possible for large scale data collection using distributed IoT terminal devices. In this paper, a condition monitoring system for machinery equipment is designed based on Narrow Band Internet of Things (NB-IoT) technique. Combined with the wavelet packet decomposition (WPD) and one-class support vector machine (OCSVM) algorithm, the abnormal data can be effectively identified. The system is verified by a small fan working at two conditions: normal and blade imbalance. The experiment results prove that the system can achieve reliable and stable online monitoring. What’s more, the low power design of the IoT terminal ensures the system’s longtime operation.

Distributed Temperature/Vibration Fiber Optic Sensor With High Precision and Wide Bandwidth

Distributed fiber optic sensor for multi-parameter measurements plays a crucial role in various applications. In this work, distributed temperature/vibration fiber optic sensor with high precision and wide bandwidth is proposed and experimentally demonstrated. To improve the signal-to-noise ratio (SNR) and sensitivity of the system, the backscattering enhanced optical fiber (BEOF) is employed to provide strong and stable sensing signal. With the assistance of the coherent phase detection, a high precision measurement of phase can be achieved. In addition, the wavelet packet decomposition algorithm is put forward to effectively discriminate the quasi-static temperature and dynamic vibration signals. In the experiment, wide bandwidth vibration signal from 0.01 Hz to 20 kHz can be successfully detected. Owing to the high sensitivity of the system, the local temperature change within the range of only 10 cm can be captured with high precision of 0.095 °C. Besides, the system has a good capability of fast response to the external temperature variation where the response time is around 0.3 seconds, which is meaningful for the early warning of centimeters-level sized fire source in the special oil and gas pipelines monitoring applications.

Fault diagnosis method of gear based on lifting wavelet packet and combined optimization BP neural network

Aiming at the problem of weak signal signature recognition of gear faults, a gear fault diagnosis method based on lifting wavelet packet and combined optimization BP neural network is proposed. The initial non-sampling prediction and update operators are calculated by Lagrange interpolation subdivision based on the principle of lifting wavelet, and the adaptive redundancy lifting wavelet packet decomposition and reconstruction algorithm is constructed. The network parameters of the number of hidden layers and the quantity of nodes, initial weights and thresholds of BP neural network are optimized by genetic algorithm (GA). The Levenberg-Marquardt (LM) algorithm is used to improve the search space of the network. Through experimental analysis, the results show that the gear fault diagnosis method proposed in this paper not only has high diagnostic accuracy, but also increase efficiency.

Wavelet-Based Kernel Construction for Heart Disease Classification

Heart disease classification plays an important role in clinical diagnoses. The performance improvement of an Electrocardiogram classifier is therefore of great relevance, but it is a challenging task too. This paper proposes a novel classification algorithm using the kernel method. A kernel is constructed based on wavelet coefficients of heartbeat signals for a classifier with high performance. In particular, a wavelet packet decomposition algorithm is applied to heartbeat signals to obtain the Approximation and Detail coefficients, which are used to calculate the parameters of the kernel. A principal component analysis algorithm with the wavelet-based kernel is employed to choose the main features of the heartbeat signals for the input of the classifier. In addition, a neural network with three hidden layers in the classifier is utilized for classifying five types of heart disease. The electrocardiogram signals in nine patients obtained from the MIT-BIH database are used to test the proposed classifier. In order to evaluate the performance of the classifier, a multi-class confusion matrix is applied to produce the performance indexes, including the Accuracy, Recall, Precision, and F1 score. The experimental results show that the proposed method gives good results for the classification of the five mentioned types of heart disease.