Research On Power Quality Disturbance Signal Recognition Method Based On Multiresolution S Transform

With the development of science and technology,a large number of high-efficiency precision electrical appliances are connected into the power system,which requires higher power quality.At the same time,more and more nonlinear power electronic devices and new energy with random fluctuation characteristics are put into use continuously,resulting in power quality problems such as signal waveform distortion,voltage instability,harmonic component increase and so on.The power pollution in power grid is becoming more and more serious,which poses a potential threat to the safety of power grid and the stable operation of power equipment.Power quality signal identification has become one of the hot topics in the industry.Aiming at the problem of complex power quality disturbance recognition,this paper based on the multiresolution S transforam theory,completes the feature extraction of the disturbance signal,and combines the knowledge of decision tree classifier and convolutional neural network to realize the fast recognition of complex power quality disturbance.Simulation and actual data processing results show that this method has advantages in recognition accuracy and timeliness,and can meet the requirements of actual power grid for disturbance signal recognition,so it has certain practical significance and application prospect.The specific research results are summarized as follows:In order to solve the problem that the time-frequency resolution of standard S transform is not flexible,this paper proposes a multiresolution S transform method.The feature of gaussian window function is adjusted by introducing adjustment factor to realize the flexible setting of time-frequency resolution and improve the time-frequency analysis ability of the algorithm.At the same time,through the optimal problem solving theory,the key parameters of multiresolution S transform are determined adaptively to meet the actual demand of power quality signal processing.On this basis,from the perspective of statistics,this paper describes the disturbance characteristics from the aspects of energy evolution and frequency change,constructs the characteristic statistics of the disturbance signal,and quantitatively characterizes the variation trend of the local energy and local frequency of the disturbance signal,thus laying a foundation for the identification of the power quality signal.Finally,combining with the classification theory of decision tree,the difference of eigenvalue of different disturbance signals is compared to realize the accurate recognition of complex disturbance signals.Combining with machine learning theory,one-dimensional convolutional neural network is applied to power quality signal recognition.On the basis of time-frequency analysis of multiresolution S transform and combining with the characteristics of two-dimensional convolutional neural network,a classification model of one-dimensional convolutional neural network is established to realize intelligent recognition of power quality disturbance signals.Specifically,the time-frequency resolution can be flexibly adjusted by using the multiresolution S transform and the advantages of the strong generalization ability of the convolutional neural network can be used to extract the characteristics of the disturbance signal.By using local energy and local frequency as input,the deep learning network is trained to accurately identify the disturbance signal and overcome the shortcomings of the traditional method,which is weak in migration and easy to produce overfitting.Finally,the method is compared with other methods,and the performance of the algorithm is verified from the aspects of recognition method reliability and operational efficiency.