Power Quality Disturbances Identification Using Efficient Time-frequency Features Extracting And Selection

Power quality is the main control target of smart grid,and it is also the basis of pricing power price by quality.Therefore,it is necessary to conduct in-depth monitoring and analysis of the power quality of distributed generation at different access points with different noise levels.Efficient and accurate identification of power quality disturbance signals is the premise of locating disturbance sources and controlling power quality pertinently,which is of great significance.In view of the shortcomings of the existing power quality disturbance signal recognition,such as low signal processing efficiency and large information storage space,the paper proposes a power quality disturbance identification method with high signal processing efficiency and low information storage space.For feature extraction,two signal processing methods are used for feature extraction in single disturbance signal scenario and complex disturbance signal respectively.(1)In order to improve the signal processing efficiency of single-type power quality disturbances,taking nine single-type power quality disturbances as analysis objects,a feature extraction method of single-type power quality disturbances based on image feature enhancement technology is proposed.Firstly,the power quality signal is converted into gray image;secondly,three feature enhancement methods,gamma correction,edge detection and peak-valley detection,are used to enhance the gray image features and obtain the binary image;finally,the disturbance features are extracted from the binary image and the original feature set is constructed.(2)In order to reduce the pressure of information storage in the analysis of complex power quality disturbances,taking17 kinds of disturbance signals including complex disturbances as analysis objects,a method for extracting the features of complex power quality disturbances based on optimal multi-resolution fast S-transform of time-domain compression is proposed.Firstly,the S-transform time-frequency matrix is compressed in time domain and frequency domain,and the intermediate matrix is obtained.Then the perturbation features are extracted and the original feature set is constructed.After obtaining the original feature set,in order to remove redundant features,feature selection method based on feature Gini importance analysis and sequence forward search strategy is adopted.Firstly,the Gini importance of all the features in the original feature set is calculated,and the importance ranking is obtained.Secondly,the classification accuracy of each sub-feature set is calculated by using the sequence forward search strategy based on thereduction of the feature importance.Finally,the optimal feature subset is determined by considering the feature dimension and accuracy.After determining the original feature subset,the optimal feature subset is used to train the random forest classifier and the rotary forest classifier to recognize the single disturbance signal and the complex disturbance signal respectively.In the process of constructing random forest classifier,aiming at minimizing generalization error,Bayesian optimization algorithm is used to optimize the parameters of the classifier.In the process of constructing rotating forest classifier,the parameters of the classifier are optimized based on the classification accuracy and two kinds of difference metrics.Finally,the optimal classifier is used to identify power quality disturbances and output the recognition results.The effectiveness of the new method in practical industrial applications is proved by the measured power quality signals of a Portuguese distribution network.The research achieves the accurate identification of power quality disturbance signal in complex noise environment.The processing efficiency of single disturbance signal is improved significantly,and the information storage pressure in the process of complex disturbance signal analysis is reduced.The new method meets the needs of practical engineering application,the application of disturbance identification technology in power quality monitoring and diagnosis of power system are further promoted,it is of great significance for ensuring the safe operation of power system.