Research On Fault Diagnosis Method For Hydroelectric Generators Based On Multi-dimensional Feature And Ensemble Classifier

With the rapid development of China’s economy,China’s demand for energy is also growing.At the same time,in the context of energy revolution and power transformation,vigorous development of renewable energy meets the national development needs and has become an important part of China’s energy structure adjustment.As the most mature renewable energy,the effective development and utilization of hydropower energy is one of the key points of energy development.As the key equipment for developing hydropower energy,it is of great significance to ensure the safe,stable and efficient operation of hydropower units.However,due to the complexity of the corresponding relationship between the faults and symptoms of hydropower units,the fault diagnosis of hydropower units has always been a research difficulty and hot spot,and the key to the fault diagnosis of hydropower units lies in the extraction of fault features and the identification of fault states.Based on this,this paper starts from two aspects of fault feature extraction and fault state recognition,establishes a fault diagnosis model of hydropower units based on multi-dimensional features and ensemble classifier,and comprehensively improves the accuracy of fault diagnosis of hydropower units from two aspects of feature extraction and fault recognition.In order to improve the accuracy of fault diagnosis of hydropower units from the aspect of feature extraction,this paper synthesizes the time domain features,frequency domain features and time-frequency domain features to form multi-dimensional features.Among them,the set empirical mode decomposition-sample entropy is extracted as time-frequency domain feature to make up for the shortcomings of traditional time-domain feature and frequency-domain feature,and the genetic algorithm is used to select multi-dimensional feature to remove redundant feature in multi-dimensional feature.Finally,the filtered multi-dimensional feature replaces the traditional one-dimensional feature as the input feature of fault recognition for fault diagnosis.In order to improve the accuracy of fault diagnosis for hydropower units from the aspect of fault recognition,this paper firstly uses three classifiers,namely,support vector machine,back propagation neural network and Naive Bayes,to diagnose the faults and obtains the preliminary diagnostic results,and then uses weighted voting method to combine the three classifiers to get the ensemble classifier,and fuses the three classifiers by the ensemble classifier.The diagnosis result is the final diagnosis conclusion.Finally,a fault diagnosis model for hydropower generating units based on multi-dimensional features and ensemble classifier is established.Rotor test-bed is used to simulate four common running states of the rotor,and the vibration signal data of the rotor are collected to verify the effectiveness of the fault diagnosis method in this paper.Using 120 sets of experimental data,the fault diagnosis method of hydroelectric generating units based on multi-dimensional features and ensemble classifier is used to diagnose the rotor misalignment,rotor unbalance,rotor rubbing and normal operation of the rotor.The accuracy of fault diagnosis can reach 98.00%.The results show that the fault diagnosis method based on multi-dimensional feature and ensemble classifier can improve the fault diagnosis accuracy of hydropower units from two aspects: feature extraction and state recognition,compared with single-dimensional feature and single classifier.