| Before diesel generator sets leave the factory,a series of performance tests need to be carried out.Test personnel judge whether the unit meets the factory standards based on the test report.However,when the generator fails,abnormal test data may cause the test report to not meet the standards,and the test personnel need to manually eliminate the fault,which will consume a lot of time and energy.Therefore,this article designs a generator set testing system and proposes a method to extract features from abnormal data and achieve fault diagnosis based on this system,so as to quickly locate faults and improve the automation level of the testing system.Firstly,in response to the actual testing requirements of the generator set,a dry load bank is used as the load for the generator set.This article designs the structure and main circuit of the load bank and selects the FX3U-16 MR PLC as the controller.The PLC receives instructions from the upper computer and controls the contactor to switch on and off,thereby achieving the purpose of load addition and reduction.The load bank works in conjunction with the computer,touch screen,and generator test instrument to realize automated testing.At the same time,all electrical parameters of the testing process are stored in the computer and corresponding test reports are generated.Secondly,in order to develop a fault diagnosis strategy,this article selects the relatively common stator winding inter-turn short-circuit fault and uniform demagnetization fault of permanent magnet synchronous generator(PMSG)during the testing process for modeling and simulation analysis.The effect of faults on PMSG output current is studied,and stator current data is selected as the simulation dataset for the fault diagnosis method proposed in this article.Furthermore,to address the issue of difficulty in extracting early fault features of generators and the potential for misjudging the degree of faults,this article proposes a feature extraction method based on the Whale Optimization Algorithm(WOA)optimized Variational Mode Decomposition(VMD)and an Improved Whale Optimization Algorithm(IWOA)optimized Support Vector Machine(SVM)pattern recognition method.The WOA optimizes the modes number K and penalty parameter ? in VMD,with sample entropy of signals as the fitness function,to determine the optimal parameter combination [K,?].WOA-VMD is used to decompose the current signal into several modal components,and based on the kurtosis criterion,a few modal components are selected to calculate their energy entropy and form feature vectors.These feature vectors are input into the IWOA-SVM pattern recognition model to classify the feature values and achieve fault diagnosis.Through comparative analysis of simulation experiments,the results show that the proposed method can accurately extract early fault features of generators,improve fault recognition accuracy,and achieve an accuracy rate of 98.75%.Finally,to verify whether the generator set meets its performance indicators,the designed generator testing system was used to perform load tests,voltage tests,and sudden load change tests on newly manufactured diesel generator sets,and test reports were generated.Based on this,the proposed fault diagnosis method was used to further process and analyze the abnormal test data,and the diagnosis accuracy reached 93.75%,confirming the feasibility and effectiveness of the method.Additionally,a diesel generator set fault diagnosis system was developed based on Matlab App Designer,which effectively reduced the workload of the testing personnel and improved the automation level of the testing system. |
