Research On Key Techniques Of Boost Circuit Fault Prediction

With the rapid development of power electronics technology,more and more electronic devices are constantly emerging and are widely used in electronic communications,household appliances and energy fields.Power electronic circuit is the main component of power supply system.If the circuit fails,it may cause problems in the entire system,and then cause huge economic losses and labor consumption.A typical power electronic circuit is mainly composed of a main circuit such as an inverter circuit,a rectifier circuit,a corresponding driving circuit,and a control circuit.In actual work,the main circuit is most prone to failure,and the main circuit is composed of basic Buck circuits and Boost circuits.The health status of these basic circuits will directly affect the working performance of power electronic circuits,and then The ability of equipment to operate safely and effectively also has an important impact.In order to reduce the impact of these basic circuit faults on electronic equipment,this paper mainly studies the fault prediction of Boost circuits.Speci fic research contents include:1.The key components that cause the Boost circuit to fail are studied,and the equivalent circuit model and failure mechanism of each key component are studied and analyzed,so that the fault characteristic parameters and failure thresholds of each key component are determined.2.The influence of each key component on the different performance parameters of the circuit is studied.Through the degradation simulation of each key component of the Boost circuit,the parameter degradation curve of each key component is established and the segmented dynamic time bending distance is determined as the system level Feasibility of fault characteristic parameters.3.The fault prediction algorithm for Boost circuit is studied.Taking electrolytic capacitors as an example,the gray components(GM(1,1))based on trend fitting,cubic exponential smoothing(CES),and the improved GM(1,1)-CES combination prediction method were used to implement circuit components.Level of fault prediction,and the feasibility of the improved GM(1,1)-CES combination prediction method was determined through a variety of different evaluation indicators.Taking the Boost circuit as an example,the data-based SVM,XGBOOST,and LSTM prediction methods are used to implement circuit-system-level fault prediction.Similarly,different parameters of the same algorithm are compared and analyzed through a variety of different evaluation indicators,and different predictions are also performed.The algorithm is compared and analyzed.Figure[27]table[29]reference[70]...