开关电源滤波电容失效的故障诊断模型构建

Construction of Fault Diagnosis Model for Filter Capacitor Failure in Switching Power Supply

高压滤波电容失效作为开关电源常见的故障之一,直接影响其稳定运行。为实现在不损坏开关电源结构的前提下,完成故障的高效诊断,本次提出一种基于改进深度置信网络的故障诊断模型(DBN-DF模型)。该模型可利用数据采集装置,对开关电源所在的电网电压进行隔离、滤波等处理,并根据电网电压畸变特性,判断开关电源滤波电容失效百分比,最终输出故障类别,完成故障诊断。经试验验证,DBN-DF模型相较于传统的DBN和BP模型,对于高压滤波电容故障诊断的准确性更高,且随着故障次数的增加,模型真确率准确率也相应提升,最终趋于稳定。说明该模型具有较高的实践应用价值。

The failure of high-voltage filtering capacitors is one of the common faults in switching power supplies,which directly affects their stable operation.To achieve efficient fault diagnosis without damaging the structure of the switching power supply,this paper proposes a fault diagnosis model based on an improved deep confidence network(DBN-DF model).This model can use data acquisition devices to isolate and filter the voltage of the power grid where the switching power supply is located.Based on the voltage distortion characteristics of the power grid,the failure percentage of the filtering capacitor in the switching power supply can be determined,and the fault category can be output to complete fault diagnosis.Through experimental verification,it has been found that the DBN-DF model has higher accuracy in diagnosing faults in high-voltage filtering capacitors compared to traditional DBN and BP models.Moreover,as the number of faults increases,the accuracy and accuracy of the model also correspondingly improve,ultimately stabilizing.This indicates that the model has high practical application value.