论文部分内容阅读
提出了一种用前向神经网络（ＢＰ网络）来诊断新型无功发生器（ＡＳＶＧ）逆变器主回路元件开路故障的方法。把逆变器输出电压波形在一个周期内分为４个区，则可以发现单个元件开路的相应故障特征。这个特征表现在某一区域的电压值异常现象（为浮动状态），这可以用前向神经网络来进行识别。通过电力电子仿真软件（ＰＳＩＭ），得到了主回路单个元件开路时逆变器输出的电压波形样本，利用这些样本可以训练神经网络。研制了用神经网络诊断故障的软件，对测试样本的识别得到满意结果。
A method to diagnose the open circuit fault of the main circuit element of a novel reactive power generator (ASVG) inverter by using a forward neural network (BP network) is proposed. The inverter output voltage waveform in a cycle is divided into four regions, you can find a single component open the corresponding fault characteristics. This feature represents anomalous voltage values in an area (floating state), which can be identified using a feedforward neural network. Through the power electronic simulation software (PSIM), the voltage waveform samples of the inverter output when the single circuit of the main circuit is open are obtained, which can be used to train the neural network. The software that diagnoses the fault by neural network is developed and the result of the recognition of the test sample is satisfactory.