Diagnosis Method of AC Filter Circuit Breaker in Converter Station Based on RBF Neural Network and Expert Experience Method

Abstract

In the DC transmission system, due to the limitations of technology, a large number of harmonics will be generated and reactive power will be consumed when the converter performs current conversion. In order not to burden the power grid, each converter station will automatically put in the corresponding AC filter bank on the AC side according to the transmission power during operation. To analyze the fault information of the circuit breaker in time, this paper proposes a fault diagnosis method for the AC filter circuit breaker of the converter station based on the combination of RBF neural network and expert experience method. For fault diagnosis based on RBF neural network, the recorded waveform of AC filter circuit breaker is used as input, and the judgement of whether the waveform is abnormal is used as output. For fault diagnosis based on expert experience method, an expert experience library is established, and the waveform of the AC filter circuit breaker is also used as the expert diagnosis input, whether the waveform is abnormal is used as the output. It is mainly based on the threshold of the current difference, the failure threshold of the closing resistance, etc. to determine whether the AC filter circuit breaker has a potential fault. The example results show that this method can find the potential fault information of the AC filter circuit breaker and issue an early warning before the protection device operates.