Research And Implementation Of Fault Diagnosis Algorithm Based On ELM

Nowadays, with the rapidly development of China electricity, as one of the key power system equipment, the power transformer and its performance when it is running affects the entire power system's stability and reliability directly. Therefore, it's very essential to prevent and reduce the failures and accidents of transformer. Based on the study and analysis of all types fault diagnosis method of power transformers, this paper makes a deep study and research to the extreme learning machine. At the same time, this paper also applied ELM to the fault diagnosis of power transformers.Firstly, the paper makes a detail introduction and research of the ELM. This algorithm has so many advantages, such as easy to implement, fast, strong generalization ability. So we often use it on the solution of all kinds of problems. Here it is applied to the fault diagnosis of transformer. Through the experimental comparison with other diagnostic methods, the performance of this algorithm is verified.Secondly, based on the transformer fault diagnosis algorithm which based on KELM, this paper proposes a fault diagnosis algorithm which the nuclear parameters of KELM is optimized by the improved bat algorithm. And on this basis, we also use the5-CV to evaluate the model. Simultaneously, the paper also gives the diagnosis implementation process. Through the comparison of several related diagnostic algorithms and the proposed diagnostic methods, we can verify that the transformer fault diagnosis based on the improved bat algorithm to optimize the KELM has the higher accuracy and efficiency.On the basis of studying transformer fault diagnosis based on the ELM, according to the research results about related literature, we get an ELM improved algorithm which based on the dynamic adjustment of the hidden layer nodes. This algorithm can exclude the no contribution hidden layer nodes effectively and improve the generalization performance of the algorithm. In this paper, this algorithm is applied to the transformer fault diagnosis and get a good diagnostic results.Finally, the design and implementation of diagnostic system is finished by using the MATLAB/GUI. According to user demand analysis, the system defines function modules. At the same time, through the software test, we can sure that the diagnosis system is useful and reliable.