Effective fault diagnosis methods can not only quickly and accurately identify the fault types of pumped storage units,but also reduce the operation and maintenance costs of pumped storage power plants.To address the problem of improper parame-ters adjustment in the relevant vector machine(RVM)leading to the improper diagnosis results,we proposed to optimize selection of the parameters in the RVM by using the pelican optimization algorithm(POA),so a classification model combined with Pelican search algorithm and relevant vector machine(POA-RVM)was constructed.After preprocessing and feature selection of the da-ta of four pumped storage units of the Xianju Pumped Plant under five states,a fault sample set was formed,and these fault sam-ples were classified by using the standard RVM,and the RVM models optimized by genetic algorithm,particle swarm optimization algorithm,and gray wolf optimizer respectively.The results showed that compared with the standard RVM and variety of RVM models optimized by genetic algorithm,particle swarm optimization algorithm and gray wolf optimizer respectively,the POA-RVM model effectively improved the accuracy of fault diagnosis of pumped storage units.
维普
万方数据