An Application of Improved SOM Neural Network in Fault Diagnosis of Wastewater Treatment

The selecting of the initial weights of the self-organizing map (SOM) neural network has an important impact on the performance of the network.In this paper,by means of an improved imperialists competitive algorithm (ⅡCA) to minimize the locally weight distortion index,the initial weights of the SOM neural network can be obtained optimally.Besides,the improved SOM neural network (ⅡCA-SOM) is applied in the clustering and fault diagnosis of the wastewater treatment process data.The experiment result demonstrates that,compared with the traditional SOM neural network,ⅡCA-SOM can attain better performance in clustering and lower misdiagnosis rate in the fault diagnosis.