Abnormal Data Cleaning Method for Municipal Wastewater Treatment Based on Improved Support Vector Machine

To reduce the impact of data noise and loss in the municipal wastewater treatment processes (WWTP), an abnormal data cleaning method was proposed based on improved support vector machine (ISVM) in this paper. First, a noise data detection method was designed to eliminate the noise data of WWTP by using the density estimation. Then, the proposed ISVM was used to design the data compensation model. This data compensation model can obtain the approximation of missing data by realizing the nonlinear fitting of the missing data. Finally, a particle swarm optimization (PSO) algorithm was adopted to optimize the parameters of ISVM to improve the precision of missing data compensation. This proposed cleaning method was applied to a real municipal WWTP, and the experimental results demonstrate that the proposed method can improve the data quality by eliminating the abnormal data and compensating the missing data.