Fault diagnosis for wasterwater treatment plant based on an improved support vector data description method

In this paper, a fault detection method based on improved support vector data description is proposed for wastewater treatment plants. First, an improved Multi-Kernel Support Vector Data Description (MKSVDD) method is presented in which multicore-kernel functions are combined with the traditional support vector data description method. This can make the model more adaptable to different input samples, allowing it to have both learning and generalization capabilities. Then, the parameters of the MKSVDD are optimizted by Sparrow Search Algorithm (SSA), which overcomes randomness and uncertainty of human manual adjustment parameters. Finally, simulation results are given to demonstrate the effectiveness of the proposed method. Nine types of fault data sets generated by the Benchmark Simulation Model 1 (BSM1) are used and simulations results show that the model is more adaptable to the samples and has better indicators for anomaly detection compared to the original algorithm.