Maintaining the discharge standards of biologically-treated wastewater is essential in order to protect receiving water bodies from secondary pollution. If unexpected faults happen in an activated sludge process that is used as biological treatment, it would affect the quality of the receiving water body. Thus, in this paper, an improved particle filter (PF) algorithm, based on a variable frequency mutation (VFM) strategy is proposed for the process faults diagnosis. This is inspired by the frequency conversion for energy-saving application in the industrial process; an adaptive frequency conversion operator has been incorporated into the mutation operation of immune algorithm to reduce the system operating costs. Then, the resampling process of PF algorithm was replaced by particle mutation based on the previously calculated information for securing the diversity and the effectiveness of the particle; lastly, a VFM-based PF algorithm for system states estimation and fault diagnosis was established. This algorithm not only effectively increases the adaptability of the particle to changes of the system state, but also conductively solves the problems of the degeneracy in the traditional PF algorithm and the diversity weakening caused by resampling operation. Simulation results show that the algorithm can effectively improve the estimation accuracy of the nonlinear system states. The application results of faults diagnosis in the activated sludge process show that it can accurately diagnose the occurrence of faults. Therefore, the proposed method has great practical significance in wastewater treatment plants to avoid problems caused by unexpected faults. (C) 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.