Author(s): Zhengchao Xie; Inchio Lou; Wai Kin Ung; Kai Meng Mok
Linked Author(s): Kai Meng Mok
Keywords: Phytoplankton abundance; Reservoir; Relevance Vector Machine; Genetic algorithm
Abstract: A hybrid intelligent model combining relevance vector machine (RVM) and genetic algorithms (GA) was developed for optimal control of parameters to predict (based on the current month’s variables) and forecast (based on the previous months’ variables) the non-linear phytoplankton dynamics in Macau Main Storage Reservoir (MSR) that is experiencing algal bloom problems in recent years. The models used the comprehensive 8 years’ monthly water quality data for training, and the most recent 3 years’ monthly data for testing. Twenty four water quality variables including physical, chemical and biological parameters were involved. The nonlinear process with measurement noise was regressed by RVM that is based on Bayesian framework, and the near-optimal set points maximizing the objective function were sought by GA. The results showed that the RVM-GA based models are valuable in prediction and optimization of the complex nonlinear processes, and are feasible in understanding the algal bloom problem and predicting/forecasting the phytoplankton abundance. This method provides a useful tool for practical algal bloom control and raw water monitoring program.
Year: 2013