Prediction Model for Production Indexes of a Flotation Circuit Based on Adaptive PCA and Composite Kernel Support Vector Regression

This An accurate prediction of important production indexes such as recovery rate and concentrate grade is essential for successful monitoring and controlling of a flotation circuit. However, due to the inherent characteristics of a flotation circuit such as time-delay, strong nonlinearities and uncertainties, these production indices are usually difficult to predict accurately. In this paper, a prediction model was developed to estimate the concentrate grade and recovery rate of a flotation circuit based on adaptive principal component analysis (APCA) and composite kernel support vector regression (CK-SVR). Firstly, the flotation prior knowledge and APCA are employed to reduce the dimension of input vector of CK-SVR. Then, considering that the characteristics of kernels have great impacts on learning and predictive results of SVR, a composite kernel SVR modeling method based on polynomial kernel and RBF kernel is adopted which hyperparameters are adaptively evolved by the particle swarm optimization (PSO) algorithm. Simulations using real operating data show that the prediction model provides the necessary accuracy for a flotation circuit.