摘要:AbstractThis paper develops a comparative study based on several modeling methods for estimating the copper concentrate grade in the copper flotation process. Back propagation neural network (BPNN) method, partial least squares (PLS) method, just-in-time learning partial least squares (JIT-PLS) method, adaptive-PLS method and window-adaptive-PLS method are proposed respectively. The prediction effects and test errors of these modeling methods are compared and analysed by using the measurement data of copper grade from X fluorescent grade analyzer in the dressing plant. The result is that window adaptive partial least squares method is the most accurate method, which can accurately predict the copper concentrate grade of the next measurement period in advance and can be used to guide operations.
关键词:Keywordsmodelingestimationcopper concentrate gradecopper flotation process