A time-difference PLS based soft sensor method of underflow concentration for thickener in hydrometallurgy

摘要

To address the real-time measurement problem of underflow slurry concentration in hydrometallurgy, a soft sensor method based on time-difference partial least square (TD-PLS) is proposed. The time difference is utilized to overcome the data misalignment problem caused by the slurry crystal on the surface of sensors. The input secondary variables of the method are time difference press value relative to the initial time, and all the press sensors located in different depth, while the output is the corresponding time difference value of underflow slurry concentration relative to the initial concentration. It is essential that the initial slurry concentration should be measured precisely. A large amount of data is used to training the soft sensor model, and the method can estimate the real-time thickener underflow slurry concentration accurately. The method is applied to a thickener in certain refinery, and the measurement error is in the allowable deviation.