Abstract
When image processing and machine vision technology are used to extract features from the image of the ore belt of the shaking table, so as to realize the analysis of the processing indictors and mapping of control parameters. To realize the adaptive optimization of the multiple control parameters of the shaking table, it is necessary to have thorough access to the parameters of the internal and external properties of the gravity shaker, such as internal control parameters and external ore zone characteristics, as well as the processing indicators. In this study, information on the multi-scale characteristics of the zone is obtained through a visual experimental system, and the data-driven model of the separation process is constructed to characterize the relationship between the properties of the internal and external parameters of the shaking table, eventually, an adaptive optimization method of control parameters of the shaking table based on maximizing beneficiation efficiency is proposed. The research results show that the data from the geometric characteristics of the ore belts obtained from practical experiments all satisfy the statistical distribution requirements. In the three optimized support vector regression (SVR) models, the sparrow search algorithm optimized SVR (SSA-SVR) has the best comprehensive performance, which overcomes the limits of data samples under objective conditions and basically meets the existing industrial requirements. With these helps, the proposed optimization method has realized the continuous optimization of multiple control parameters of the shaking table, and the optimization results have a good guarantee.