A novel dynamic history-based algorithm for flotation process optimisation

In this paper, a novel dynamic history-based optimisation algorithm (DHOA) has been proposed and applied in the optimisation of two different copper flotation processes. The proposed algorithm dynamically optimizes flotation process for stable recovery, using historic patterns and predictive models. Performance of the proposed DHOA algorithm has been compared with naïve search-based algorithm (NSA), genetic algorithm (GA) and particle swarm optimisation (PSO). Experimental results show that DHOA offers much faster recovery time and copper savings compared to other methods (NSA, GA and PSO). NSA led to significant loss of time and copper during optimisation. PSO also showed some improvement over GA for the average optimisation time and copper savings. Integration of multilayer perceptron showed better results compared with extreme gradient boosting and decision tree in the proposed DHOA algorithm. Details of the proposed novel algorithm and detailed results from the different applications have been presented.