Energy Consumption Optimization Of Vertical Stirred Mills Based On Machine Learning

The vertical screw mixer mill is a commonly used crushing equipment known for its advantages of higher grinding efficiency,lower energy consumption,and smaller volume compared to other equipment.However,in the field of crushed ore milling,the vertical screw mixer mill is not as widely utilized as traditional ball mills and other grinding equipment.As a result,there is insufficient research on its structural parameters and process parameters.Furthermore,due to the coupling between different process parameters in the grinding process,establishing an accurate mathematical model has proven difficult.Consequently,in actual production processes,working parameters of the vertical screw mixer are adjusted based on experience.This approach not only fails to fully utilize the performance of the mixer mill but also risks energy wastage.To address these issues,this paper focuses on the vertical screw mixer mill as the research object.It adopts a combined method of discrete element simulation and experimentation and utilizes relevant machine learning algorithms to establish prediction models for each process parameter,energy consumption,and product powder yield in the grinding process.Additionally,a multi-objective optimization algorithm is employed to obtain the optimal combination of process parameters.The specific research content is outlined as follows:The grinding mechanism of the vertical screw mixer mill and the factors affecting its work efficiency are analyzed.The main process parameters to be studied are determined,and the working process of the vertical screw mixer mill is simulated using discrete element simulation to identify a relatively optimal process parameter range under the defined simulation conditions.Based on the simulation,experimental data is collected for energy consumption and powder yield under different process parameter combinations.The predictive performance of various machine learning regression algorithms on energy consumption and powder yield is studied and analyzed,with a comparison made to determine the best BP neural network prediction model.The paper also analyzes the shortcomings of the multi-objective optimization algorithm NSGA II and proposes an improved approach that combines Logistic-Tent chaotic mapping,adaptive crossover and mutation probability,elite reverse learning,and a new crowding comparison strategy.A comparison between the improved algorithm and the original NSGA II algorithm using test functions demonstrates the enhanced optimization capabilities of the proposed approach.Finally,using the BP neural network prediction model as the objective function,the improved NSGA II algorithm is applied to identify the optimal combination of process parameters for the vertical mixer mill under the defined conditions.A comparison with previous optimization data shows a 5.8% reduction in energy consumption while maintaining powder yield,providing valuable insights for energy-saving and consumption reduction in the mining and beneficiation industry.