DOU-DCGAN: Optimization of cement grinding operational metrics in a dynamic environment

The coordinated optimization of unit power consumption and cement specific surface area is of great significance in improving the production efficiency of cement grinding. Targeting the problems of multiple equivalent optimal solutions and easy to fall into local optimal solutions in Pareto-based optimization algorithms, this paper proposes a new optimization decision model based on DCGAN, taking unit power consumption and cement specific surface area as the optimization objectives. A new generator structure is proposed to solve the optimization problem in a dynamic environment. Production indicators are incorporated into the loss function, and a global optimal solution is selected as the target value based on historical experience, getting rid of the decision-making problems of local optimal solutions and multiple equivalent optimal solutions. Finally, experiments have proven that the proposed model not only achieves the goal of low power consumption and high quality, but also improves the stability of the production process. [Display omitted] •A multi-objective optimization framework for cement grinding process is proposed.•This model can generate real-time operational indicators for decision-making.•Improving the loss function to guide production towards global optimization.•Combining U-Net with DCGAN as a generator structure.•The generated decision values can stably reduce power consumption and ensure cement quality.