For the cement production process, the optimization method of the grate cooler is important in reducing energy consumption and ensuring product quality. As a complicated and slow control process, there are several control objectives of the grate cooler, which are determined by design parameters. To compute the time delay of the design parameters automatically, we propose an improved long short-term memory with adaptive computation time (ACT-LSTM) model for objective prediction. An improved multi-objective optimization algorithm named bounded stable non-dominated sorting genetic algorithm II (BS-NSGA-II) is proposed to solve the optimal solutions. With the proposed methods, the average electricity consumption is reduced by 13.2%, the secondary air temperature is increased, the clinker outlet temperature is stabilized in a reasonable range, and the design parameters change smoothly. The experiment results have indicated that the proposed method is effective in the optimization of objectives and the stability operation of the equipment.
Available capacity computation model based on long short-term memory recurrent neural network for gelled-electrolyte batteries in golf carts
