It is significant to ensure the temperature stability of the test tank, which has a direct impact on reducing production energy consumption, improving tank quality and ensuring experimental results. However, the high nonlinearity and long delay of the temperature control process in the test tank make it difficult to satisfactorily control the temperature by traditional control methods. To solve the problem, this paper designs a temperature prediction model for the test tank based on backpropagation neural network (BPNN), which has a good fitting ability for nonlinear systems. The proposed model was coupled with improved generalized predictive control (GPC) into a new test tank temperature control method, namely, BPNN-based stepped GPC. Simulation results show that the new control method could reduce the prediction error of the BPNN and effectively control the temperature of the test tank.

确保测试罐的温度稳定性非常重要，这对降低生产能耗，提高罐质量和确保实验结果具有直接影响。然而，测试罐中温度控制过程的高度非线性和长时延使得难以通过传统的控制方法来令人满意地控制温度。为解决这一问题，本文设计了一种基于BP神经网络的试验箱温度预测模型，该模型对非线性系统具有良好的拟合能力。该模型与改进的广义预测控制（GPC）相结合，成为一种新的试验箱温度控制方法，即基于BPNN的阶梯式GPC。