基于改进粒子群算法的RBF神经网络磨机负荷预测研究

Research on Mill Load Prediction of RBF Neural Network Based on Improved Particle Swarm Optimization Algorithm

摘要:

磨机负荷是评价磨机运行状态和预测磨机行为的重要指标，针对粉磨机磨矿过程中负荷难以检测和不能准确判断负荷状态的问题，提出了一种基于改进型粒子群算法（Improved particle swarm optimization, IPSO）优化径向基神经网络（Radial Basis Function，RBF）参数的磨机负荷预测模型（IPSO-RBF），使惯性权重因子在迭代过程中非线性下降，平衡局部搜索能力与全局搜索能力之间的矛盾，该算法能快速准确地找到最优解，提高粉磨机磨机负荷的预测精度。通过水泥厂的实测数据实验对比，结果表明，基于IPSO-RBF模型的预测精度最高，其预测结果与真实值相比较，均方根误差（Root Mean Square Error，RMSE）、均方误差（Mean Square Error，MSE）、平均绝对误差（Mean Absolute Error，MAE）、平均绝对百分比误差（Mean Absolute Percentage Error，MAPE）和决定系数（coefficient of determination，）分别为0.210 2、0.044 2、0.161 7、1.778%和0.978 2。

Abstract:

Mill load is an important index to evaluate the running state of the mill and predict the behavior of the mill. Aiming at the problem that the load is difficult to detect during the grinding process of the grinding mill and the load state cannot be accurately determined, an improved particle swarm optimization algorithm (IPSO) Optimized the radial load-based neural network (RBF) parameters of the mill load prediction model (IPSO-RBF), so that the inertia weight factor decreases nonlinearly in the iterative process, balancing the local search ability and the global The contradiction between search capabilities, the algorithm can quickly and accurately find the optimal solution, improve the prediction accuracy of the mill mill load. Through the experimental comparison of the measured data of the cement plant, the results show that the prediction accuracy based on the IPSO-RBF model is the highest, and the prediction results are compared with the real values. Root Mean Square Error (RMSE) and mean square error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and coefficient of determination () were 0.210 2, 0.044 2, 0.161 7, 1.778%, and 0.978 2, respectively.