基于高斯搜索的改进粒子群优化在磨矿预测控制中应用

磨矿车间工业现场在保证控制效果的同时，一般要求控制变量具有较小的变化率。提出一种基于高斯搜索的改进粒子群优化算法，该算法以高斯分布来初始化粒子群，并改进粒子速度更新公式，将所提算法融合到最小二乘支持向量机预测控制中。针对选矿厂磨矿过程，给出了基于最小二乘支持向量机的预测控制系统，以及基于高斯搜索的改进粒子群优化算法步骤。对实际磨矿过程进行仿真实验，结果表明该算法在保证控制效果的同时，能大幅度减小控制量的变化率，具有良好的性能指标和应用前景。

Application of improved particle swarm optimization based on Gaussian search to grinding predictive control

It is necessary for the industrial field of grinding plant to ensure the control effect and its control variables with less variation rate at the same time. An improved particle swarm optimization algorithm based on Gaussian search is proposed, where the particle swarm is initialized by the characteristics of Gaussian distribution, and the particle velocity update formula is modified. The proposed algorithm is combined with a least square support vector machine (LS-SVM)-based predictive control. Aiming at the grinding process of a concentration plant, a predictive control system based on LS-SVM is designed, and the steps of the improved particle swarm optimization algorithm are also provided. The results of simulation experiments on actual grinding process demonstrate that the proposed algorithm can greatly reduce the control variable changing rate while ensuring the control effect, and have good performance index and application prospects.