基于改进粒子群优化的磨矿过程预测控制

发布时间：2019-06-25 21:27

【摘要】：磨矿过程在选矿工业中起着至关重要的作用,选矿厂的工艺指标和经济效益与磨矿产品的优劣直接相关。磨矿过程是一个具有生产过程缓慢、非线性、大时滞、多变量耦合、生产状况变化大及环境干扰强、噪声严重等特点的复杂生产过程,很难对其进行系统优化与控制。如何对磨矿过程进行良好的控制吸引了国内外学者的广泛关注。国内外对磨矿过程的控制方法中,对控制量变化率的约束并没有得到充分重视,而控制量的频繁且大范围变化显然对工业设备不利,不但浪费能源也可能带来安全隐患。针对磨矿过程,本文建立了一种基于最小二乘支持向量机的非线性预测控制器,提出了基于高斯搜索的改进粒子群优化算法(G-IPSO),将其作为非线性预测控制中最佳控制量的优化求解方法。该算法以高斯分布来初始化粒子群,将前一个计算周期的控制量作为粒子群的初始中心,并将其引入到粒子的速度更新中,改进粒子的速度更新方式,使得在迭代寻优的过程中粒子具有向初始中心靠拢的趋势,加强初始中心附近的搜索强度,从而减小控制量的变化率。将该优化方法融合到最小二乘支持向量机预测控制中,并简约求解最佳控制量的适应值函数。以磨矿分级过程为研究对象,对其应用本文所建立的预测控制器,并给出本文提出的G-IPSO优化求解控制量的详细步骤。为验证该方法在预测控制效果及对控制量变化率约束方面的优势,应用本文所提出的G-IPSO优化方法建立的最小二乘支持向量机的非线性预测控制器,针对经典的多峰值函数和某工况下的磨矿工业过程进行计算机仿真验证。实验结果从控制效果和被控量变化率两方面进行了对比,表明了本文所提出的G-IPSO优化算法在约束控制量变化方面具有良好的性能指标和应用前景。
[Abstract]:The grinding process plays a very important role in the ore dressing industry, and the process index and economic benefit of the ore dressing plant are directly related to the quality of the mineral products. The grinding process is a complex production process with the characteristics of slow production process, non-linearity, large time lag, multi-variable coupling, large production condition, strong environmental disturbance and serious noise. How to control the grinding process has attracted the wide attention of the domestic and foreign scholars. In the control method of the grinding process both at home and abroad, the restriction of the rate of change of the control quantity is not given full attention, and the frequent and large variation of the control quantity is obviously disadvantageous to the industrial equipment, and not only is the waste of energy, but also the potential safety hazard. In this paper, a nonlinear predictive controller based on the least square support vector machine is established for the grinding process, and a modified particle swarm optimization algorithm (G-IPSO) based on the Gaussian search is proposed, which is used as the optimal solution for optimal control in the nonlinear predictive control. in that method, a particle swarm is initialized with a Gaussian distribution, a control amount of a previous calculation period is taken as an initial center of a particle swarm, Such that the particles have a tendency to close to the initial center during the iterative optimization, to enhance the search intensity in the vicinity of the initial center, thereby reducing the rate of change of the amount of control. The optimization method is integrated into the least-squares support vector machine prediction control, and the optimal control amount adaptive value function is simply solved. In this paper, the prediction controller set up in this paper is applied to the grinding and classification process, and the detailed procedure of the G-IPSO optimization solution control is given in this paper. In order to verify the advantages of the method in predicting the control effect and the constraint of the rate of change of the control quantity, the nonlinear predictive controller of the least square support vector machine set up by the G-IPSO optimization method proposed in this paper is applied, The computer simulation and verification are carried out for the classical multi-peak function and the grinding industry process under certain working conditions. The results show that the G-IPSO optimization algorithm proposed in this paper has good performance index and application prospect in the variation of the constraint control amount.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TD921.4;TP18

【参考文献】