论文部分内容阅读
针对粒子群优化算法(particle swarm optimization,PSO)在高维空间复杂曲面寻优时易于陷入局部最小值的问题,分组扰动粒子群优化算法(partially-perturbed particle swarm optimization,PPSO)结合问题特征,采用启发式规则,实施参数分组扰动策略,对PSO算法进行改进,从而增大了跳出局部极小的可能性。本文主要研究PPSO在精馏塔模型参数闭环辨识上的应用,分别针对模型参数可辨识性,参数的不同分组,鲁棒性进行分析验证;并在其他精馏塔模型上进行了相应的验证。仿真实验表明,PPSO辨识算法比序列近似法等其它辨识算法具有更高的辨识精度,并且具有较强的鲁棒性;在其他精馏塔模型参数辨识上PPSO算法也同样取得了很好的辨识精度。实验结果证明了PPSO算法在精馏塔模型参数闭环辨识上的可行性和有效性。
Aiming at the problem that Particle Swarm Optimization (PSO) tends to fall into local minimum when finding complex surfaces in high-dimensional space, Part-perturbed particle swarm optimization (PPSO) Heuristic rules, the implementation of parameter grouping disturbance strategy, the PSO algorithm is improved, thus increasing the possibility of jumping out of local minima. In this paper, the application of PPSO in the closed-loop identification of distillation column model parameters is studied. The identifiability of the model parameters, the different groups of parameters and the robustness are analyzed and verified respectively. The corresponding verification is carried out on other distillation column models. The simulation results show that the PPSO algorithm has higher recognition accuracy than other approximate algorithms, such as sequence approximation method, and has strong robustness. The PPSO algorithm has also been well identified in other parameters of the distillation column model Accuracy. Experimental results demonstrate the feasibility and effectiveness of the PPSO algorithm in closed loop identification of distillation column model parameters.