【摘要】 为优化精馏塔系统控制器的性能，提出一种添加正态分布函数的非线性递减惯性权重和对加速因子进行异步时变调节的改进策略，优化了粒子群算法的搜索效率和精度。并利用该算法对PID神经网络的初始权值进行训练，提高其性能。设计控制器并进行仿真，结果表明：训练后的PID神经网络控制器性能有较大提升，控制器的抗干扰能力和反应速度得到极大改善，有效提高了精馏塔的控制效果。更多还原

【Abstract】 For purpose of optimizing performance of the distillation column’s controller, a nonlinear decreasing inertia weight with added normal distribution function and an improved strategy for the asynchronous time-varying adjustment to the learning factor were proposed to optimize both search efficiency and accuracy of the algorithm, including making use of this algorithm to train initial weights of PID neural network and improve its performance. Designing and simulating the controller show that, the performance of the trained PID neural network controller is greatly improved; the controller’s anti-interference ability and response speed can be greatly improved together with effectively improved control effect of the distillation column.更多还原