论文部分内容阅读
针对纸浆漂白过程温度控制的大滞后、大惯性、对象参数时变的特点,传统的Smith预估器在理论上解决了纯滞后系统控制问题,但其依赖于被控对象精确的数学模型,在应用中存在缺陷。本文采用PID神经网络(PIDNN)与Smith预估相结合的算法,利用Smith对纯滞后系统进行预估补偿以及PIDNN的自适应、自学习和在线调整控制器的参数功能。仿真结果表明,Smith-PIDNN算法简单、稳定且收敛速度快,能有效的解决系统大滞后、大惯性及蒸汽压力时变等问题。
In view of the characteristics of large delay, large inertia and time-varying object parameters in the temperature control of pulp bleaching process, the traditional Smith predictor solves the problem of purely hysteretic system in theory, but it depends on the precise mathematical model of the controlled object There are flaws in the application. In this paper, the algorithm of PID neural network (PIDNN) combined with Smith prediction is used in this paper. It uses Smith to estimate and compensate PIDNN, self-learning and on-line parameter adjustment of controller. The simulation results show that the Smith-PIDNN algorithm is simple, stable and fast convergence, and can effectively solve the system lag, large inertia and steam pressure time-varying problems.