摘要
针对温室温度控制系统所存在的大惯性、非线性等问题,对基于重新参数化的B样条神经网络以及考虑到早熟现象的改进粒子群算法的B-BP-PSO-PID控制器进行研究;提出由PSO寻优找到最适合的β因子,得到适合权值搜索的最佳B样条基函数;提出由考虑早熟处理的改进粒子群算法取代传统BP后向传播算法来作为学习算法,有效克服传统算法易于陷入局部最优的缺点;仿真中以某温室20个不同时间的温度数据作为测试样本,结果表明,新型控制器的平均超调量为9.9%,平均响应时间为8.7s,可以实现对温室温度的最优化控制。
In view of the problems of nonlinear and great inertia existing in the temperature control system of greenhouse, the B-BP PSO-PID controller based on re parameterization B-spline neural networks and improved PSO considering prematurity is studied. The controller can find the best suitable β factor by PSO, receive the best B-spline basic functions fit for searching weights, at the same time, substitute improved PSO considering prematurity for conventional backward-propagation （BP） algorithm which overcome the shortcomings of easily going into local optimum. With 20 test samples of temperature data at different time in a greenhouse, the experiment results indi cate the B BP-PSO-PID controller＇average overshoot is 9.9%, the mean response time is 8.7s, it can achieve the optimal control for the temperature of greenhouse.
出处
《计算机测量与控制》
CSCD
北大核心
2012年第9期2423-2425,2428,共4页
Computer Measurement &Control
基金
国家自然科学基金(60977038)
江苏省科技支撑项目(BE2009100)
淮安市科技支撑计划农业项目(SN1161)
关键词
温室
温度控制
B样条函数
重新参数化
粒子群算法
PID
greenhouse
temperature control
b-spline function
re-parameterization
PSO algorithm
PID
