中文摘要：

针对温室温度控制系统所存在的大惯性、非线性等问题,提出一种基于改进型BP神经网络PID控制器（BP-PSO-PID）的温室温度控制技术。该控制器由经典PID控制器及神经网络构成,通过神经网络的自学习、加权系数的调整,使系统输出最优控制下的PID控制器参数Ki,Kp,Kd,并利用粒子群算法作为其中神经网络的学习算法,实现了对神经网络的改进,有效克服了传统BP算法的收敛速度慢、存在局部极小值等问题。仿真实验表明：相对常规PID以及BP-PID,该BP-PSOPID控制器大大改善控制过程的响应速度、调节时间、超调量、误差等性能,且在加入干扰的情况下,该控制器的调节时间最短,波动最小,表现出更强的抗扰能力及适应性,从而大大提高温度控制过程的稳定性、精确性与鲁棒性。

英文摘要：

In view of the problems of nonlinear and large inertia existed in the temperature control system of greenhouse,the BP-PSO-PID controller based on improved BP neural networks is proposed for the greenhouse＇temperature control.The controller is composed by classic PID controller and Neural Networ. Through the self-learning of neural network and the adjustment of weighting coefficients,the system outputs the optimal PID parameters Ki,Kp,Kd,and the improved neural network is realized by using particle swarm algorithm. The learning algorithm can overcome effectively the shortcoming of standard BP algorithm,such as slowly convergence and easily immerging in partial minimum,and can provide better assurance for the parameters＇ adaptive adjustment of the BP-PID. The simulation experiment indicates the performances such as response speed,adjust time,overshoot,error have been greatly improved by the BP-PSO-PID controller compared with the conventional PID and BP-PID. a The controller also exhibits the stronger anti-disturbance ability and adaptability in the presence of interference. It can shorten regulation time and reduce the fluctuation. The controller can greatly improve the stability,accuracy,obustness in the temperature control of greenhouse.