基于改进LMBP神经网络的地源热泵空调控制研究

Research of Ground Source Heat Pump Air Conditioning Control Based on Improved LMBP Neutral Network

摘要:

针对BP神经网络训练速度慢并且容易陷入局部最小点的缺点,研究了LM优化算法；为解决LM算法中求解逆矩阵这一最耗时因素的问题,提出将共轭斜量法与LMBP算法结合起来,并通过Matlab编程语言实现,并将其应用到地源热泵冷冻水控制回路中;通过仿真对比,结果表明这种通过算法改进的控制器比PID控制器调节时间快约10 s、超调量小0.4,稳态误差小,具有明显的控制效果。

Abstract:

For BP neural network training is slow and is easy to fall into local minimum point , a optimized LM algorithm was improved. To solve the most time-consuming issue of solving the inverse matrix of factors in LM algorithm, combination of conjugate gradient method and LMBP algorithm was improved, and it was solved through the Matlab programming language, and it was applied to the ground source heat pump chilled water control loop. The simulation results show that the controller improved with algorithm is 5 seconds faster than PID controller and reduces about 0.4 overshoot, having steady-state error and control effect was very obvious.