基于神经网络的变风量空调解耦控制

Research on Decoupling Control of VAV Air Conditioning System Based on Neural Network

目的分析变风量空调运行时各变量之间的耦合关系,针对变风量空调参数多变、强耦合的特点,提出一种变风量空调系统改进型误差反播神经网络解耦控制方法,对变风量空调温湿度控制系统进行解耦.方法把整个系统的解耦目标分解为N个子目标,每个子目标仅仅对一个回路通道进行解耦,其结构与指标函数简单,易于实现;并将模糊神经网络控制器与解耦控制器有机结合.结果解耦成功后,控制响应速度快、超调量几乎为零,达到期望温度后温度曲线保持不变,而此过程中湿度值基本没有变化,整个控制过程调节响应快,稳态误差小,解耦效果明显,有很强的控制精度和鲁棒性.结论 BP神经网络解耦控制算法具有很强的自学习功能和自适应解耦能力,能取得良好的解耦控制效果.

The paper aims to analyze the coupling relationship between every variable when VAV air conditioning operation,as the VAV air conditioning has variable parameters and strong coupling characteristics.We propose a variable air volume air conditioning system improved error reverse sowing neural network decoupling control method to decouple VAV air conditioning temperature and humidity control system.Decomposed decoupling target of the whole system for N goals,every sub goal only decouple a loop channels,its structure and index function is simple,easy to be implemented;and combined the fuzzy neural network controller and decoupling controller.After decoupling,control response becomes speed and overshoot is almost zero,temperature curve remains unchanged after the temperature reach the expected,among the process humidity value basically does not change,the whole control process adjustment response is fast,steady state error is small,the decoupling effect is obvious,has very strong control accuracy and robustness.The BP neural network decoupling control algorithm has strong self-learning function and adaptive decoupling ability,can obtain good decoupling control effect.