| Limiting the power consumption of air conditioning is an effective way to solve the problem of the peak and valley power consumption of power grid.Fuzzy and PID control strategies are often used in the current inverter air conditioner compressor control.Temperature and temperature change rate are collected by dry bulb and wet bulb sensors on the indoor side of air conditioner as the data basis of control.According to the temperature setting value and the actual measured value,the fuzzy control constitutes the temperature deviation.Firstly,the temperature deviation and the temperature change rate are processed by fuzzy quantification,then the fuzzy rules written by the expert experience are used for reasoning,and finally the fuzzy value obtained by the reasoning is inverted and output to the controlled object;the PID control method generates control signal by linear combination of proportion,integral and differential of temperature deviation to control the object.Aiming at the problem that existing in current control methods,such as slow response speed,large temperature overshoot,and continuous temperature oscillation in the later stage,this paper studies the fuzzy neural network control method of inverter air conditioner combined with big data processing,which can achieve more accurate and rapid control of the temperature of inverter air conditioner,conserve energy and solve the problem of the peak and valley power consumption.The main research contents are as follows:(1)Setting up the power consumption database and storing the power data collected at the power input terminal.The ARIMA model method of time series analysis is used to analyze and preprocess the massive power data,establish the autoregressive moving average model,analyze the power consumption law and predict the power consumption trend as the data basis for limiting the power consumption.(2)Establishing the fuzzy Elman neural network controller.Based on the widely used fuzzy control rules of inverter air conditioner,data points are extracted evenly on the surface of control rules as training samples of neural network.At the same time,aiming at that the negative gradient algorithm used in the network training has the problems of slow convergence speed and easy to fall into local minima in the later training period,an improved artificial bee colony(IABC)algorithm is proposed,by combining the simulated annealing algorithm with the artificial bee colony algorithm,to train the network.Then the well trained network is used in the control of the inverter air conditioner compressor.(3)On the basis of the above research,the idea of predictive control is introduced,and the neural network is used to model and predict the indoor cooling and heating load as the control basis to limit the temperature overshoot and fluctuation,so as to achieve more accurate on-line predictive control of inverter air conditioner.Since on-line control requires higher training speed of neural network,the neural training method based on cubature Kalman filter is used to train the neural network.Simulation and analysis show that,compared with fuzzy control and PID control methods,the fuzzy Elman neural network control method of inverter air conditioner based on IABC algorithm can obviously solve the problem that the neural network is easy to fall into local minima in the later period of training,and effectively improve the training and control accuracy,and limit the temperature overshoot in the control process,which shows that the method is accurate and feasible and it can effectively reduce the energy consumption of air conditioning.On this basis,the on-line control by introducing the cubature Kalman filter can effectively solve the problem of temperature overshoot and temperature fluctuation,improve the control accuracy,and be more user-friendly. |
PDF Full Text Request