Research On Control Strategy Of Heat Exchange Station Based On Heat Network Decoupling Technology

In view of the problem of poor control effect in the central heating system due to the coupling of the two loops of heat exchange station quality and quantity regulation,the use of traditional un-decoupled quality control may lead to high energy consumption,slow regulation speed,poor system controllability and poor controllability.A neural network decoupling implicit generalized predictive control strategy based on thermal network quality and tuning is proposed.There is a strong nonlinear coupling time-delay relationship between the variables of this type of system,and its accurate mathematical model is not easy to identify.Therefore,its control should fully consider the temporal relationship between data,and the adaptability and robustness of the control strategy.First of all,establish the control relationship for the quality regulation of the thermal power station.The type and order of the heat exchange station model are determined by mechanism modeling;the model parameters are identified by the experimental modeling method based on the soaring experiment and the least square fitting,and the deterministic partial model of the heating process is obtained;the controlled autoregressive sliding The Integral Average Model(CARIMA)time series modeling method determines the stochastic part of the model;synthesis gives its complete dynamic description.By building an actual physical system,a special study is carried out on a small double-input double-output(TITO)device to simulate the heat exchange station system and verify the control effect of the strategy in this thesis.Secondly,there is a coupling relationship between the quality control and quantity control channels of the heat exchange station.According to the characteristics of the heat exchange station,the coupling channel is established,and the time-delay recurrent neural network decoupling controller for the heating process is designed based on the feedforward idea.There are some disturbances and abnormal point data in the thermal network control,which lead to the unreasonable selection of the hidden nodes of the neural network.The K-means algorithm is used to improve it based on density clustering.The coupled nonlinear function is designed to verify its effect,and the recursive time series processing is compared.As a result,the mean square error(MSE)and absolute error(MAE)of the improved RBF neural network decoupler training were reduced by 41.1% and 24.2%,respectively,and its decoupling performance was analyzed combined with the heat exchange station process simulation.Thirdly,combined with heat exchange station process design control strategy.The quality regulation channel has time-varying characteristics.The Implicit Generalized Predictive Control(IGPC)strategy is introduced,and the rolling optimization and online correction links are used to overcome the time-varying influence of the quality regulation channel.,to find the adaptive control rate.The experimental results show that under the same disturbance conditions,the stability time of the strategy in this thesis is reduced by 136 s and the overshoot is reduced by 17.2% compared with the generalized predictive control(GPC)strategy.The quantity regulation channel control model has hysteresis characteristics,and the Bang-Bang algorithm is used to control its channel.Finally,based on the architecture of cyber-physical system(CPS),a new heat network control strategy system is established to realize the real-time online decoupling control of the heat exchange station system.Verified by the physical experiment platform and the industrial field,this scheme can accurately control the temperature difference and flow of the secondary network of the heat exchange station online.Through the application of the heating area of about 800,000 m2 in a university in Dalian,the annual cost saving is about 280,200 yuan.Moreover,the strategy in this thesis has good adaptability and robustness,and can be extended to the control research of coupled time-varying time-delay systems similar to industrial sites.