Research On AGC System Of Power Grid Based On Model Predictive Control

This thesis was supported by National Natural Science Foundation of China (51277184) and independent research project of State Key Laboratory of Power TransmissionEquipment&System Security and New Technology(2007DA10512709209).Automatic Generation Control (AGC) system is an important secondary frequencymodulation tool in modern interconnected power system dispatch center. In recent yearslarge-scale distributed new energy power sources accessed to grid,these power sourcewith forms of volatility, intermittent and randomness make the grid active power outputrandom fluctuations increase drastically. With the development of industry and economy,unstable and impact load capacity increased dramatically in grid. These problems putforward higher requirements for the dynamic response performance of the AGC systemthat is responsible for grid load frequency control. With the development of intelligentand information grid, networked AGC systerm based on integrated information networkbecomes a research focus for its openness, a high degree of information sharing, controlflexibility and low cost. AGC system in the networked environment will face theproblems of communication delay, packet wrong sequence,loss and other issues, whichtraditional AGC system based on dedicated line does not need to consider. Therefore, itis quite significant to research for improving dynamic response performance of theAGC system and new control method for networked AGC. This article researched inapplying model predictive control in the interconnected grid AGC system. In traditionaldedicated line communication environment, this article establishes dynamic matrixmodel predictive control method for AGC system; In networked control environment,this article proposes a network AGC system control method which based on delaymodel predictive control algorithm. The main contents of the paper are as follows:Basic principle of automatic generation control has been introduced. Thecharacteristic of six kinds of interconnected power grid AGC system control modes,which come from three kinds of single area control methods composite applications, aredetailedly analyzed. Through detailedly analyzeing of the "three principles", whichinclude prediction model, rolling optimization and feedback correction, the basicprinciples of model predictive control is elaborated. It establishs the base theory formodel predictive control application to AGC.Dynamic matrix model predictive control algorithm is applied to the traditional type AGC system, by building model predictive controller to replace PI controller. Inthis method, Prediction model Dynamic is based on efficient matrix formation thatcomes form state equations of dynamic model of interconnected grid AGC system; ittakes the variation of governor as the input, the load variation as disturbance inputs, thefrequency variation of regions, tie-line exchange power and the area control error (ACE)as predictive outputs. Rolling optimization objective function coordinated controlincremental and the desired output target to get control law; Output prediction errorsfeedback to the prediction model to form a closed-loop controller. The simulation modelof two-area interconnected network AGC system based on model predictive control hasbeen established in Matlab/Simulink. Two compared simulation cases of dynamicresponse performance and robustness proves that the proposed control method iseffective and robustness.On the basis of analysis the impacts of the communication system to the networkedAGC system, this paper proposed networked AGC system control mothed based ontime-delay model predictive control algorithm. In the proposed method, utilizing theoptimal weighting forecast communication delay, predictive model takes thecommunication delay into account. The improved rolling optimization objectivefunction,which adopts similar PI structure form by increase the incremental error offorecast output, improves tracking capability. Besides a buffer is installed on theactuator, which stored the control variables’ prediction sequences and pick out the bestpredictive control variable according to the different communication condition.Simulation model based on the proposed control method of interconnected gridnetworked AGC system is built in Matlab/Simulink. Through two simulation cases ofrandom communication delay simulation and sustained extreme maximumcommunication delay, it proved the the proposed control method is correct andadaptable.