Research On Nonlinear Voltage Predictive Control Algorithm Of Regional Power Grid

In the age of huge network, ultra-high voltage, heavy loads and long-distance transmission of power system, voltage stability becomes a key work affecting the safe operation. The traditional method usually rules the reactive power equipments and the controllers scattered across the system by local information to maintain local stability. And setting voltage reference for nodes by power flow calculations can hardly consider system dynamic and constraints. We adopt the nonlinear system model and combine hierarchical idea and model predictive control into voltage control to achieve global dynamic stability.This paper presents a hierarchical nonlinear voltage predictive control method for regional network. It includes the low-layer and the predictive control layer. The predictive control layer takes the dynamic model of power system and related constraints as the optimization problem and calculates out the optimal voltage references of the notes of controllers in the low-layer. The low-layer includes the original controllers of the major equipments of the regional power grid, such as excitation controllers of generators, voltage/ reactive power controllers of substations and so on. The low-layer uses the voltage references from the predictive control layer as the inputs of local controllers.Firstly, we use the Power System Toolbox(PST) to simulate power system in the low-layer and on this basis on-load tap changer model, switched capacitor model and exponential recovery load model are added in the system. Those models are proved correct by simulation examples. Then, the paper introduces the generator mode, the load model and networking method to get the nonlinear voltage predictive model in the high-layer. After that, we use the least-square residual of system inputs and states as the objection function, using the nonlinear predictive model as the equality constraints and the limits of system states, outputs and inputs as the inequality constraints to establish the rolling optimization problem of voltage p-step predictive control. At last, interior-point method is adopted to solve the optimization problem obtain optimal voltage references. Before sending references, we have to verify that whether they meet the sending time interval and the error band to avoid negative effects.The proposed method is programmatic implemented in MATLAB. We take advantage of IEEE New England 39-bus system and 10-bus system to verify the effectiveness of the method by comparing the system responses with and without the predictive control layer.