Research On High Reliability Divided-Period Reactive Power Optimization In Regional Power System

In practice, since on-line load is constantly fluctuating, the reactive power optimization is needed to be conducted constantly in theory in order to keep optimal state of the power system. By regulating the taps of transformers, the reactive power compensators and controlling reactive power of the generator, reactive power optimization is to distribute reactive power reasonably to improve the quality of voltage and decrease the losses of network.In the process of reactive power optimization, there are two problems needed to be solved: the first problem is that how to determine the time of reactive power optimization. The traditional method chooses each time of the daily 24 times as the fixed time of reactive power optimization. Because of the daily power load data at the different times are unfix, the traditional method easy to cause the frequent operation of compensation devices at the time when the reactive power needed not to be optimized, thus the traditional method easy to reduce the number of switching operations of compensation devices. Choosing a reasonable method to determine the time of reactive power optimization plays a key role in reducing the number of switching operations of compensation devices. The second problem is that which node should be chosen to determine the time of reactive power optimization. In the power system, Because of the power load data in different nodes are difference, the time of reactive power optimization of each node is different, thus choosing a node which can reflect the voltage level of network to determine the time of reactive power optimization.The divided-period method based on multi-fractal characteristic parameters was proposed to determine the time of reactive power optimization. The daily power load data at the same time are not identical or periodic, but the tendencies of the daily power load are similar. So this situation accords with the fractal theory in nonlinear dynamics. Considering the maximum allowable daily number of switching operations of compensation devices, the compensation times are classified adaptively through clustering 24 different times which have the similar multi-fractal characteristic parameter in one day. This method which conducted reactive power optimization in every compensation time greatly reduces the number of switching operations of compensation devices.The method of selecting the pilot node to determine the time of reactive power optimization was proposed. The pilot node can reflect the voltage level of network and have a close relationship with other nodes.The main works in this paper includes three parts:First, we select the pilot node using sensitivity matrix, this method well achieve the coordination between every time of reactive power optimization of each node.Second, the divided-period method based on multi-fractal characteristic parameters was proposed to determine the time of reactive power optimization to reduce the number of switching operations of compensation devices. Using the fractal theory, considering the maximum allowable daily number of switching operations of compensation devices, the compensation times are classified adaptively using hierarchical clustering method through clustering 24 different times which had the similar multi-fractal characteristic parameter in one day, this method which conducted reactive power optimization in every compensation time greatly reduces the number of switching operations of compensation devices.Finally, we wrote the program of the divided-period method to carry out the simulation and debugging. Through analyzing the classification result, we proved that the divided-period method based on multi-fractal characteristic parameters proposed in this paper is effective. The methods are further proved effective by putting them into application.