Research On Key Technology Of Active Power Control In The PV Station

Striving to develop renewable energy such as PV generation is the importa-nt foundation of realizing clean alternative and power alternative. However, the intermittence and stochastic fluctuation of PV generation would bring negative i-nfluence to the operation of power system. This paper conducts the research ab-out the related key technology of PV station active power control from the pers-pective of dispatching end and station end, aiming at improving the system active power dispatching management with PV station and the PV station active power control level, promoting PV station to participate in system active power dispatching, and enhancing the security and economy of system operation.Aiming at the problem that the forecast of BP neural network method based on similar days is not accurate enough in non-ideal weather in the PV station short-term power forecast, the short-term power forecast method is proposed based on the gray system correction-wavelet neural network which is on the basis of adjacent days’ power sample. Aiming at the ultra-short term power forecast, The ultra-short term power forecast model using ARIMA(2,1,1) is proposed with time-series approach chose as the object of study, on the basis of the analysis of time-series characteristics of PV station historical power sample in Hefei. The forecast results show that the short-term forecast method and the ultra-short term forecast method proposed in this paper both have good index values under the condition of using the CORR, MAE, RMSE of forecast results and actual values as evaluation criterion.Aiming at PV array, the diagnostic method of PV array fault types based on fuzzy c-means clustering is proposed, in order to reduce the influences of weather fluctuations i n the process of fault diagnosis. The PV array fault types of are classified by different a pplication targets using the method of two-stage fault type. The first stage fault classifi cation is applied to power data amendment, and the second stage fault classification is applied to fault location. Combined with station supervision system, the fault character istic quantities are designed to apply to fault diagnosis. The fault diagnosis samples are obtained by using MATLAB to build PV array model and to simulate it. The example analysis shows that the proposed method not only can be used in the fault type diagnos is which applies to the large-scale PV array, but also can judge the fault types and forec ast the potential faults. On the basis of the PV array fault diagnosis results, the amendm ent methods of station power historical data and station ultra-short term power forecast results aiming at the fault diagnosis area are proposed with the specific strategy of stat ion output faults and based on amending PV module cross-power based values.Aiming at the requirement of the station control mode for PV station active power control system based on MAS, A power system active power control scene with PV station is extracted by taking the system operating states as the first classification standard and the role of PV stations participating in system active power dispatching as the second classification standard. Control objective of PV station active power is obtained based on the demand analysis to the control scene. On the basis of functional orientation of analysis dispatching end and station end, a power interval control mode is proposed under the dispatching mode of fat dispatching end and thin station end. In the perspective of load role and dispatching role, mathematical models of PV stations participating in system dispatching under two kinds of roles are built, and the general model of PV station based on it is proposed. General model unifies the dispatching models of PV stations under different roles, and provides the foundation for building the system dispatching model.In consideration of the differences between the station power forecast accuracy under different time scales, the system active power dispatching model under composite time scale is built, which is composed of the day-ahead generation plan and actual generation plan. On the basis of PV station ultra-short term forecast result, the difficulty of PV station power control and thermal power generation unit power regulation is reduced, using the modification method of actual generation plan to day-ahead generation planned values. And the reserve costs required by system side are reduced with the method that thermal power generation unit and PV station assume shared responsibility to PV station power fluctuation. The example analysis indicates that using the proposed model and method to dispatch PV stations can reduce the call volumes of system spinning reserve and improve the system economic efficiency. The model can satisfy the system operation requirements under the multiple conditions of different load ratio, station short term forecast error changes and system random faults.Aiming at the characteristic of PV station primary system distribution and the deficiency of power control system based on two-layer centralized structure, the solution of PV station power control system based on MAS is proposed, and the function model of each generation agent in MAS is designed. The control agent allocation strategy based on token ring and greedy algorithm is proposed in order to improve the reliability and operating efficiency of composite structural MAS. The power allocation method based on improved ant colonies algorithm on the basis of PV agent completion rates in order to improve the PV station power control efficiency. The two-stage power control strategies led by different agents aimed at unitary type and composite type of PV station are proposed respectively. The example analysis indicates that the control agent allocation strategy, the power allocation and the two-stage power control strategies proposed in this paper can improve the operating reliability and efficiency of MAS and satisfy the demands of station power control.The solutions of related key technology aiming at PV station power control proposed in this paper not only have certain academic value, but also have promotional effects of the economic and secure operation of power system and the two alternatives in energy territory.