| With the rapid growth of power load demand and the access of large scale of distributed generations (DGs) into distribution network, traditional planning method and operation control strategy is difficult to meet the increasing voltage quality demand of users. Under the background, this paper focused on the problems of reactive power planning and operation in distribution network, containing the reactive power configuration of substation,the voltage and reactive power control in distribution network with DG. The main content is as follows.(1) An optimal grouping method of shunt capacitors in substation is proposed, considering the temporal characteristics of reactive power demand curves. Firstly, the profile is partitioned into several segments; then a modified fuzzy C-means clustering (FCM) algorithm is presented to cluster the results of partitioning. The number of clusters is the number of groups, and the distance between adjacent cluster centers is the capacity of each group. Furthermore, the effect of the number of profile partitioning on the clustering results is analyzed while the number of capacitor groups is known. At last, the technical indicators such as voltage quality, energy loss and the the number of operations of capacitors are comprehensively considered to evaluate the optimization effect of different grouping schemes based on the nine field control.(2) In order to solve the voltage and reactive power control problem in distribution network containing large number of distributed generations,this paper proposed a coordinated control framework considering the spatiotemporal coupling characteristics of distribution network. Based on the analysis of the voltage and reactive power characteristics of common DG and the control characteristics of different types of voltage and reactive power equipment, a two-level coordinated voltage and reactive power control mode in distribution network is proposed to realize the spatial decoupling, and the two control levels continuously correct voltage and power actor of boundary gateway to achieve global optimization. Then a multi-time scale control method is presented to decouple the voltage and reactive power control in temporal scale. The influence of load and DG output forecast error on system control is progressively reduced by the rotational correction of three time scales, containing day-ahead, short-time and real-time, and finally the optimal control of voltage and reactive power in distribution network is realized.(3) A day-ahead reactive power optimization method in distribution network is proposed, to solve the problem of coordinated optimization of DG, on-load tap cganger (OLTC) and capacitors. The static optimal output of OLTC and capacitors are clustered in order by means of the optimal partition method,which realize the time decoupling of control variables and meet the constraints of maximum number of switching times. Based on the voltage and reactive power control characteristics of DG, OLTC and capacitors, the specific process of coordinated control is presented. Firstly, the swithing time and position of OLTC is optitmized; then capacitors operation time is determined; and the last step is the joint optimization of the output of capacitors and DG. Finally, case study is provided to verify the rationality and effectiveness of the proposed method.(4) As the existing reactive power optimization algorithm is not easy to converge and the calculation speed is slow, an entropy weight optimun seeking method (EWOSM) is proposed, which is suitable for the real-time reactive power optimization in distribution network. The method contains two steps, load distribution matching and entropy weight method. Firstly, based on the analysis of the relationship between the data source and the optimization schemes in distribution network, a load distribution matching method based on the typical scenario partitioning and Pauta criterion is proposed,which can select out several alternatives quickly and accurately from the massive historical data. As the reactive power optimization is a multi-objective problem, the multi-attribute decision making method is transformed to a single attribute decision making problem based on entropy weight method, and then select out the optimal scheme from the alternatives. Besides, in order to solve the problem that the historical database is not complete or some historical schemes are not optimized, a hybrid algorithm based on the combination of EWOSM and the conventional method is proposed to ensure the optimality of the calculated results. Finally, case study based on several standard and numerical systems including IEEE 33 nodes system,PG&E 69 nodes system, a practical 173 nodes system and 292 nodes system is provided, and genetic algorithm and sequential quadratic programming method are compared with the proposed method to verify the validity and effectiveness. |
PDF Full Text Request