Study On Model And Parallel Processing For Reliability Assessment Of Bulk Power Systems Based On PC Cluster

Along with the development of economy, the further innovation in electric power industry and the form of the electrical energy market, the role of bulk power system reliability assessment(BPSRA) becomes more important in future economic life. Due to the expansion in size of power system, the BPSRA is increasingly great difficult work. The systematical studies on theory and application of AC load flow method of bulk power system adequacy assessment are thoroughly conducted in this dissertation. Many remarkable works on the models and algorithms of AC load flow method of bulk power system adequacy assessment are under intensive study to improve accuracy and efficiency of calculation methodologyThe main results in detail are included in the following:(1) Independent outages and common mode outages of two transmission lines in parallel are included and dependent outages is not included in traditional models. In order to involve conveniently the kinds of outage modes of these lines in BPSRA, utilizing the theory of Markov stochastic process, a reliability assessment generalized model for two transmission lines in parallel is deduced, which is more sophisticated than existing methods. Independent outages, common mode outages and dependent outages can be include, as well as various effects of repair mode on operation reliability of two transmission lines in parallel can be considered in the model. By predigesting the generalized model, the quite simple model is deduced. It is very easy that using the generalized model a series of reliability model of two transmission lines in parallel can be deduced according to different require of practice applications. For this reason, there is important value not only in theory but also in practice engineering application. The physical concept of its predigested model is definite and it is used conveniently and easy. Example analysis shows that the proposed model takes on better adaptability and effectiveness.(2) In an effort to calculate and analyze the effect of two transmission lines in parallel on BPSRA, a new algorithm including independent outages, common mode outages and dependent outages corresponding to its generalized model is presented. Two transmission lines in parallel including common mode outages and dependent outages can be really simulated as a multi-state component. While reliability indices are calculated, the probability and frequency of independent outages, common modeoutages and dependent outage of a system failure state can be calculated at the same time, and computation requirements are reduced. The proposed model and algorithm are applied to the RBTS system, IEEE One Area RTS96 system and a real power system for reliability computation and analysis. Results show that the algorithm is credible and validity.(3) The enumeration method for BPSRA requires a large amount of CPU time. In order to reduce such computation requirements, a functioned link neural network(FLNN) classification model identifying contingencies in power systems is proposed. For the sake of gaining post-accident information of system states, a group of performance index(PI) is designed according to the performance character relative to the changes of base case. Moreover, a neural network classifier is constructed. A simplified load flow model which is able to include both of active and reactive effects is constructed using an AC load flow algorithm. The models can be employed to calculate quickly PIs that are used to form input pattern sets of a FLNN. A contingency screening based reliability assessment method of power system taking advantage of features of the FLNN classification model combined with the simplified load flow model is also presented. Application examples are tested on the IEEE One Area RTS96 system and the results show the correctness and validity of the developed method in a decrease in computation time with a maintained accuracy. At the same time, it can be learned from the outcome that an improvement of computation accuracy in an algorithm is more effective than an inclusion o...