The Artificial Neural Network Model And Its Realization For Power System Reliability Evaluation

This thesis is the summary of achievements in which the study of theory studies form the basis of major technological projects such as "An application of neural network to electrical network reliability management." This thesis researches methods and realization technologies of searching for minimal cutset of a graph by neural network theory. These methods and technologies would greatly alleviate the "Calculation catastrophe" facing on the reliability evaluation of power systems. Then they would afford the powerful implement and scientific fundament of the realization of the real-time and on-time analysis and of evaluation for electric network reliability.These main studies include the following:1). For the first time, presenting concepts of the feasible flow, weakness degree and weakness area and proving the validity of searching for K-weak cutset and weakness degree of a network with DASWA algorithm.2). On the basis of the researching the neural network theory, presenting the neural network model (N-SearCut model) for searching for minimal cutset of a graph, proving the stability of this model, analyzing the dynamic behavior of this model, deducing a series of corresponding theorems and affording an academic fundaments for solving optimization problem based on this model.3). Presenting the HBM method and proving the applicability and validity of judging the connectivity of a graph with this method by analyzing the searching rules of minimal cutset of electric network. Then, This HBM method develops the application prospect of the N-SearCut model and improves the practical project value of this model when searching minimal cutset of a graph using this method with N-SearCut model.4). Setting forth the NNCS and NNSP algorithms for realizing the N-SearCut model by software and giving some concrete steps of these algorithms. It is proved that these methods alleviate "calculation catastrophe" facing on reliability analysis of power system at some extend with the application of NNCS and NNSP algorithms to the IEEE-RTS.5). Setting forth the realization method by hardware for searching minimal cutset of a graph with N-SearCut model. Then this method affords theory and practice fundament for solving this puzzle of searching for minimal cutset in reliability evaluation of power system.6). Completing the realization scheme of searching for minimal cutset of a network with N-SearCut model by neural calculation system. Then, the N-SearCut model could be applied to practical project authentically, the puzzle of searching for minimal cutset in reliability evaluation of power system could be perfected and the reliability evaluation technology of power system could be developed adequately.7). According to the work principle of system and the requirement of reliability evaluation, developing the reliability management system of electric network (or ESMIS system) on the basis of such three elements as opening, reliability and object oriented programming. It is the theoretical bases and powerful implement of calculation and management for the reliability of a practical electric network.