| Modern power grids are always characterized by a large scale and a sophisticated structure. Besides, most of them are directly exposed to natural environment so that they are vulnerable to natural disasters. Various natural disasters respectively influence the safety of power grid in different transfer ways. Since the safety of power grid is vital to society stability and economy development, natural disasters’ impacts on the safe operation of power grid cannot be ignored. To mitigate these impacts, it is necessary to extend disaster prevention framework to natural disaster early-warning in power grid. Then the focus of research on natural disaster prevention in power grid will be shifted from emergency management during or after a disaster to early-warning of risk management before a disaster. Therefore, research on coupling effect and risk transfer path, establishment of a multi-stage, multi-agent and multi-hierarchy model for early-warning management, design and application of early-warning decision support system are all profound to make power grid operation safe. The outputs of research in this dissertation are shown below.Firstly, from the perspective of catastrophology and to take the coupling system between power grid and natural disasters as the research object, natural disasters in power grid have been analyzed by the means of the concept, elements and mutual transformation of elements in this dissertation. According to various mechanisms how disaster factors impact on disaster bearing body, disaster factors have been classified. Then direct & gradually-occurring disasters and indirect & gradually-occurring ones have been proposed as the research objects of early-warning management for natural disaster in power grid. Based on evolution and risk transfer mechanisms of natural disasters, early-warning management has been divided into three stages (latent stage, disaster developing stage and critical stage). From four dimensions on research objects as early-warning management, evolution stage of natural disaster in power grid, early-warning management method and information technology, a 4D model of early-warning management for natural disasters in power grid has been built from two aspects of theory and application. Then the dimension of evolution has been taken as the basic line to realize systematically research on early-warning management of natural disasters in power grid.Secondly, with regard to early-warning management of natural disasters in latent stage, it has been proposed in this dissertation to build a hidden risk assessment model by means of starting with recognizing the signs of natural disasters in power grid, taking the disaster system as the research object, analyzing the coupling relationships among elements of natural disasters in power grid, and basing on two aspects of disaster factors’ dangerousness and disaster bearing body’s vulnerability. Taking pollution flashover as the typical one of direct & gradually-occurring disasters in this dissertation, a hidden risk recognition & assessment model has been built based on subjectively and objectively modified gray attributes. Taking forest fire as the typical one of indirect & gradually-occurring disasters in this dissertation, a hidden risk assessment model has been built based on trapezoidal fuzzy numbers and matter-element extenics.Thirdly, with regard to early-warning management in the disaster developing stage, prediction and early-warning models for both direct & gradually-occurring disasters and indirect & gradually-occurring ones have been built by means of starting with analyzing the developing stage of natural disasters in power grid, demonstrating relationships of and characteristics of monitoring& early-warning and prediction& early-warning in early-warning process, analyzing the difference of early-warning patterns between direct & gradually-occurring disasters and indirect & gradually-occurring ones for different early-warning objects. Taking icing as the typical one of direct& gradually-occurring disasters, a prediction &early-warning model for icing thickness has been built based on BP neural network optimized by improved fruit fly algorithm. Taking forest fire as the typical one of indirect &gradually-occurring disasters, an early-warning process in the disaster developing stage has been established, which includes data collection, model selection and early-warning analysis, meanwhile both self-adaptive selection strategy based on multi-models and early-warning system for forest fire along transmission lines based on multi-dimensions have been discussed deeply.Fourthly, with regard to early-warning management and decision in the critical stage, an optimization scheme model to deal with a critical situation of natural disasters in power grid has been established to realize optimal scheme selection combining static and dynamic characteristics by means of starting with posing the problem to optimize emergency scheme, analyzing dual (static and dynamic) characteristics in the critical stage, suggesting to roughly select scheme based on static characteristic attributes and to optimize emergency scheme based on dynamic characteristic attributes.Finally, with regard to the application of early-warning management and decision for natural disasters in power grid, train of thoughts to build early-warning decision support system for natural disasters in power grid has been put forth in this dissertation. Architecture system based on four libraries (data library, model library, method library and knowledge library) has been built to realize functions as hidden risk assessment, monitoring & early-warning, prediction &early-warning, decision-making and system management by applying key technologies of SOA, J2EE, and so on. Meanwhile, a practical application of early-warning management and decision support system in a provincial power grid has been given to demonstrate that this system can help electrical enterprises to prevent disasters more effectively, to save more times, and to improve the safety of power grid operation. |
PDF Full Text Request