| As an indispensable part of emergency control, fast and reasonable service restorationreconfiguration is an important basis of power grid dispatching and reliability evaluation.Objective and quantitative reliability evaluation of power system can explore the weaknessand point out a way for power system planning, operation and maintenance, in order toachieve the goal of keeping security, stability and reliability of power system operation,realizing the integration of reliability and economy. Among, distribution network at theterminal of power system is a significant element impacting power supply reliability andpower quality, access point of distributed energy system, implement site of flexible interactiveintelligent electricity utilization. According to incomplete statistics, more than80percent inthe service interruption is caused by distribution system faults at present. Distributed energysystem with the features of generating flexible, environmental friendly, power supplyreliability and power quality improvement, is increasingly connected to the distributionsystem. In the meanwhile, there is a far-reaching influence in structure and operation of thedistribution system, which will bring the traditional distribution system from a simple passivenetwork into a complex active network. Its complexity represents in below:1) power flowchanges from unidirectionality to bidirectionality, and uncertainty of system operation state isgreatly increased;2) power output of distributed generation whose primary energy is variousis of volatility and intermittency;3) operation scheme switch problem, including connectedscheme and islanding scheme, and so on. For this reason, the dissertation discussed servicerestoration and reliability evaluation of distribution network with distributed generations. Themain work is as follows:1. Service restoration is converted to constraint satisfaction problem. First, severalBoolean functions are defined, including power source constraint, radial constraint, switchoperation constraint and power balance constraint, etc. They are composed as ordered binarydecision diagram in order to reduce solution space. Then, each solution in the space ischecked bit by bit and sorted. In the accordance to sort result, the optimal scheme of servicerestoration is determined. The proposed model and its approach can reduce the complexity ofthe solving process significantly, and improve the credibility of the decision results. If thecalculations of constraint Boolean function proceed in quasi real-time environment, the mainwork is composing constraint and check analysis in the decision-making stage. That will helpreduce real-time computation and ensure the efficiency of real-time decision-making application. Finally, the numerical results of IEEE33system indicated that the proposedmodel and its solution are effective and feasible.2. Fast and reasonable service restoration of interrupted area affected by faults is onebasic requirement of the self-healing control of distribution system with distributed generation(DG). A service restoration model and its approach of distribution system with distributedgeneration based the hybrid decision diagram are established. The solution is two-stagemethod which converts service restoration problem to constraint satisfaction one. It definesBoolean functions, as power source constraint, radial constraint, switch operation constraintand DG islanding constraint, considering interruptible load. The scale of the solution space isreduced substantially by the hybrid decision diagram composed of several constraints. eachsolution in the space is checked bit by bit and sorted, in order to obtain the optimal scheme ofservice restoration stably. Modified IEEE33system is used to verify the proposed model andits approach.3. The output power of distributed generation (DG) system presents strong fluctuationand obvious correlation. Meanwhile the correlation among load, as well as between load andDG output, is obvious. In order to study the impact of energy fluctuation and correlation onsystem reliability, this paper analyzes transferring probability and its influence factors, basedon traditional reliability analytic approach of distribution network. Incorporating with theprobability distributions of DG output and load, as well as their correlation matrix, Nataftransformation technology is employed to generate the sample space of related randomvariable, and a reliability evaluation approach of distribution network considering correlationof DG output and load is established. The proposed algorithm overcomes the difficulties ofsampling DG outputs which obey correlated non-normal distributions. Finally, the casestudies of the improved RBTS BUS6system are given to demonstrate the feasibility of theproposed model and its solution.4. Four reliability evaluation approaches of distribution system with distributedgeneration are compared, which are Monte-Carlo simulation, hybrid method, Cornish-Fisherexpansion and Edgeworth expansion, considering the effects of random variable correlationon reliability indices. The transfer supply probability distribution and reliability indices byMonte-Carlo simulation is taken as benchmark, comparing the efficiency of the other threeapproaches. The numeral results indicate that Monte-Carlo simulation can take thesophisticated factors and achieve simply and inefficiently; the hybrid method can give therespective potentials of analysis methods and simulation methods to full play, whosetime-consuming is shorter than the former. cumulant method requires to simplify the model, whose efficiency is the highest, to gain the indices whose error in the acceptable range.Cornish-Fisher expansion is more suited to handle non-normal distribution of randomvariables than Edgeworth expansion, so it can be applied decision-making of distributionsystem with reliability.5. As control objects of distribution automation, automation switches play an importantrole in security, reliability and economy of distribution system operation. Their upgrade is oneimportant contents from traditional distribution network to smart distribution one. Theoptimization programming model of distribution network automation switch upgrade is builtto absorb DG output, which maximizes social economic benefits and takes load-pointavailability as constraints. This model can consider volatility and correlation of DG output,automation scheme influence in reliability indices. And genetic algorithm is employed tosolve the model. The modified RBTS BUS6system is took as a case to verify the proposedmodel.6. In accordance with the continuous variation regularities of distributed generator outputand load power, a mixed integer programming model for service restoration dynamicreconfiguration of distribution network with distributed generators (DG) is proposed. Themodel splits the restoration process into several stages, and is converted into a dynamicprogramming problem with multi stages. It considers black start ability and power adjustfeature of DG, charge/discharge characteristics of energy storage system. A heuristic methodis applied to solve the model and achieve the restoration scheme which satisfied securityconstraints and switching times constraints. The numerical results of modified IEEE33systemdemonstrate the availability and feasibility of the model and its solution proposed.This dissertation is supported by National Key Technology Research and DevelopmentProgram of the Ministry of Science and Technology of China (No.2013BAA01B00),high-tech industrialization key project in Guangdong (No.2010A010200005), and ResearchFund for the Doctoral Program of Higher Education of China (No.20120172120042). |
PDF Full Text Request