Inverse Problem Of Composite Power System Reliability Evaluation And Its Algorithms

The safe and reliable operation of the power system is of great significance to the development of national economy and people's production and living.Power system reliability evaluation starts from reliability parameters of components(generating units,lines,transformers,etc.),electrical parameters and system structure parameters,and calculates the system or node reliability indices through the process of reliability modeling and system state analysis,which can guide the decision-making of power system planning and operation.Among them,component reliability parameters(CRPs)are the basis and key of the reliability evaluation.On the one hand,due to the complexity of statistical work and the influence of human factors,the omissions and errors of component reliability parameters are almost inevitable,which thereby leads to wrong reliability evaluation results and then affects the planning decision of power systems.On the other hand,abundant outage information is accumulated during the operation of the system,so the reliability indices of the system / nodes formed by the analysis is reliable.Based on the above background,the concept of “inverse problem of reliability evaluation(IPRE)” came into being,which refers to the process of obtaining component reliability parameters starting from the known reliability indices of the system / nodes.The inverse problem of reliability evaluation expands and completes the theory of conventional reliability evaluation,which is an important potential research direction in the field of power system reliability.At present,the theory of IPRE is still in its infancy,and its model,method and application are not mature,which urgently needs more systematic and in-depth research.This thesis is supported by the National Natural Science Foundation of China(51677011).Considering three engineering scenarios including the omissions,errors and optimization of component reliability parameters,this thesis conducts the research on the modeling and algorithms for the corresponding three kinds of IPRE based on the composite power system.The main work is as follows.In the process of solving the IPRE,it is extremely time-consuming to carry out the reliability evaluation repeatedly for unknow CRPs.To solve this problem,based on sequential Monte Carlo simulation,an analytical model of the reliability index with respect to CRPs is proposed.Using the full probability formula,the system events with load loss generated by sequential Monte Carlo simulation are divided into a complete event group.Furthermore,based on the conditional probability criterion,CRPs are separated from the reliability index calculation formula of Monte Carlo simulation,and then the analytical calculation model of the reliability index is established.The sensitivity formulas of the reliability index to CRPs are derived based on the proposed model.The proposed model is applied to IEEE-RTS and a 91-bus provincial power system.Case studies show that the analytical model can quickly calculate the corresponding system reliability index for the different values of CRPs,and the error is less than 2% compared with sequential Monte Carlo simulation.In engineering practice,some CRPs are unknown or missing.It is difficult to get the accurate values of the CRPs by conventional optimization algorithms which depend on the initial value,because only the rough range of the CRPs can be obtained.To solve this problem,a unified model of IPRE for calculating unknown CRPs and the interval algorithm are proposed based on the above reliability index analytical model.Firstly,the analytical nonlinear equations of IPRE is constructed,and the conditions for multiple solutions of IPRE are analyzed.Secondly,the set of nonlinear equations is transformed into an optimization problem,and the unified model of IPRE for calculating unknown CRPs is established,which can consider three cases where the number of known reliability indices is greater than,equal to or less than the number of unknown CRPs.Then,interval numbers are used to represent the unknown CRPs so that the optimization model of IPRE is transformed into an interval nonlinear optimization problem.Finally,interval Krawczyk-Hansen operator and hull consistency technique are used to judge the existence of solutions in a certain interval and delete the interval without solutions.Thus,all solutions of IPRE can be obtained.The proposed method is applied to IEEE-RTS and the 91-bus system.Numerical results show that unknown CRPs including failure rates and repair rates can be accurately obtained by the proposed method for the three cases mentioned above.In some engineering scenarios,although the statistical values of CRPs are known,there may exist some incorrect values.To solve this problem,this chapter proposes the model of IPRE for wrong CRPs calibration,and corresponding identification and calibration algorithms based on the initial value estimation strategy.Firstly,assessment indices for the identification and calibration effects are proposed in terms of the identification accuracy and the calibration error.Secondly,an improved particle swarm optimization algorithm with global optimization ability is used to estimate the sensitive CRPs.Then,a rolling estimation algorithm based on the dynamic correction for the initial values of CRPs is adopted to estimate all CRPs,and wrong CRPs are identified according to the deviations between the estimated values and the statistical values.Finally,wrong CRPs are corrected by the interval algorithm.The proposed method is applied to IEEE-RTS and the 91-bus system.Numerical cases show that almost all the wrong CRPs can be identified by the proposed method and the errors of calibration results are less than 1%,when the number of components with wrong reliability parameters accounts for less than 9% of the system.As a kind of IPRE,the reliability parameters optimization(RPO)problem needs to calculate the system reliability index repeatedly for variable CRPs in the solution process,which leads to the inferior solution and low computational efficiency.To solve this problem,this chapter proposes a model of IPRE for reliability parameters optimization.Firstly,the analytical model of the reliability index is established based on the hybrid stratified sampling method.Secondly,the analytical model of the reliability index is embedded into RPO problem by matrix operation and introducing auxiliary variables.As a result,the model of IPRE for reliability parameters optimization is constructed,which minimizes the system EENS index.Then,the model is decomposed into a bi-level problem to facilitate the solution finding.According to the analytical sensitivity of the objective function to the component investment cost,the upper level uses greedy strategy to determine the set of components to be optimized.The lower level employs the interval optimization algorithm to determine the investment cost and unavailability of the components to be optimized.The proposed method is applied to IEEE-RTS and a modified IEEE-RTS96 system.The numerical cases show that compared with the RPO method based on the maximum risk index,the EENS index obtained by the proposed method is reduced by 65%;compared with the RPO method based on the reliability expectation index,the proposed method greatly improves the computation efficiency.