Reliability Forecasting Models And Algorithms For Electrical Distribution Systems

As the connection of transmission systems and power consumers, Electrical Distribution System(EDS) reliability is becoming more and more important in power system planning, operation, production management and customer service. Currently, considerable efforts have been devoted to study the EDS reliability statistics, reliability evaluation and reliability optimization, but few researchers focus on the field of EDS reliability forecasting and its large-scale applications in pratical EDSs. Therefore, reliability forecasting models and algorithms for EDSs are studied in this thesis, which is helpful to effectively grasp the development trend of EDS reliability, determine the EDS investment and model the electricity price associated with the EDS reliability performance.Power transformer is one of the most important components in EDS, which has the function of transforming voltage, distributing power and transmitting power. Based on the reliability model of transformer subsystems and logical relationships among them, this paper proposed a full-state reliability forecasting model using the condition-based maintenance and Markov process. The transformer risk assessment matrix and the model for forecasting the transformer reliability were built by combining with the failure frequency model for transformer subsystems and the defect severity of transformers using entropy weight fuzzy method. The results of case studies on real transformes show that the model is suitable for the overall risk assessment of the subsystems and whole transforms in a period of time in a area. The proposed method provides a new thinking for distribution transformer risk prediction and maintenance strategies, is an important part of distribution system reliability forecasting, and also lays the foundation for EDS reliability forecasting.Considering the features that EDS average service availability index(ASAI) has a monotone increase step by step and is no more than 1, this paper built a grey forecasting model and the least-absolute-deviation forecasting model based on the data transformation. Firstly, an appropriate transformation, such as the reliability index from the interval [0, 1] to [1, + ?], is selected to transform the domain of EDS reliability indices into a given interval. Secondly, the grey forecasting method and the least-absolute forecast model are used to forecast the reliability indices after data transformation. Finally, the foreacsted EDS reliability can be obtained by inversing the transformation. Two EDSs were used as numerical examples to forecast ASAI, the results show that compared with forecasting model without data transformation, the forecasting model with data transformation can be overcome the issue of predicted value is larger than 1.Combinational forecasting model can be used to improve the prediction accuracy due to the complete consideration of the disadvantages, features and data information of all single forecasting models. Therefore, based on grey forecasting model, leastabsolute-deviation forecasting model and Logistic forecasting model, combinational forecasting model for EDS reliability was built considering the sum of error squares, mean absolute error as the prediction accuracy decision criteria, which was solved the weight coefficients of single models in combinational forecasting model by using a particle swarm algorithm. One EDS was analyzed as a numerical example, and the results show that the proposed combinational forecasting model has higher prediction accuracy compared with single models, which verified the effectiveness of the proposed model.EDS reliability is a concentrated reflection of the whole power system structure, equipment and the operations. In order to accurately forcast the reliability of an EDS, it is necessary to analyze the influencing factors of EDS reliability. This paper proposed two methods for seperately recognizing the key influencing factors of EDS reliability by the grey relational analysis and the redundent influencing factors by the correlation analysis, which provide the basis for determining the priorities of EDS reliability improvement measures. Through classifying more than 20 EDS reliability influencing factors, the key EDS reliability influencing factors and redundanc influencing factors were identified, and then the key influencing factors of EDS reliability were determined. Case studies show that among the numerous influencing factors of EDS reliability, ratio of available tie lines, ratio of insulated feeders, average number of customers in each section, ratio of tie lines, average sections of each feeder, and average load factor of feeders, are the most significant influencing factors of EDS reliability.Traditional methods use the statistical reliability parameters of components for many years and the determined EDS structure for evaluating the EDS reliability. In the beginning stage of investment, it is very difficult to evaluate the EDS reliability by using traditional methods due to EDS topology not being fully determined. This paper presented a comprehensive model for forecasting EDS reliability, which is built separately into two parts, i.e. models for EDS failures and planned outages. Firstly, a three-layer artificial neural network(ANN) model is proposed to forecast the EDS reliability considering EDS component failures. Each neuron in the ANN input layer represents a key influencing factor of EDS component failures, and the output layer is TCOH-F(Total Customer Outage Hours due to Component Failures) index. The proposed ANN is trained by using the historical reliability data of an EDS, and the EDS reliability indices can be forecasted using the trained ANN. In addition, a planned outage reliability model is also built according to the investment and type of planned outage. Therefore, the total forecasting EDS reliability indices can be obtained. Case studies of practical EDSs illustrate the efficiency and applicability of the proposed techniques.