| The reliability of power systems depend on the system load demands, electricalparameters of power system components and power system structure. Due to thedistribution characteristics of system load, it is nearly impossible to evaluate thereliability of a power system using a single system load level, such as the peak load andaverage load. The power system load rate will be changed when the demand sidemanagement is introduced to a power system, hence the reliability of power system willbe changed. Therefore, the studies on the relationship between the system reliability andthe system load rate is critical for power systems, in addition, the study on the fastevaluation method for the power system reliability is also important.The thesis is supported in part by the Natural Science Foundation Project (No.50777067), which is titled “Recognizing the noncoherent mechanism and noncoherentcomponents of power systems”. The main achievements in this thesis are as follows:A novel method based on the Support Vector Regression (SVR) is presented in thepaper to forecast the short-term load, which can overcome the defects of the traditionalmethods, such as poor generalization capability, easy to fall into local minimum valueand slow convergence. The Particle Swarm Optimization with Extended Memory(PSOEM) has the following merits: a clearance of the searching direction, fastconvergence and particle accumulation of individual experience. A model based onSVM and PSOEM is proposed to forecast the short-term load of power system, inaddition, the influence on the accuracy of SVM caused by the insensitive loss function εand penalty factor c is investigate, finally optimization on the parameters is carried out.Then the model is used to forecast the short-term load in a Chinese city, and resultsindicated that the PSOEM-SVR model is more accurate than the traditional BP neutralnetwork method, and the calculated errors for the summer load and winter load are2.24%and2.29%respectively. The model can also be used to adjust the strategy ofsystem load rate and the forecast of system load.The rules between the reliability indices of power system, such as: LOLP (loss ofload probability), LOLF (loss of load frequency), EENS (expected energy not supplied),and the system load were analyzed. The relationship model between the power systemreliability and system load rate is modeled using the cubic splines, and the model can besolved according to the interpolation boundary conditions, the interpolation points and derivative information. Therefore, a fast evaluation technique based on the proposedmodel for the reliability of power systems can be obtained, which can avoid theduplication of reliability evaluation with multi load levels. The proposed model isapplied to the RBTS, IEEE-RTS79, and IEEE-RTS96system. Results indicate that therelationship between the system reliability and load rate is non-linear variation. Theaverage error for the evaluation on the reliability in these cases is about2.0%when theload changes.Load rate adjustment strategies based on reducing peak and filling valley and theproportion of load shifting are presented, and the algorithm for the load adjustment isalso presented. The relationship between the system reliability and the load rate ismodeled using the spline interpolation method. Both the two load rate adjustmentstrategies can adjust the load demand between the peak and valley periods, the systemload curve on the premise of satisfying the electricity demand for the peak load. In thisway the system load rate can be decreased. The proposed model and method are used tothe RBTS and IEEE-RTS79, and results indicate that with the increase of the systemload rates, the system reliability increase in a non-linear way, which also means thereliability performance of power systems can be improved by increase the system loadrate. However, when the loading rate increased to a certain extent, saturation will arisein the system reliability.With the increase of system load rate, the power system reliability will beimproved, which is usually associated with a higher cost of system expense. Therefore,the load rate adjustment cost and benefits of reliability improvement should be takeninto account in the model of optimizing the system load rate. The optimum load ratewith a high efficiency and low economic cost can be obtained. Based on the changingrules of the power system reliability and cost with the adjustment of system load rate, anon-linear model for the optimization of the system load rate is built, in which thedifference for the electricity price during the peak and valley, and the costs of load pointoutages have been considered. The golden section algorithm was used to solve theproposed model. Case studies on the RBTS, IEEE-RTS79indicate that great economicbenefits can be obtained by optimizing the system load rate, and the adjustment strategyfor the system load rate have a great influence on the optimum load rate. |
PDF Full Text Request