Probabilistic Analysis And Optimal Operation Method For Wind Energy Integrated Power Systems

The integration capacity of wind power has been growing rapidly in power system,which brings inevitable challenges to both academic study and engineering practice inpower system analysis and optimal operation. In addition to randomness, thecharachteristics of wind sources in boundedness, correlation, difficulties in forecastingand fluctuation arevery diffirent from traditional sources and cannot be handled byexisting analysis and optimal operation methods. This thesis developed new methods ofprobabilistic power flow, probabilistic optimal power flow and probabilistic generationdispatching to address these problems caused by wind sources.A point estimation method (PEM) based probabilistic power flow (PPF) techniqueconsidering bounded wind speeds is proposed. The domain of wind speed is alwaysbounded. Unfortunately, the locations computed from PEM may fall outside the domain.In this case, the PPF cannot be calculated accurately or even cannot be calculated at all.A universal transformation and a power transformation are combined with thetraditional PEM to address this issue. The IEEE14-bus system and several wind speedcurve data are used to demonstrate the effectiveness of the presented technqiue. Thepresented transformation method can be extended to deal with the boundedness of otherrandom variables such as loads.A probabilistic optimal power flow (POPF) technique considering the correlationsbetween wind speeds following arbitrary probability distributions using PEM isproposed. Correlated wind speeds following different distributions are transformed intorandom variables following correlated normal distributions and then into thosefollowing independent normal distributions so that the traditional2m+1PEM can beapplied to solve the probabilistic optimal power flow with wind speed correlations. TheIEEE14-bus system, IEEE118-bus system and an actual utility system in the southwestof China with additional wind farms are used to demonstrate the effectiveness of thepresented method.A chance constrained programming generation dispatching model consideringforecasting errors of wind power is developed and a genetic algorithm combined withthe PEM is proposed to solve the presented model. The expeted objective function andprobabilistic constrains are obtained using the point estimation method in evaluating thefitness function. To reduce computaional burdens in power flow analyses required in implementation, a new power flow algorithm for composite meshed and radial systemsis also presented using the topological division of system structure as a part of theimproved genetic algorithm. The effectiveness of the presented model and algorithm isdemonstrated using several systems.A probabilistic model of daily generation scheduling for reducing losses of thermalunits’ lives under the environment of fluctuated wind sources is developmed. Animproved multi-population genetic algorithm (IMPGA) is presented to solve the modelbased on load curve segmentation. The cost for regulation of thermal units is estimatedfirst. A chance constrained programming model for reducing units’ regulationfrequencies is then proposed by introducing the regulation cost in the objective functionand regulation frequency limits in the constraints. Load points on a load curve areaggregated to form an equivalent multiple-level load curve representation. The globaloptimization is reached with coordination between the multiple-level load model andmulti-population strategy of GA. The effectiveness of the presented model andalgorithm is demonstrated using the IEEE30-bus and IEEE118-bus systems.In this thesis, the efficient probabilistic analysis and optimal operation methods areproposed to address the boundedness, correlation, difficulties in forecasting andfluctuation of wind sources. These methods provide a useful package in helping ensuresecurity, high quality and economic operation of power systems.