Study On Ultra-short Term Prediction And Grid-connected Optimal Scheduling Model Of Wind Power

Due to the high uncertainty,anti-peak regulation and low schedulability of wind power output,large-scale wind power integration poses severe challenges for the safe and stable operation of power systern Accurate wind power predictions are conductive to making reasonable unit commitment(UC)decision,adjusting generation plan,reducing the impact of wind power integration on power grid and improving the safety and reliability of power system operation.As the capacity of grid-connected wind energy increases,the traditional extensive scheduling mode in which power generation tracks load demand cannot meet the needs of safe and reliable operation for power system in conditions of high uncertainty.Furthermore,anti-peak regulation of wind power output,the on/off status and the adjustment ability of conventional units have restrictive effect on large-scale wind power absorption for power system.Under this background,this paper starts from deterministic and interval short-term wind power forecasting methods for short-term wind power.On this basis,with the purpose of operation reliability,economy and low-carbonization simultaneously,the multi-time scale coordination scheduling model involving multiple types of power supply and flexible load and its solving algorithm is studied.The main work and research results are as follows:(1)In order to reduce the nonstationarity of the original wind power series and anaiyze its multi-scale behavior characteristics,the ensemble empirical mode decomposition and variational mode decomposition are introduced to decompose the original series into several mode functions at different time scales.Then a novel ultra-short term wind power hybrid forecasting model based on signal processing technology and machine learning method is proposed.Compared with the traditional forecasting methods and the base models based on chaotic phase space reconstruction,the proposed model has higher prediction accuracy.(2)Unable to select the optimal one among the ultra-short term wind power hybrid forecasting models,a multi-leveled combined forecasting model with adaptive weight update by integrating with three different machine learning methods is proposed.Moreover,a multi-strategy self-adaptive differential evolution algorithm is developed to update the weight matrix dynamically.The case study demonstrates that the proposed combined forecasting model can further improve the accuracy and stability of ultra-short term wind power prediction compared with the hybrid forecasting models.In addition,the interval forecasting model based on quantile regression average is established to estimate the fluctuation of wind power under a given confidence level,which can provide information support for optimal scheduling problem of power system with wind farm(3)The dynamic economic emission dispatch(DEED)model with day-ahead UC for power system with wind power integration is studied.A probabilistic scenario-based framework is firstly constructed to obtain the UC solution.Then an enhanced multi-objective particle swarm optimization(EMOPSO)algorithm and an iterative combination method are proposed to solve the stochastic DEED problem with the reduced scenario set.For EMOPSO,the dual population evolution mechanism and the hierarchical elitism updating mechanism are introduced to improve its optimization performance.Moreover,the elitism preserving strategy based on crowding entropy is implemented to optimize the distribution of Pareto optimal solutions.The case study of the modified IEEE 10-unit 24-bus system demonstrates that the proposed approach can satisfy the demand of power system operation considering system operation economy,emission and reliability in harmonious cooperation,which can provide an effective and multi-perspective measured solution for the day-ahead unit commitment dispatch with wind power integration.(4)Based on the study about the day-ahead UC optimal dispatch with wind power integration,combined with the deterministic and interval prediction results of ultra-short wind power,the multi-time scale coordination scheduling model involving high energy load considering wind power prediction accuracy difference with different forecasting period is established,which is solved by intelligent optimization algorithm step by step.The case study shows that the cooperation dispatch involving high energy load car prominently reduce the total operation cost and improve the wind power absorption,which verifies the rationality and effectiveness of source-load coordination scheduling model(5)For the coordination and interaction between different power generation resources and controllable load resources with wind power and flexible load integration,hydro-thermal-wind combined dispatch model involving electric vehicle(EV)fleet is established.Moreover,the constraint handling strategies for different types of power generation resources,EV fleet and system operation limits.The case study demonstrates that smart scheduling of flexible resource can promote wind power absorption and verifies the feasibility and rationality of the proposed model,which can provide strong support for the coordinated optimal scheduling containing different types of power generation resources and controllable load resources.