Research On Optimal Power Flow Algorithm And Its Application For AC/DC Hybrid Power System Considering Wind Power Uncertainty

With the integration of large-scale sustainable energy,operation of high voltage direct current(HVDC)transmission lines,effective interconnection between regional power grids,China's power grid has gradually formed as a large-scale and ultrahigh voltage AC/DC hybrid system,with high voltage level,high transmission capacity and high load density.Compared to traditional power grid,the modern power system's operating conditions are more complex and changeable.Facing such a large-scale and complex modern production system,how to improve its operation efficiency will become the focus of attention.Under the given operating conditions,optimal power flow(OPF)calculation means that power system can achieve the best stable operation state by adjusting the control variables available,which considers power balance as well as security constraints of the operating process.The traditional OPF model and its algorithm will be difficult to apply the modern power grid due to taking all kinds of actual requirements into account.Therefore,it is of great research value and practical siginificance to study OPF model with modern power grid characteristics and its algorithm.Based on the steady-state model of AC/DC hybrid system including wind power,the model,target,algorithm,and application of AC/DC OPF are studied in this dissertation.The main jobs of this dissertation are as follows.(1)There exists some issues about non-convergence and divergence when conventional power flow algorithm to solve AC/DC power flow equations.In this dissertation,two kinds of Levenberg-Marquardt(LM)method with high order convergence are applied to solve AC/DC power flow,including LM method with trust region technique and LM method with line search technique.Under the well-conditioning,the proposed LM method is compared with Newton method to verify its correctness and effectiveness.Under the ill-conditioning,such as heavy load,heavy DC transmission power,multiple DC transmission lines,the alternate iteration model combined with the proposed LM method with line search technique shows its good convergence performance.When conventional power flow method fails to find solution under ill-conditoning,the least square solution obtained by the proposed LM method can be used as comprehensive reference materials for power department to adjust power flow and improve its convergence.(2)In order to avoid great changes to the existed OPF program when the control mode of HVDC line changes,a decoup ling optimization algorithm is proposed to solve AC/DC OPF including wind power.On the basis of the existed OPF program,AC/DC OPF program can be derived just by modifying DC system equations program accoding to different DC control modes.Thus,the AC/DC OPF model is divided into two parts: AC OPF model and DC system model,and alternate iterative method is employed to solve AC/DC OPF.The AC OPF is solved by path following interior point method,while the DC system equation is solved by the proposed LM method.Simulation results show that the decoupling optimization method has fast computation speed and strong convergence,which can satisfy practical engineering requirements.The impacts of wind speed,location of wind farms,and DC control mode on AC/DC OPF results are analyzed.(3)The existed mathematical optimization method will be difficult to solve AC/DC OPF problem as considering actual operation condition,such as valve-point loading effects of generators,prohibited operating zones,carbon tax,since these conditions can make AC/DC OPF model become as discontinuous,nonconvex and non-differentiable.Based on the proposed LM method with high order convergence,an adaptive simplified human learning optimization method(ASHLO)with strong global convergence is applied to solve AC/DC OPF.Numerical results suggest that ASHLO method has good convergence and robustness.The impacts of wind speed and location of HVDC transmission line on OPF results are systematically analyzed.(4)The existed AC/DC OPF model just takes economy as objective function and fails to consider other objective functions involved in the model.A multi-objective AC/DC OPF model is established,where includes three objective functions,operating cost,power loss,and pollutant emission.Besides,the valve-point cost and carbon tax are also involved in the operating cost.Based on the proposed LM method with high order convergence,combining the non-dominated sorting technique and the crowding distance index,a multi-objective human learning optimization method(MOHLO)is derived and applied to solve AC/DC multi-objective OPF problem.To demonstrate the applicability and effectivenesss of the MOHLO method,a comparison between the MOHLO method and non-dominated sorting genetic algorithm II(NSGAII)is made.The results show that the MOHLO method has excellent search efficiency and powerful ability of searching optimal.Meanwhile,MOHLO method can obtain more complete and uniform pareto front.(5)In order to take uncertainties and complex dependences of random variables into account during the operation process of AC/DC hybrid power system including wind power,such as wind speed and load,an AC/DC probabilistic OPF model(AC/DC POPF)is established.Monte carlo simulation method is applied to solve AC/DC POPF.The high dimensional dependences of wind speed among wind farms are considered by pair-copula method,and dependences of load are considered by cholesky factorization method.The probability density curve and cumulative distribution curve of input and output variables are obtained by kernel density estimation method.Numerical results suggest that the dependences of wind speed among wind farms have great influences on the standard deviation of output variables.The average relative error of AC/DC POPF results obtained by pair-copula method is smaller than that by multivariate copula function.Under the case of high requirements for calculation accuracy,it is a feasible scheme for using pair-copula method to construct dependences of wind speed to calculate AC/DC POPF.(6)The issue of AC/DC POPF considering demand response is also studied.Firstly,the AC/DC POPF model aiming at minimizing total operating cost as objective function is established,where the dependences of wind speed and load are considered by pair-copula method and cholesky factorization method separately.Quasi monte carlo simulation method is operated to solve AC/DC POPF.For a specified sample,the locational margional price(LMP)can be obtained by decoupling optimization method since it is equivalent to the lagrange multipliers of the corresponding power flow equations.Therfore,the demand response model based on LMP can be constructed,and the statistical characteristics of the load,operating cost,and power flow distribution after demand response are calculated.Simulation results show that demand response based on LMP has positive effects on reducing the total operating cost,while it has no obvious effects on DC variables.The implementation of demand response can reduce the LMP.At the same time,it can also reduce the possibility of transmission congestion,and plays an important role in promoting the security,stability and economic operation of power grid.