Research On Modeling And Optimization Of The Reference Network

Power grid plays an important supporting role in the balance of generation and load,and the capability and margin of the balance are related to the characteristics of the network topology.The energy crisis and environmental pollution promote the trend of the energy substitution of terminal users' and the clean substitution of the generation side,and thus the fluctuation of generation and load side is deteriorating,which makes it difficult to grasp the nodal injection rule.Besides,due to the imbalance spatial and temporal distribution of generation and load,the power grid will be in unbalanced development and face challenges in supporting the balance of generation and load.It is of realistic significance in how to evaluate and quantify the balance of generation and load to guide the operation and planning of power grid.The reference network can provide a benchmark that the existing one can be objectively compared.Optimal transmission line capacity and optimal congestion cost can be quantified and compared with those in the real network to identify the new investment.However,the reference network mathematical model for a real large power system can be quiet complex and difficult to solve.The original reference network is based on a certain network topology,and generators and loads are unchanged,uncertainties in load levels and operation modes are not included.The decision making for new investment needed is usually conservative and less economic.Thus,an improved reference network model and its algorithm is raised in this paper,considering uncertainties in future load levels,predicting the development trend of transmission network,combined with possible operation modes.The main research works are shown as follows:(1)A reference network model considering N-1 security analyze is raised in this paper.However,the number of these constraints is tremendous and this makes the impediment when reference network model is applied for real large power systems.In this paper a new method is proposed to solve reference network efficiently based on line outage distribution factor.Firstly,the power flow without contingency events is calculated.Next,the line flow under contingency events is calculated based on LODF.For each contingency event,the line flow constraint which possesses the most severe overload level is added into the reference network model.The optimization procedure proceeds until all the overload flows are eliminated.Compared with other method,the efficiency and validity of the proposed method is verified by IEEE test systems and a real large power system.(2)The existing reference network models rarely take the performance of real power grid and its operation modes into consideration.Based on the load level forecasted in the schedule time scale and candidate branch in annual network planning strategy,a model is built to minimize the total annual generation cost and transmission cost,providing guidance for transmission flexible planning,the security constraints such as transmission line constraints under normal and N-1 contingency conditions should be respected.After dealing with the interval form constraints,the complex model is converted into a linear programming model,in which the linear programming method is used to find the optimal capacity of each transmission line.Finally,the effectiveness of the proposed model is illustrated based on study case.(3)Considering the uncertainties of future load levels and the accuracy of load forecasting is not guaranteed,an optimal method of reference network model with load interval is built in this paper.This new model is raised to minimize the annual generation cost and annuitized cost of transmission,based on the load level forecasted in the schedule time scale and candidate branch in the annual network planning strategy,considering normal and N-1 contingent conditions.A mixed-integer nonlinear programming model is formed after transforming the interval form constraints.Then a mixed-integer nonlinear method is used to solve the model,which is proved to be effective and accurate in the case study.