面向两个细则考核的风-储场站日前-日内两阶段功率优化上报策略

Two-Stage Day-Ahead and Intra-Day Optimized Power Reporting Strategy of Wind-Storage Stations for the‘Two Detailed Rules’Assessment

针对两个细则考核风功率上报准确率的问题,提出一种可有效降低风-储场站风险考核电量的日前-日内两阶段功率优化上报策略。首先,提出面向两个细则的风-储场站日前-日内功率协同上报框架;然后,以期望考核电量最小化为目标,建立计及储能调节的日前-日内两阶段风-储功率优化上报模型,模型中,日前、日内风功率场景集通过基于预测误差分箱的Cornish-Fisher级数与Cholesky分解法生成;最后,针对模型的多重非线性特征与不同时间尺度的求解效率要求,制定基于小步长线性化迭代的求解算法。算例仿真结果表明,相较其余上报策略与求解算法,该文提出的日前-日内协同上报策略及相应算法有效地提升了上报准确率及求解性能。

The rational optimization of the power reported by wind farms to the grid is a critical routine to improve the benefits of wind farms under the assessment of the‘two detailed rules’.However,most of the traditional wind power reporting studies report the predicted power directly,or report the expected wind power obtained by superimposing the forecast error distributions with the predicted power.In addition,the sales revenue,the compensation for ancillary services are employed as the main optimization goals in these studies.Hence,it is difficult to apply the traditional reporting strategies directly under the background of the‘two detailed rules’assessment.To address these issues,this paper proposes a two-stage day-ahead and intra-day optimized power reporting strategy of wind-storage stations that fits the mechanism of the‘two detailed rules’assessment.Firstly,the framework of the proposed strategy is constructed,where the day-ahead reporting results are embedded in the intra-day reporting model as the key to synergy,in order to achieve the optimal power reporting on both two time scales for the‘two detailed rules’assessment.Secondly,the specific assessment mechanism of the‘two detailed rules’in central China is used as an example to formulate the day-ahead and intra-day optimized wind power reporting model respectively.Particularly,the expected assessment power of day-ahead wind power scenarios is regarded as the objective function of the day-ahead reporting model.Besides,the reported wind power series obtained by the day-ahead model is utilized in the intra-day reporting model.With the energy storage as the medium,the total assessment power of all intra-day wind power scenarios on the two time scales can be minimized in the intra-day model.As for the scenario generation,the day-ahead and intra-day historical errors are binned according to meteorological conditions and different forecast time steps respectively.The Cornish-Fisher series and Cholesky decomposition are applied to generate and reorder the error scenarios considering temporal correlations.By superimposing the reordered error scenarios with the predicted power,the day-ahead and intra-day wind power scenarios can be obtained to refine the wind power reporting model.Owing to the requirements of different time scales for the model solving,a two-stage algorithm is proposed.In the day-ahead stage,due to the small scale of the model,the improved bald eagle search(IBES)algorithm with strong global search ability is adopted.In the intra-day stage,in order to meet the requirements of 15-minutes rolling reporting for solving efficiency,a small step linearized iteration-based solving algorithm is developed to transform the intra-day model with multiple nonlinear constraints into linear models.Simulation results on an actual wind farm in central China show that,the difference between the correlation coefficients of reordered error scenarios and historical error samples are no more than 0.0228 on average.Furthermore,compared with reporting the forecast power and reporting the optimized power based on independently solving the day-ahead and intra-day model,the two-stage collaborative power reporting strategy can reduce the assessment power by more than 50%in both two time scales.In addition,the small step linearized iteration-based solving algorithm can efficiently solve the intra-day model on average 30 s,which is less than other types of intelligent algorithms.Meanwhile,compared with the regulation of the energy storage based on the predicted power,the proposed reporting strategy saves about half of the discharge electricity.The following conclusions can be drawn from the simulations:(1)The reordered wind power scenarios is more in line with the correlation characteristics of historical samples,which lays a foundation for the modeling of reporting.(2)Compared with other reporting strategies,the two-stage collaborative reporting strategy can better fit the trend of actual wind power and reduce the assessment power.(3)The small step linearized iteration-based solving algorithm balances the optimization effect and solution efficiency,which is worthy of practical application.(4)The proposed energy storage optimization scheme is more energy-efficient than the scheme based on the predicted power,and it helps to extend the life of energy storage.