基于改进评价指标的波动性负荷短期区间预测

针对传统点对点预测难以适用于波动性较大、不确定性较强的负荷的问题,提出了一种基于改进评价指标的区间预测方法,从区间宽度和累计误差2个角度对现有区间预测评价指标做出改进,提高了预测结果的合理性。在此基础上,从各个评价指标的自身特性及其对预测结果的影响程度进行考量,建立了区间预测综合评价指标,并利用神经网络构建了区间预测模型,以综合评价指标最优为目标,采用粒子群优化算法对网络结构参数进行训练优化,从而取得理想的波动性负荷区间预测效果。仿真中通过对某波动性较强的历史负荷数据进行预测分析,并与传统的点预测和区间预测方法进行对比,验证了所提方法的有效性和优越性。

Improved Evaluation Index Based Short-term Interval Prediction of Fluctuation Load

To solve the problem that traditional point-to-point prediction method is not applicable to the load with large fluctuation and uncertainty, this paper implements a prediction interval method based on improved evaluation index to improve existing forecasting evaluation index from two aspects of interval width and cumulative error, which enhances the reasonableness of prediction results. On this basis, weighing the characteristics and importance of each evaluation index for the influence on prediction results, the comprehensive evaluation index for interval prediction is established, and the interval prediction model is constructed by using neural network. Aiming at the optimization of the comprehensive evaluation index, the particle swarm optimization algorithm is used to train and optimize the structure parameters, so as to achieve ideal effect of interval prediction for fluctuating load. The historical load data with strong uncertainty is used to validate the proposed method. Comparing with the traditional point-to-point and interval forecasting methods, the results and analysis of the improved interval prediction verifies the effectiveness and superiority of the method.