当前用户侧接入大量新能源,通常采用实际电力负荷减去新能源发电功率后的负荷(称为“净负荷”)进行预测研究。由于新能源发电随机性强,净负荷具有不确定性强、规律性差的特点,难以准确预测。为此,提出了一种基于特征加权改进的Stacking集成学习净负荷预测方法。首先,通过对不同单一预测模型的预测性能和差异性的分析,优选出长短期记忆网络、Elman神经网络、随机森林树和最小二乘支持向量机作为Stacking集成的学习器。其次,针对传统Stacking集成预测由于忽略学习器之间差异性所导致的预测能力不足问题,根据预测精度对不同学习器进行特征加权,以修正各学习器所带来的整体预测误差。最后,基于德国TENNET区域实测数据进行算例分析,结果表明,相比于单一预测模型、传统Stacking集成预测方法,基于特征加权Stacking集成学习的净负荷预测方法在晴天、多云、多雨、多雪等天气情况下均具有更高的预测精度。

At present, a large amount of new energy is connected to the user side, and the actual power load minus the power generated by the new energy (hereinafter referred to as “net load”) is usually used for prediction research. Due to the strong randomness of new energy power generation, net load has strong uncertainty and poor regularity, which makes it difficult to predict accurately. To this end, this paper proposes a net load prediction method based on the feature-weighted Stacking ensemble algorithm. Firstly, through the analysis of the prediction performance and difference of different prediction models, this paper chooses Long Short-Term Memory Network, Elman Neural Network, Random Forest Tree and Least Squares Support Vector Machine as stacking ensemble learners. Secondly, because the traditional Stacking ensemble prediction model ignores the differences between learners, the model’s prediction ability is insufficient. Therefore, this paper weights the features of the learners according to the prediction accuracy to correct the prediction error introduced by different learners. Finally, the measured data in the German TENNET area is analyzed as an example. The simulation results show that, compared with the single forecasting model and the traditional Stacking integrated forecasting method, the payload prediction method based on feature-weighted stacking ensemble learning has higher forecasting accuracy in sunny, cloudy, rainy, snowy and other weather conditions.

电化学储能的循环寿命受到充放电次数和放电深度的影响,为了更加准确地在新能源电力系统中规划储能电站,提出基于Kullback-Leibler（KL）散度的储能电站分布鲁棒规划方法。根据电化学储能循环寿命的幂函数,建立基于等效全循环次数的储能电站寿命模型,考虑储能电站寿命模型约束和系统运行约束,以储能电站的全寿命周期成本和机组运行成本最小为目标来构建储能电站的规划模型。进一步,将基于KL散度的风电出力不确定性集嵌入到储能电站规划模型中,通过样本平均近似法将储能电站分布鲁棒规划模型转化混合整数线性规划模型求解。在含2个风电场的改进IEEE-30节点系统中验证了所提出的储能电站分布鲁棒规划方法的优势。

The cycle life of electrochemical energy storage is affected by the number of charges and discharg and the depth of discharge. In order to more accurately plan the energy storage plants in the renewable energy power system, this paper proposes a distributionally robust planning method based on Kullback-Leibler （KL） divergence. According to the power function of cycle life of electrochemical energy storage, the life model of energy storage plants with equivalent full cycles times is established. Considering the life model constraints of the energy storage plant and system operating constraints, the planning model of energy storage plants is constructed with the life cycle cost and units’ operating cost minimum as the objective. Furthermore, the uncertainty set of wind power output based on KL divergence is embedded into the planning model of energy storage plants, and the distributionally robust planning model of energy storage plants is transformed into mixed integer linear programming model by sample average approximation method. In the modified IEEE-30 mode system with two wind farms, the advantages of the distributionally robust planning method proposed in this paper are verified.

What is the difference between a prediction that is made with a causal model and that with a non-causal model? Suppose that we intervene on the predictor variables or change the whole environment. The predictions from a causal model will in general work as well under interventions as for observational data. In contrast, predictions from a non-causal model can potentially be very wrong if we actively intervene on variables. Here, we propose to exploit this invariance of a prediction under a causal model for causal inference: given different experimental settings (e.g. various interventions) we collect all models that do show invariance in their predictive accuracy across settings and interventions. The causal model will be a member of this set of models with high probability. This approach yields valid confidence intervals for the causal relationships in quite general scenarios. We examine the example of structural equation models in more detail and provide sufficient assumptions under which the set of causal predictors becomes identifiable. We further investigate robustness properties of our approach under model misspecification and discuss possible extensions. The empirical properties are studied for various data sets, including large-scale gene perturbation experiments.

相对于系统级负荷,用户负荷具有基数小、波动性与随机性更强的特点,加大了用户负荷预测的难度。文章借助互信息与深度学习理论,提出了一种基于最大相关最小冗余(max-relevance and min-redundancy, mRMR)和长短期记忆网络(long-short term memory networks, LSTM)的用户负荷短期预测模型。首先,采用mRMR算法对特征变量进行排序并选取合适的输入变量集合,mRMR既可以保证输入变量与目标值间互信息值最大,又使得变量间冗余性最小。接着,对选取的输入变量集合建立LSTM预测模型,LSTM能较好处理和预测延迟较长的时间序列,且不会存在梯度消失和梯度爆炸现象。最后,通过算例验证了所提算法的有效性。

Compared with the system-level load, consumer load has the characteristics of small base, stronger volatility and randomness, which increases the difficulty of consumer load forecasting. With the help of mutual information and deep learning theory, this paper proposes a short-term consumer load forecasting model based on max-relevance and min-redundancy (mRMR) and long short-term memory network (LSTM). Firstly, the mRMR algorithm is used to sort the characteristic variables and select a suitable set of input variables. mRMR can not only ensure the maximum mutual information value between the input variable and the target value, but also minimize the redundancy between the variables. Secondly, the LSTM forecasting model is established for the selected set of input variables. LSTM can better process and forecast time series with long delays, and there will be no gradient disappearance and gradient explosion. Finally, an example is used to verify the effectiveness of the algorithm in this paper.