基于多尺度卷积神经网络的短期电力负荷预测

In order to improve the accuracy of short-term load forecasting of power system, a multi-scale information fusion convolutional neural network(MS-ConvNet)model based on deep learning technology was proposed. A full convolution network structure and causal logic constraints were introduced to enhance the expression of time series features; a multi-scale convolution was utilized to extract the relationship among time domain data of different lengths for obtaining more abundant series features; a residual network structure was designed to increase the network depth, which increased the acceptance domain of outputneurons and enhanced the prediction accuracy. The results show that the accuracy and stability of MS-ConvNet model is better than those of multi-layer perceptron machine, long-short term memory network and gated recurrent unit network, indicating that the as-proposed model has a good application prospect in power load forecasting.     

日前电能市场与深度调峰市场联合出清模型

以采用全电量集中竞价模式的日前电能交易和东北深度调峰交易融合运行、联合出清为背景,建立一种日前电能量市场和深度调峰服务市场联合出清模型。模型以系统电能量费用、启动费用、调峰费用与新能源限电惩罚费用之和最小为目标函数,以系统电力平衡与备用、火电机组和新能源2类发电单元运行边界为约束条件,实现新能源优先消纳基础上的2市场整体最优;模型通过日前电能量市场预出清与2市场联合再出清的分步计算方法,在可接受的时间内,实现2市场的联合出清。     

基于离散傅里叶变换的微电网混合储能容量优化

针对并网型微电网中由蓄电池和超级电容组成的混合储能系统进行容量的优化配置。光伏发电和负荷之间产生的净负荷功率由大电网和混合储能装置来共同进行平抑。建立一个俩阶段混合储能容量优化的数学模型,利用离散傅里叶变换对微电网中产生的净负荷功率进行分解,第一阶段在满足联络线功率波动要求的基础上来选取联络线功率和混合储能系统功率的分界点使得联络线利用率最高的;第二阶段以混合储能容量配置的经济成本最低为目标选取蓄电池功率和超级电容功率的分界点;从而得到联络线、蓄电池和超级电容的功率分配。利用遗传算法对混合储能容量的优化模型进行求解,得到最优的混合储能容量的配置。通过算例进行了验证分析。