基于连云港市的电力系统短期负荷预测研究

发布时间：2019-01-05 01:44

【摘要】：负荷预测工作是电力企业调度、用电、计划、规划等部门的重要工作内容之一,电力负荷预测水平的高低也是衡量现代电力发展程度的重要标志。提高电力系统负荷预测水平,有利于计划用电管理,节约一次能源和降低发电成本,提高电力系统的经济效益和社会效益。论文基于连云港市的实际负荷情况,首先对其做了比较详细而有条理的分析说明,对影响负荷预测的诸多因素,例如历史负荷数据、温度高低、天气状况等——考虑进负荷预测的模型中。为了提高预测准确率,对负荷数据和其他样本做了大量的预处理,以便于数据平滑而易于被模型所辨识。随后介绍了误差反向传播算法即BP算法的结构和原理,将BP算法用于负荷预测,简单高效可行,但由于该算法收敛的时间较长、且容易陷入局部极小点,故提出了用粒子群算法优化BP神经网络的新型优化算法,该算法能够有针对性地优化网络结构中的权值和阈值,在不断迭代的情况,使得预测误差向减小的方向训练,从而使预测结果准确率有了较大程度的提高,满足了负荷预测的基本要求。
[Abstract]:Load forecasting is one of the most important tasks in the power enterprise such as dispatching, power consumption, planning, planning and so on. The level of power load forecasting is also an important symbol to measure the development of modern electric power. Improving the load forecasting level of power system is beneficial to the management of planned power consumption, saving primary energy and reducing the cost of power generation, and improving the economic and social benefits of the power system. Based on the actual load situation of Lianyungang City, the paper first makes a detailed and orderly analysis of the load forecasting factors, such as historical load data, temperature, Weather conditions, etc.-in models that take into account load forecasting. In order to improve the prediction accuracy, a lot of preprocessing is done to the load data and other samples to make the data smooth and easy to be identified by the model. Then it introduces the structure and principle of error back-propagation algorithm, that is, BP algorithm. The BP algorithm is simple, efficient and feasible for load forecasting. However, because of its long convergence time and easy to fall into local minimum point, Therefore, a new optimization algorithm based on particle swarm optimization (PSO) algorithm for BP neural network is proposed. The algorithm can optimize the weights and thresholds in the network structure, and train the prediction error in the direction of decreasing the prediction error in the case of continuous iteration. Therefore, the accuracy of forecasting results has been improved to a large extent and the basic requirements of load forecasting have been met.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM715

【相似文献】