【作者】 邹政达；

【导师】 孙雅明；

【作者基本信息】 天津大学 ， 电力系统及其自动化， 2005， 硕士

【摘要】 本论文是探索蚁群优化算法(ACOA)用于电力系统领域中短期负荷预测和输电网络扩展规划两个重要课题的研究。负荷预测已有很长期的研究历史,但负荷预测的精度仍然是不能满足运行要求;随着电力系统规模的扩大,对输电网络扩展规划优化研究也提出新问题。论文是在基于神经网络(NN)原理的负荷预测研究已取得成果的基础上进行的,早期研究的NN预测模型都采用BP算法或改进BP算法,但对BP算法的易陷入局部极小点、收敛速度慢、NN推广能力较差等实质性问题都未能彻底摆脱,导致预测精度不能有效的保证,因此研究NN预测模型的学习算法也是提高负荷预测精度的重要方面。本论文首次提出基于ACOA递归神经网络的短期负荷预测模型。通过对实际负荷系统日、周预测研究和仿真测试,证明了对提高预测精度的有效性,特别对工作日和休息日都具有良好的稳定性和适应能力,无须分别建预测模型,预测精度明显优于BP算法和遗传算法。本论文根据输电网络扩展规划研究中出现的停滞现象以及搜索时间过长的问题,论文提出基于带扰动的自适应ACOA的输电网络扩展规划方法,建立了相应的模型及其算法,通过2个算例系统的计算表明所提出方法能有效的提高计算速度,并具有良好的收敛性。本论文提出基于ACOA的电力系统负荷预测和带扰动的自适应ACOA的输电网络扩展规划两方面的研究,经仿真测试,证明所提出的方法都具有明显优势,能有效的解决研究问题,本论文的研究是有理论意义和实用价值的。更多还原

【Abstract】 Study work of this paper includes two sections: the study of short-term loadforecasting (STLF) based on recurrent neural network (NN) using ant colonyoptimization algorithm (ACOA) and the study of self-adaptive ACOA withperturbation for transmission network expansion planning. Although STLF researchhas been done for a long history, the precision is seldom high enough for the demandof power system;With the development of power system, there are some newproblems about transmission network expansion planning.This paper is studied based on NN principle for STLF. Back Propagation (BP)and improved BP were widely used in NN forecasting model, but it has someshortcomings, such as slow convergence rate, easy to fall into local minimum and lowgeneralization ability of NN, which result in low forecasting precision. So the study oftraining algorithm for NN forecasting model is an important impact to enhance theprecision of STLF.In this paper the STLF based on recurrent NN model using ACOA is firstproposed. The simulation results of daily and weekly loads forecasting for actual powersystem show that the proposed forecasting model can effectively improve the accuracyof short-term load forecasting (SLTF) and this model is stable and adaptable for bothworkday and rest-day, in addition, its forecasting performance is far better than that ofBP-RNN and GA-RNN.To overcome the stagnation and long searching time appeared in transmissionnetwork expansion planning, a self-adaptive ACOA with perturbation is presented inuse for transmission network expansion planning, the corresponding mathematicalmode is established and solution algorithms are developed. The presented method hasbeen tested on two systems, and the results show its advantage on computing speedand convergence.In this paper, the study of STLF based on recurrent NN using ACOA and thestudy of self-adaptive ACOA with perturbation for transmission network expansionplanning are presented. Through testing, the proposed method has the obviousadvantage and can solve the problem effectively. This paper’s study is significative intheory and is worthful in the practice.更多还原