【作者】 王晶；

【导师】 张国立；

【作者基本信息】 华北电力大学（河北） ， 计算机应用技术， 2006， 硕士

【摘要】 蚁群算法(Ant Colony Algorithm,ACA)是由意大利学者Dorigo、Maniezz等人在20世纪90年代初提出的模拟蚁群搜索食物过程的一种搜索算法,该算法具有一定的全局搜索能力。然而,蚁群算法收敛速度较慢,并且容易出现停滞现象,针对这个不足,本文引入自适应免疫算法来提高蚁群算法的全局搜索能力和加快蚁群算法的收敛速度,提出了一种改进蚁群算法。本文将改进蚁群算法作为人工神经网络的学习算法,建立了蚁群神经网络模型,通过与BP算法、模拟退火算法、进化算法训练的人工神经网络进行比较,验证了蚁群神经网络具有较强的全局寻优能力和较快的收敛速度。进而利用蚁群神经网络进行电力系统短期负荷预测,建立了蚁群神经网络预测模型。实验结果表明,用蚁群神经网络进行短期负荷预测,可以提高预测精度。更多还原

【Abstract】 Ant Colony Algorithm (ACA) is by Italian scholar Dorigo, Maniezz and so onproposes in the the beginning of 1990s. It is simulates the ant group search food process onekind of search algorithm. This algorithm has the certain overall situation search ability.However, ACA convergence rate slow, and is easy to appear stagnates the phenomenon. Inview of this insufficiency, this article introduces Adaptive Immune Algorithm (AIA) to speedup the convergence rate of ACA. This article proposed an improved ACA, takes improvedACA to be the study algorithm of Artificial Neural Network (ANN), established a ACANeural Network (ACAN) model. Through with the BP algorithm, the simulation degenerationalgorithm, the evolution algorithm training ANN carries on the comparison, confirmedACAN have the strong overall situation to seek the superior ability and the quick convergencerate. Then carries ACAN on the electrical power system short-term load forecasting, hasestablished the ACAN forecast model. The experimental result indicated that, carries on theshort-term load forecast with ACAN, may increase the forecast precision.更多还原