佳木斯大学学报：自然科学版
佳木斯大學學報：自然科學版
가목사대학학보：자연과학판
Journal of Jiamusi University(Natural Science Edition)
2012年
1期
94-98
,共5页
架空输电线路%可靠性评估指标%遗传算法%BP神经网络
架空輸電線路%可靠性評估指標%遺傳算法%BP神經網絡
가공수전선로%가고성평고지표%유전산법%BP신경망락
overhead transmission line%reliability assessment index%genetic algorithm%BP neural algo-rithm%reliability
为了改善当前输电线路可靠性评估中的人为主观局限性，文中首次将遗传神经网络运用于输电线路可靠性评估中．利用BP神经网络的自学习、自适应、强容错性，通过遗传算法（GA）优化BP神经网络的权值和阈值．弱化了评价体系中的人为因素，解决了BP神经网络易陷入最小值点及收敛速度慢的问题，相比神经网络，遗传神经网络的仿真次数减少了3014次．实例仿真研究表明，该评价方法在输电线路的可靠性评价中取得了很好的评价效果，很好地克服了人为主观局限性，其精度达到了90％以上．
為瞭改善噹前輸電線路可靠性評估中的人為主觀跼限性，文中首次將遺傳神經網絡運用于輸電線路可靠性評估中．利用BP神經網絡的自學習、自適應、彊容錯性，通過遺傳算法（GA）優化BP神經網絡的權值和閾值．弱化瞭評價體繫中的人為因素，解決瞭BP神經網絡易陷入最小值點及收斂速度慢的問題，相比神經網絡，遺傳神經網絡的倣真次數減少瞭3014次．實例倣真研究錶明，該評價方法在輸電線路的可靠性評價中取得瞭很好的評價效果，很好地剋服瞭人為主觀跼限性，其精度達到瞭90％以上．
위료개선당전수전선로가고성평고중적인위주관국한성，문중수차장유전신경망락운용우수전선로가고성평고중．이용BP신경망락적자학습、자괄응、강용착성，통과유전산법（GA）우화BP신경망락적권치화역치．약화료평개체계중적인위인소，해결료BP신경망락역함입최소치점급수렴속도만적문제，상비신경망락，유전신경망락적방진차수감소료3014차．실례방진연구표명，해평개방법재수전선로적가고성평개중취득료흔호적평개효과，흔호지극복료인위주관국한성，기정도체도료90％이상．
The subjective limitation problems exist in traditional transmission line rellal3mty assessment. To solve the problems, an transmission line reliability assessment method based on neural network whose thresh- olds and connection weights optimized by genetic algorithm（ GA -NN） was first proposed. Using the characteris- tic of self - learning, self - adaptive and efficient fault tolerant of BP neural networks, the problems of trapping into local minimum point and low convergence speed of the BP neural network were solved. The method weak- ened the human factors. Comparing the neural network, the simulation times of genetic - neural network was re- duced to 3014. Example simulation shows that the method can obtain very well effectiveness evaluation in over- head transmission line reliability assessment, and the accuracy of assessment can be more than 90%.
