电网技术
電網技術
전망기술
POWER SYSTEM TECHNOLOGY
2015年
1期
130-135
,共6页
遗传算法%序列二次规划-遗传算法%变压器%局部放电%超声波定位%非智能算法
遺傳算法%序列二次規劃-遺傳算法%變壓器%跼部放電%超聲波定位%非智能算法
유전산법%서렬이차규화-유전산법%변압기%국부방전%초성파정위%비지능산법
genetic algorithm%SQP-GA%transformer%partial discharge%ultrasonic localization%non-intelligent algorithm
针对基本遗传算法(genetic algorithm,GA)局部寻优能力较差和易出现早熟收敛现象,提出了一种改进的遗传算法,即序列二次规划-遗传算法(sequential quadratic programming- genetic algorithm,SQP-GA)。该混合优化算法SQP-GA在基本GA中引入序列二次规划(sequential quadratic programming, SQP)算法,经典算法SQP作为一个元算子有效地提高了基本GA的局部搜索能力,并克服了早熟收敛现象。函数仿真测试结果表明,SQP-GA混合优化算法在收敛速度和寻优精度上都优于基本 GA,表明所提出的算法的有效性。同时,利用提出的SQP-GA求解变压器局部放电超声波定位问题,并将其定位结果与GA和非智能算法的定位结果进行比较;算例结果表明,基于 SQP-GA 的变压器局部放电超声定位法能有效地防止结果陷入局部最优,该方法的定位效果理想。
針對基本遺傳算法(genetic algorithm,GA)跼部尋優能力較差和易齣現早熟收斂現象,提齣瞭一種改進的遺傳算法,即序列二次規劃-遺傳算法(sequential quadratic programming- genetic algorithm,SQP-GA)。該混閤優化算法SQP-GA在基本GA中引入序列二次規劃(sequential quadratic programming, SQP)算法,經典算法SQP作為一箇元算子有效地提高瞭基本GA的跼部搜索能力,併剋服瞭早熟收斂現象。函數倣真測試結果錶明,SQP-GA混閤優化算法在收斂速度和尋優精度上都優于基本 GA,錶明所提齣的算法的有效性。同時,利用提齣的SQP-GA求解變壓器跼部放電超聲波定位問題,併將其定位結果與GA和非智能算法的定位結果進行比較;算例結果錶明,基于 SQP-GA 的變壓器跼部放電超聲定位法能有效地防止結果陷入跼部最優,該方法的定位效果理想。
침대기본유전산법(genetic algorithm,GA)국부심우능력교차화역출현조숙수렴현상,제출료일충개진적유전산법,즉서렬이차규화-유전산법(sequential quadratic programming- genetic algorithm,SQP-GA)。해혼합우화산법SQP-GA재기본GA중인입서렬이차규화(sequential quadratic programming, SQP)산법,경전산법SQP작위일개원산자유효지제고료기본GA적국부수색능력,병극복료조숙수렴현상。함수방진측시결과표명,SQP-GA혼합우화산법재수렴속도화심우정도상도우우기본 GA,표명소제출적산법적유효성。동시,이용제출적SQP-GA구해변압기국부방전초성파정위문제,병장기정위결과여GA화비지능산법적정위결과진행비교;산례결과표명,기우 SQP-GA 적변압기국부방전초성정위법능유효지방지결과함입국부최우,해방법적정위효과이상。
In allusion to the premature convergence phenomena and insufficient local searching ability of the basic genetic algorithm(GA), an improved GA, namely the sequential quadratic programming-genetic algorithm (SQP-GA), is proposed. In SQP-GA the sequential quadratic programming(SQP) algorithm is led into the basic GA, and as a child operator the classic algorithm SQP can effectively improve the local searching ability of the basic GA and the premature convergence phenomenon is overcome. Test results of function simulation show that in both aspects of convergence speed and searching accuracy the SQP-GA hybrid algorithm is superior to the basic GA, so it is shown that the proposed algorithm is effective. Meanwhile, using the proposed SQP-GA hybrid algorithm to solve the ultrasonic localization of partial discharge (PD) inside the power transformer and comparing the obtained localization results with those obtained by GA and non-intelligent algorithm, the comparison results show that the ultrasonic localization method of PD inside the power transformer based on the SQP-GA can effectively prevent the measured results fallinginto the local optimum, and the localization results by the proposed method are satisfactory.