中国电机工程学报
中國電機工程學報
중국전궤공정학보
ZHONGGUO DIANJI GONGCHENG XUEBAO
2014年
32期
5779-5787
,共9页
量子粒子群算法%双量子粒子群算法%数据挖掘%多变量系统%系统辨识
量子粒子群算法%雙量子粒子群算法%數據挖掘%多變量繫統%繫統辨識
양자입자군산법%쌍양자입자군산법%수거알굴%다변량계통%계통변식
quantum particle swarm optimization%double quantum particle swarm optimization%data mining%multivariable system%system identification
针对智能算法与历史大数据相结合进行多变量系统辨识过程中不能精确量化每个子系统数学模型的问题,提出了一种有效的数据并行优化计算的解决方案。在辨识过程中,为了解决量子粒子群算法(quantum particle swarm optimization ,QPSO)收敛速度和寻优精度方面的不足,提出了一种改进的 QPSO 算法--双量子粒子群算法(double quantum particle swarm optimization ,D-QPSO)。该算法对粒子种群编码和原有的进化搜索策略同时进行了量子化处理,经过测试函数实验,改进的算法在搜索能力上优于 PSO 和QPSO算法。最后利用现场运行历史数据,通过D-QPSO算法进行参数估计,将设计的解决方案应用于热力发电厂负荷控制系统的传递函数辨识中,得到的模型为控制器的设计与优化奠定了基础。
針對智能算法與歷史大數據相結閤進行多變量繫統辨識過程中不能精確量化每箇子繫統數學模型的問題,提齣瞭一種有效的數據併行優化計算的解決方案。在辨識過程中,為瞭解決量子粒子群算法(quantum particle swarm optimization ,QPSO)收斂速度和尋優精度方麵的不足,提齣瞭一種改進的 QPSO 算法--雙量子粒子群算法(double quantum particle swarm optimization ,D-QPSO)。該算法對粒子種群編碼和原有的進化搜索策略同時進行瞭量子化處理,經過測試函數實驗,改進的算法在搜索能力上優于 PSO 和QPSO算法。最後利用現場運行歷史數據,通過D-QPSO算法進行參數估計,將設計的解決方案應用于熱力髮電廠負荷控製繫統的傳遞函數辨識中,得到的模型為控製器的設計與優化奠定瞭基礎。
침대지능산법여역사대수거상결합진행다변량계통변식과정중불능정학양화매개자계통수학모형적문제,제출료일충유효적수거병행우화계산적해결방안。재변식과정중,위료해결양자입자군산법(quantum particle swarm optimization ,QPSO)수렴속도화심우정도방면적불족,제출료일충개진적 QPSO 산법--쌍양자입자군산법(double quantum particle swarm optimization ,D-QPSO)。해산법대입자충군편마화원유적진화수색책략동시진행료양자화처리,경과측시함수실험,개진적산법재수색능력상우우 PSO 화QPSO산법。최후이용현장운행역사수거,통과D-QPSO산법진행삼수고계,장설계적해결방안응용우열력발전엄부하공제계통적전체함수변식중,득도적모형위공제기적설계여우화전정료기출。
To overcome the drawback of imprecise quantification for each subsystem during the process of multivariable system identification with intelligent algorithm and historical data, an effective optimized data parallel computing solution was proposed. In order to improve the convergence speed and precision of quantum particle swarm optimization (QPSO) during the identification, a new improved QPSO algorithm named double quantum particle swarm optimization (D-QPSO) was presented. The particle’s encoding mechanism and the evolutionary search strategy were simultaneously quantized in D-QPSO algorithm. Several benchmark test functions were used to test the proposed D-QPSO algorithm, which verified that the new algorithm was superior to standard PSO and QPSO in search capabilities. Finally, the proposed scheme was used in transfer function identification of the thermal power plant load control system, and the parameters of the model were estimated by the D-QPSO algorithm and historical data. The identification results laid the foundation for load control system design and optimization.