计算机工程与设计
計算機工程與設計
계산궤공정여설계
Computer Engineering and Design
2015年
10期
2733-2737
,共5页
动态称重%引力搜索算法%全局搜索%BP网络%惯性权重
動態稱重%引力搜索算法%全跼搜索%BP網絡%慣性權重
동태칭중%인력수색산법%전국수색%BP망락%관성권중
dynamic weighing%gravitational search algorithm (GSA)%global search%BP network%inertia weight
为提高动态称重数据处理的精确度和速度，研究动态称重数据处理相关方法，提出一种改进的引力搜索算法（GSA）对BP神经网络进行优化的方法。通过引入改进黑洞因子（BH）和自适应惯性权重，提高GSA的搜索能力，优化BP神经网络的初始权值和阈值，使动态称重数据的处理速度更快、精确度更高。将改进GSA‐BP算法和BP算法、GA‐BP、GSA‐BP进行仿真对比，对比结果表明，改进的GSA具有优秀的全局搜索能力，经其优化的BP网络对动态称重数据的处理结果更加精确、性能更好。
為提高動態稱重數據處理的精確度和速度，研究動態稱重數據處理相關方法，提齣一種改進的引力搜索算法（GSA）對BP神經網絡進行優化的方法。通過引入改進黑洞因子（BH）和自適應慣性權重，提高GSA的搜索能力，優化BP神經網絡的初始權值和閾值，使動態稱重數據的處理速度更快、精確度更高。將改進GSA‐BP算法和BP算法、GA‐BP、GSA‐BP進行倣真對比，對比結果錶明，改進的GSA具有優秀的全跼搜索能力，經其優化的BP網絡對動態稱重數據的處理結果更加精確、性能更好。
위제고동태칭중수거처리적정학도화속도，연구동태칭중수거처리상관방법，제출일충개진적인력수색산법（GSA）대BP신경망락진행우화적방법。통과인입개진흑동인자（BH）화자괄응관성권중，제고GSA적수색능력，우화BP신경망락적초시권치화역치，사동태칭중수거적처리속도경쾌、정학도경고。장개진GSA‐BP산법화BP산법、GA‐BP、GSA‐BP진행방진대비，대비결과표명，개진적GSA구유우수적전국수색능력，경기우화적BP망락대동태칭중수거적처리결과경가정학、성능경호。
To improve the accuracy and speed of dynamic weighing data processing ,related methods for dynamic weighing data processing were studied ,an improved gravitational search algorithm (GSA) to optimize the BP neural network was proposed .A new operator was presented called black hole (BH) and the adaptive inertia weight was introduced .The exploitation capability was improved ,initial weights and thresholds of the BP neural network were optimized .The computation efficiency and the accu‐racy of the dynamic weighing data processing were also improved .The simulation was carried out to verify the algorithm .The result of the simulation ,which was compared to other well‐known algorithms ,such as traditional BP algorithm ,GA‐BP and GSA‐BP ,indicates that the improved GSA has better capability on global search ,and the BP network was optimized using the improved GSA is more accurate and has better performance on processing of dynamic weighing data .
