机械设计与研究
機械設計與研究
궤계설계여연구
MACHINE DESIGN AND RESEARCH
2010年
2期
99-101
,共3页
张强%毛君%李守巨%田大丰
張彊%毛君%李守巨%田大豐
장강%모군%리수거%전대봉
反问题%神经网络%优化设计%硬岩掘进机
反問題%神經網絡%優化設計%硬巖掘進機
반문제%신경망락%우화설계%경암굴진궤
inverse problem%neural network%optimization design%hard rock machine
硬岩掘进机优化设计过程中设计变量较多的问题,采用应力分布为设计变量,再通过反问题计算得到截割头采间接对截割头进行参数化;针对评价函数计算量太大的问题,根据试验设计理论安排训练样本,采用神经网络建立设计变量与目标函数间的复杂的响应关系,并且详细研究了径向基函数网络在对评价函数进行预测过程中的应用,建立了一种新的截割头优化设计方法.与传统的优化方法相比,其设计变量数目较少,利用此方法对硬岩掘进机截割头的能耗和效率进行优化,所得截割头破碎性能良好,从而验证了此方法的有效性.
硬巖掘進機優化設計過程中設計變量較多的問題,採用應力分佈為設計變量,再通過反問題計算得到截割頭採間接對截割頭進行參數化;針對評價函數計算量太大的問題,根據試驗設計理論安排訓練樣本,採用神經網絡建立設計變量與目標函數間的複雜的響應關繫,併且詳細研究瞭徑嚮基函數網絡在對評價函數進行預測過程中的應用,建立瞭一種新的截割頭優化設計方法.與傳統的優化方法相比,其設計變量數目較少,利用此方法對硬巖掘進機截割頭的能耗和效率進行優化,所得截割頭破碎性能良好,從而驗證瞭此方法的有效性.
경암굴진궤우화설계과정중설계변량교다적문제,채용응력분포위설계변량,재통과반문제계산득도절할두채간접대절할두진행삼수화;침대평개함수계산량태대적문제,근거시험설계이론안배훈련양본,채용신경망락건립설계변량여목표함수간적복잡적향응관계,병차상세연구료경향기함수망락재대평개함수진행예측과정중적응용,건립료일충신적절할두우화설계방법.여전통적우화방법상비,기설계변량수목교소,이용차방법대경암굴진궤절할두적능모화효솔진행우화,소득절할두파쇄성능량호,종이험증료차방법적유효성.
The ordinary parameterization programs for cutting head of hard rock machine contain a great deal of design variables, so an indirect parameterization method was proposed ,where the stress distribution was treated as de-sign variable. The blade was calculated by inverse design method and neural networks was adopted to construct the response relationshie between the design variables and the objective function. The sample data used to train neural network was schemed according to the theory of experiment design, the application of radial basic function network was investigated in detail, and a new optimization method was proposed. Compared with the traditional optimization method, here the design variable number is few, the calculation time is shortened obviously; and in this toothed, the energy consumption and the efficiency was selected as the objective functions for optimization. Finaly the validity of this newly proposed method is confirmed.