王建維， 陳建政
（西南交通大學(xué)牽引動力國家重點(diǎn)實(shí)驗(yàn)室，四川 成都 610031）
摘 要：測力輪對（Instrumented wheelset）是測量輪軌力最直接最準(zhǔn)確的方法。利用有限元法建立測力輪對的有限元模型，并通過計(jì)算機(jī)模擬研究載荷作用點(diǎn)位置對測力輪對輻板應(yīng)力的影響，在此基礎(chǔ)上，建立加載位置的BP神經(jīng)網(wǎng)絡(luò)模型。實(shí)踐表明，采用BP網(wǎng)絡(luò)模型仿真的結(jié)果可以得到更理想的輸出波形，解決或減少了傳統(tǒng)組橋過程中存在的一系列問題，證明了所設(shè)計(jì)的網(wǎng)絡(luò)具有足夠的精度，所建立的網(wǎng)絡(luò)也是具有應(yīng)用價值的。
關(guān)鍵詞：測力輪對；有限元法；神經(jīng)網(wǎng)絡(luò)；BP算法；橫垂向橋
中圖分類號：U213.2+12 文獻(xiàn)標(biāo)識碼：A 文章編號：1672-4984（2007）01-0054-03
Research on solution of wheel/trail force load position based on BP neural network
WANG Jian-wei， CHEN Jian-zheng
（Traction Power State Key of Laboratory，Southwest Jiaotong University，Chengdu 610031，China）
Abstract：Instrumented wheelset is the most direct and precise method to measure wheel/trail force.A finite element （FE） model of Instrumented wheelset was researched，and the effect of load position to the stress of the Instrumented wheelset was simulated by computer.A BP neural network model was set up based on simulation result.It turned out that a more perfect output wave form can be gotten using the BP neural network.This method solvesd or reduces a series of problems which produced in traditional bridge.The model’s simulation data have proved the high precision and application value of the neural network.
Key words：Instrumented wheelset；Finite element method；Neural network；Back propagation algorithm；Lateral and vertical bridge
