端到端機(jī)器學(xué)習(xí)代理模型構(gòu)建及其在爆轟驅(qū)動(dòng)問(wèn)題中的應(yīng)用
打開(kāi)文本圖片集
中圖分類號(hào)：0389 國(guó)標(biāo)學(xué)科代碼：13035 文獻(xiàn)標(biāo)志碼：A
Abstract： Artificial intellgence/machine learming methods candiscover hidden physical patters in data.Byconstructing an end-to-end surogate model between state parameters and dynamic results， many complex engineering problems such as strong coupling，nonlinearity，and multiphysicscan be eficiently solved.Inthe fieldof highlynonlinearexplosion and shock dynamics，a clasic detonation driving problem was chosen asthe research object.Using numerical simulation results as trainingdatafor machine learningsurrogate models，and combining forward simulation and reversedesign organicall. Based on deepneural network technology，anend-to-end surogatemodel wasconstructed between feature position velocity profiles， material dynamic deformation，and engineering factors.And the calculation accuracyof the surrogate model was provided， verifyingtheabilitytoinvertengineering factorsfromvelocityprofiles.Theresearchresultsindicatethattheend-to-end surrogate model has high predictive ability，with relative errors of less than 1 % in both velocity profile prediction and enginering factorestimation.Itcanbeappliedtotherapiddesign，high-precisionprediction，andagileiterationof highly nonlinear explosion and impact dynamics problems.
Keywords：computational explosionmechanics； detonation drive；artificial inteligence； machine learning； end-to-end surrogatemodel； deep neural network
人工智能（artificial inteligence，AI）是能夠和人一樣進(jìn)行感知、認(rèn)知、決策和執(zhí)行的人工程序或系統(tǒng)，是新一輪科技革命和產(chǎn)業(yè)變革的重要驅(qū)動(dòng)力量。（剩余12596字）