應(yīng)用 Ω_t. -SNE流形學(xué)習(xí)方法的盾構(gòu)刀盤磨損信號(hào)降維

打開文本圖片集

關(guān)鍵詞：盾構(gòu)刀盤；性能評(píng)估： ;t- 分布隨機(jī)鄰域嵌入；信號(hào)降維中圖分類號(hào)：U455 文獻(xiàn)標(biāo)志碼：A 文章編號(hào)：1671-5276（2025）03-0096-04

Dimension Reduction of Shield Cutterhead Wear Signal by t -SNE Manifold Learning Method

HUANG Yuhua1，LI Feng2 （1.School of Architecture and Electrical Engineering，Hezhou University，Hezhou 532899，China; 2.School of Civil Engineering，Guangxi University of Science and Technology，Liuzhou 545oo6，China）

Abstract：Inordertoenhancetheabilityofmonitoringtherunningstabilityofshieldcuterhead，amethodfordimensionality reduction and performance evaluation of torque vibration signal of shield cutterhead based on t- SNE manifold learning was designed.Acording totheactual parametersof shieldequipment，theidentificationofcuterstateparameters wasrealizedby data driven.The results show that compared with ISOMap，LLE，KPCA，etc.，using t -SNEmanifold dimensionalityreduction can obtain a longer t -distribution，so that thelow-dimensional far end pointswillhave a larger low-dimensional interval，and canaccuratelyclasifynormalanddegeneratesamples，andaccuratelyidentifyhigh－dimensionaldatacontaining low一 dimensional manifold.Thevariationof normaloperatingconditionsandfaultmaintenancesampling interval in Markov spaceis small，andtheuseof time-dependent Markovdistance measure toevaluatetoolperformancecanobtain higherprecision， efctivelyeliminatethecorrelationbetween diferentdimensions，and havebeterperformancethanEuclideandistance.

Keywords：cutter head of shield tunneling；performance evaluation； t- distributed random neighborhood embedding；signal dimension reduction

0 引言

刀盤在盾構(gòu)裝備中是一個(gè)核心受力部件，能夠?qū)崿F(xiàn)掘進(jìn)并維持開挖面穩(wěn)定，直接影響掘進(jìn)效率[1-2]。（剩余4443字）