Mukufambira mberi kwakakosha mumunda wekuongorora mashandiro emuviri, chidzidzo chitsva chakaratidza kushanda kwekubatanidza modulation signal bispectrum (MSB) ne convolutional neural networks (CNN) pakuongorora kukanganisa kwemasero.magiya esimbi anotendereraMaitiro aya matsva anovimbisa kunyatsojeka, kuonekwa nekukurumidza, uye sisitimu yekuongorora yakangwara yemagiya emagetsi anoshanda zvakanyanya anoshandiswa mumashandisirwo endege, mota, uye maindasitiri.

Kutendereramagiya ebhevelzvinhu zvakakosha zvekutumira zvinhu zvinowanikwa mumichina ine torque yakawanda, maherikoputa, masisitimu ekufambisa emugungwa, uye zvinoderedza simba remaindasitiri. Nekuda kwehunhu hwazvo hwakaoma uye mamiriro ekushanda, kuona nekukurumidza zvikanganiso zvegiya zvakaita sekunge maburi, kupera, uye kutyoka kwezino kuchiri dambudziko rehunyanzvi. Maitiro echinyakare ekugadzirisa masaini anowanzo tambura nekukanganiswa kweruzha uye hunhu hwekutadza kusina kurongeka.

Nzira itsva iyi inoburitsa hurongwa hwekuongorora zvikanganiso mumatanho maviri. Kutanga, zviratidzo zvevibration zvinogadzirwa ne operating gear system zvinoongororwa uchishandisa modulation signal bispectrum (MSB) inova nzira yepamusoro yekuongorora spectral iyo inobata zvinobudirira zviratidzo zvisiri zvemutsara uye zvisiri zveGaussian zvechiratidzo. MSB inobatsira kuratidza hunhu hwezvikanganiso hwakagadziriswa hushoma hunowanzovanzwa mu standard frequency spectra.

Tevere, data rechiratidzo rinogadziriswa rinoshandurwa kuita mifananidzo yenguva uye rinoiswa mu convolutional neural network (CNN) modhi yekudzidza yakadzika inokwanisa kubvisa otomatiki zvinhu zvekutadza kwepamusoro uye kuisa muzvikamu mamiriro egiya. Iyi modhi yeCNN yakadzidziswa kusiyanisa pakati pegiya rine hutano, zvikanganiso zvidiki, uye kukuvara kwakanyanya mumamiriro akasiyana emutoro nekumhanya.

Mhedzisiro yekuyedza, yakaitwa pa spiral bevel gear test rig yakagadzirwa nehunyanzvi, inoratidza kuti nzira yeMSB CNN inowana kururama kwekupatsanura kwe97%, ichipfuura nzira dzechinyakare dzakadai seFFT based analysis uye kunyange dzimwe nzira dzekudzidza zvakadzama dzinoshandisa data re vibration. Uyezve, iyi hybrid modhi inoratidza kusimba kwakasimba kune ruzha rwekumashure, zvichiita kuti ive yakakodzera kushandiswa mumaindasitiri chaiwo.

Kubatanidzwa kwe modulation signal bispectrum neCNN hakungobatsiri chete pakuona zvikanganiso asi kunoderedzawo kuvimba nekushandisa manual feature engineering, iyo inowanzotora nguva uye inoenderana nehunyanzvi. Nzira iyi inogona kuwedzerwa uye inogona kushandiswa kune zvimwe zvikamu zvemuchina zvinotenderera, zvakaita semabearings nemagiya epuraneti.

Tsvagiridzo iyi inomiririra nhanho kufambira mberi mukugadzirwa kwehurongwa hwekuongorora zvikanganiso hweIndustry 4.0 uye munda wakakura wekugadzira zvinhu zvakangwara. Sezvo otomatiki uye kuvimbika kwemuchina zvichiri kukosha,