Ekuthuthukisweni okubalulekile emkhakheni wokuxilonga ngomshini, ucwaningo olusha lubonise ukusebenza kahle kokuhlanganisa i-modulation signal bispectrum (MSB) namanethiwekhi e-convolutional neural (CNN) ekuxilongweni kwamaphutha e-amagiya e-bevel ajikelezayoLe ndlela entsha ithembisa ukunemba okuthuthukisiwe, ukutholwa okusheshayo, kanye nohlelo lokuxilonga oluhlakaniphe kakhulu lwama-gearbox asebenza kahle kakhulu asetshenziswa kuizinhlelo zokusebenza zezindiza, izimoto, kanye nezimboni.

I-Spiralamagiya e-bevelyizingxenye ezibalulekile zokudlulisa ezitholakala emishinini enamandla amakhulu, izindiza ezinophephela emhlane, izinhlelo zokuhambisa zasolwandle, kanye nemishini yokunciphisa izimboni esindayo. Ngenxa ye-geometry yazo eyinkimbinkimbi kanye nezimo zokusebenza, ukutholakala kusenesikhathi kwamaphutha egiya njengokufaka imigodi, ukuguguleka, kanye nokuphuka kwamazinyo kuseyinselele yobuchwepheshe. Amasu okucubungula isignali yendabuko avame ukuba nezinkinga zokuphazamiseka komsindo kanye nezici zephutha ezingezona eziqondile.

Le ndlela entsha yethula uhlaka lokuhlonza amaphutha ezigaba ezimbili. Okokuqala izimpawu zokudlidliza ezikhiqizwa uhlelo lwegiya lokusebenza zihlaziywa kusetshenziswa i-modulation signal bispectrum (MSB) okuyisu lokuhlaziya i-spectral eliphezulu elibamba ngempumelelo izici ezingezona eziqondile nezingezona ze-Gaussian zesignali. I-MSB isiza ekwembuleni izici zephutha ezishintshiwe ezicashile ezivame ukufihlwa kuma-spectra emvamisa ejwayelekile.

Okulandelayo, idatha yesignali ecutshungulwayo iguqulwa ibe izithombe zemvamisa yesikhathi bese ifakwa kunethiwekhi ye-convolutional neural (CNN) imodeli yokufunda okujulile ekwazi ukukhipha ngokuzenzakalelayo izici zamaphutha aphezulu kanye nokuhlukanisa izimo zegiya. Le modeli ye-CNN iqeqeshwe ukuhlukanisa phakathi kwamagiya anempilo, amaphutha amancane, kanye nomonakalo omkhulu ezimweni ezahlukene zomthwalo kanye nesivinini.

Imiphumela yokuhlolwa, eyenziwe ngegiya lokuhlola eliyindilinga elenziwe ngokwezifiso, ikhombisa ukuthi indlela ye-MSB CNN ifinyelela ukunemba kokuhlukaniswa okungaphezu kuka-97%, isebenza kangcono kunezindlela zendabuko ezifana nokuhlaziywa okusekelwe ku-FFT kanye namanye amasu okufunda okujulile ancike kudatha yokudlidliza eluhlaza. Ngaphezu kwalokho, le modeli ehlanganisiwe ibonisa ukuqina okuqinile komsindo wangemuva, okwenza ifaneleke ukusetshenziswa kwezimboni zangempela.

Ukuhlanganiswa kwe-bispectrum yesiginali yokuguquguquka ne-CNN akugcini nje ngokuthuthukisa ukusebenza kokuqaphela amaphutha kodwa futhi kunciphisa ukuthembela kubunjiniyela bezici ezenziwe ngesandla ngokwesiko inqubo ethatha isikhathi futhi encike kobuchwepheshe. Le ndlela ingandiswa futhi ingasetshenziswa kwezinye izingxenye zemishini ezijikelezayo, njenge-bearings kanyeamagiya eplanethi.

Lolu cwaningo lumelela isinyathelo esiqhubekela phambili ekuthuthukisweni kwezinhlelo zokuxilongwa kwamaphutha ezihlakaniphile ze-Industry 4.0 kanye nensimu ebanzi yokukhiqiza okuhlakaniphile. Njengoba ukuzenzakalela kanye nokuthembeka komshini kuba yinto ebaluleke kakhulu,