Fault detection and predictive maintenance system for electric fans

Z zajemom in analizo vibracij je mogoče pridobiti pomembne informacije o stanju naprave, hkrati pa identificirati kritična področja delovanja, pri katerih je pospešen proces njenega staranja. Kontinuirano obdelovanje podatkov o vibracijah je tako lahko podlaga za prediktivno vzdrževanje naprav, ki je zaradi svoje učinkovitosti vse pogosteje uporabljeno na najrazličnejših področjih. V okviru magistrskega dela je bil zasnovan in izdelan sistem za analizo vibracij aksialnega ventilatorja z elektronsko komutiranim motorjem. Sistem je sestavljen iz pospeškometra in mikrokrmilnika, ki sta vgrajena v pogonsko elektroniko ventilatorja in povezana preko vodila I2C. Razvit je bil algoritem, ki na podlagi izračuna efektivnih vrednosti prebranega pospeška preko celotnega hitrostnega območja ventilatorja določi resonančna območja, ki se jim nato ventilator izogiba. S tem je preprečeno krajšanje življenjske dobe ventilatorja, saj je izpostavljen nižjemu nivoju vibracij. Prav tako je bil realiziran algoritem, ki na podlagi frekvenčne analize zaznava neuravnoteženje rotorja, ki se izrazi v obliki povečane amplitude komponente, ki ustreza hitrosti vrtenja. Na ta način so bili postavljeni temelji prediktivnega vzdrževanja tovrstnih ventilatorjev. Izvedene so bile meritve v laboratorijskem okolju, ki so potrdile ustrezno zasnovo in delovanje algoritmov, s čimer je bila omogočena njihova dejanska implementacija v produkte podjetja Hidria d.o.o.Vibration measurement and analysis can provide important information about the device’s current condition. It also enables the identification of the device’s critical operating points, where increased vibration accelerates its ageing. Therefore, continuous vibration data inspection can serve as a foundation for predictive maintenance, which is increasingly used in various fields due to its efficiency. A system for vibration analysis of axial fans with an electronically commutated motor was designed. It comprises an accelerometer and a microcontroller, which are implemented on the fan’s printed circuit board and connected via the I2C bus. An algorithm was developed, which identifies resonance operating points of the fan and does not allow their use. It is based on the calculations of the root-mean-square value of the read acceleration signal over the fan’s whole speed range. Consequently, the life span of the fan increases due to reduced vibration exposure. Furthermore, another algorithm was developed, which detects unbalance of the rotor, based on the frequency analysis. Unbalance causes the increase in the amplitude of the frequency component, corresponding to the rotating frequency. In that way, the foundations for predictive maintenance of axial fans were laid. Measurements in the laboratory environment confirmed appropriate design and correct operation of the developed algorithms, which enabled their actual implementation in the products of the company Hidria d.o.o