A Non-Intrusive Deep Learning Based Diagnosis System for Elevators

With the ever-growing number of elevators coupled with aging workforce, diminishing new installations and limited use maintenance technology, it is increasingly challenging for owners responsible parties to maintain safe reliable operation lift systems. To address this issue, a non-intrusive artificial intelligence (AI) based diagnosis system, aiming at providing fault detection potential prediction multi-brand lifts without intervening existing circuitry installations, proposed in paper. The system employs multivariate long short term memory fully convolutional network (MLSTM-FCN) learn analyze measured signals from elevators. It capable (i) giving advance clear warnings corrective actions prevent major equipment breakdowns, (ii) indicating just-in-time enhancing reliability low cost. implementation provided. design diagnostic algorithm elaborated. Both simulations experiments commercial elevator have been conducted verify effectiveness system.