Systems and methods for self-learning artificial intelligence of things (AIOT) devices and services

A method of monitoring or predicting a service event for an industrial asset using an artificial intelligence of things (AIoT) device without the need for historical data, the AIoT device being in communication with a cloud-based processor, the AIoT device comprising one or more sensors and an inference engine, the cloud processor comprising a narrow-band classifier and a machine-learning classifier, the method comprising:

• during a training mode:
• gathering data only from the AIoT device, the data being collected by the one or more sensors and periodically providing the data to the cloud based processor;
• using the data to train an inference engine and a machine learning model;
• storing the inference engine on the AIoT device to provide edge processing capability;
• storing the machine learning model accessible by the cloud-based processor;
• during an operational mode:
• using the inference engine to determine if the data includes an anomaly, and if so, sending at least some of the data to the cloud-based processor;
• at the cloud-based processor, receiving at least some of the data from the AIoT device and classifying the anomaly using a narrow-band classifier, or a machine-learning classifier;
• if the anomaly is classified, determining alerts, corrective actions, and/or maintenance requests associated with the anomaly;
• if the anomaly is not classified, providing at least some of the data to a technician or work order system;
• if the technician or work order system classifies the anomaly retraining the machine-learning classifier based on such classification;
• if the technician or work order system determines that the anomaly determined by the inference engine is a false positive, retraining the inference engine based on such determination.