Predictive Maintenance

May 3, 2019


The real-time service able to continuously monitor the current status of the plant and perform the real-time failure prediction.

The I-EM’s system PREDICO is able to predict inverter faults status before they actually occur, allowing the customer to take actions to reduce plant under performances.

Key features

> Event Detection – prediction of inverter deviations from nominal behavior

> Advanced Fault Recognition – prediction of a specific class of faults

Key benefit

> Keep optimal performances of PV plants over time

> Avoid Loss production period

> Allow efficient O&M procedure planning


The i-EM’s predictive maintenance service is based on efficient combination of Advanced Artificial Intelligence algorithms.

Output interface

> Different warning levels (green, yellow, orange and red) corresponding to 4 different inverter fault severities

> Quality Index trend warning time-evolution and corresponding inverter errors

> Fault class and severity of the inverter alarm

> Action-to-solve (if available from inverter manufacturer)

The service is highly customized on customer’s plant characteristic.

> Inverter technology and Plant Configuration represent relevant input of the model ensuring customized solution to customer plant.

> Historical manual and automatic alarms logbook-based models training ensures to follow customers’ plant specific behavior


The PREDICO Service reliability has been widley assessed through the whole software development cycle as well as during the real-time services to customers.

The performance are continuously assessed by mean of the usual classification metrics such as Sensitivity, Specificity, Accuracy and ROC curve.

i-EM solutions are completely customizable on demand and are provided as Platform as a Service (PaaS)