SatADAPT - Satellite Enhanced Adaptive Predictive Maintenance

You are here

Objectives of the service

The unexpected stop of a machine that is part of a larger system or process is usually orders of magnitude more costly in time and money than planned, preventive maintenance of the machine would have been. The SatDAPT system can learn the correct patterns of behavior through classification of acquired values during normal functioning, and can distinguish situations where acquired values combine into a set that indicate the potential for abnormal operation. The goal is to detect an increased probability of failure in the future, that is a possible future equipment failure, based on combinations of parameter values which may still be within the operational limits, and then to classify these situations in categories linked to the type of defect forecasted and the kind of maintenance needed. While the classification itself is done by the AI component of the system, a human expert may still be involved in the process of mapping the various classes of values determined by the AI to actual real failures that have at some point occurred in operation in some other deployment.

Users and their needs

The energy production sector and the manufacturing industry are the first intended beneficiaries of the solution, with end users from both segments helping to refine the system requirements.

Energy companies need a mechanism for avoiding failure of equipment, which can lead to the shutdown of an entire facility, production interruptions and in some cases can have catastrophic consequences with loss of human lives and environmental damage. To avoid such failures, equipment is checked regularly according to maintenance schedules and manuals

Companies that manufacture equipment containing moving parts such as pumps or power generators need to provide warranty and maintenance services to their customers. In general, the more reliable a piece of equipment is, the more efficient is providing these services with the added benefit of a higher customer satisfaction level and a better brand perception in the market.

Railway, maritime and ground transportation are all dependent of the uninterrupted functioning of large engines, and so are other industries, such as aerospace, manufacturing, pulp&paper, food processing, etc.

The geographic area of the user base is worldwide.

The following user needs in respect to the predictive maintenance solution have been identified:

  • High performance in predicting failure and learning new patterns
  • Good RoI
  • Compliance with local laws and regulations
  • Reliability in harsh industrial environment
  • Ease of installation
  • Ease of operation
  • Open architecture and ease of integration in larger Industry 4.0 systems

Service/ system concept

In a field installation there are one or several Wireless Gateways operating as edge devices, each of them creating a Wireless Sensor Network - WSN around them by connecting with a number of vibration sensors using an open standard industrial wireless protocol. Each Gateway connects to a server in the cloud which runs the Central Management System  - CMS, that includes an AI engine with Deep Learning capabilities and also includes a central configuration tool for managing all the Gateways.

The Gateways are connected over Internet to the CMS running in the cloud by cabled Ethernet, 3G/4G cellular connection or satellite connection which can be either using a LEO constellation or a geosynchronous communications satellite (VSAT). Each Gateway includes a GNSS modem for location mapping and for serving as a time reference.

Each Gateway runs an embedded Neural Network - NN which receives its functioning parameters, i.e. the network weights, from the AI engine that is part of the CMS running in the cloud. As the central AI engine is evolving, the updates are continuously transferred back to the NN's running on the Gateways.

Space Added Value

The communication between the CMS and the Gateways is implemented over the Internet via a 3G/4G cellular or a satellite connection which can be either using a LEO constellation or a geosynchronous communications satellite (VSAT). Especially for mobile equipment such as power generators for construction sites or water systems for disaster zones, a satellite connection is the preferred and sometimes the mandatory means of connectivity to the cloud CMS. Also for offshore installations 3G connections can get really expensive so satcom is a better choice. When choosing between a LEO and a VSAT connection, a trade off should be made between high speed and low latency provided at a high cost and with expensive and large equipment (in the case of VSAT) and lower speed with increased latencies provided at a lower cost with cheap and relatively small equipment. The architecture proposed which is heavily relying on the concepts of "fog computing" and "edge AI" eliminates the need for a high speed low latency connection to the cloud and thus allows for the use of low cost LEO satellites, with the added benefit of the reduced size and cost of the communication equipment.

In addition to satcom, GNSS timing is another space asset critical for the correct functioning of the system. Correct, synchronized timing is essential for aligning timestamps of sensor data acquired by different field deployed Gateways. 

Current Status

The project has started in June 2020 and to date the service and system architecture have been defined and requirements are in the process of being validated with selected end users. The Mid Term Review was conducted successfully and a proof of concept system is being built for validation in a real world setup.

Prime Contractor(s)

Status Date

Updated: 08 October 2020 - Created: 08 October 2020