Pre-emptive strikes on military equipment faults
decisionLab developing Artificial Intelligence and machine learning to analyse and predict the condition of military assets and equipment
Challenge
The Ministry of Defence (MOD) is experiencing a significant growth in the number of sensor platforms it employs. The sheer volume and complexity of data from these sensors and other information sources is a real challenge for operators and military decision makers. For example, over the course of a day, a Type 45 Destroyer can produce up to 10 million lines of Data. If MOD is to get the most out of their people and systems, it needs to continually increase an individual’s ability to analyse and utilise such data to make informed, effective and timely decisions.
Solution
Under a Defence and Security Accelerator (DASA) led themed competition to ‘Revolutionise the human information relationship for Defence’, decisionLab submitted a proposal to develop Artificial Intelligence and machine learning as a prediction tool for potential component failure.
This resulted in DASA funding of £524,651. decisionLab’s capability uses Artificial Intelligence to analyse and predict the condition of military assets and equipment, reliably predicting potential component failure. This intelligence allows the Military to take pre-emptive action to replace or fix a component quickly and efficiently. This could potentially save MOD millions of pounds in terms of reduced maintenance and improved scheduling.
Ian Griffiths, Chief Strategist, decisionLab said:
DASA funding has transformed our business. It has enabled us to grow by 50% and it has provided a reputational boost which has brought further business investment meaning we can build an effective data science capability that now has potential in many sectors.
Originally focused on aircraft, the same technology has now been applied to use on board one of the most capable and complex of all UK warships, the Royal Navy’s Type 45 Destroyer. The Type 45 is fitted with thousands of sensors, providing a wealth of big data on which to trial this capability. Working closely with the Royal Navy, Programme Nelson and HMS Diamond Marine Engineers, decisionLab conducted a series of workshops to collaboratively develop new machine learning and anomaly detection models to calculate a system health score based solely on sensor data. A user friendly interface then allows Marine Engineers to interact with the AI model and feedback the actual system state, enabling the model to learn. The AI capability has been developed rapidly and has been deployed onboard HMS Defender for trials.
Chris Smith, Navy IW Artificial Intelligence and Data, Programme NELSON, Royal Navy stated:
The data engineering and analysis developed in response to the Type 45 dataset have opened up the possibility of expanding the approach to more naval systems.
Benefits
The capability improves maintenance planning by predicting failing components ahead of time so that action can be taken. This has potentially multiple benefits:
- Reducing costs and improving platform availability
- Enabling a better understanding of asset degradation (linking cause and effect) which can allow changes to usage that reduces degradation or improved risk management and improved longterm failure prediction
Next Step
decisionLab is looking for opportunities to meet potential collaborators and will be continuing to push exploitation in other sectors.
Contact
Ian Griffiths: ian.griffiths@decisionlab.co.uk