W. Edwards Deming has famously said, “In God we trust, all others bring data” Nowadays, data is available in abundance. Especially in the wake of IoT and Industry 4.0, millions of machines operating in the industrial sector generate a lot of operational data in a time-series format. But like Ronald Coase said once, 

this huge operational data is only useful if it is analyzed well and squeezed out for every ounce of insight it has for the benefit of the industry.

Modern philosophies of asset maintenance and operations include concepts such as predictive and proactive approach that says 'fix it at the right time'. Even for defense operations, concepts like Condition-Based Maintenance Plus (CBM+) bring out the need for inserting technologies such as artificial intelligence to analyze the real-time asset condition data for accurate estimates of impending failures and enhancements in the logistical responses.

If you agree with me that the data tells stories and every asset has its own story, you will also agree to the fact that these 'data stories' could be laid out to map the behavior of each asset separately. Once you have mapped a usual or 'normal' behavior of the asset, you can identify the difference between success and failure i.e. you can observe the asset behavior to identify its 'non-normal' behavior. Asset Digital Fingerprinting uses time-series data that exists almost everywhere in the manufacturing industry and defense operations. Especially in defense operations and maintenance, it holds a good value as all major capital assets or platforms have to undergo maintenance or refit cycles after a fixed period. The beauty in using asset fingerprinting in defense operations is that it uses historical and real-time asset behavior not only to monitor the asset remotely but also to identify its behavior after any major overhaul/refit. Using state-of-the-art artificial intelligence and machine learning techniques, the temporal data associated with the asset and its sub-systems is exploited fully to understand the normal operating condition of the asset. In the usual case of predictive maintenance or remote asset monitoring, AI is used to identify anomalies and discover precursors to any adverse condition much ahead of the actual failure event. However, the same feature may be utilized for asset fingerprinting and analysis during the post-overhaul assessment of the asset. The maintainer here will be assisted by the time-series AI platform to identify the behavioral state of the asset before and after the overhaul. (Interesting read: Vision 2022: What's next for smart manufacturing?)

Recently, USS Gerald R. Ford, the largest aircraft carrier was tested for ship shocks during its sea trials to determine shock qualifications of the ship. Operational and combat situations were simulated by detonating ordnance at predetermined distances from the ship. The data gathered during such tests might be in the range of a few Terabytes to tens of Terabytes per test. This time-series data can be fed to AI to analyze it in near real-time and identify various anomalies in the thousands of sub-systems of the carrier, post shocks. This will centralize the efforts, get quicker results, and most importantly, will provide accurate assessment deep at the component level of each system.

During such sea trials for any ship, the overhauled ship's systems are put to test in the actual operating conditions at sea and their responses are recorded or observed by the operators. These observations are then compared with the normal parameters or limits to determine system responses. Even in military aviation, the air assets i.e. aircraft, helicopters have finite operating limits, mostly in terms of operational engine hours. This is usually termed calendar-based maintenance operations or preventive maintenance. Some of the Western-origin aircraft follow flex-ops maintenance concepts where each sub-system or component falls due for maintenance at its own independent schedule. You may imagine the complexities that the maintenance PMO has to face in planning maintenance and operational readiness. Once the aircraft is overhauled, its systems are tested first in a standalone mode using available STTE and test kits by the maintenance personnel. Then the systems are tested on the ground under light operating load to observe normal operating behaviors before being subjected to full power tests. Post confirmation of on-ground satisfactory test results, it is subjected to a Check Test Flight (CTF) where a qualified and experienced pilot tests the aircraft on the runway (Go/No Go) as well as in the actual flight. Most of these tests are based on observations, readings on the ground, and their validation against known 'normal' parameters. A similar process is performed during the production of a new commercial and military aircraft.

Asset fingerprinting using AI can help in reducing total test time to a great extent with accurate results. It doesn't mean that the need for SMEs is eliminated as their operational expertise is still needed, however, the digital behavior of the asset at a system level and also at components level will be evaluated in real or near real-time. AI applications can find patterns across multiple sources of data and map them against a known 'normal' pattern to discover anomalies. Such correlation and similarity calculation techniques can save not only cost but valuable operational time to deploy with increasing test accuracy.

Like it is said before, data needs to be squeezed out for every insight and every possible application. Asset digital fingerprinting allows the use of data for intelligent analysis through the asset lifecycle thereby saving time, cost, effort, and improving test accuracy multifold.