N22A-T026 TITLE: Low-Cost, Low-Power Vibration Monitoring and Novelty Detection

OUSD (R&E) MODERNIZATION PRIORITY: Artificial Intelligence (AI)/Machine Learning (ML);Autonomy;Microelectronics

TECHNOLOGY AREA(S): Electronics;Ground / Sea Vehicles;Materials / Processes

OBJECTIVE: Develop a device to bring the benefits of machine health and usage monitoring to a broad spectrum of Navy and Marine Corps assets, especially those of lower value that cannot afford full-up Health, Usage, and Monitoring System (HUMS) systems by developing powerful, inexpensive processing hardware at a target price of less than $100.00 per node.

DESCRIPTION: Lower cost USN/USMC platforms (especially land systems) cannot afford conventional HUMS sensors/processors typically priced at over $1,000.00 per node. Direct sensing of relevant features and the extraction of "actionable" information may be accomplished by purpose-built signal processing hardware. On-chip integration of neural networks (trained offline) holds the promise for self-contained smart sensors that are both extremely powerful and affordable for all platforms. This capability is vital for those platforms deployed and operating at (or beyond) the tactical edge. Very high risk with extremely high payoff is possible if successful. The envisioned device (or family of devices) is expected to be self-contained in a rugged package able to be permanently installed on vehicle components.

This STTR topic seeks innovation in the development of onboard analytics (e.g., neural networks) that operate at the component level and are able to detect and identify anomalous signatures. State of the art is to attach sensors to the component and wire them to conventional signal conditioning hardware in data acquisition components. Digital Signal Processing (DSP) and other computations are done to convert the raw sensor values into information on centralized processors. Some sensors are directly connected to serial buses on the platform with analog-to-digital (A/D) inside the sensor package. The intent is to push the processing into the sensor package, leveraging integration of neural networks and other Artificial Intelligence/Machine Learning (AI/ML) tools at the chip scale to combine the data acquisition and health determination into a single, low-cost device.

PHASE I: Define and develop a concept for a compact device able to monitor, detect, and identify symptoms of failure on typical rotating mechanical equipment. Vibration, temperature, and electrical current signature are typical measurands of interest. The device should be inobtrusive in size and rugged to the ground vehicle�s under-hood environment. Approximately 1 cubic inch volume and less than $100 unit cost. The intent is for the device to be self-contained conducting measurement, analysis, and communications within the package. Ideally it should be environmentally powered or contain energy storage capable of design operation for 1 to 3 years. It should support wired (e.g., CAN bus) or wireless (e.g., IEEE 1451) communications. Perform modeling and simulation to provide an initial assessment of the concept and exercise alternatives. Develop a Phase II plan.

PHASE II: Develop a Phase II prototype for evaluation based on the results of Phase I. The prototype will be evaluated to determine its capability to meet the performance goals defined in the Phase II Statement of Work (SOW) and the Naval need for detection and diagnosis of typical faults in military ground vehicles. In production, the device will be a part of an integrated system of similar devices monitoring different symptoms of faults on a single machine, other similar devices on other machines, and additional control system parametric data captured from existing onboard buses or traditional sensors. The intent is to detect early stage faults at a component level and merge the information to understand the impact of the faults on the mission capability of the platform. Conduct further evaluation of the feasibility of the prototype to evolve into a hardened device capable of surviving in the target environment, meeting required cost targets, and performing the necessary analytics. The device should support other third party analytics as well as provide native analytic capability. A family of devices with different processing, memory, and sensing capacity for different applications is anticipated. Testing will be performed on laboratory equipment at the proposer's facility to demonstrate performance. Cybersecurity is a key attribute; "cyber-invisible" is the goal. Formal approval is not to be sought during Phase II, but the design must consider the cyber environment from the outset and incorporate the ability to be properly secured when produced.

PHASE III DUAL USE APPLICATIONS: The technology developed in this effort is intended to comprise a part of an onboard, health monitoring and processing system providing Autonomic Readiness Management (ARM) applicable to all types of naval vehicles. The ARV acquisition program is an ideal target for a rapid maturation and integration into the production process. The FFG-62 Mission Readiness Support System (MRSS) is another acquisition program with need for CBM+ and ARM to which this device could apply.

Commercial uses of the device are everywhere. Interest in condition monitoring for all classes of vehicles is high and lack of an affordable implementation has limited the deployment of the capability. The device developed here is an inherent member of the Internet of Things (IoT) and could be adapted to a variety of applications beyond condition monitoring for vehicles. The fundamental capability to measure, monitor, detect, and project are capabilities that have broad applications across the IoT.

Specific commercial industries/markets that could use and benefit from the technology include: commercial trucking, heavy construction equipment, manufacturing, aircraft and related equipment, commercial maritime, and infrastructure monitoring (e.g., bridges, locks, damns).

REFERENCES:

1. Liobe, J.; Fiscella, M.; Moule, E.; Balon, M.; Bocko, M. and Ignjatovic, Z. "DS Sentry: an acquisition ASIC for smart, micro-power sensing applications." Proceedings Detection and Sensing of Mines, Explosive Objects, and Obscured Targets XVI, Vol. 8017, 2011, p. 80170H. https://www.spiedigitallibrary.org/conference-proceedings-of-spie/8017/80170H/DS-Sentry--an-acquisition-ASIC-for-smart-micro-power/10.1117/12.884253.short?SSO=1.
2. Liobe, J.; Ignjatovic, Z. and Bocko, M. "Ultra-low overhead signal acquisition circuit for capacitive and piezoelectric sensors." 2008 51st Midwest Symposium on Circuits and Systems, August 2008, pp. 33-36. https://ieeexplore.ieee.org/document/4616729.
3. Japkowicz, N.; Myers, C. and Gluck, M. "A novelty detection approach to classification." IJCAI'95: Proceedings of the 14th International Joint Conference on Artificial Intelligence, Volume 1, August 1995, pp. 518�523. https://dl.acm.org/doi/10.5555/1625855.1625923.

KEYWORDS: Condition-based maintenance; CBM; Internet of Things; IoT; neural network chips; wireless sensors; integrated processing; anomaly detection