Initiative Defines Requirements for Applying Predictive Maintenance Capabilities to Autonomous Vehicles

In recent years, maintenance and sustainment professionals have increasingly been advancing predictive maintenance capabilities to catch and solve problems before they cause catastrophic failures. As the software and hardware used in vehicle systems becomes more complex, reactive vehicle maintenance becomes more challenging and expensive, necessitating increased investment in improving proactive maintenance.

Recently, the CTMA Program launched an initiative designed to boost predictive maintenance in the commercial and defense ground vehicle sectors. The initiative, Advanced Diagnostics and Prognostics Systems for Autonomous Vehicles, includes the efforts of the US Army and two industry partners: Global Strategic Solutions and HII Unmanned Systems (HII). The team is developing a state-of-the-art integrated vehicle health management (IVHM) system that will incorporate advanced diagnostics and prognostics capabilities into emerging and future US Army robotic and semi-autonomous ground vehicles.

Manned and unmanned vehicles that travel for long durations in remote areas face several challenges, including wear and tear on mechanical parts and electromechanical components, along with the chance of unexpected faults and a high risk of catastrophic failure while deployed. An effective IVHM system will be able to monitor the status of such vehicles and analyze future performance. Design goals include capabilities such as identifying faults, performing real-time prediction of reliability and remaining useful life (RUL), and gathering data to improve future vehicle engineering. If achieved, these capabilities will lower vehicle maintenance and sustainment costs, increase asset availability, accelerate repair timelines, and reduce vehicle failures.

To accomplish these objectives, the group is working on five main deliverables. First, they are documenting a process that characterizes health-ready subsystems and system-level IVHM architectures for ground combat vehicles and ground robotic vehicles. The team is specifically focusing on a Bradley-type vehicle that can be operated in a semi-autonomous mode or manned with a crew. Second, they are creating IVHM system specifications for vehicles and subsystems that will include the capability levels for selected vehicles per IVHM standards developed by the Society of Automotive Engineers (SAE). Third, they are conducting use case analyses on several vehicle subsystems: the insulated-gate bipolar transistor (IGBT); the diesel engine (cylinder/valve); the solid-state power distribution (SSPD) and supported line replaceable units (LRUs); and the coolant loop subsystem’s valves and appropriate controls.

Fourth, they are designing a framework called Self-Adaptive, Predictive Prognostics and Health Optimization System (SAPPHOS) that will deliver prognostics and health monitoring capabilities for vehicles. This model fuses two prognosis methods. First, the system is model-based; it is grounded in a mathematical physics-based model of the system and incorporates expert advice for the development of a hard-coded, rule-based representation. Second, the model is data-driven, using historical data to learn a model of system/subsystem behavior and an analysis of data feeds from operational assets. The team has conducted analyses on several datasets on varying engines and fault modes to develop useful health index and remaining useful life values. Additionally, they have evaluated autoencoder neural network and statistical analysis methodologies, concluding that the best approach is to combine statistical and advanced machine learning methodologies. The SAPPHOS framework contains monitoring, fault detection, and diagnosis, all within a prognosis-oriented framework that aligns with the goals of CBM+. Finally, the team is developing a plan of potential future initiatives that will outline next steps and possibilities for technology evolution.

“The principles behind SAPPHOS can be applied to predictive maintenance needs across manned and unmanned platforms,” said Dave Summer, Program Manager, Unmanned Systems at HII. “For the Army and other DOD services, this has the potential to benefit more than just ground vehicles—the techniques outlined through this project can be applied to platforms across domains, analyzing the data and identifying trends that inform predictive maintenance models.”

The collaboration is working toward achieving SAE’s IVHM Capability Level 5, Self-Adaptive Health Management. This capability level, which is critical to robotic vehicle applications, integrates the vehicle health management capability with vehicle control functions to provide autonomous, real-time, self-adaptive control and optimization to extend vehicle operation and enhance mission completion and safety.

“This project can act as the foundation for the development of applications to implement the Army Optionally Manned Fighting Vehicle strategy to capitalize on prognostics/predictive maintenance (PPMx) capabilities and leverage artificial intelligence,” said Summer. “These capabilities can be used to monitor and manage at both the enterprise fleet level and individual combat vehicle level to determine the optimum maintenance cycles, diagnose and predict equipment health, optimize asset utilization, and improve performance. PPMx is an operational tool for commanders to make dynamic decisions based on near real-time combat platform readiness conditions.”

The project’s framework will also serve as a road map for the commercial industry by providing systems to optimize vehicle maintenance data. The initiative’s replicable, standardized framework will enable commercial industries—especially ground transportation, oil, gas, and power companies—to use this project’s methodology for monitoring system health, identifying system faults, and performing real-time prediction of reliability and remaining useful life of electronics-rich systems. As a result, commercial industry will be able to produce more reliable and more easily maintainable manned and autonomous vehicles. The process will be particularly applicable and timely for the next generation of electrical air and ground vehicles that will have a higher degree of autonomy.