A formal framework for predictive maintenance

This paper begins the process of specifying a particular predictive maintenance (PM) software system using formal methods. Specifically, schema diagrams in formal Z notation are used to specify an integrated set of black box PM operations devoid of implementation details. The schema diagrams describe the input-output behavior of the PM operations and employ high-level abstract mathematical data types. In total, the PM operations serve as a framework upon which to base the implementation of a provably correct PM software system. This highly-assured system would add substantial value to automatic test equipment and test program sets by providing the capability to 1. detect anomalies and incipient failures and isolate existing/impending failures of a unit-under-test (UUT) based on system-level data, 2. forecast remaining useful life of a UUT based on system-level data, 3. reduce substantially the number of false alarms attributed to automatic test equipment and test program sets, and 4. Enable condition-based and prognostic-based maintenance and inventory process decision triggers based on the outcomes of 1-3. During implementation of the PM software system, Floyd-Hoare logics can be used to verify whether particular algorithms meet the finished black-box PM specifications. This paper also presents the state transition diagram that corresponds to the framework.