Life extension module for machinery diagnostic systems

Machinery diagnostic systems detect and monitor faults but are unable to predict remaining life. The RiskMAN^TM process developed as one of the modules of the Neural Network System Health Usage and Monitoring System (NNS/HUMS), developed under JAHUMS contract administered by NSWC/CD, provides the means for converting the outputs of a diagnostic system into parameters usable in a set of life extension mathematical predictors. Loads and fault severity indicators are incorporated into a risk probability of survival [P_s(t)] equation which varies over time as the machinery loads and faults vary. A measure of the machinery reliability [R(t)] is employed in the derivation of P_s(t) and in modified forms as bases of comparison for P_s(t). The lower confidence limit on R(t) and a safety level R(t) are used in conjunction with P_s(t) to indicate alarm conditions. RiskMAM^TM techniques and mathematical models have been developed as put of the NNS/HUMS real time diagnostic system for Navy and Army military aircraft (Giuntini, 1997). The NNS/HUMS is a critical link in the transition from a safe life maintenance (SLM) philosophy to a condition-based maintenance (CBM) philosophy. RiskMAN^TM receives the output parameters from the neural network module which are load state, fault class, fault severity, fault probability and severity probabilities. Then it converts these parameters into meaningful information that will detect an imminent failure or will enable life of the system to be extended beyond its normal/typical overhaul time. Though being developed for military aircraft applications, RiskMAN^TM is also applicable to general industrial machinery