Artificial Intelligence Applications to Maintenance

Abstract : The maintenance of modern military systems employs a variety of automation. Built-In-Test provides on-line fault detection and some isolation, Automatic Test Equipment is indispensable at intermediate and depot repair stations, and automated maintenance aids and trainers abound. These developments were designed to speed maintenance and to compensate for declining skill levels in the maintenance force. They are currently far from satisfactory. Modern maintenance is characterized by excessive false alarms and unnecessary removals at all levels of maintenance. The results of these deficiencies are long maintenance times, resources wasted in unnecessary for inefficient maintenance actions, and systems out of action which need not be. Correcting these problems would therefore provide both an economic advantage and a force multiplier. To create quantum improvements in maintenance will require the application of radical changes to the technology. One possibility is the application of Artificial Intelligence (AI) techniques to maintenance. AI is beginning to see application to practical problems in many disciplines, and hence is potentially capable of relatively rapid implementation into military systems. At present, DoD efforts in applying AI to maintenance are small and exploratory. The task of the Artificial Intelligence Applications committee was to examine the opportunities for applying AI to maintenance, assess the costs, risks, and development time required, and provide recommendations to the DoD for action.