During the past decade, there has been a significant increase in the development, application and use of advanced diagnostics and artificial intelligence technology in field service. Research in the 1980s and 1990s showed that both remote and central service problem diagnostics could result in avoidance of between 30 and 35 percent of all on-site field service calls, and that in-depth diagnostic valuation could significantly reduce the number of "broken" field service calls through more intelligent dispatch and assignment of both parts and service engineers. Since that discovery, a great deal of work has been done to both develop and apply advanced diagnostics technology in the field service industry.
In traditional breakdown maintenance and emergency repair, the information used to diagnose the equipment or system fault is usually obtained after the event occurs. Preventive maintenance, regularly scheduled, provides significant protection from catastrophic equipment failure; however, such schedules might make costs unreasonable from actions such as parts replacement, overhauls and pulling equipment out of service too often.
Predictive maintenance is condition-based--service provided when the equipment's operating condition is shown to be deteriorating. Condition-monitoring methods are dependent upon the criticality of the equipment and its function, the ability to apply sensors to monitor remote or central conditions and to communicate that data and the information needed to make competent service decisions and predictions.
In predictive maintenance, remote/on-board or centralized condition-monitoring techniques are primarily used to assess the operating health of a particular system or piece of equipment on board. Condition-monitoring typically uses the following measures:
Vibration analysis (fast fourier spectrum analysis)
Electric current fluctuation
The predictive maintenance approach, using remote or on-board diagnostics, might lead to service being rendered when it is not necessary, increasing costs. The condition-monitoring approach, while incurring a cost to implement, could optimize the usage of service, providing maintenance and repair only when necessary to maintain the equipment's integrity. In general, preventive and predictive maintenance is best for mechanical and electromechanical equipment that fails gradually. Repair on demand works best for equipment that is either working or not working (as in electronic technology).
Work in advanced general diagnostics and artificial intelligence in the field service industry initially focused on decision techniques based upon either rule or heuristic models and processes. Within the past few years, there has been an extension to include self-organizing systems and processes.
Within explicit active central or general diagnostic advisory systems are more specific mechanisms, including:
General fault models
Deep casual models
Rule-based expert systems either involve flow charts, procedural rules or a decision-tree to diagnose a repair situation. The decision tree uses a tree-like diagram to illustrate the relationship of the machines or products to the sequence of customer complaints, symptoms, causes and corrective action recommendations. Reliability and maintainability engineers can carry out such analysis before any repair action is taken. Alternatively, a call handling, dispatch and call closeout process can be established that uses the same decision tree structure on an after-the-fact basis. This is easily done by linking information from incoming service calls by product or machine to complaints and symptoms, and then tying this information to a closeout call that ties the cause to the symptom and leads to the actions that ultimately result in a fix or repair.
Using problem-resolution diagnostics in central help desks and remote field service situations clearly shows the potential to improve the use of people and parts resources to increase service force productivity and efficiency.