A form of maintenance which uses a combination of certain devices and techniques to monitor the performance as well as condition of various machines to predict the requirement of maintenance could be a simple definition of predictive maintenance. This has been widely used in various kinds of industry since the 1990s. This system uses machine learning which falls under artificial intelligence ((KI) in German)) through the integration of Internet of Things (IOT) to achieve its results.

The various technologies in PDM:

• IR Thermography – Infrared Thermography is one of those technologies which is not prone to be destructive or intrusive. Personnel can make use of IR cameras to detect abnormal high temperatures in equipment which would indirectly point them toward any worn out or improperly functioning electrical circuits because these parts typically emit heat and are displayed as hotspots on thermal images. A quick identification of hotspots would help to find out any abnormal condition and prevent costly repairs and non-productive time in an industry. This is one of the most versatile forms of technologies used in PDM and is useful in the analysis of both individual parts of machines to even complete buildings and plant systems. This technology also finds its uses in the detection of thermal anomalies and faulty process systems which rely on the retention and/or transfer of heat.
• Acoustic Monitoring – Detection of leaks of gas and liquid on a sonic or ultrasonic level is possible with the use of acoustic technologies. Sonic technologies are cheaper than ultrasonic technologies. However, their uses are restrictive on mechanical equipment. Ultrasonic technology can be more dependably applied to the detection of mechanical issues because it converts sounds beyond the hearing capability of human beings into audible or visible signals. These sounds are exactly the frequencies of leaked valves, bearings which are not well-lubricated, malfunctioning electrical equipment and so on.
• Vibration analysis – The use of this technology enables a technician to analyse the vibrations of a machine by the help of real time sensors pre-installed in the machines or some handheld analyser. In this form of analysis, complete dependence on machine learning is not possible because it needs some skilled technician in the determination of faulty components after the comparison with known vibration patterns of parts when they are faulty. These parts generally include certain motor problems, misalignment unbalanced components and so on. Individuals trained in vibration analysis are a prerequisite in the use of this system of predictive maintenance because of the use of vibration patterns. This application is perhaps the best suited to the definition of PDM.
• Oil analysis – This form of technology in a PDM allows a technician to analyse the condition of the oil from the various tests which determine the viscosity, water content, count of particles, metals, and the acidic or basic nature. Since the primary test(s) would set a benchmark for a new machine, a correctly performed oil analysis would help make the PDM successful with its outcome(s).
• Other technologies – There are several technologies which are a part of PDM. These include motor condition analysis used to analyse the condition of the motors’ running and operating. Changing thickness of tube walls in boiler systems and centrifugal chillers is identified by the use of eddy current analysis. It should be noted that having several IOT devices which fall in the definition of PDM along with artificial intelligence and KI, however, does not reject the requirement of the human element in the whole process.

Conclusion:

Despite all the various things available in IOT, their integration is perhaps the most important thing in determining the success of the PDM. This would lead to better accuracy and performance.