Try Search GlobalSpec or Google for basics

How can you say that it could be worse if I will be doing preventive maintenance

Thank you

Predictive methods for equipment or processes requires some understanding of the "new" or "rebuilt" condition and how that compares to the condition just before failure. There are several ways to specify the point where something should be done and they are based on historical performance under known conditions.

Consequently, for some things (like automobiles) there is enough known history to be able to use "rules of thumb" like changing your oil every 3000 miles. But with other machines there may be little or no historical information available except from the manufacturer. Generally you can ask them about preventative maintenance.

Machines or systems or processes that you build will require observations from the point that it is first placed into service until it fails. From that you will be able to build a database of information called "mean time to failure". However, if you modify some part to reduce frequent failures then you must begin a new database.

Thermographic scanning is useful for picking out hot spots without touching anything. Placing your hand on equipment that is running hot is also a good detection method if you don't burn yourself. The same is true of vibration except that the measuring equipment is different. In the case of vibration, excess noise may be your indicator if you know what it should sound like.

There is no software that knows how to do all of this. It takes experience or history to create the ability to predict failure or breakdown. If there is no data from the people that made this equipment, you will have to collect it and process it using statistical methods. If you want to talk about statistical methodology, please start a new thread.