Hello Yanco,
GusD is completely correct.
My suggestion, would be to create a specialized scenario, where you know exactly how the motor will be used. Collect all variable data not only of the machine in question, but its working environment as well. Example: if a motor is to sink "X" amount of thermal energy to the ambient air, the higher the ambient air temperature, the less thermal transfer will occure, possibly causing premature failure in the windings. In that scenario, it would be wise to program a PIC type circuit to measure the motor temp and compare it to ambient air temp. Of course employ the typical voltage and current measures. Vibration measure etc. What you choose to analyse fully depends on the operating conditions of the motor, its load and its environment. If you can cut down the amount of factors needed per application, you can determine a rate of failure more easily and employ a sensor type circuit to observe and warn of possible failure conditions.
So, first you must know the normal operation data of the specific motor. Then, you must know the working load and environment. From this data, you can construct an observation tool that could possibly determine what it is that you wish to know.
You chose an exceptionaly difficult task. I hope this helps you. Also keep in mind that this tool would only be useful in determining when, under what conditions, a certain motor should be changed or checked/repaired. It is purely ballpark.