Design And Realization Of Cooling Fan’s Prognostics And Health Management (PHM) System

The heat dissipation effect of electronic equipment is directly dependent on whether fanrunning state is in good condition since it serves as a main radiator. But most of the electronicdevices are in an unattended operation mode. It will be difficult for us to know the faultinformation in time if fan had a breakdown or failed to rotate. So, design and realization ofcooling fan’s prognostics and health management (PHM) system can inform us to give areplacement without delay, which is significant to maintain electronic equipment’s normalthermal dissipation.Paper first briefs the present situation of the research on prognostics and healthmanagement (PHM) system of cooling fan, then we find that failure detection is our studyfocus rather than fault prediction. In view of this, the thesis decides to put emphasis on fanfailure prognostic. We classified failure modes after looking up a lot of fan failure statisticsliterature. The dissertation points out that the gradual failure is the research priority and givesout the parameters that need to be monitored according to the structure and principles of fan.Many fault experiments show that the difference between parameter and its initial value canreflect the relevant fan fault. The factors that influence the fan operation state are divided intothree modules, including environment factor, mechanical factor and electronic factor. Analgorithm idea is put forward which says that a way of data fusion can get the final runningstatus after three modules has given their results, respectively.We have made a set of hardware device based on DSP which can acquire fan’sparameters correctly and give us failure prognostic information, and implemented the stateprediction algorithm using C language. A fan running state prediction result can be workedout after the data that was measured by the hardware device was processed through DSP. Themonitoring interface is designed by the language of VB which reads and writes data with database, and we can watch the fan prediction status in real time through it. There may be errorsin the result obtained by measuring device compare with the real situation, but the generaltrend is the same.