Reciprocating compressors are key equipments in many industries. The results of the faults of the reciprocating compressors will be disastrous. It is necessary for reciprocating compressors to build up a condition monitoring system. This thesis introduced a design of condition monitoring system for reciprocating compressors and a diagnosing method for non-linear system based on system parameter identification. The methods and processes to implement the system are suitable for many other applications in reciprocating machine. First of all, after the introduction of the principle of reciprocating compressors, this thesis analyses the methods that represent the condition of reciprocating compressors and introduces the means of monitoring typical faults of reciprocating compressors. Secondly, two different ways, namely, on-line and off-line ways that are adopted to implement the condition monitoring system are compared. According to the practice, the latter is adopted. The condition monitoring system for reciprocating compressors is built up after selecting variables and samples. Thirdly, the software of the condition monitoring system is discussed. In order to accomplish high-speed data acquisition and real time analysis, we realized a portable off-line compressor analysis instrument based on embedded system. Fourthly, in order to deal reciprocating system’s great impulse phenomenon, we built a non-linear model and designed a software to deal acquired data . By establishing auto-regressive model (TAR) and bilinear model (BM), we can identify system’s parameters, and process failure diagnosing. By analysis acquired and simulated data, testified the diagnosing method. Finally, the merit and modified method of the system was discussed, and several suggestions were given for further research work in the future.
↧