Abstract
In reciprocating compressors, unexpected accidents have been reported under long-term operation. Although these machines are managed by monitoring and regular maintenance, it is difficult to predict a sudden accident or a damaged part far from preventing a fatal trouble in advance. The purpose of this study is to establish a new monitoring technique to detect the state of sliding and connecting parts, where it is known that some failure often occurs. In this paper, we examined the feasibility of estimation for the state of the sliding parts in a reciprocating compressor using both the vibration characteristics in the experiment for the small apparatus and in the eigenvalue analysis based on the mathematical model. To estimate the state of the sliding portion, each spring constant is identified solving an optimum problem with constraint. Moreover, we introduce an individual parameter identification method, where assuming that the initial value for each spring constant is larger, parameter identification is carried out. As a result, it is found that we can estimate the position that has undergone changes in the rigidity of the sliding parts comparing each error function for the individual parameter identification.
