A hybrid pso-svm based method for degradation process prediction of reciprocating seal

Abstract

Reciprocating machines such as piston pumps, compressors and internal combustion engines are widely used in a wide range of manufacturing industries including automotive, aerospace, paper, oil and gas, etc. Reciprocating seal located directly on the rod/piston of a reciprocating equipment is used for preventing leakage and reducing wear between two parts that are in relative motion. Seal failure is one of the foremost causes of breakdown of reciprocating machinery and such a failure can be catastrophic, resulting in costly downtime and large expenses. Assessment of reciprocating seal is extremely important in the manufacturing industry to avoid fatal breakdown of reciprocating equipment and machines. Prediction of time series using predictive maintenance practices and tools to estimate the evolution of the future condition of the system is of great interest to the operators for taking timely and appropriate maintenance decisions. In this paper, we have built and trained a hybrid PSO-SVM model to predict the degradation process of reciprocating seals. Particle swarm optimization is used to optimize the penalty factor and kernel parameter of SVM model. Controlled experiments are designed and performed, and data collected from a dedicated experimental set-up is used to validate the proposed approach.