Taking the typical friction pair cylinder liner piston system of internal combustion engine as the research object, the condition monitoring test bench of cylinder liner piston system of internal combustion engine is designed and built. Aiming at the shortcomings of the traditional maximum entropy method to analyze the wear particle monitoring data, an improved fractional moment maximum entropy method (FM-MEM) is proposed, and the key parameters are optimized by combining the carnivorous plant algorithm (CPA). The threshold division of the monitoring data of lubricating oil abrasive particles is carried out to evaluate the health status of the internal combustion engine. Then, the theory and experiment are combined to monitor the running state of the cylinder liner piston system of the internal combustion engine from three aspects: lubricating oil abrasive particles, physical and chemical indexes and surface morphology. The wear health status and the change trend of abrasive particle concentration in each time period of the cylinder liner piston system under low-speed condition are analyzed. Through the on-line abrasive particle monitoring test of the internal combustion engine, it is proved that this method can realize the on-line real-time condition monitoring of the cylinder liner piston system of the internal combustion engine.