Oil Detection Fault Tree Analysis Based on Improved Expert’s Own Weight–Aggregate Fuzzy Number

Abstract

Oil detection technology improves the reliability of machinery or equipment. The physical and chemical indicators of the fluid can reflect the cause of the failure in various aspects, which can prevent major accidents to the greatest extent by setting up a fault tree. Owing to the lack of data, it is difficult to accurately obtain the basic event probabilities, which makes it difficult to diagnose faults. The expert evaluation method and aggregated fuzzy numbers are used to exact the failure probability, where the event probability is evaluated as the subjective will of the expert. To improve the probabilistic accuracy, weights are improved by the combined assignment method as well as the reasonableness analysis. A fault tree diagnostic model is constructed for qualitative and quantitative analysis, taking the ship engine oil viscosity high fault as an example. According to the results, the model can provide a comprehensive analysis of physical and chemical indicators. Experts’ own weights have a large impact on the failure probability, with their weight changes leading to a change in the failure ranking. From the discrimination, following a Bland–Altman analysis of the results, the selected combined empowerment method improved the discrimination of the results by 4.8% compared to the traditional method, with 100% data consistency, which proved that the improvement was reliable and effective. The structure of this fault diagnosis model is clear, which can quickly give the fault cause and probability reference value.