A Prognosis Method Using Age-dependent Hidden Semi-Markov Model And A Systematic Maintenance Policy

Since the past decade, the quickly development of manufacture techniques have greatly increased the demand for efficient equipment maintenance management. Many new maintenance policies have be developed and condition based maintenance (CBM) is one of them. The main idea of CBM is to monitor the health of critical machine components and system almost continuously during operation and maintenance actions based on the assessed condition. If done correctly, CBM has the benefits such as reducing catastrophic failures, minimizing maintenance and logistical cost, maximizing system security and availability and improving platform reliability. A CBM system usually has four functional modules: feature extraction, diagnostics, prognostics and decision support.Health monitoring and prognostics of equipment is a basic requirement for CBM in many application domains where safety, reliability, and availability of systems are considered mission critical. Conducting successful prognosis, however, is more difficult than conducting fault diagnosis. A much broader range of asset health related data, especially those related to failures, must be collected. The asset health progression can then be possibly extracted from the congregated data, which has proved to be very challenging. In this paper, a large number of literatures that focus on the machinery prognostic have been reviewed. Generally, prognostic models can be divided into four categories: physical model, knowledge based model, data-driven model and combination model. Various relative techniques and algorithms have been categorized depending on what model they usually adopt. From the literature review work, advantages and disadvantages of these methodologies are been discussed and some increasing treads appeared in the research field of machinery prognostic have been summarized. Furthermore, the future research directions have been explored. Based on the analysis of the current existed prognosis techniques, the following work are finished in the proposed paper:(1)Proposed a hidden semi-Markov model (HSMM) based prognosis method for the prediction of equipment health. In order to characterize the deterioration of equipment, three types of aging factors that discount the probabilities of staying at current state while increasing the probabilities of transitions to less healthy states will be introduced. The assumption is that the probability of making transition to a less healthy state could increase with the age. The performances of the HSMMs with aging factors are compared by using historical data colleted from hydraulic pumps. With the equipment health prognosis, the behavior of the equipment condition can be predicted.(2) To solve the missing data problem, the paper proposed a forward-backward grey model to compute the estimator of missing data. Even more, an iterative algorithm combined with HSMM is developed to improve the precision of estimator. In the case study, statistical indicators like average error and MSE of error are used to evaluate the performance of the proposed missing data process method. The statistical results show that the grey method has a better performance on precision and stability.(3)In order to improve the possibility that the prognosis result could help optimize the maintenance policy in real application, the paper design a double layer dynamic programming model for the maintenance policy decision making. The objective function is comprised with maintenance total cost and equipment available rate. Besides the tradition maintenance fees such as operation fee and maintenance fee, spare part inventory cost is creatively introduced into the total maintenance cost function. The goal of optimization is minimize objective function. The renew of operation fee, maintenance fee and spare part inventory fee based on the maintenance action are according to the health state transition matrix from previous proposed HSMM prognosis method.This research is an important part in the CBM maintenance field. The 3 part of research contact closely with each other and could be combined together to be a systematic scheme for the CBM maintenance. It develops the data pre-process techniques about the missing data and the RUL prognosis method for equipment health monitoring and predict. It also provides a decision support for maintenance scheduling of production systems which using the information from the prognosis model. The proposed maintenance decision theory with consideration of spare part inventory cost,which is an extension of CBM intelligent maintenance, can be a useful guidance for the manufacturing industry to organize the maintenance activities of production systems.