Research On Preventive Maintenance Strategy For NC Equipment Based On Multi-failure Mechanism And Accuracy Degradation Analysis

As the core productivity of modern manufacturing, the numerical control (NC) equipment has been playing a key role in all of the military, aviation, aerospace, shipbuilding, energy, transportation, metallurgy, machinery enterprises. The NC equipment is a mechatronic products that integrates themechanical, electrical, hydraulic, pneumatic, microelectronics, information and many other technologies. Considering the characters of small amount, complex structure, variable degradation and numerous failure, it is a great challenge to apply the preventive maintenance (PM) to NC equipment. This dissertation makes efforts to set up an effective PM strategy for NC equipment based on multi-failure mechanism and accuracy degradation analysis. Firstly, FMECA is adopted to identify the numerous failure modes. The graph model is then proposed to represent the complicated relationship between failures. On the other hand, the accuracy of NC equipment undergoes the explicit degradation process. Thus, the accuracy degradation pattern is analyzed and the future degradation and global failure time are then predicted. The identification of failure mechanisms and accuracy degradation pattern helps to derive the optimal PM strategy.This dissertation provides supports for PM scheduling optimization from the perspectives of both multi-failure mechanism representation and degradation-based failure time prediction, including the extended FMECA based failure mode identification and failure mechanism representation, accuracy degradation analysis and failure time prediction, PM scheduling under multiple failure modes. Specifically, the contents includes:(1) The principle of preventive maintenance for NC equipment is defined and its key technology consists of four aspects:failure diagnosis, degradation modelling, failure time prediction and PM scheduling optimization. The art of the state of the relative methodologies is reviewed in detail. Consequently, the challenge of implementing the preventive maintenance to NC equipment is revealed. Then, the framework of preventive maintenance strategy is proposed.(2) The failure and failure mode are specifically re-defined. Considering the weaknesses of traditional FMECA, the extended FMECA is proposed to identify all the failure modes of NC equipment, including failure severity, detectability, criticality, category, risk, failure time distribution and cause failures. Furthermore, the failure propagation graph model is built to depict the relationship between failures. The graph can be represented by matrices so that the graph updating algorithm and fault dianogsis process are easily realized.(3) The accuracy degradation pattern is identified with the degradation measurement sequence. Firstly, the accuracy items are divied into two groups:non-repetitive accuracy and repetitive accuracy. For non-repetitive accuracy degradation, the term of’local degradation pattern’ is introduced which can be expressed by exponential function. The fitted local degradation curves are treated as samples of global degradation pattern.These curves will produce the future degradation samples and also the lifetime samples. With these samples, the parametric empirical Bayes method is used to estimate both the priori and posterior distubition which will used to predict the future degradation and global lifetime distributions. On the other hand, the repetitive accuracy degradation is treated as integration of the mean and variance degradation processes. For either degradation process, a wiener process with drift is built. The drifts in both models is considered to be from some bivariate normal distribution. As a result, the double-weiner process based repetitive accuracy degradation model is built. EM algorithm is adopted for parameter estimation where the particle filter method is used in E-step to derive the expectation of latent variables. With the established model, the degradation and lifetime distributions are derived using Mento-Carlo method.(4) The proposed PM scheduling takes into accout the key failure modes of NC equipment. According to the equipment’s structure and functions, the key failure modes are identified which are then categorized into degradation-type failure and function-type failure. After analyzing the maintenance processes of NC equipment, the assumptions are reasonably made for preventive maintenance, corrective maintenance, mediate maintenance and major maintenance are. The cost function is formulated including the physical cost, soft-failure cost, ourage cost and fix cost. The general PM scheduling is built with the minimal long-term cost rate. To overcome the superfluous maintenance of the general PM scheduling, the improved PM solution is proposed by introducing two Boolean matrices. These two matrices are used to judge whether each failure is maintained at each mediate maintenance. The cost function is adjusted with the Boolean matrices and then the improved preventive maintenance scheduling is realized by the GA based approach.(5) The failure can never be eliminated. When unexpected failure occurs, the symptom-driven failure propagation graph (FPG) will be built with the extended FMECA information. According to the FPG, the probabilities of all the cause failures are calculated and then the risk of the failed NC equipment can be derived. Furthermore, the risk based binary decision tree is proposed to determine the failure ascertainment order. According to the tree, each ascertainment task will bring the maximal risk reduction so that all the cause failures can be rapidly ascertained.(6) The preventive maintenance scheduling based on multi-failure mechanism and accuracy degradation is applied to the RAM feed subsystem of FB260boring machine. Comparasion study is made to validate the proposed approaches.Finally, a summary of the entire dissertation is proposed and the future research directions are pointed out as well.