Research On Key Technologies Of Maintenance Management For Military Information Equipment Based On Condition Monitoring

Maintenance is an important way to guarantee the support effectiveness of military information equipment during its whole life cycle.Maintenance for military information equipment mainly adopts time-scheduled manner at present,which easily results in over-or under-maintenance and has a high maintenance costs.With in-depth advancing of military modernization,the dramatic increase of the variety and quantity of military information equipment,together with its technologyintensive and complicated construction characteristics,cause maintenance work becoming more demanding and difficult.How to maintain and restore good technical condition with least maintenance costs is an important issue for maintenance management of military information equipment.Condition-Based Maintenance(CBM),based on condition monitoring,is appropriate for precise maintenance actions through obtaining maintenance requirement determined by condition assessment and fault prognosis,reducing downtime and maintenance costs effectively.As an advanced maintenance theory,CBM is one of research directions of military equipment maintenance.This thesis discusses application of CBM for military information equipment maintenance with emphasis on key technologies related to its procedure.The main contents and contributions of the thesis are summarized below.Firstly,according to the concept of CBM,a maintenance management model based on condition monitoring for military information equipment named is developed,ranging from concepts to framework,data modeling and process modeling.is an extension of CBM,not only including three key steps,i.e.condition monitoring,condition assessment and fault prognosis,but also taking maintenance decision-making support into account so as to provide essential supports to maintenance management of military information equipment.By comparison with OSA-CBM,we argue that is reasonable and more specific.The is the principle foundation of our following researches.Secondly,taking network cryptographic equipment as an example,the process of building monitoring index system is explained.Through analyzing the structure and characteristics of network cryptographic equipment,its technical condition can be evaluated by encrypting/decrypting speed and network performance.To achieve a fine trade-off between information consistency and communication cost during monitoring data transferring,a modified push and pull cooperative algorithm named MP&P is proposed,switching push and pull model intelligently.The experiments show MP&P has greater robustness than other similar algorithms.Thirdly,a fault classifier with confidence level based on Support Vector Machine,called-for condition assessement is presented.The classifier can distinguish the normal condition and fault condition,and sequentially uses the distance between the data point indicating unknown condition mapped in feature space and the optimal separating hyperplane as a baseline to evaluate the approximation of unknown condition to fault condition by fault probability.To enhance the performance of-,an adaptive differential evolution algorithm with two-mutation strategy called is designed for parameter optimization,which is a modification to MGBDE in BBDE family.Inspired by adaptive differential evolution,it chooses the preferred mutation strategy for each individual in population with fitness-based so that increases the probability to find out the best solution.Additionally,in order to decrease the risk of falling into local optimum,it brings in a well-designed perturbation mechanism focusing on the situation where search is trending to stagnate.The statistical analysis shows that is a competitive evolutionary algorithm.Lastly,regarding fault probability prediction as time series analysis,a fault probability predicting method based on Generalized Regression Neural Network named-is proposed,which trains the network using historical fault probability data.Aiming at giving maintenance interval and maintenance actions for maintenance program,a method named-is proposed,which determines the maintenance interval in the criterion of maintenance costs minimum per unit time under the assumption that maintenance costs is the total of precautionary maintenance costs and corrective maintenance costs only.Then,the suitable maintenance actions are suggested by the result of fault situation and maintenance interval.