Research On Fault Diagnosis Method Of Linear Transmission Mechanism Of Ship Storage And Supply System Based On Data-driven

Ship equipment manufacturing industry is an important industry related to the development of national economy.Any delay in the ship mission system may cause disruptive consequences.Therefore,it is urgent to improve the support capability by advanced technology represented by fault prediction and health management.With the improvement of equipment task requirements,coupled with the rapid development of modern sensing and detection technology,multi-source information fusion,signal analysis and processing and computer technology,intelligent maintenance,fault prediction and health management are widely used in ship equipment to achieve ’ active maintenance ’.This paper mainly studies the fault diagnosis method of linear transmission mechanism of ship storage and supply system,which is mainly divided into the following four parts.The structure of the key components of the ship storage and supply system is analyzed and studied,and the weak links in the storage and supply system are determined.Then,the typical failure modes of the weak links in the availability of key components are found out by using the failure mode and impact analysis.Then,the operating conditions are determined according to the actual application scenarios,and the construction of the test bench is completed.On this basis,the deployment of the key components measuring point layout scheme of the linear transmission mechanism of the storage and supply system is completed.Aiming at the difficulty of extracting non-stationary signal features obtained from the practical application scenarios of two linear transmission mechanisms of rack and pinion drives and ball screw pair,two data preprocessing processes of autonomous segmentation and adaptive down-sampling are firstly carried out to realize the standardization of multisource heterogeneous data.Then,the time domain,frequency domain and fuzzy entropy features of the standardized data are extracted.Finally,based on the local binary pattern theory,multi-scale analysis is fused to extract more features.In order to solve the difficulty of fault diagnosis of rack and pinion drives under lowspeed and speed fluctuation conditions,a fault diagnosis algorithm based on sparse representation and Ada Boost is designed.Firstly,the obtained multi-scale local binary pattern features are sparsely represented to construct a fault feature dictionary and perform dictionary learning.Then,the samples are sparsely represented under the fault feature dictionary to obtain sparse vectors.Then,based on Ada Boost theory,a strong classifier composed of multiple weak classifiers is trained by sparse vectors.Finally,the sparse vector of the test sample is input into the strong classifier to obtain the fault pattern.For the problem of large deviation of evaluation results in different application scenarios,distance-based and probability similarity-based methods are used to design performance evaluation algorithms respectively.Based on the evaluation of distance type,firstly,manifold learning is carried out based on the extracted multi-source features,then the manifold distance is measured in the feature space,and finally the manifold distance is transformed into confidence value to obtain health degree.Based on the probabilistic similarity evaluation,the signal features are first sparsely represented,and then the obtained sparse representation coefficient vector elements are matched with the corresponding subjective evaluation scores,and finally the health degree is accumulated.Finally,the effect of each part of the algorithm is verified by the ball screw test bench and the rack and pinion drives test bench.The actual data proves the feasibility and superiority of the fault diagnosis and performance evaluation algorithm proposed in this paper,which provides the theoretical and technical support for the condition-based maintenance under the real-time monitoring state,and has strong engineering application value.