The existence of the failure transitivity of machine tool components makes the fault transfer probability of components demonstrate dynamic time-variability, which affects the importance of components and further affects the machine maintenance cycle. Therefore, studying fault transfer probability and the importance of machine tool components is necessary. In this article, the fault transfer probability of component is defined according to component fault propagation directed graph and component independent fault and related fault model based on time correlation. Assuming that the fault propagation follows the Markov process, the improved LeaderRank algorithm is applied to evaluate the importance of components by introducing background node and calculating failure impact degree of component on the basis of out-degree. Finally, the specific application is verified by taking the fault information of a certain type of machine as an example.

Identifying the key components of CNC lathe and analyzing the fault propagation behavior is a powerful guarantee for the fault diagnosis and health maintenance of CNC lathe. The traditional key component identification studies are mostly based on the feature parameter evaluation of the fault propagation model, disregarding the dynamics and influence of fault propagation. Therefore, this paper proposes a key component identification method based on the dynamic influence of fault propagation. Based on the CNC lathe architecture and fault data, the cascaded faults are analyzed. The improved Floyd algorithm is used to iterate and transform the direct correlation matrix expressing the cascaded fault information, and the fault propagation structure model of each component is constructed. The coupling degree function is introduced to calculate the dynamic impact degree between components, and the dynamic fault propagation rate of each component is calculated with the dynamic fault rate model. Based on this, the dynamic influence value of fault propagation is obtained by using the improved ASP algorithm. The key components of the system are identified by synthesizing the fault propagation structure model and the dynamic influence value of fault propagation. Taking a certain type of CNC lathe as an example, the proposed method is verified to be scientific and effective by comparing with the traditional identification method of key components based on fault propagation intensity.

No abstract

Aiming at demands based on abnormal condition detection and warning in welding machine monitoring system of IOT, put forward a burst detection algorithm for multi-population firefly, achieve optimization selection and configuration of sliding window size and improve the processing speed and detection performance of burst detection model through collaborative work of different species of fireflies. Simulation results show that burst detection algorithm based on multi-population firefly has less processing time and higher accuracy and recall rate than the traditional burst detection algorithm under the condition of same burst probability or maximum sliding window size.