In recent years, deep learning (DL)-based fault diagnosis methods have demonstrated significant success in various domains due to their high accuracy. Similar to other data-driven techniques, DL models are application-specific and strongly depend on the data which they are developed upon. Different DL models need to be designed for different tasks. In addition, manual-tuning of the DL structures and associated parameters as part of the model design is a trial-and-error optimization process, which is time-consuming and sensitive to changes. To address these limitations, this article presents a reinforcement learning (RL) and neural architecture search (NAS)-based automatic modeling framework (RL-NAS) for fault diagnosis of machinery. The RL-NAS method can adaptively and automatically design high-accuracy network models according to the intended diagnosis tasks. A weight-sharing mechanism has been developed to alleviate the bottleneck of NAS where the computation time increases exponentially with the expansion of the search space and the deepening of the model structure. To evaluate its effectiveness, efficiency, and reproducibility, the proposed framework is validated on five datasets from different institutes, spanning the applications of rolling element bearings, gears, and ball screws. The results show that for all five applications, the RL-NAS method successfully searched out diagnostics models with over 97% accuracy within a short time, from 40 s to 30 min depending on the complexity of the diagnosis task, showing good performance of the proposed method.

Gear transmission systems are widely used in various fields. The occurrence of gear cracks, tooth pitting, and other faults will lead to the dynamic characteristics deterioration of the transmission system. In order to calculate the meshing stiffness of faulty gear pairs more effectively and precisely, this article improves the loaded tooth contact analysis (LTCA) method by analyzing the influence of different fault types on gear deformation, including bending-shearing deformation and contact deformation, which combines the accuracy of the finite element method (FEM) and the rapidity of the analytical method (AM). The improved LTCA method can model the fault areas accurately and optimize the deformation coordination equation under the actual meshing situation of the faulty gear tooth, making it suitable for calculating the meshing stiffness of faulty gears. Based on the calculation results of the finite element method, the accuracy of the improved meshing stiffness calculation method has been verified, and the sensitivity of different fault type parameters on meshing stiffness has been studied.

The dynamics analysis of a planetary gear set transmissions requires the creation of completely different models for different gears, which is very tedious. In this paper, a generalized dynamics modeling process is proposed for a three planetary gear set transmissions, and a generalized dynamic model for multiple gears is established by using the lumped mass method. The analysis of meshing force characteristics is carried out for the second gear position, and the meshing frequency coupling phenomenon between the meshing forces of the three planetary gear sets is investigated. The results show that, for the current gear set of meshing force, the meshing frequency components of other gear sets only appear in a part of the speed, and with the increase in speed, certain low-frequency components of other sets that exist at low speed will decrease or even disappear, and the coupling relationship between the meshing forces of different planetary gear sets is not symmetrical.