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.

Monitoring surface quality during machining has considerable practical significance for the performance of high-value products, particularly for their assembly interfaces. Surface roughness is the most important metric of surface quality. Currently, the research on online surface roughness prediction has several limitations. The effect of tool wear variation on surface roughness is seldom considered in machining. In addition, the deterioration trend of surface roughness and tool wear differs under variable cutting parameters. The prediction models trained under one set of cutting parameters fail when cutting parameters change. Accordingly, to timely monitor the surface quality of assembly interfaces of high-value products, this paper proposes a surface roughness prediction method that considers the tool wear variation under variable cutting parameters. In this method, a stacked autoencoder and long short-term memory network (SAE–LSTM) is designed as the fundamental surface roughness prediction model using tool wear conditions and sensor signals as inputs. The transfer learning strategy is applied to the SAE–LSTM such that the surface roughness online prediction under variable cutting parameters can be realized. Machining experiments for the assembly interface (using Ti6Al4V as material) of an aircraft’s vertical tail are conducted, and monitoring data are used to validate the proposed method. Ablation studies are implemented to evaluate the key modules of the proposed model. The experimental results show that the proposed method outperforms other models and is capable of tracking the true surface roughness with time. Specifically, the minimum values of the root mean square error and mean absolute percentage error of the prediction results after transfer learning are 0.027 μm and 1.56%, respectively.