For Prognostics and Health Management (PHM) of Lithium-ion (Li-ion) batteries, many models have been established to characterize their degradation process. The existing empirical or physical models can reveal important information regarding the degradation dynamics. However, there are no general and flexible methods to fuse the information represented by those models. Physics-Informed Neural Network (PINN) is an efficient tool to fuse empirical or physical dynamic models with data-driven models. To take full advantage of various information sources, we propose a model fusion scheme based on PINN. It is implemented by developing a semi-empirical semi-physical Partial Differential Equation (PDE) to model the degradation dynamics of Li-ion batteries. When there is little prior knowledge about the dynamics, we leverage the data-driven Deep Hidden Physics Model (DeepHPM) to discover the underlying governing dynamic models. The uncovered dynamics information is then fused with that mined by the surrogate neural network in the PINN framework. Moreover, an uncertainty-based adaptive weighting method is employed to balance the multiple learning tasks when training the PINN. The proposed methods are verified on a public dataset of Li-ion Phosphate (LFP)/graphite batteries.

To ensure the safety of electrical drive systems, fault detection and diagnosis (FDD) has become an active approach over the past two decades. Multivariate analysis is a popular method in FDD, among which canonical correlation analysis (CCA) has been widely applied and studied. However, most CCA-based fault detection (FD) methods assume that the signal is Gaussian and that there is a linear relationship between the variables. Since the electrical drive systems are nonlinear, these CCA-based FD methods are not optimal. With the help of the reversible residual network, this paper proposes a reversible residual network-aided CCA (RRNCCA) for fault diagnosis. The main work is as follows: 1) The objective function of RRNCCA is reformulated; 2) RRNCCA-based FDD is first designed for electrical drive systems; 3) Through the difference in FD results, fault diagnosis is directly achieved. The effectiveness of the proposed method is verified via an electrical drive system.

With the steady improvement of advanced manufacturing processes and big data technologies, modern industrial systems have become large-scale. To enhance the sensitivity of fault detection (FD) and overcome the drawbacks of the centralized FD framework in dynamic systems, a new data-driven FD method based on Hellinger distance and subspace techniques is proposed for dynamic systems. Specifically, the proposed approach uses only system input/output data collected via sensor networks, and the distributed residual signals can be generated directly through the stable kernel representation of the process. Based on this, each sensor node can obtain the identical residual signal and test statistic through the average consensus algorithms. In addition, this paper integrates the Hellinger distance into the residual signal analysis for improving the FD performance. Finally, the effectiveness and accuracy of the proposed method have been verified in a real multiphase flow facility.

The heat exchanger made up of several T-shaped joints is a critical component in energy equipment. A split induction coil was designed for T-shaped aluminum pipe brazing, and the Taguchi method was combined with multi-physical simulation to solve the multi-factor optimization of the split coil. The results showed that the multi-physical simulation model had high precision. The melting of filler metal was almost consistent with the spatial distribution of the temperature field, and the average simulation error was approximately 5.753 °C. The optimal coil parameters were obtained with a turn number of 3.5, a turn space of 7 mm, a heating distance of 15.6 mm, a coil diameter of 8 mm, and a coil length of 9 mm. Three well-formed T-shaped joints could be obtained at one time via the optimal split coil. During the induction brazing, the skin effects and the proximity effects induced a high magnetic field intensity around the joint, which had a significant relationship with the coil length and coil diameter. The high magnetic field intensity promoted a high eddy current density in this place, and as a result, the high Joule heat could be generated around the joint. In addition, the significant decrease in the heating rate at high temperatures promoted the homogenization of the temperature and the melting and filling of the filler metal as well as avoided local overheating.

This article presents a data-driven framework for fault detection and diagnosis (FDD) in permanent magnet synchronous motor (PMSM) drive systems using conditional invertible neural network (CINN). Unlike conventional deep models that operate as closed boxes, the proposed CINN framework establishes a bidirectional and control-theoretically consistent mapping between process data and residual signals, enabling explainable fault reasoning. First, an analytically interpretable formulation of the CINN-based residual mapping is derived using composite operators, and two sets of residual signals are designed to enhance FDD. Second, a Bayesian classification-based fault diagnosis strategy is implemented through a detailed analysis of the residual signals. The proposed FDD scheme offers the following advantages: 1) the system modeling and residual construction are integrated into the CINN; and 2) its design is grounded in control-theoretic principles to ensure the theoretical soundness of the learning mechanism. Experimental results verify that the proposed method achieves fast and accurate fault detection along with effective diagnosis for various fault types in PMSM drive systems.