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.

Industrial Internet of Things has significantly boosted predictive maintenance for complex industrial systems, where the accurate prediction of remaining useful life with high-level confidence is challenging. By aggregating multiple informative sources of system degradation, information fusion can be applied to improve the prediction accuracy and reduce the uncertainty. It can be performed on the data-level, feature-level, and decision-level. To fully exploit the available degradation information, this paper proposes a hybrid fusion method on both the data level and decision level to predict the remaining useful life. On the data level, Genetic Programming is adopted to integrate physical sensor sources into a composite health indicator, resulting in an explicit nonlinear data-level fusion model. Subsequently, the predictions of the remaining useful life based on each physical sensor and the developed composite health indicator are synthesized in the framework of belief functions theory, as the decision-level fusion method. Moreover, the decision-level method is flexible for incorporating other statistical data-driven methods with explicit estimations of the remaining useful life. The proposed method is verified via a case study on NASA’s C-MAPSS data set. Compared to the single-level fusion methods, the results confirm the superiority of the proposed method for higher accuracy and certainty of predicting the remaining useful life.

In this paper, we propose a downlink intelligent reflecting surface (IRS) aided non-orthogonal multiple access (NOMA) for millimeter-wave (mmWave) massive MIMO with lens antenna array, i.e., IRS-aided mmWave beamspace NOMA, where the single-antenna users without direct-link but connected to the base station (BS) with the aid of the IRS are grouped as one NOMA group. Considering the power leakage problem in beamspace channel and the per-antenna power constraint, we propose two multi-beam selection strategies for the BS-IRS link under two channel models, i.e., 2-dimension (2D) channel model and 3-dimension (3D) channel model, respectively, where two corresponding RF chain configuration strategies are designed, respectively. Then, we formulate and solve the optimization problem for maximizing the weighted sum rate by jointly optimizing the active beamforming at the BS and the passive beamforming at the IRS, where we propose the alternating optimization (AO) method to solve the above joint optimization problem. Especially, different from the stochastic method, based on the beam-splitting technique, we propose the method to initialize the feasible solution for the proposed AO method, where the conditional transmit power minimization problem is formulated and solved. Through simulations, the weighted sum rate performance of the proposed IRS-aided mmWave beamspace NOMA is verified.

With the continuous increase of location-aware wireless applications, various indoor localization techniques based on radio frequency (RF) electromagnetic waves,^1,2 ultrasound,³ and infrared⁴ (IR) have been proposed. For low-frequency RF-based approaches, such as wireless local networks^4,5 (WLAN) and ultra-wideband,⁶ the localization accuracy is highly related to the density and distribution of the routers, and the wireless RF signals are liable to suffer from RF interference.⁷ The average localization error for RF localization is several meters, and this accuracy level cannot be applied to the traditional trialeration process to localize user devices correctly due to multipath propagation.⁸ Ultrasound-based approaches can provide a high localization accuracy of 10 cm but require expensive and specialized infrastructures.⁹ IR-based indoor localization can achieve relatively high localization accuracy of centimeter-level, but its transmission distance (1 m) is shorter than that of VLC (10 m).¹⁰ It also leads to a higher cost since it is necessary to install large numbers of IR nodes in the room. Recently, visible light communication (VLC) localization methods have drawn much attention of researchers and industries. Unlike conventional techniques, VLC-based localization is immune to electromagnetic interference and can be used in RF-restricted zones¹¹ (e.g., airport and hospital). It can achieve centimeter-level accuracy¹² 13. 14. 15. 16. 17. 18. 19. 20. 21. ^{– 22} with lower cost infrastructures by remodifying the existing illumination system to VLC transmitters. Moreover, it offers a higher security for users since visible light signals cannot transmit across the walls and may not be accessed by eavesdroppers from other rooms.¹²

This paper presents a new general method for evaluating the reliability function and the mean residual life of degrading systems subject to condition monitoring and random failure. In the proposed method, the degradation process of the system is characterized by a continuous-time Markov chain, which is then incorporated into the proportional hazards model as a stochastic covariate process to describe the hazard rate of the time to system failure. Unlike the conventional method based on conditioning, which is applicable only for a small number of degradation states, the proposed method is capable of tackling the case with a general number of degradation states. Using the developed approximation techniques, closed-form formulas for related reliability characteristics are obtained in terms of the appropriate transition probability matrix. The proposed evaluation algorithm is computationally efficient and embeddable to support real-time reliability assessment of the system subject to condition monitoring for developing the optimal maintenance policy. The effectiveness and the accuracy of the method are validated by a numerical study and compared with the conventional method. A general case where the degradation path can be discretized up to ten states is also studied to illustrate the appealing general features.

Active reconfigurable intelligent surface (RIS) draws great attentions recently due to its unique nature of “changing” the wireless propagation channels via simultaneously adjusting the phase and amplifying the amplitude of signals, and is proved to effectively overcome the double fading effect when the signals are reflected by RIS compared with the widely investigated passive RIS with only adjusting the phase of the signals. Motivated by this great advantage, in this paper, we apply active RIS into a multi-user system and focus on enhancing the secure transmission performance in the presence of a malicious eavesdropper (Eve). Specifically, two types of imperfect channel state information (CSI) between RIS and Eve, namely imperfect CSI with bounded estimation error and with statistical distribution of estimation error, are considered, and we focus on enhancing the worst-case secrecy rate (SR) as well as weighted sum-SR of this system under each CSI condition. In particular, for the study of weighted sum-SR maximization, a more serious situation is assumed where Eve can filter out all the interference signals as well as the external noise generated by RIS when decoding each user's information. Given these settings, we propose a robust alternating optimization (AO)+line search combined algorithm to optimize the worst-case SR, and a dual-successive approximation algorithm to optimize the weighted sum-SR. Numerical examples demonstrate that compared with the benchmark scheme under passive RIS aided case, the proposed algorithms for the active RIS aided system greatly enhance both the worst-case and weighted-sum SR under these CSI conditions.