Recently, Deep Reinforcement Learning (DRL) has been applied to learn priority dispatching rules (PDRs) for solving complex scheduling problems. However, existing works face challenges in dealing with flexibility, which allows an operation to be scheduled on one out of multiple machines and is often required in practice. Such one-to-many relationship brings additional complexity in both decision-making and state representation. This paper considers the well-known Flexible Job-shop Scheduling Problem (FJSP), and addresses these issues by proposing a novel DRL method to learn high-quality PDRs end-to-end. The operation selection and machine assignment are combined as a composite decision. Moreover, based on a novel heterogeneous graph representation of scheduling states, a Heterogeneous Graph Neural Network based architecture is proposed to capture complex relationships among operations and machines. Experiments show that the proposed method outperforms traditional PDRs and is computationally efficient, even on instances of larger scales and different properties unseen in training.

Nowadays the industrial Internet of things (IIoT) has been successfully utilized in smart manufacturing. The massive data in IIoT promotes the development of deep learning-based health monitoring for industrial equipment. Since monitoring data for mechanical fault diagnosis collected on different working conditions or equipment have domain mismatch, models trained with training data may not work in practical applications. Therefore, it is essential to study fault diagnosis methods with domain adaptation ability. In this paper, we propose an intelligent fault diagnosis method based on an improved domain adaptation method. Specifically, two feature extractors concerning feature space distance and domain mismatch are trained using maximum mean discrepancy (MMD) and domain adversarial training respectively to enhance feature representation. Since separate classifiers are trained for feature extractors, ensemble learning is further utilized to obtain final results. Experimental results indicate that the proposed method is effective and applicable in diagnosing faults with domain mismatch.

Remaining useful life (RUL) prediction is one of the key technologies of condition-based maintenance (CBM), which is important to maintain the reliability and safety of industrial equipment. Massive industrial measurement data has effectively improved the performance of the data-driven-based RUL prediction method. While deep learning has achieved great success in RUL prediction, existing methods have difficulties in processing long sequences and extracting information from the sensor and time step aspects. In this article, we propose dual-aspect self-attention based on transformer (DAST), a novel deep RUL prediction method, which is an encoder–decoder structure purely based on self-attention without any recurrent neural network (RNN)/convolution neural network (CNN) module. DAST consists of two encoders, which work in parallel to simultaneously extract features of different sensors and time steps. Solely based on self-attention, the DAST encoders are more effective in processing long data sequences and are capable of adaptively learning to focus on more important parts of the input. Moreover, the parallel feature extraction design avoids the mutual influence of information from two aspects. Experiments on two widely used turbofan engines datasets show that our method significantly outperforms the state-of-the-art RUL prediction methods.

Massive industrial data collected from industrial Internet of things (IIoT) assets improves data-driven methods for prognostics and health management (PHM) systems. As an important role in PHM, remaining useful life (RUL) prediction is essential to maintain the reliability and safety of industrial manufacture. However, recent data-driven approaches for bearing RUL prediction do not weight the contributions of data from different sensors and time steps, which decreases the efficiency in the big data era. In this context, we present a deep learning-based RUL prediction method with attention mechanism to weight sequence data’s representations. Specifically, the proposed method weights different industrial sensors and time steps respectively based on distributed attention mechanism. Then temporal convolution modules with the shared weights are used for feature extraction of time series. Finally, performance of the proposed method is verified using a popular dataset C-MAPSS. Experimental results reveal that the proposed method has high accuracy and is efficient in practical applications.

Economic dispatch problem (EDP) is a fundamental optimization problem of power systems. With the penetration of renewable energy sources in microgrids, this paper proposes a distributed algorithm based on consensus theory to solve the EDP with a quadratic cost function. The method takes advantage of the fact that incremental costs need to be equal for all buses at optimal output power values. Thus, the incremental cost of each bus is selected as a consensus variable and the local mismatch between total demand and generation is assigned as a feedback variable to meet demand constraints. Unlike the existing related works, the feedback gains for the feedback variables are different and time-varying. Using multi-parameter perturbation theory and graph theory, the convergence of the algorithm is proved. Furthermore, the upper bounds of the feedback gains are given theoretically, and we obtain that the algorithm converges to the optimal solution at a rate governed by the second largest eigenvalue of the system matrix. Meanwhile, the algorithm is a fully distributed algorithm without a leader or a virtual command node. The simulation results illustrate the effectiveness of the algorithm even though there is a demand change or some generator is damaged. Some simulation results intuitively illustrate how the convergence speed changes with the feedback gains.

It is a significant and challenging problem to coordinate multiple microgrids (MMGs) belonging to different entities and achieve their excellent energy-sharing performance to ensure the stability of electricity markets. This article studies a grid-oriented energy bidding problem for MMGs with peer-to-peer (P2P) energy trading under uncertainty. A stochastic Cartel game (SCG) based strategy is developed. A stochastic Cartel nonlinear programming model is formulated to characterize the joint energy bidding, the energy production, and P2P energy transactions while minimizing the total cost for MMGs under uncertainty. A diagonal quadratic approximation method is employed to linearize quadratic terms, and the SCG problem for MMGs is further decomposed into subproblems for individual MGs based on a surrogate Lagrangian relaxation method. The equivalence of problem transformation is proved and equilibrium solutions are derived in an iterative and distributed manner. Comparisons for different strategies, models, and solution algorithms are conducted to testify the rationality and validity of the proposed strategy.