Smart manufacturing that involves tight integration of the physical system and cyber system is a hot topic in both industry and academia in the era of the Internet and big data. However, the dynamic and uncertain manufacturing environment introduces a significant adaptive issue of production scheduling, which is one of the pivotal tasks for smart manufacturing. This paper focuses on this problem and proposes a closed-loop adaptive scheduling solution based on the Cyber-Physical Production System (CPPS) with four phases: production data acquisition (PDA), dynamic disturbance identification (DDI), scheduling strategy adjustment (SSA), and schedule scheme generation (SSG). In the DDI phase, in view of the disturbance classification, a disturbance identification procedure based on CPPS monitoring is studied to ensure real-time response. In the SSA phase, an industrial big-data-driven scheduling strategy adjustment method is proposed, which consists of GA-based offline knowledge learning and KNN-based online adjustment, to enhance the system adaptability. We apply and verify the proposed adaptive scheduling solution on an experimental semiconductor manufacturing system, and the results demonstrate that the proposed method outperforms the dynamic scheduling method in terms of multiple objectives under different disturbance levels.
The dynamic scheduling problem of semiconductor manufacturing systems (SMSs) is becoming more complicated and challenging due to internal uncertainties and external demand changes. To this end, this paper addresses integrated release control and production scheduling problems with uncertain processing times and urgent orders and proposes a convolutional neural network and asynchronous advanced actor critic-based method (CNN-A3C) that involves a training phase and a deployment phase. In the training phase, actor–critic networks are trained to predict the evaluation of scheduling decisions and to output the optimal scheduling decision. In the deployment phase, the most appropriate release control and scheduling decisions are periodically generated according to the current production status based on the networks. Furthermore, we improve the four key points in the deep reinforcement learning (DRL) algorithm, state space, action space, reward function, and network structure and design four mechanisms: a slide-window-based two-dimensional state perception mechanism, an adaptive reward function that considers multiple objectives and automatically adjusts to dynamic events, a continuous action space based on composite dispatching rules (CDR) and release strategies, and actor–critic networks based on convolutional neural networks (CNNs). To verify the feasibility and effectiveness of the proposed dynamic scheduling method, it is implemented on a simplified SMS. The simulation experimental results show that the proposed method outperforms the unimproved A3C-based method and the common dispatching rules under the new uncertain scenarios.
The fault diagnosis problem of prognostics and health management (PHM) is becoming more complicated and challenging due to the increasing demand for timeliness and adaptability under dynamic working conditions. This article studies a new adaptive fault diagnosis (AdaFD) scheme and proposes a deep learning-based model and algorithm to enhance both the timeliness of fault diagnosis and its adaptability to dynamic conditions. The primary AdaFD model is based on a convolutional neural network (CNN). The two key modules of AdaFD are: 1) adaptive and independent learning rate design for model training and 2) adaptive structure optimization approach for model application. During the training phase, we design an adaptive learning rate strategy by distinguishing the weight and bias parameters in the CNN model to accelerate the training process. In the application phase, the adaptive batch normalization (AdaBN) algorithm is adopted to improve the adaptability of the primary CNN model to dynamic working conditions and data noise. Finally, a case study is conducted to validate the superiority of AdaFD based on a benchmark rolling bearing dataset. The results show that the proposed fault diagnosis method has higher accuracy and better adaptability than other common methods and possesses advantages in terms of training efficiency.
Various factors and constraints should be considered when developing a manufacturing production schedule, and such a schedule is often based on rules. This paper develops a composite dispatching rule based on heuristic rules that comprehensively consider various factors in a semiconductor production line. The composite rule is obtained by exploring various states of a semiconductor production line (machine status, queue size, etc.), where such indicators as makespan and equipment efficiency are used to judge performance. A model of the response surface, as a function of key variables, is then developed to find the optimized parameters of a composite rule for various production states. Furthermore, dynamic scheduling of semiconductor manufacturing is studied based on support vector regression (SVR). This approach dynamically obtains a composite dispatching rule (i.e., parameters of the composite dispatching rule) that can be used to optimize production performance according to real-time production line state. Following optimization, the proposed dynamic scheduling approach is tested in a real semiconductor production line to validate the effectiveness of the proposed composite rule set.