NEURAL NETWORK BASED REAL-TIME PRODUCTION SCHEDULING FOR INDUSTRIAL PROCESSES

Real-time production scheduling in a multi-grade PET resin plant under demand uncertainty

Recent advancements in production scheduling have arisen in response to the need for adaptation in dynamic environments. This paper addresses the challenge of real-time scheduling within the context of sustainable production. We redefine the sustainable distributed permutation flow-shop scheduling problem using an online mixed-integer programming model. The proposed model prioritizes minimizing makespan while simultaneously constraining energy consumption, reducing the number of lost working days and increasing job opportunities within permissible limits. Our approach considers machines operating in different modes, ranging from manual to automatic, and employs two real-time scheduling strategies: predictive-reactive and proactive-reactive scheduling. We evaluate two rescheduling policies: continuous and event-driven. To demonstrate the model's applicability, we present a case study in auto workpiece production. We manage model complexity through various reformulations and heuristics, such as Lagrangian relaxation and Benders decomposition for initial optimization as well as four problem-specific heuristics for real-time considerations. For solving large-scale instances, we employ simulated annealing and tabu search metaheuristic algorithms. Our findings underscore the benefits of the predictive-reactive scheduling strategy and the efficiency of the event-driven rescheduling policy. By addressing dynamic scheduling challenges and integrating sustainability criteria, this study contributes valuable insights into real-time scheduling and sustainable production.

Spatial-Temporal Finite Element Analytics for Cyber-Physical System-enabled Smart Factory: Application in Hybrid Flow Shop

• A new R-DTFJSS idea was developed to realise high-efficiency production. • Edge computing-based R-DTFJSS framework can enhance the data processing ability. • A variable time window-based RS strategy was used to achieve operation allocation. • An IHA-based solve method was designed to reduce the computational complexity. Production scheduling is the central link between enterprise production and operation management and is also the key to realising efficient, high-quality and sustainable production. However, in real-world manufacturing, the frequent occurrence of abnormal disturbance leads to the deviation of scheduling, which affects the accuracy and reliability of scheduling execution. The traditional dynamic scheduling methods (TDSMs) cannot solve this problem effectively. This paper presents a real-time digital twin flexible job shop scheduling (R-DTFJSS) method with edge computing to address the issue. Firstly, an overall framework of R-DTFJSS is proposed to realise real-time scheduling (RS) through real-time interaction between physical workshop (PW) and virtual workshop (VW). Secondly, the implementation process of R-DTFJSS is designed to realise real-time operation allocation. Then, to obtain the optimal RS result, an improved Hungarian algorithm (IHA) is adopted. Finally, a case simulation from an industrial case of a cooperative enterprise is described and analysed to verify the effectiveness of the proposed R-DTFJSS method. The results show that compared with the TDSMs, the R-DTFJSS method can effectively deal with unexpected and frequent abnormal disturbances in the production process.

Simulation can be an approximate performance evaluation method of real-time scheduling strategies. This method is useful when the analytical evaluation is too complex to be put in practice. In this paper we use simulation to evaluate the performance of Rate Monotonic and FIFO scheduling algorithms in different multiprocessor scheduling approaches. We present the design of our real-time scheduling simulation tool and the loose clustered scheduling approach as a result of our research.

The graduation development such as employment and graduate school admission of college students are important tasks. However, mining the factors that can affect the development of graduation remains challenging, because the most important factor “course” is not independent and inequality, which are always ignored by previous researchers. Furthermore, traditional structured methods cannot handle the complex relationships between courses, and attention networks cannot distinguish the weights of compulsory and elective courses with different distributions. Therefore, we present a Graph-based Hierarchical Attention Neural Network Model with Elective Course (GHANN-EC) for the prediction of graduation development in this study. Specifically, we use graph embedding that captures the unstructured relationships between courses and hierarchical attention that assigns the importance of the courses to excavating course information that represent students’ independent interests, and can more accurately understand the relationship between graduation development and academic performance. Experimental results on the real-world datasets show that GHANN-EC outperforms the existing popular approach.

Non-point source pollution from agricultural activities poses a significant threat to water quality by introducing excess nutrients like nitrogen into aquatic ecosystems, leading to issues such as eutrophication and groundwater contamination. In agricultural watersheds, nitrate transport involves intricate physical, chemical, and biological processes influenced by meteorological conditions, hydrological features, and spatial topologies, making accurate short-term predictions challenging. Traditional data-driven deep learning models often fail to incorporate physical constraints and complex spatiotemporal dynamics, limiting their interpretability and predictive accuracy. Here we show a hierarchical transformer and graph neural network model that accurately predicts watershed nitrate concentrations by integrating multi-source data and simulating pollutant migration. The model captures nonlinear multivariate temporal patterns through hierarchical transformers, fuses global meteorological and local hydrological features via neural networks, and models runoff topologies with physically constrained graph neural networks. For predicting the concentration changes of pollutants discharged from watersheds, it outperforms baselines like multi-layer perceptrons, recurrent neural networks, and long short-term memory networks, with state-of-the-art performance in root mean square error, mean absolute error, and R2. Ablation studies confirm the essential roles of multi-source data integration and watershed topological modeling in enhancing performance. This method of directly modeling physical processes by leveraging the characteristics of different neural network architectures opens up a new path for addressing the interpretability problem in neural earth system modeling, apart from the process-guided deep learning and differentiable modelling methods.

A reactive piloting policy for Flexible Assembly Lines (FAL) is proposed, where the sequencing of the operations as well as the assignment of tasks to workstations are not pre-determined but driven by the actual state of the FAL. For each work-in-process coming out from a workstation, the next destination is determined by minimizing a temporal criterion taking in account: the time needed to reach the destination, the load of the workstation to reach, the duration of the operation to be completed in the destination station, the disponibility of product components in this station.

Training in the mining industry received considerable attention since the Federal Mine Safety and Health Act of 1977 mandating thorough health and safety training. The response to a large extent has been one in which individual facilities developed their own training “programs” rather than sharing in the development of appropriate programs or using another's proven materials. With the basic achievement of health and safety training requirements, the industry has turned its attention to training to improve job performance; here again, independent and relatively informal approaches have been developed. Woodward Associates, Inc., San Diego, has developed, under contract to the U.S. Bureau of Mines, structured equipment operator training programs which teach how to (1) operate the machine and (2) perform the task. The WAI Six-Element approach provides the structure to guide the learning process to (1) ensure exposure to information, (2) promote knowledge and skill acquisition, (3) aid transfer of knowledge and skills to new situations, and (4) assess knowledge and skill acquisition and application. Because of the potential severity of accidents in the mining (and construction) industries, particular attention has been directed to helping future operators to cope with not only normal but also with the abnormal (machine malfunction) and emergency (machine failure) conditions which tend to result in accidents. The Six-Element approach has taken this concern and addressed it through (1) classroom instruction in the cues and procedures for handling abnormal and emergency conditions, and (2) structured on-machine training which employs a truck as task trainer and an OBSAC (On-Board Simulator of Abnormal Conditions) which induces apparent malfunctions, and actual system degrades to allow trainees to experience abnormal and emergency situations under safe conditions and under an instructor's watchful eye. This paper discusses the purpose, benefit, and challenges of introducing and using structured operator training in the mining industry. An evaluation of the experience related to the Haulage Truck Training Program will be discussed. Applications related to training operators of other mobile surface mining machines in normal and abnormal operating conditions will also be discussed.

The paper has presented a practical application result of large scale linear programming to real-time optimal production scheduling in actual pulp and paper mill. Real-time application brings in several restrictions and requirements. The system based on a 32-bit minicomputer has been in operation since March 1986, providing optimal production schedules every 8 hours. The reduction in production cost is estimated to be more than 1% on the average. We believe that this is one of a few practical and successful real-time application cases of large scale linear programming realized on a mini-computer which also runs several hundreds real-time tasks.

Integrated real-time production scheduling of a multiple cryogenic air separation unit and compressor plant

This article describes the architecture of a typical mobile health application. Anyoneusingthesystemcansettheirownpersonalizedprivacylevel.Thisarticledescribeshowsocialnetworkingcommunicationscanbemademoresecureandprivateinamulti-cloudenvironment,especiallyformedicalapplications.The purpose of this study is to present a practical approach for extracting individual-level AH from GSV data. We have developed an entirely new hierarchical deep learning method that can identify AH based on global and local visual features. The objective of this study was to present a practical technique for separating individual-level BP from GSV data. To determine AH using global and local visualization features, we created a new hierarchical deep learning system. Thesesystemsbecomemorevulnerabletohackingastechnologyadvances.The research presented in this white paper aims to show the feasibility and utility of artificial intelligence-based hierarchical deep neural network models to protect data in motion. A new aspect of this research is the synergistic formation of multiple clouds using different complex models, which shortens the formation time and increases the accuracy of detecting unidentified attacks. receive. Asaresult, the core cloudmodelconverges much lesstime(6-8)andfarfewerepochsthantheboundarycloudmodel, which takes 35-40epochstotrain.Extensiveevaluationhas demonstrated that theMUSEsystemcantrainhugemerged models faster thanindividuallybuiltunmergedmodelsonthemaincloud.Afteraseriesofruns,wefoundthatthecombinedmodelreducedtrainingtimebyanaverageof26.2%.

Selection and corrosion evaluation tests of candidate SCWR fuel cladding materials

The design of systems for intelligent control of urban traffic is important in providing a safe environment for pedestrians and motorists. Artificial neural networks (ANNs) (learning systems) and expert systems (knowledge-based systems) have been extensively explored as approaches for decision-making. While the ANNs compute decisions by learning from successfully solved examples, the expert systems rely on a knowledge base developed by human reasoning for decision making. It is possible to integrate the learning abilities of an ANN and the knowledge-based decision-making ability of the expert system. This paper presents a real-time intelligent decision-making system, IDUTC, for urban traffic control applications. The system integrates a backpropagation-based ANN that can learn and adapt to the dynamically changing environment and a fuzzy expert system for decision-making. The performance of the proposed intelligent decision-making system is evaluated by mapping the adaptable traffic light control problem. The application is implemented using the ANN approach, the FES approach, and the proposed integrated system approach. The results of extensive simulations using the three approaches indicate that the integrated system provides better performance and leads to a more efficient implementation than the other two approaches.

A system for approximate number discrimination has been shown to arise in at least two types of hierarchical neural network models—a generative Deep Belief Network (DBN) and a Hierarchical Convolutional Neural Network (HCNN) trained to classify natural objects. Here, we investigate whether the same two network architectures can learn to recognise exact numerosity. A clear difference in performance could be traced to the specificity of the unit responses that emerged in the last hidden layer of each network. In the DBN, the emergence of a layer of monotonic ‘summation units’ was sufficient to produce classification behaviour consistent with the behavioural signature of the approximate number system. In the HCNN, a layer of units uniquely tuned to the transition between particular numerosities effectively encoded a thermometer-like ‘numerosity code’ that ensured near-perfect classification accuracy. The results support the notion that parallel pattern-recognition mechanisms may give rise to exact and approximate number concepts, both of which may contribute to the learning of symbolic numbers and arithmetic.