Flexible Job Shop Scheduling Problem (FJSSP) stands as one of the most intricate challenges in intelligent manufacturing, demanding coordinated decision-making across interdependent operations. Deep Reinforcement Learning (DRL) methods, such as Deep Q-Network (DQN) and Proximal Policy Optimization (PPO), typically rely on single-step action selection based on the current system state. Although these methods have shown effectiveness in short-term scheduling, their one-step decision paradigm limits their ability to maintain temporal consistency across extended scheduling trajectories. This motivates the need for a DRL framework with explicit multi-step decision modeling for complex shop-floor environments. To overcome these limitations, this study proposes an Action Chunking Transformer (ACT)-enhanced DRL framework, which formulates FJSSP as a sequential decision process endowed with multi-step temporal inference. In this framework, the ACT module predicts coherent action chunks from recent state trajectories, allowing agents to plan and execute temporally consistent decisions rather than isolated one-step actions. A confidence-gated control mechanism adaptively balances ACT-generated sequences and conventional DRL policy outputs, ensuring reliable behavior even under environmental uncertainty. A joint optimization strategy is then introduced, integrating DRL learning loss and ACT prediction loss, to promote stable convergence and bidirectional adaptation between the decision and prediction modules. The proposed ACT-DRL architecture is implemented and evaluated using benchmark FJSSP datasets. Comparative experiment results reveal that the ACT-DRL achieved faster convergence, smoother scheduling trajectories, and robust decision.

This paper is based on the construction of a quality traceability system, with the background of tobacco material quality management. Through a unified coding rule for tobacco materials and relying on scanning association and label recognition technology for purchased filter rods, it establishes a system for tracking and managing process information from supplier production to workshop machine usage. This reduces the scope of quality traceability for purchased filter rods and finished cigarettes, improves quality traceability efficiency and reduces traceability costs, reduces the impact of quality factors on timely sales and supply, and achieves error prevention and precise traceability management for purchased filter rods.

The feature recognition of cigarette box identification codes is of great significance in avoiding packaging errors of cigarette products. This article designs an automatic recognition system for cigarette box identification codes based on optical character recognition (OCR) technology, which can achieve brand recognition of old cardboard boxes. The system consists of industrial cameras, light sources, industrial control computers, etc. After capturing the image of the cigarette box with an industrial camera, the OCR algorithm is used to identify the identification code in the image and compare it with the brand information in the product library to determine whether the cigarette box is abnormal. The application results show that the recognition accuracy of reused cardboard boxes with blurred identification is 98.9%, which can significantly improve the reuse rate of old cardboard boxes.

Automated Guided Vehicles (AGVs) and other autonomous conveyance vehicles play a pivotal role in enhancing efficiency and precision in packaging areas. Traditional manual methods are inefficient and prone to errors and delays. To overcome these challenges, this paper proposes an AVCS that is specifically designed for use in packaging zones. The system also incorporates AGVs and other autonomous vehicles through a control system for vehicle allocation and path interference. They include AGVs that have laser-based positioning, another vehicle with visual recognition or SLAM-based positioning, central processors, unified dispatch depending on the priority set, map database that stores positioning and path planning data. Some of the components in the system include wireless communication modules that enable communication between the vehicles and the central control unit. Implementation includes placing vehicles on the map and installing communication modules.

2D drawing hotspot links refer to visualizing the BOM structure of a device in a 2D drawing document, and adding more detailed device information to an electronic drawing in the form of hotspot links. According to the model, the management system divides BOM into sub equipment, systems, mechanisms, devices, assemblies, components, and other hierarchical contents according to the hierarchical node structure requirements of different parts. It establishes drawing configuration information corresponding to the equipment parts, and establishes information related to spare parts information corresponding to BOM parts at all levels. 2D drawing hotspot link optimizes the management of drawings and parts data of production equipment, and realizes the tree structure information management of parts at all levels under equipment classification and model. By adding hotspot links to 2D drawings, it is convenient for maintenance personnel to understand the structure of special parts and search for professional terms, saving the usual time of searching through a large amount of materials. In addition, the hot link method can also be directly clicked to apply for spare parts during troubleshooting.

Understanding how the magneto-Rayleigh–Taylor instability (MRTI) scales to the current rise time is vital for Z-pinch dynamic hohlraum driven inertial confinement fusion. Wang et al discovered in prior theoretical work that the perturbation amplitude of MRTI before stagnation increases linearly with the current rise time when the implosion velocity of Z-pinch plasma is held constant. In the present work, three types of wire-array experiments with similar implosion dynamics and constant implosion velocity are performed on an 8 MA pulse power generator to investigate the scaling of MRTI to the rise time. It is successfully accomplished for the first time to obtain the similar wire-array Z-pinch implosions in which the current rise time is scaled up to three times on the generator by controlling the trigger time of its 24 modules. Both the experimental results, which include x-ray radiation pulses and x-ray images of imploding plasmas, and the related numerical analysis have shown that the MRTI before stagnation grows linearly with the rise time, as predicted by the theoretical model.

With the continuous improvement of the complexity and intelligence level of production equipment, the production capacity of production equipment is also increasing. Equipment maintenance and fault warning have become the key to ensuring the normal operation of production equipment units and reducing enterprise operating costs. Due to the complex and closely connected electromechanical structure of production equipment units, there are chain reactions and mutual influences between production equipment unit components, which increases the difficulty of abnormal fault warning for production equipment unit components. However, the accuracy of fault warning methods for production equipment unit components is currently low, resulting in maintenance personnel often not having time to make appropriate equipment rotation maintenance plans.

The content of this paper is to study the business processes of manually covering purchased filter rods, entering the filter rod elevated warehouse, entering and exiting the flat warehouse, manually loading the rod at the emission end, and manually loading the rod on the cigarette machine, in order to achieve accurate traceability of purchased filter rods. Exploring the integration of handheld, WMS, WCS, electronic control, data acquisition, MES and other systems by relying on the production information encoding of 30 manufacturers of purchased filter rods to associate RFID. The entire process of reading and writing codes is used to read, associate, and verity the relevant information of purchased filter rods, achieving full process tracking, error prevention, and traceability of purchased filter rods. The research purpose of this paper is to effectively improve the quality management level of outsourced filter rods in the workshop in real time, reduce consumption and costs, and enhance the level of lean management in the workshop. It can provide reference and reference for the industry to promote precise traceability of filter rods.

The automatic waste smoke recycling vehicle has electrical equipment such as an onboard weighing system and a radio transceiver system. Adequate and stable power supply is an important factor in achieving the functionality of electrical equipment during its use. In production practice, the situation where electrical equipment cannot work properly due to insufficient electrical energy may bring inconvenience to workshop production. In order to improve electrical energy reserves, the capacity of the battery is often increased, and the volume and weight of the battery will also significantly increase, resulting in a significant increase in the load of waste smoke trucks. How to balance the volume, weight, and functional implementation of electrical equipment is also the problem addressed in this paper.

In order to solve the problems of untimely storage of filter rods in the high-speed station platform of the filter rod automatic stereoscopic library (ASRS) and slow response of the space hopper group, a task management system of stacker crane is developed, which specifically involves the task priority management system based on the stacker crane with one rail and two vehicles. The system is based on the stacker crane task management system SCS, combined with the upper management system WMS and stacker crane plc control system, and sets the corresponding priority for the inbound and outbound platforms of different vehicle speeds, so as to realize the task priority management of the stacker crane. The automatic stereo library for filter rods (ASRS) is tested as the object, and the results show that the system can realize the task priority management of the storage platform of the stacker, and ensure the priority response of the high-speed machine platform.

In order to solve the problems of the color of different auxiliary material and the reflection of the material itself, for each auxiliary material, the visual inspection of light source is designed to improve the accuracy of obtaining the original picture. In this paper, the FOCKE-F5 double aluminum packaging machine is taken as the experimental machine, and take a certain brand for example, through the comprehensive design and research on the color of light source, intensity of light source and mechanism angle to achieve the best detection effect. Finally, the system structure is reasonable, parameters can be adjusted, stable and reliable.

针对国内7～8MA脉冲功率装置驱动条件,通过耦合等效电路模型和McBride等人发展的半解析模型,研究了MagLIF总体能量学过程及中子产额随关键参数的变化规律,获得了中子产额大于1010的参数设计区间.结果表明:7～8 MA驱动条件、套筒材料、负载高度、燃料半径与密度、预热能量、外加轴向磁场等多因素共同决定了燃料的最终压缩状态;预热能量越大,燃料初始升温以及滞止时刻升温越高,中子产额越高;轴向磁场增加,热传导能量损失减小,但燃料收缩比也会减小,因此存在优化轴向磁场以获得较高中子产额;杂质质量分数超过10％,中子产额开始显著下降.燃料密度0.7 mg/cm3、外加轴向磁场27 T、预热能量200 J、杂质质量分数小于50％的条件下,可以获得3.5×1010中子产额,从而有望在7～8 MA条件下建立MagLIF关键问题研究平台.

In order to realize the inspection of cigarette appearance size, an online inspection method based on machine vision is proposed to realize the intersection inspection of cigarette outer diameter, cigarette total length and tipping paper length. The weighted average method is used to grayize the image to remove useless parts and reduce the size of the original image collected; The noise points in the image are removed by bilateral filtering to reduce the error of edge extraction, and the image is binarized by the maximum variance between classes method; The improved Canny operator is used to extract the contour of cigarette image, and the least square method is used to fit the pixel size; By calibrating the dimensions of the standard parts, the cigarette outline measurement dimensions are obtained. The experimental object is a cigarette with a length of 84mm and a diameter of 7.7mm. This method has the characteristics of high efficiency and high accuracy, and has high practical value. Through experiments, this method can quickly detect the appearance size of cigarettes, and the detection accuracy is higher than 99.5%, meeting the detection standard requirements of actual production.

Cigarette industry enterprises establish a Manufacturing Execution System (MES) to collect and process the operating parameters of equipment in each production workshop. In the local area network, the local time of each equipment and the MES system time will have a large shift, which affects the accuracy of the data collected by the MES system. A time synchronization method of devices in local area network based on MES is proposed to solve this problem. Through the technical solution of Eclipse+Jave, the device can be connected to the MES system to obtain the MES system time which will be set as the local time of the device, so that can greatly reduce the shift between the local time of device and the MES system time. 10 devices were selected to count 8 weeks data of the time shift between them and MES system time before and after using the time synchronization method separately. After using the time synchronization method, the mean value of the time shift between the device local time and the MES system time is reduced by 99.12%, and the overall standard deviation is 0.315, which is much smaller than 37.401 that obtained before using the time synchronization method. Meanwhile, the time shift of the device local time and the MES system time is small, and is stable at about 0.497s.

In fast Z-pinches, rise time of drive current plays an important role in development of magneto-Rayleigh-Taylor(MRT) instabilities. It is essential for applications of Z-pinch dynamic hohlraum (ZPDH), which could be used for drivinginertial confinement fusion (ICF), to understand the scaling of rise time on MRTs. Therefore, a theoretical model for nonlinear development of MRTs is developed according to the numerical analysis. It is found from the model that the implosion distance L = r 0 – r mc determines the development of MRTs, where r 0 is the initial radius and r mc is the position of the accelerating shell. The current rise time τ would affect the MRT development because of its strong coupling with the r 0. The amplitude of MRTs would increase with the rise time linearly if an implosion velocity is specified. The effects of the rise time on MRT, in addition, are studied by numerical simulation. The results are consistent with those of the theoretical model very well. Finally, the scaling of the rise time on amplitude of MRTs is obtained for a specified implosion velocity by the theoretical model and numerical simulations.

In recent years, the industrial automation of cigarette enterprises has made great progress, and the underlying data acquisition, centralized control and condition monitoring systems have been established. The management has also established a large number of application systems. There are a large number of equipment data resources in these industrial automation systems and application systems, but incremental data value mining is difficult to really solve the problem. The exploration of big data and machine learning has gradually encountered bottlenecks and ceilings, Lack of integration of domain knowledge, it is difficult to mine high-value application scenarios. From the point of view of "semantic analysis" and "equipment management", it is necessary to establish a unified method of data collection and application, which is the basis of effective data collection and management. From the perspective of "semantic analysis" and "equipment management", it is necessary to establish a unified method of data collection and application.

Cigarette length deviation is an important index of cigarette quality control. Domestic cigarette machines have no real-time online detection function of cigarette length. The tipping paper used in cigarette manufacturing has the phenomenon of position offset due to the influence of tension or the axial movement of tipping paper tray frame during the operation of the equipment. In order to solve this problem and ensure product quality, this paper designs an intelligent deviation correction device of tipping paper based on visual detection of cigarette length, which can detect the position change of tipping paper in the process of transportation and cutting in real time. The automatic deviation correction and accurate control of tipping paper are carried out through the control circuit, and the automation of position detection and adjustment of tipping paper is realized.