Machine vision monitoring system for industrial production progress management

Abstract

The invention relates to the technical field of industrial production progress management, in particular to a machine vision monitoring system for industrial production progress management, which comprises a data acquisition module, an image processing module, a progress management module, a production progress monitoring module and a user interface module. According to the invention, the image processing module realizes high-precision object recognition and state analysis by using a convolutional neural network in combination with a histogram weighting algorithm; the progress management module predicts the production progress based on the long and short term memory network in combination with an attention mechanism, optimizes task scheduling through an adaptive genetic algorithm, and ensures seamless connection among the processes; the production progress monitoring module displays a production progress state in real time, realizes fault early warning by adopting a fuzzy logic algorithm, and dynamically adjusts an early warning rule according to an actual demand; through the functions of efficient data storage and retrieval, real-time monitoring and feedback, automatic report generation and the like, the production efficiency and the product quality are remarkably improved, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial production progress management, and in particular to a machine vision monitoring system for industrial production progress management. Background Art

[0002] In traditional industrial production environments, monitoring of production progress and equipment status usually relies on manual inspection and recording. This approach has significant limitations. First, manual monitoring requires a lot of time and manpower, especially in large-scale production lines, and it is difficult to achieve comprehensive coverage. Second, manual monitoring is easily affected by subjective factors, resulting in inaccurate data or missing key information. In addition, manually recorded data often lags behind, and it is impossible to detect and handle abnormal situations in the production process in a timely manner. Therefore, traditional manual monitoring methods can no longer meet the high-efficiency, accurate and real-time requirements of modern industrial production.

[0003] With the development of technology, some automated monitoring systems have been introduced into industrial production. These systems mainly include sensor-based monitoring systems, video monitoring systems and Industrial Internet of Things (IIoT) systems. Sensor-based monitoring systems collect environmental data on the production line in real time by installing various sensors (such as temperature sensors, pressure sensors, etc.), and transmit these data to the central control system for analysis, and adjust production parameters according to the results. However, the limitation of such systems is that they mainly focus on environmental data and lack effective monitoring methods for the state of objects in the production process (such as position, shape, quality, etc.). Therefore, although they can provide information on some production environments, they cannot fully reflect the overall status of the production line. Video monitoring systems deploy multiple cameras to monitor various areas on the production line in real time, and store video data on the server for subsequent viewing and analysis. analysis; although this system can improve monitoring efficiency to a certain extent, it still relies on manual observation of videos, lacks automated image analysis capabilities, and is difficult to achieve real-time monitoring and early warning; in addition, video surveillance systems usually generate a large amount of data, which increases the complexity of storage and management, and is also unable to respond quickly to emergencies; the Industrial Internet of Things (IIoT) system connects production equipment through IoT technology to achieve data exchange and collaborative work between devices; managers can monitor and control production equipment in real time through remote terminals, improving production flexibility and controllability; however, although the IIoT system can achieve device interconnection and remote control, for complex production scenarios, it is still necessary to combine other technical means (such as machine vision) to improve monitoring accuracy and effectiveness; in addition, when processing large amounts of data, the IIoT system may face problems such as data transmission delays and bandwidth bottlenecks.

[0004] In recent years, the application of machine vision technology in industrial production has gradually increased, providing new ideas for solving the above problems. Machine vision technology uses computer algorithms and image processing technology to extract useful information from images, and has the following advantages: High-precision recognition: It can accurately identify the position, shape, color and other characteristics of objects, and is suitable for detection tasks in complex production environments; Strong real-time performance: can quickly process large amounts of image data and achieve real-time monitoring and feedback; High degree of automation: image acquisition, processing and analysis can be completed automatically without manual intervention, greatly improving work efficiency; However, although machine vision technology performs well in certain specific application scenarios, its application in industrial production schedule management still faces some challenges. For example, how to seamlessly integrate machine vision technology with production management systems, how to ensure the accuracy and reliability of data, and how to cope with complex production environment changes are all technical problems that need to be solved urgently. Therefore, a machine vision monitoring system for industrial production schedule management is proposed to address the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a machine vision monitoring system for industrial production schedule management to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: A machine vision monitoring system for industrial production schedule management, comprising: Data acquisition module: used to collect image data on the production line in real time and transmit the image data to the image processing module; Image processing module, used to clean, enhance, extract features and identify objects in image data, and output production status analysis results; The progress management module is used to perform production progress forecasting, task scheduling and fault warning based on the production status analysis results, and dynamically adjust the production management strategy; Production progress monitoring module, used to display the production progress status in real time and trigger abnormal alarms according to preset rules; Data storage module, used to store image data, progress data and scheduling records during the production process; The user interface module provides a visual interface to support interactive operations of production personnel.

[0007] As a preferred solution, the cleaning and enhancement functions of the image processing module are implemented by a preset image processing algorithm, which includes a non-local mean filtering method that introduces gradient information optimization, wherein the pixel similarity weight The calculation formula is: ; in, and Represent the positions of two pixels respectively. and Represents pixel points and The gray value vector of and Represents pixel points and The gradient of is the gradient weight coefficient, Ensure that the weights sum to 1 for the normalization factor; is a smoothing parameter that controls the scale of similarity.

[0008] As a preferred solution, the progress prediction function of the progress management module is implemented through a time series prediction algorithm, which includes a long short-term memory network model optimized by the attention mechanism, in which the attention-weighted hidden state The calculation formula is: ; in, represents the sequence length, Indicates The hidden state of time steps, Indicates Time step The attention weight of the time step reflects the importance of different time steps.

[0009] As a preferred solution, the task scheduling function of the progress management module is implemented through an optimization algorithm, which includes a genetic algorithm with adaptive crossover probability and mutation probability. and mutation probability The calculation formulas are: ; ; in, is the maximum fitness value in the population, is the average fitness value of the population, is the fitness value of the parent individual, is the fitness value of the individual to be mutated, , , , is a constant used to adjust the crossover and mutation probabilities.

[0010] As a preferred solution, the early warning rule engine of the production progress monitoring module is implemented through a fuzzy logic algorithm. The input variables include production progress deviation and equipment parameter deviation, and the output is a dynamically adjusted early warning level. The fuzzy rule is expressed as: ,in, and They represent the production schedule deviation and equipment parameter deviation respectively. and They represent the fuzzy sets of input variables, is an index variable used to distinguish different production schedule deviation fuzzy sets. is the index used to distinguish the input variables The corresponding fuzzy sets are is the output variable, Represents the fuzzy set of the output variable.

[0011] As a preferred solution, the fault warning function of the progress management module is realized through the support vector machine model with parameter optimization. The parameter optimization adopts the particle swarm algorithm to dynamically adjust the particle speed. and location The update formula is: ; ; in, Indicates The particle in Speed ​​in dimension; is the inertia weight; and is the acceleration factor; and is a random number; is the historical optimal position of the particle; is the global optimal position; Indicates The particle in The new position in dimension, is the particle number, Indicates the number of iterations.

[0012] As a preferred solution, the feature extraction and object recognition functions of the image processing module are realized through a deep learning model, which includes a composite algorithm combining a convolutional neural network and a weighted directional histogram, wherein the feature descriptor The calculation formula is: ,in, is the direction histogram, which represents the direction distribution of each area in the image; is the weighting coefficient, which is used to adjust the weights in different directions; Represents an element-wise multiplication operation.

[0013] As a preferred solution, the schedule management module uses a reinforcement learning algorithm to dynamically optimize the production strategy. The reinforcement learning algorithm includes a deep Q network based on prioritized experience replay, in which the priority of the experience samples is The calculation formula is: ,in, is the temporal difference error, which measures the difference between the current estimate and the target value; is an index used to distinguish different experience samples; A small positive number to prevent the priority from being zero; It is a priority control parameter that determines the sensitivity of the priority.

[0014] As a preferred solution, the progress forecast of the progress management module is optimized in combination with multi-dimensional production factors, including raw material supply, personnel scheduling and equipment maintenance plan.

[0015] It can be seen from the technical solution provided by the present invention that the machine vision monitoring system for industrial production schedule management provided by the present invention has the following beneficial effects: 1. Improve production efficiency Real-time monitoring and feedback: By collecting and analyzing image data on the production line in real time, problems in the production process can be discovered and corrected in a timely manner, reducing downtime and waste, thereby improving overall production efficiency; Optimize task scheduling: Use adaptive genetic algorithms and reinforcement learning technologies to dynamically optimize task allocation and resource scheduling to ensure seamless connection between processes and maximize production efficiency; 2. Enhance production stability Fault warning and prevention: Fault warning is performed based on the support vector machine (SVM) model and the particle swarm algorithm (PSO), identifying potential problems in advance and issuing alarms to prevent production interruptions caused by equipment failures and ensure production continuity; Multi-level alarm mechanism: Set different alarm levels according to the severity of the abnormality, and notify relevant personnel through various means (such as sound, light, SMS, email, etc.) to ensure that the problem is handled in a timely manner; 3. Improve product quality High-precision object recognition: Use convolutional neural network (CNN) combined with directional histogram weighted composite algorithm for object recognition, ensuring accurate identification of target objects even in complex environments, and improving the accuracy of product quality inspection; Production status analysis: By analyzing the extracted features and recognition results, a detailed production status report is generated to help managers adjust the production process in a timely manner to ensure the consistency and stability of product quality; 4.Data management and decision support Efficient data storage and retrieval: Adopt a combination of relational databases, non-relational databases and time series databases to efficiently store and manage the large amount of data generated during the production process; support fast and flexible data query and statistical analysis, and provide scientific decision-making basis for management; Automated report generation: Automatically generate production progress reports, equipment maintenance reports, etc., and support export to multiple formats such as PDF and Excel to facilitate management personnel to review and analyze. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The figure is a schematic diagram of the overall structure of a machine vision monitoring system for industrial production schedule management according to the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0019] like Figure 1 As shown, an embodiment of the present invention provides a machine vision monitoring system for industrial production progress management, including a data acquisition module, an image processing module, a progress management module, a production progress monitoring module, a data storage module and a user interface module.

[0020] In this embodiment, the data acquisition module is used to collect image data on the production line in real time and transmit the image data to the image processing module; The data acquisition module is the basic component of the entire machine vision monitoring system for industrial production progress management. It is responsible for real-time acquisition of image data on the production line and transmits it to the subsequent processing module. The design and implementation of this module is crucial to ensure that the system can accurately and efficiently obtain key information in the production process. The following is a detailed description of its functions, composition and technical details. 1. Functional Overview The main functions of the data acquisition module include: Real-time image acquisition: Capture image data in real time from cameras or other image sensors on the production line; Data preprocessing: Perform preliminary processing on the original image (such as format conversion, compression, etc.) to facilitate efficient processing of subsequent modules; Data transmission: The processed image data is transmitted to the image processing module via the network or local bus.

[0021] 2. Components The data acquisition module usually consists of the following key components: 2.1 Image Sensor Type: Can be a CCD (charge-coupled device) or CMOS (complementary metal oxide semiconductor) sensor; Resolution: Select the appropriate resolution based on the application scenario to ensure a balance between image quality and processing efficiency; Frame rate: A high frame rate helps capture fast-changing production scenes, but it also increases the amount of data and the processing burden; 2.2 Camera Fixed cameras: installed in fixed locations to monitor production conditions in specific areas; Mobile camera: can be mounted on a robot arm or other mobile device, and can flexibly adjust the viewing angle to cover a wider production area; Multi-camera system: multiple cameras work together to provide a full range of monitoring perspectives; 2.3 Data acquisition card / interface Hardware interface: supports multiple standard interfaces, such as USB, GigEVision, CameraLink, etc., for connecting image sensors and computer systems; Driver: Provide compatible operating system drivers to ensure that image data can be stably and efficiently transmitted to the computer system; 2.4 Data Preprocessing Unit Image format conversion: converting raw image data into a standard format suitable for subsequent processing (such as JPEG, PNG, TIFF, etc.); Image compression: Use appropriate compression algorithms (such as JPEG, H.264, etc.) to reduce data volume and transmission bandwidth requirements; Timestamp: Add accurate timestamps to each frame of the image to facilitate time synchronization during subsequent analysis; 2.5 Network Transmission Unit Local Area Network (LAN): transmits image data to a central server or processing node via wired or wireless LAN; Wide Area Network (WAN): Data transmission via the Internet or a dedicated network when remote monitoring is required; Data cache: Provides a temporary storage mechanism to prevent data loss when the network is unstable.

[0022] 3. Technical Details 3.1 Real-time: Low latency: To ensure real-time production monitoring, the data acquisition module must have low latency characteristics to minimize the time interval from image capture to transmission and processing; Priority scheduling: When the network is congested, you can set priorities to ensure timely transmission of key data packets; 3.2 Reliability: Redundant design: Improve system reliability through dual-machine hot standby or multiple redundant transmission paths to avoid data loss caused by single point failure; Error detection and recovery: built-in verification mechanism (such as CRC verification) and automatic retransmission mechanism to ensure data integrity and consistency; 3.3 Scalability: Modular design: supports on-demand expansion, and can add more image sensors or cameras according to actual production needs; Software-defined: Dynamically adjust acquisition parameters (such as resolution, frame rate, etc.) through configuration files or API interfaces to adapt to different production environments.

[0023] 4. Application Examples Assume that in an automobile manufacturing plant, the data acquisition module is deployed to monitor the welding process on the assembly line; the specific application is as follows: Image sensor: Multiple high-resolution cameras installed around the welding robot capture images of the welding process in real time; Data acquisition card: transmits image data to the central server through the GigEVision interface; Data preprocessing unit: compresses the image and adds a timestamp to ensure that the data volume is moderate and convenient for subsequent analysis; Network transmission unit: The image data is transmitted to the image processing module in real time through the factory's internal LAN for further analysis of welding quality.

[0024] In this embodiment, the image processing module is used to clean, enhance, extract features and identify objects on the image data, and output the production status analysis results; The image processing module is one of the core components in the machine vision monitoring system for industrial production schedule management. It is responsible for cleaning, enhancing, feature extraction, object recognition and other operations on the image data obtained from the data acquisition module, and outputs the production status analysis results. The technical implementation of this module directly affects the overall performance and accuracy of the system. The following is a detailed description of its functions, composition and technical details. 1. Functional Overview The main functions of the image processing module include: Image cleaning: remove noise and interference to improve image quality; Image enhancement: Improve the contrast and clarity of images through various algorithms to facilitate subsequent processing; Feature extraction: extract key features (such as edges, corners, textures, etc.) from the image for further analysis; Object recognition: Identify target objects in images using deep learning models or traditional computer vision methods; Production status analysis: Based on the extracted features and recognition results, analyze the production status on the production line and generate corresponding reports.

[0025] 2. Components Image processing modules usually consist of the following key components: 2.1 Image Preprocessing Unit Denoising: Use filters (such as Gaussian filtering and median filtering) to remove noise from the image; Non-local mean filtering: The non-local mean filtering method based on gradient information optimization is introduced, and its calculation formula is: ; in, and Represent the positions of two pixels respectively. and Represents pixel points and The gray value vector of and Represents pixel points and The gradient of is the gradient weight coefficient, Ensure that the weights sum to 1 for the normalization factor; is a smoothing parameter that controls the scale of similarity; 2.2 Image Enhancement Unit Contrast adjustment: Improve image contrast through histogram equalization, adaptive contrast enhancement and other methods; Sharpening: Use the Laplacian operator or other sharpening filters to enhance image edges; Color correction: adjusts the color balance of the image based on ambient light conditions to ensure consistency; 2.3 Feature Extraction Unit Edge detection: Use edge detection algorithms such as Sobel and Canny to extract edge information from images; Corner detection: Use Harris corner detection or FAST algorithm to extract corners in the image; Direction Histogram (HOG): Calculates the direction distribution of each region in the image as part of the feature descriptor. The calculation formula is: ,in, is the direction histogram, which represents the direction distribution of each area in the image; is the weighting coefficient, which is used to adjust the weights in different directions; Represents an element-by-element multiplication operation; 2.4 Object Recognition Unit Convolutional Neural Network (CNN): Use pre-trained deep learning models (such as ResNet, VGG, etc.) for object recognition; Composite algorithm of weighted combination of directional histogram: Combine convolutional neural network with directional histogram to improve recognition accuracy; Template matching: In some specific scenarios, template matching methods can be used to identify known target objects; 2.5 Production Status Analysis Unit Status classification: Based on the extracted features and identification results, the production status is classified using classification algorithms such as support vector machine (SVM) and random forest; Anomaly detection: Detect anomalies on the production line by setting thresholds or using anomaly detection algorithms (such as Isolation Forest, DBSCAN, etc.); Report generation: Organize the analysis results into easy-to-understand reports for management reference.

[0026] 3. Technical Details 3.1 Cleaning and enhancement: Non-local mean filtering: This method not only considers the grayscale difference between pixels, but also introduces gradient information, which makes the denoising effect better while retaining more detail information; Adaptive contrast enhancement: Automatically adjust the contrast according to the brightness of different areas to avoid overexposure or underexposure; 3.2 Feature Extraction Multi-scale feature extraction: extract features at different scales to deal with target objects of different sizes; Rotation invariance: By using rotation invariant feature descriptors (such as SIFT and SURF), the feature consistency under different viewing angles is ensured; 3.3 Object Recognition Deep learning model: uses convolutional neural network (CNN) for end-to-end learning, automatically extracting high-level features and performing classification; Transfer learning: Leverage pre-trained models and fine-tune them on specific tasks, reducing training time and data requirements; 3.4 Production status analysis Real-time: By optimizing algorithms and hardware acceleration (such as GPU), the analysis results can be fed back to the management system in real time. Explainability: Provides detailed analysis reports, including the location, type, and severity of detected anomalies, to help managers make timely decisions.

[0027] 4. Application Examples Assume that in an electronic component manufacturing plant, the image processing module is deployed to monitor the soldering quality of printed circuit boards (PCBs); the specific application is as follows: Image cleaning: Use non-local mean filtering to remove noise in the image and ensure the clarity of the welding points; Image enhancement: Improve the visibility of welds through contrast adjustment and sharpening; Feature extraction: Use edge detection and orientation histogram to extract key features of welding points; Object recognition: Use convolutional neural networks to identify whether there are defects in the welding points (such as cold welding, leaking welding, etc.); Production status analysis: Generates welding quality reports based on the identification results, marks problematic welding points, and recommends corresponding corrective measures.

[0028] In this embodiment, the progress management module is used to perform production progress prediction, task scheduling and fault warning based on the production status analysis results, and dynamically adjust the production management strategy; The schedule management module is a key component in the machine vision monitoring system for industrial production schedule management. It is responsible for executing production schedule prediction, task scheduling and fault warning based on the production status analysis results provided by the image processing module, and dynamically adjusting the production management strategy. This module uses a variety of advanced algorithms and technical means to ensure that the production process runs efficiently and stably, and can respond to abnormal situations in a timely manner. The following is a detailed description of its functions, composition and technical details; 1. Functional Overview The main functions of the progress management module include: Production progress forecast: predict future production progress based on historical data and real-time data; Task scheduling: Optimize task allocation and resource scheduling based on current production status and forecast results; Fault warning: By monitoring equipment parameters and production progress deviations, potential problems can be discovered in advance and early warnings can be issued; Dynamically adjust production strategies: Dynamically adjust production plans and management strategies based on actual conditions to respond to emergencies or optimize production efficiency.

[0029] 2. Components The progress management module usually consists of the following key components: 2.1 Production schedule forecasting unit Time series prediction algorithm: Use the long short-term memory network (LSTM) combined with the attention mechanism to predict the production progress; the calculation formula is: ; in, represents the sequence length, Indicates The hidden state of time steps, Indicates Time step The attention weight of the time step reflects the importance of different time steps; Multi-dimensional production factor optimization: further optimize multi-dimensional factors such as raw material supply, personnel scheduling and equipment maintenance plan to improve forecast accuracy; 2.2 Task Scheduling Unit Adaptive genetic algorithm: Task scheduling is performed through a genetic algorithm with adaptive crossover probability and mutation probability; adaptive crossover probability and mutation probability The calculation formulas are: ; ; in, is the maximum fitness value in the population, is the average fitness value of the population, is the fitness value of the parent individual, is the fitness value of the individual to be mutated, , , , is a constant used to adjust the crossover and mutation probabilities; Reinforcement learning algorithm: Uses a deep Q network (DQN) based on prioritized experience replay to dynamically optimize production strategies; Priority of experience samples The calculation formula is: ,in, is the temporal difference error, which measures the difference between the current estimate and the target value; is an index used to distinguish different experience samples; A small positive number to prevent the priority from being zero; It is the priority control parameter, which determines the sensitivity of the priority; 2.3 Fault warning unit Support vector machine model: Fault warning is performed through a parameter-optimized support vector machine (SVM) model; parameter optimization uses a particle swarm algorithm (PSO), and the update formula is: ; ; in, Indicates The particle in Speed ​​in dimension; is the inertia weight; and is the acceleration factor; and is a random number; is the historical optimal position of the particle; is the global optimal position; Indicates The particle in New position in dimension; is the particle number, Indicates the number of iterations; Fuzzy logic algorithm: The early warning rule engine is implemented through fuzzy logic algorithm. The input variables include production schedule deviation and equipment parameter deviation, and the output is the dynamically adjusted early warning level. The fuzzy rule is expressed as: ,in, and They represent the production schedule deviation and equipment parameter deviation respectively. and They represent the fuzzy sets of input variables, is an index variable used to distinguish different production schedule deviation fuzzy sets. is the index used to distinguish the input variables The corresponding fuzzy sets are is the output variable, represents the fuzzy set of output variables; 2.4 Dynamic Adjustment Unit Real-time monitoring and feedback: Continuously monitor production progress and equipment status, and dynamically adjust production plans and management strategies based on real-time data; Adaptive adjustment: Dynamically adjust warning rules and management strategies according to different production stages and product types to adapt to diverse production needs.

[0030] 3. Technical Details 3.1 Production schedule forecast: Attention mechanism: By introducing the attention mechanism, the model can pay more attention to important time steps and improve prediction accuracy; Multi-dimensional factor optimization: Combine factors such as raw material supply, personnel scheduling and equipment maintenance plan, comprehensively consider the impact of multiple dimensions, and improve the comprehensiveness and accuracy of the forecast; 3.2 Task Scheduling: Adaptive genetic algorithm: By adaptively adjusting the crossover probability and mutation probability, the search capability and convergence speed of the genetic algorithm are enhanced; Reinforcement learning: Using the priority experience replay mechanism, the model can learn effective strategies more quickly, improving the efficiency and effectiveness of task scheduling; 3.3 Fault warning Support Vector Machine (SVM): The support vector machine model with optimized parameters can effectively identify complex fault modes and provide high-precision fault warnings; Particle Swarm Optimization (PSO): Optimize the parameters of SVM through the particle swarm algorithm to improve the generalization ability and robustness of the model; 3.4 Dynamic Adjustment Real-time monitoring: By real-time monitoring of production progress and equipment status, abnormal situations can be discovered and responded to in a timely manner to ensure the continuity and stability of production; Adaptive adjustment: Flexibly adjust warning rules and management strategies according to different production stages and product types to meet diverse needs.

[0031] 4. Application Examples Assume that in an automobile manufacturing plant, the progress management module is deployed to monitor the production progress and equipment status on the assembly line; the specific application is as follows: Production schedule prediction: Based on the historical data of the past few months, LSTM combined with the attention mechanism is used to predict the production schedule for the next week. At the same time, factors such as raw material supply, staff scheduling, and equipment maintenance plans are considered to further optimize the prediction results. Task scheduling: Based on the current status of the production line and the prediction results, an adaptive genetic algorithm is used to optimize task allocation and resource scheduling to ensure seamless connection between various processes; Fault warning: By combining the support vector machine model with the particle swarm algorithm, the equipment parameters and production progress deviations can be monitored in real time to detect potential fault hazards in advance and issue warnings; Dynamic adjustment: Dynamically adjust production plans and management strategies based on real-time monitoring data to ensure efficient operation of the production process and stable product quality.

[0032] In this embodiment, the production progress monitoring module is used to display the production progress status in real time and trigger abnormal alarms according to preset rules; The production progress monitoring module is an important component in the machine vision monitoring system for industrial production progress management. It is responsible for displaying the production progress status in real time and triggering abnormal alarms according to preset rules. This module uses a variety of advanced algorithms and technical means to ensure that the production process can be monitored efficiently and accurately, and issue alarms in time when abnormalities occur, so that managers can take corresponding measures. The following is a detailed description of its functions, composition and technical details; 1. Function Overview: The main functions of the production progress monitoring module include: Real-time display of production progress: Display the status and progress of the current production line with a visual interface; Abnormality detection and alarm: Detect abnormal situations that may occur in the production process and issue alarms based on preset rules and real-time data; Dynamically adjust warning rules: Flexibly adjust warning rules according to different production stages and product types to adapt to diverse production needs.

[0033] 2. Components The production progress monitoring module usually consists of the following key components: 2.1 Real-time display unit Visual interface: Provides an intuitive user interface to display the status and progress of the production line in real time; data can be presented in the form of charts, dashboards, etc. Multi-dimensional display: Supports display of production progress from different dimensions (such as time, process, equipment, etc.), helping managers to fully understand the production status; 2.2 Anomaly Detection Unit Fuzzy logic algorithm: The early warning rule engine is implemented through fuzzy logic algorithm. The input variables include production schedule deviation and equipment parameter deviation, and the output is the dynamically adjusted early warning level. The fuzzy rule is expressed as: ,in, and They represent the production schedule deviation and equipment parameter deviation respectively. and They represent the fuzzy sets of input variables, is an index variable used to distinguish different production schedule deviation fuzzy sets. is an index used to distinguish input variables The corresponding fuzzy sets are is the output variable, represents the fuzzy set of output variables; Threshold setting: Based on historical data and experience, set a reasonable threshold to determine whether the production progress is normal; if the threshold is exceeded, an alarm is triggered; 2.3 Alarm unit Multi-level alarm mechanism: Set different alarm levels (such as minor, medium, and severe) according to the severity of the abnormality, and notify relevant personnel through different methods (such as sound, light, SMS, email, etc.); Customized alarm rules: Allow users to customize alarm rules according to specific needs, such as alarm conditions within a specific time period, alarm conditions for specific equipment, etc. 2.4 Dynamic Adjustment Unit Rule engine: The rule engine based on fuzzy logic algorithm can dynamically adjust the warning rules according to different production stages and product types; for example, in the production process of certain products with high precision requirements, more stringent warning thresholds can be set; Feedback mechanism: Through the real-time feedback mechanism, the production progress and equipment status are continuously monitored, and the early warning rules and management strategies are dynamically adjusted according to the actual situation.

[0034] 3. Technical Details 3.1 Real-time display Data visualization technology: Use chart libraries (such as ECharts, D3.js, etc.) and dashboard tools (such as Grafana) to visualize data and provide intuitive production progress display; Multi-dimensional analysis: Supports display of production progress from multiple dimensions (such as time series, process, equipment, etc.), helping managers to fully understand the production status; 3.2 Anomaly Detection Fuzzy logic algorithm: Fuzzy logic algorithm is used to process complex input variables (such as production schedule deviation and equipment parameter deviation) to generate dynamically adjusted warning levels; the applicability calculation formula of fuzzy rules is: ; Or using the product method: ; in, and Respectively represent the membership function values ​​of the input variables; Indicates the applicability of the rules; Threshold setting and optimization: Set reasonable thresholds through historical data analysis and expert experience, and make dynamic adjustments based on actual operating conditions to improve detection accuracy; 3.3 Alarm Mechanism Multi-level alarm: Set different alarm levels according to the severity of the abnormality, and notify through various means (such as sound, light, SMS, email, etc.) to ensure that relevant personnel can respond in time; Customized alarm rules: Allow users to customize alarm rules according to specific needs, such as enabling or disabling certain alarm conditions during a specific time period, or setting special alarm conditions for specific devices; 3.4 Dynamic Adjustment Rule engine: The rule engine based on fuzzy logic algorithm can dynamically adjust the warning rules according to different production stages and product types; for example, in the production process of certain products with high precision requirements, more stringent warning thresholds can be set; Feedback mechanism: Through the real-time feedback mechanism, the production progress and equipment status are continuously monitored, and the early warning rules and management strategies are dynamically adjusted according to the actual situation to ensure the continuity and stability of production.

[0035] 4. Application Examples Assume that in an electronic component manufacturing plant, the production progress monitoring module is deployed to monitor the assembly process of printed circuit boards (PCBs); the specific applications are as follows: Real-time display: The status and progress of each production line are displayed in real time on the visual interface, the completion rate of each workstation is displayed using a line chart, and the overall production efficiency is displayed using a dashboard; Anomaly detection: Monitor production progress deviations and equipment parameter deviations through fuzzy logic algorithms; for example, if the welding process completion rate of a production line is lower than the set threshold (such as 80%), a minor alarm is triggered; if the welding quality inspection failure rate exceeds 5%, a serious alarm is triggered; Alarm mechanism: When an abnormal situation is detected, the system will sound an alarm through sound, light, etc., and automatically send SMS or email to notify relevant management personnel; at the same time, the abnormal location and related information will be highlighted on the visual interface; Dynamic adjustment: The warning rules can be adjusted dynamically according to different production stages and product types. For example, in the production process of high-precision products, the completion rate threshold of the welding process is set to 90%, while in the production process of conventional products, it can be set to 80%. In addition, the warning rules can be adjusted according to the equipment maintenance plan to ensure that the production equipment is in the best working condition.

[0036] In this embodiment, the data storage module is used to store image data, progress data and scheduling records during the production process; The data storage module is a key component in the machine vision monitoring system for industrial production schedule management. It is responsible for storing image data, schedule data and scheduling records in the production process. The design and implementation of this module directly affects the system's data management capabilities, query efficiency, and data security and integrity. The following is a detailed description of its functions, composition and technical details. 1. Functional Overview The main functions of the data storage module include: Data storage: Efficiently and reliably store various data obtained from the data acquisition module and image processing module (such as image data, production status analysis results, etc.); Data retrieval: Support fast and flexible data retrieval functions to facilitate subsequent analysis and use; Data backup and recovery: ensure data security and provide regular backup and disaster recovery mechanisms; Data compression and optimization: Improve storage efficiency and query performance through compression and indexing techniques.

[0037] 2. Components The data storage module usually consists of the following key components: 2.1 Database Management System (DBMS) Relational databases: such as MySQL and PostgreSQL, used to store structured data (such as production progress data, task scheduling records, etc.); Non-relational databases: such as MongoDB and Cassandra, used to store unstructured or semi-structured data (such as image data, log files, etc.); Time series databases: such as InfluxDB and TimescaleDB, which are specifically used to store time series data (such as sensor data, production status changes, etc.); 2.2 Storage Engine File system: used to store large-capacity image data and video files; distributed file systems (such as HDFS and Ceph) can be used to improve storage scalability and reliability; Object storage: such as Amazon S3, MinIO, etc., which provide high availability and scalability storage services and are suitable for large-scale data storage needs; In-memory databases: such as Redis and Memcached, used to cache hot data and improve reading speed; 2.3 Data Compression and Optimization Unit Data compression: Use JPEG, PNG and other formats to compress image data to reduce storage space; for text data, use Gzip, Snappy and other compression algorithms; Indexing mechanism: Create indexes for commonly used query fields to improve query efficiency; for example, create indexes for fields such as production date and equipment number; Partition strategy: partition and store data based on time, device number and other dimensions to improve query performance and management efficiency; 2.4 Data backup and recovery unit Regular backup: Set up scheduled tasks to regularly back up data to off-site storage or cloud storage to ensure data security; Incremental backup: only backs up the data that has changed since the last backup, reducing backup time and storage space; Disaster recovery: Develop a detailed disaster recovery plan to ensure that data can be quickly restored in the event of data loss or corruption; 2.5 Data Security Management Unit Access control: through user authentication and permission management, limit the access rights of different users to ensure data security; Encrypted storage: Encrypt sensitive data to prevent data leakage; Audit log: records all data access and operation logs for easy traceability and review afterwards.

[0038] 3. Technical Details 3.1 Data Storage Multi-source data integration: Support multiple data sources (such as sensors, cameras, production line equipment, etc.) and store them in one or more AI-friendly data warehouses; Hybrid storage architecture: Combining the advantages of relational databases and non-relational databases, choosing the appropriate storage method according to different types of data; for example, structured data is stored in a relational database, and unstructured data is stored in an object storage; 3.2 Data Retrieval Efficient query: Ensure efficient data retrieval through reasonable index design and query optimization; for example, use B+ tree or hash index to speed up query operations; Complex query support: Supports complex query conditions and aggregation operations to meet diverse data analysis needs; for example, query production progress data within a specific time period and calculate statistical information such as average value and maximum value; 3.3 Data Compression and Optimization Image compression: Use efficient image compression algorithms (such as JPEG2000 and WebP) to reduce storage space while ensuring image quality; Data partitioning: partition and store data based on dimensions such as time and device number to improve query performance and management efficiency; for example, partition and store data by day or month; 3.4 Data backup and recovery Automatic backup: Set up automatic backup tasks to regularly back up data to offsite storage or cloud storage to ensure data security; Incremental backup: only backs up the data that has changed since the last backup, reducing backup time and storage space; Disaster recovery: Develop a detailed disaster recovery plan to ensure that data can be quickly recovered in the event of data loss or corruption; for example, use snapshot technology and redundant storage to improve data recovery capabilities; 3.5 Data Security Management Access control: through user authentication and permission management, the access rights of different users are restricted to ensure data security; for example, ordinary users can only view data, while administrators can modify and delete data; Encrypted storage: Encrypt sensitive data to prevent data leakage; for example, use encryption algorithms such as AES to encrypt important data; Audit log: records all data access and operation logs for easy traceability and review; for example, records the time, user, and operation content of each data query.

[0039] 4. Application Examples Assume that in an automobile manufacturing plant, a data storage module is deployed to store various data on the production line; the specific applications are as follows: Data storage: Store image data obtained from the data acquisition module in object storage (such as Amazon S3), and store production progress data and task scheduling records in a relational database (such as MySQL); Data retrieval: Managers can query production progress data within a specific time period through a visual interface and generate reports; for example, query the production completion rate of a production line in the past week and generate a line chart to show the trend; Data compression and optimization: JPEG compression is performed on image data to reduce storage space usage; at the same time, indexes are created for commonly used query fields such as production date and equipment number to improve query efficiency; Data backup and recovery: Set up daily automatic backup tasks to back up data to off-site storage; when data is lost or damaged, it can be quickly restored through backup data; Data security management: through user authentication and permission management, limit the access rights of different users; encrypt and store sensitive data to prevent data leakage; record all data access and operation logs for subsequent tracing and review.

[0040] In this embodiment, the user interface module provides a visual interface to support interactive operations of production personnel; The user interface module is a key component in the machine vision monitoring system for industrial production schedule management. It is responsible for providing a visual interface to support the interactive operations of production personnel. Through an intuitive and user-friendly user interface (UI), this module enables managers to monitor production schedules, view analysis results, adjust production strategies, and receive alarm information in real time. The following is a detailed description of its functions, composition, and technical details. 1. Functional Overview The main functions of the user interface module include: Real-time monitoring: Display the status and progress of the production line to help managers fully understand the current production situation; Data analysis and report generation: Provide data query, statistical analysis and report generation functions to facilitate managers to make decisions; Alarm notification: When an abnormal situation is detected, an alarm is issued to the user in a timely manner and detailed abnormal information is provided; User interaction: Supports users to configure the system, adjust parameters and perform operations such as task scheduling and equipment maintenance planning.

[0041] 2. Components The user interface module usually consists of the following key components: 2.1 Real-time monitoring unit Dashboard: Provides a comprehensive dashboard showing key performance indicators (KPIs) such as production efficiency, completion rate, failure rate, etc. Chart display: Use line charts, bar charts, pie charts and other charts to display production progress, equipment status and other information; Map view: For factories with multiple locations or production lines, the status and location of each production line can be displayed through the map view; 2.2 Data Analysis and Reporting Unit Query tool: supports users to query historical data based on time range, equipment number, process and other conditions; Statistical analysis: Provides basic statistical analysis functions, such as calculating the average, maximum, minimum, standard deviation, etc.; Report generation: Automatically generate production progress reports, equipment maintenance reports, etc., and support export to PDF, Excel and other formats; 2.3 Alarm Notification Unit Real-time alarm: When an abnormal situation is detected, the system will immediately display the alarm information on the interface and remind the user through sound, light, etc.; Alarm records: record the time, type, severity and other information of all alarm events to facilitate subsequent analysis and review; Customized alarm rules: Allow users to customize alarm rules according to specific needs, such as setting alarm conditions within a specific time period, alarm conditions for specific devices, etc. 2.4 User Interaction Unit Task scheduling: supports users to manually adjust task scheduling plans, such as reallocating tasks, adjusting priorities, etc. Equipment Maintenance Plan: Allows users to create and edit equipment maintenance plans and synchronize them to the system; Parameter configuration: Provides a system parameter configuration interface, allowing users to adjust the system's operating parameters according to actual needs.

[0042] 3. Technical Details 3.1 Real-time monitoring Front-end technology stack: Use modern front-end frameworks (such as React, Vue.js) to build dynamic and responsive user interfaces to ensure a good user experience; Data update mechanism: Use WebSocket or polling mechanism to achieve real-time data update to ensure that the interface can reflect the latest production status in a timely manner; Chart library: Integrates powerful chart libraries (such as ECharts, D3.js), supports multiple chart types and rich visualization effects; 3.2 Data Analysis and Reporting Query optimization: Improve query efficiency through indexing and caching mechanisms to ensure fast response under large data volumes; Statistical analysis algorithms: implement common statistical analysis algorithms, such as mean, variance, correlation analysis, etc., to meet diverse data analysis needs; Report template: Provides predefined report templates, users can select the appropriate template and generate reports according to their needs; supports exporting to multiple formats (such as PDF, Excel); 3.3 Alarm Notification Real-time push: Push alarm information to the user's interface through WebSocket or other real-time communication protocols to ensure instant notification; Grading alarm: Set different alarm levels (such as minor, medium, and severe) according to the severity of the abnormality, and notify relevant personnel through different methods (such as sound, light, SMS, email, etc.); Custom rule engine: allows users to customize alarm rules according to specific needs, such as alarm conditions within a specific time period, alarm conditions for specific devices, etc. 3.4 User Interaction Task scheduling interface: Provides an intuitive task scheduling interface that allows users to drag and drop tasks, adjust priorities, reallocate resources, etc. Equipment maintenance plan management: Provides an equipment maintenance plan management interface to support users to create, edit and delete maintenance plans and synchronize them to the system; Parameter configuration interface: Provides a system parameter configuration interface, allowing users to adjust the system's operating parameters according to actual needs, such as threshold settings, alarm rules, etc.

[0043] 4. Application Examples Assume that in an automobile manufacturing plant, a user interface module is deployed to manage and monitor various activities on the production line; the specific application is as follows: Real-time monitoring: The status and progress of each production line are displayed in real time on the dashboard, the completion rate of each workstation is displayed using a line chart, and the location and status of different workshops are displayed using a map view; Data analysis and reporting: Managers can use query tools to query historical data by time range, equipment number and other conditions, and generate production progress reports and equipment maintenance reports; for example, query the production efficiency of a production line in the past month and generate a line chart to show the trend; Alarm notification: When the welding quality failure rate exceeds 5%, the system will display a red alarm on the interface and remind the management personnel through sound, light, etc. At the same time, the system will record the time, type and severity of the alarm event for subsequent analysis and review; User interaction: Managers can manually adjust task allocation in the task scheduling interface, such as scheduling certain high-priority tasks in advance or deferring low-priority tasks; in addition, they can also create and edit equipment maintenance plans in the equipment maintenance plan management interface to ensure that production equipment is in the best working condition.

[0044] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A machine vision monitoring system for industrial production progress management, characterized by: include: Data acquisition module: used to collect image data on the production line in real time and transmit the image data to the image processing module; An image processing module, used for cleaning, enhancing, extracting features and identifying objects of the image data, and outputting production status analysis results; A progress management module is used to perform production progress prediction, task scheduling and fault warning based on the production status analysis results, and dynamically adjust the production management strategy; Production progress monitoring module, used to display the production progress status in real time and trigger abnormal alarms according to preset rules; Data storage module, used to store image data, progress data and scheduling records during the production process; The user interface module provides a visual interface to support interactive operations of production personnel.

2. The machine vision monitoring system for industrial production progress management according to claim 1, characterized in that: The cleaning and enhancement functions of the image processing module are realized by a preset image processing algorithm, which includes a non-local mean filtering method that introduces gradient information optimization, wherein the pixel similarity weight The calculation formula is: ； in, and Represent the positions of two pixels respectively. and Represents pixel points and The gray value vector of and Represents pixel points and The gradient of is the gradient weight coefficient, Ensure that the weights sum to 1 for the normalization factor; is a smoothing parameter that controls the scale of similarity.

3. The machine vision monitoring system for industrial production schedule management according to claim 1, characterized in that: The progress prediction function of the progress management module is realized by a time series prediction algorithm, which includes a long short-term memory network model optimized by introducing an attention mechanism, wherein the attention-weighted hidden state The calculation formula is: ； in, represents the sequence length, Indicates The hidden state of time steps, Indicates Time step The attention weight of the time step reflects the importance of different time steps.

4. The machine vision monitoring system for industrial production schedule management according to claim 1, characterized in that: The task scheduling function of the progress management module is implemented by an optimization algorithm, which includes a genetic algorithm with adaptive crossover probability and mutation probability. and mutation probability The calculation formulas are: ； ； in, is the maximum fitness value in the population, is the average fitness value of the population, is the fitness value of the parent individual, is the fitness value of the individual to be mutated, , , , is a constant used to adjust the crossover and mutation probabilities.

5. The machine vision monitoring system for industrial production schedule management according to claim 1, characterized in that: The early warning rule engine of the production progress monitoring module is implemented by a fuzzy logic algorithm. The input variables include production progress deviation and equipment parameter deviation. The output is a dynamically adjusted early warning level. The fuzzy rule is expressed as: ,in, and They represent the production schedule deviation and equipment parameter deviation respectively. and They represent the fuzzy sets of input variables, is an index variable used to distinguish different production schedule deviation fuzzy sets. is the index used to distinguish the input variables The corresponding fuzzy sets are is the output variable, Represents the fuzzy set of the output variable.

6. The machine vision monitoring system for industrial production progress management according to claim 1, characterized in that: The fault warning function of the progress management module is realized by a support vector machine model with parameter optimization, and the parameter optimization is dynamically adjusted by a particle swarm algorithm, wherein the particle speed and location The update formula is: ； ； in, Indicates The particle in Speed ​​in dimension; is the inertia weight; and is the acceleration factor; and is a random number; is the historical optimal position of the particle; is the global optimal position; Indicates The particle in The new position in dimension, is the particle number, Indicates the number of iterations.

7. The machine vision monitoring system for industrial production progress management according to claim 1, characterized in that: The feature extraction and object recognition functions of the image processing module are realized by a deep learning model, which includes a composite algorithm combining a convolutional neural network and a weighted directional histogram. The calculation formula is: ,in, is the direction histogram, which represents the direction distribution of each area in the image; is the weighting coefficient, which is used to adjust the weights in different directions; Represents an element-wise multiplication operation.

8. The machine vision monitoring system for industrial production progress management according to claim 1, characterized in that: The schedule management module uses a reinforcement learning algorithm to dynamically optimize the production strategy. The reinforcement learning algorithm includes a deep Q network based on priority experience playback, where the priority of the experience samples is The calculation formula is: ,in, is the time difference error, which measures the difference between the current estimate and the target value. is an index used to distinguish different experience samples; A small positive number to prevent the priority from being zero; It is a priority control parameter that determines the sensitivity of the priority.

9. The machine vision monitoring system for industrial production schedule management according to claim 1, characterized in that: The schedule forecast of the schedule management module is optimized in combination with multi-dimensional production factors, which include raw material supply, personnel scheduling and equipment maintenance plan.

Images