Intelligent control method for container painting oven based on visual detection and related equipment

Abstract

The embodiment of the application provides a kind of container painting oven intelligent control method and related equipment based on visual detection, belong to industrial machine vision technical field.The method comprises: the image acquisition of current batch of container is carried out, and box image is obtained;According to box image, workpiece state analysis processing is carried out, and structured workpiece state data is obtained, wherein, workpiece state data includes workpiece category and statistical quantity;Get the recent data of oven environment, and with energy consumption minimization as target, according to oven environment recent data and workpiece state data, energy consumption prediction and parameter optimization are carried out, and optimal control parameter is obtained;According to optimal control parameter, drive the action of actuator in oven, change oven operating state.The application can realize the fine management and on-demand distribution of energy, improve the foresight and response speed of control, enhance the self-adaptive ability to complex working condition.

Description

Technical Field

[0001] This application relates to the field of industrial machine vision technology, and in particular to an intelligent control method and related equipment for container coating drying ovens based on visual inspection. Background Technology

[0002] In container coating production lines, the drying oven is a crucial step in ensuring paint curing and the formation of a high-quality paint film; it is also one of the most energy-intensive units in the production line. Currently, the industry generally employs the following two mainstream technologies for controlling drying ovens: Fixed-program timing control: The operator presets a fixed temperature-time curve in the control system according to the process specifications. Regardless of whether there are workpieces in the drying chamber, or the type and quantity of workpieces, the system operates strictly according to this curve. This is an open-loop control, lacking perception and feedback of actual operating conditions.

[0003] Environmental parameter-based feedback control: A small number of temperature sensors are installed in the drying room, and the classic PID (proportional-integral-derivative) control algorithm is used to dynamically adjust the opening of the natural gas valve or the burner power according to the deviation between the set temperature and the actual temperature, so as to maintain the stability of the ambient temperature.

[0004] Whether using sequential or PID control, the object of their control is the ambient air temperature of the drying chamber, not the workpiece itself. Because the material, paint film thickness, and initial temperature of the workpiece (box) are spatially unevenly distributed, a uniform, "one-size-fits-all" environmental control will inevitably lead to over-drying in some areas (wasting energy) and under-drying in others (quality risk). When the production line needs to switch to producing containers of different specifications and paint types, a fixed control program or a single PID parameter is difficult to adapt quickly, resulting in fluctuating control performance and an inability to consistently maintain optimal energy efficiency.

[0005] Furthermore, the aforementioned control methods cannot obtain the workpiece's drying status online and in real time. The determination of dryness depends on manual periodic sampling or offline laboratory measurement. The adjustment of control commands lags far behind changes in actual process requirements and cannot achieve real-time optimization. Summary of the Invention

[0006] The main objective of this application is to propose a visual detection-based intelligent control method and related equipment for container painting drying ovens, aiming to achieve refined energy management and on-demand allocation, improve the foresight and response speed of control, and enhance the adaptability to complex working conditions.

[0007] To achieve the above objectives, one aspect of this application proposes an intelligent control method for container painting drying ovens based on vision detection, the method comprising the following steps: Images of the containers in the current batch are acquired to obtain images of the container bodies; The workpiece status is analyzed based on the box image to obtain structured workpiece status data, which includes workpiece category and statistical quantity. Acquire current environmental data of the drying chamber, and with the goal of minimizing energy consumption, perform energy consumption prediction and parameter optimization based on the current environmental data of the drying chamber and the workpiece status data to obtain the optimal control parameters; The actuators inside the drying chamber are driven to move according to the optimal control parameters, thereby changing the operating state of the drying chamber.

[0008] In some embodiments, prior to the step of performing workpiece state analysis processing based on the box image to obtain structured workpiece state data, the method further includes the following steps: The box image is preprocessed to obtain a preprocessed image, wherein the preprocessing includes lens distortion correction, illumination equalization and noise filtering; The step of performing workpiece state analysis processing based on the box image to obtain structured workpiece state data includes the following steps: The preprocessed image is used to perform workpiece state analysis to obtain structured workpiece state data.

[0009] In some embodiments, the step of performing workpiece state analysis processing based on the box image to obtain structured workpiece state data includes the following steps: Target detection is performed based on the box image to obtain detection results, wherein the detection results include category labels, confidence scores, and location information of several targets in the box image; The detection results are processed by structured transformation to obtain workpiece status data, which includes workpiece category and workpiece quantity.

[0010] In some embodiments, the step of minimizing energy consumption by predicting energy consumption and optimizing parameters based on current data of the drying chamber environment and workpiece status data to obtain optimal control parameters includes the following steps: Obtain future production information for future batches; The current environmental data of the drying room, the status data of the workpiece, and the future production information are input into a pre-trained prediction model for nonlinear mapping to obtain the predicted energy consumption under different control parameters. With the goal of minimizing energy consumption, the parameters are optimized based on the predicted energy consumption to obtain the optimal control parameters.

[0011] In some embodiments, the training method of the prediction model includes the following steps: Acquire historical data on workpiece status, drying oven environment, and historical energy consumption; Based on the historical data of the workpiece state, the historical data of the drying room environment, and the historical energy consumption data, a neural network model is trained using a machine learning algorithm to learn nonlinear mapping relationships. A loss function with the goal of minimizing energy consumption is introduced to optimize the training of the neural network model, resulting in a well-trained prediction model.

[0012] In some embodiments, the loss function includes a predicted energy consumption, a paint film hardness penalty, an adhesion penalty, and a surface defect penalty, and the expression for the loss function is: Loss = argmin( E_total+λ * Penalty(Q) ); Where Loss is the loss function, E_total is the predicted energy consumption, Q represents the drying quality, Penalty(Q)=α×(1-Hardness_score) +β×(1-Adhesion_score) +γ×Surface_defect_score, Hardness_score is the paint film hardness penalty term, Adhesion_score is the adhesion penalty term, Surface_defect_score is the surface defect penalty term, and λ, α, β, and γ are the weights.

[0013] In some embodiments, driving the actuators in the drying chamber to operate according to the optimal control parameters and changing the operating state of the drying chamber includes the following steps: The instruction is converted according to the optimal control parameters to obtain the control instructions to be sent to the programmable logic controller. The control command is subjected to upper and lower limit amplitude checks and change rate limits based on the preset safety interlock logic of the programmable logic controller to obtain a safety command; The safety instructions drive the actuators inside the drying chamber to change the operating state of the drying chamber.

[0014] To achieve the above objectives, another aspect of this application proposes a vision-based intelligent control system for container painting drying ovens, the system comprising: The visual perception module is used to acquire images of the containers in the current batch and obtain images of the container bodies; The status analysis module is used to perform workpiece status analysis processing based on the box image to obtain structured workpiece status data, wherein the workpiece status data includes workpiece category and statistical quantity. The parameter control module is used to acquire the current environmental data of the drying room and, with the goal of minimizing energy consumption, perform energy consumption prediction and parameter optimization based on the current environmental data of the drying room and the workpiece status data to obtain the optimal control parameters. The drying oven drive module is used to drive the actuators inside the drying oven to operate according to the optimal control parameters, thereby changing the operating state of the drying oven.

[0015] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0016] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0017] To achieve the above objectives, another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0018] The embodiments of this application include at least the following beneficial effects: This application provides a visual inspection-based intelligent control method and related equipment for container painting drying ovens. This scheme acquires images of the containers in the current batch to obtain container images; performs workpiece state analysis processing based on the container images to obtain structured workpiece state data, wherein the workpiece state data includes workpiece category and statistical quantity; acquires current environmental data of the drying oven, and, with the goal of minimizing energy consumption, performs energy consumption prediction and parameter optimization based on the current environmental data and workpiece state data to obtain optimal control parameters; drives the actuators within the drying oven to change the operating state of the drying oven according to the optimal control parameters. This application can achieve refined energy management and on-demand allocation, improve the foresight and response speed of control, and enhance the adaptability to complex working conditions. Attached Figure Description

[0019] Figure 1 This is a flowchart of the intelligent control method for container painting drying oven based on vision detection provided in the embodiments of this application; Figure 2 This is a schematic diagram of the box image provided in the embodiments of this application; Figure 3 This is an architecture block diagram of the pure back-end core control system provided in the embodiments of this application; Figure 4 This is a flowchart of a visual detection-based intelligent control method for container painting drying ovens provided in another embodiment of this application; Figure 5 This is a schematic diagram of the structure of the intelligent control system for container painting drying room based on vision detection provided in the embodiments of this application; Figure 6This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of systems and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0022] Before providing a detailed description of the embodiments of this application, some of the nouns and terms involved in the embodiments of this application will be explained first. The nouns and terms involved in the embodiments of this application are subject to the following interpretations.

[0023] (1) Programmable Logic Controller (PLC): A programmable logic controller is a general-purpose industrial automatic control device that is based on a microprocessor and integrates computer technology, automatic control technology and communication technology. It can load control instructions into memory at any time for storage and execution.

[0024] (2) YOLO algorithm: The YOLO algorithm is a real-time object detection algorithm that puts the entire image into an instance and predicts the bounding box coordinates and the probability of the class to which these boxes belong.

[0025] (3) Current batch: The current batch refers to the workpieces that have been identified in the camera's field of view and are about to enter the drying room.

[0026] (4) Future batch: Future batch refers to the sequence of workpieces that immediately follow the current batch in the production plan.

[0027] In related technologies, both timing-based and PID-based control methods suffer from problems such as misalignment of the controlled object, lack of specificity, control lag, insensitive response, poor process adaptability, and insufficient flexibility.

[0028] In view of this, this application provides a visual inspection-based intelligent control method and related equipment for container painting drying ovens. This method acquires images of the containers in the current batch to obtain container images; performs workpiece state analysis processing based on the container images to obtain structured workpiece state data, including workpiece category and quantity; acquires current environmental data of the drying oven, and, with the goal of minimizing energy consumption, performs energy consumption prediction and parameter optimization based on the current environmental data and workpiece state data to obtain optimal control parameters; and drives the actuators within the drying oven to change the oven's operating state based on the optimal control parameters. This application enables refined energy management and on-demand allocation, improves the foresight and response speed of control, and enhances the adaptability to complex operating conditions.

[0029] The intelligent control method for container coating drying ovens based on vision detection provided in this application relates to the field of industrial machine vision technology. This intelligent control method for container coating drying ovens based on vision detection can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle-mounted terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing the intelligent control method for container coating drying ovens based on vision detection, but is not limited to the above forms.

[0030] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0031] Figure 1 This is an optional flowchart of the intelligent control method for container painting drying ovens based on vision detection provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S104.

[0032] Step S101: Image acquisition is performed on the containers of the current batch to obtain container images.

[0033] Step S102: Perform workpiece status analysis processing based on the box image to obtain structured workpiece status data, wherein the workpiece status data includes workpiece category and statistical quantity.

[0034] Step S103: Obtain current environmental data of the drying chamber, and with the goal of minimizing energy consumption, predict energy consumption and optimize parameters based on the current environmental data of the drying chamber and the workpiece status data to obtain the optimal control parameters.

[0035] Step S104: Drive the actuator in the drying chamber to move according to the optimal control parameters, and change the operating state of the drying chamber.

[0036] In this embodiment, to achieve precise control of on-demand functions, machine vision is introduced, enabling the control system to "see" and "identify" each workpiece about to enter the drying chamber. This transforms the control strategy from "controlling the environment" to "controlling the environment based on the workpiece status," thereby achieving refined energy management and on-demand allocation.

[0037] Specifically, an industrial camera array is deployed at the entrance of the drying chamber to perceive the workpiece in real time. When the container is transported to the designated position, the industrial camera array collects images of the current batch of containers and captures images of the container body that is about to enter the drying chamber.

[0038] like Figure 2 As shown, the box image serves as the information source for subsequent workpiece identification. Through intelligent recognition algorithms, the workpiece status of the box image can be analyzed, identifying information such as the box wheel chocks and surface coating status. This allows for the analysis of workpiece category and the statistical quantity of each workpiece category, providing structured forward-looking information for subsequent intelligent decision-making.

[0039] Intelligent decision-making can solve energy-saving control problems in multi-variety, small-batch production modes by automatically identifying different types and quantities of workpieces and calling the optimal control model that matches them, thereby improving the overall intelligence level and flexibility of the production line.

[0040] For example, by continuously collecting and recording environmental data such as temperature T(t) and wind speed V(t) inside the drying room through sensors deployed in the drying room, the current and past environmental data of the drying room are collected to obtain the current environmental status data of the drying room.

[0041] Furthermore, the current environmental data of the drying chamber and the workpiece status data are input into a pre-trained intelligent optimization model. This model has established a mapping relationship by learning from a large amount of historical data. With the goal of minimizing predicted energy consumption, it extrapolates the energy consumption under different operating conditions and finds a set of optimal control parameters. This allows for early prediction and adjustment of the drying chamber to the best operating conditions, transforming the original "post-event correction" that required waiting for the workpiece to enter the drying chamber into proactive "pre-event planning". This greatly shortens the response time and improves the foresight and response speed of the control.

[0042] Optionally, the entire drying process can be divided into multiple time segments, each corresponding to a different temperature-wind speed combination. In addition, the drying room can also be divided into different zones, each corresponding to a different temperature-wind speed combination.

[0043] Finally, control commands are sent to each actuator in the drying chamber according to the optimal control parameters, driving the burner, fan and other actuators to change the operating state of the drying chamber, and finally completing the closed-loop control of the entire drying chamber's operating conditions.

[0044] Through close collaboration at these four levels, this embodiment achieves a complete technology chain from "seeing the workpiece" to "understanding the needs" and then to "precise control," providing a new and efficient solution to the industry challenge of energy-saving control in drying rooms.

[0045] In some embodiments, prior to step S102, the intelligent control method for container painting drying oven based on vision detection may include, but is not limited to, step S201.

[0046] Step S201: Preprocess the box image to obtain a preprocessed image. The preprocessing includes lens distortion correction, illumination equalization, and noise filtering.

[0047] Step S102 may include, but is not limited to, step S202.

[0048] Step S202: Perform workpiece state analysis processing based on the preprocessed image to obtain structured workpiece state data.

[0049] In this embodiment, in order to eliminate lens distortion, uneven lighting caused by workshop ambient lighting, and image noise, a series of preprocessing steps are performed after acquiring the image stream from the camera to improve the accuracy of subsequent detection.

[0050] Specifically, the preprocessing includes lens distortion correction, illumination equalization, and noise filtering. Lens distortion correction is necessary because the optical characteristics of the camera lens can cause the container to appear curved at the edges of the image, thus affecting the subsequent positioning and size estimation of the container. Therefore, after obtaining the container image, lens distortion correction is first performed on the container image. The camera's intrinsic parameter matrix and distortion coefficients are obtained through camera calibration. Then, the distortion correction algorithm is used to process the container image to obtain a distortion-free image.

[0051] Next, considering the possibility of strong overhead lighting causing localized overexposure in the image, as well as shadows caused by equipment obstruction in the workshop environment, the distortion correction process was further enhanced by lighting balance. Brightness was suppressed in overexposed pixel areas, and contrast was enhanced in shadow areas, thereby restoring the details and textures in the image and obtaining a balanced image.

[0052] Finally, noise filtering is performed on the illumination-equalized image to filter out randomly distributed noise particles, resulting in a clear, reliable, and stable preprocessed image. This provides geometrically accurate and uniformly illuminated input data for subsequent workpiece state analysis.

[0053] In some embodiments, step S102 may include, but is not limited to, steps S301 to S302.

[0054] Step S301: Target detection is performed based on the box image to obtain detection results, which include the category labels, confidence scores, and position information of several targets in the box image.

[0055] Step S302: Perform structured transformation processing based on the detection results to obtain workpiece status data, which includes workpiece category and workpiece quantity.

[0056] In this embodiment, after receiving the box image, the box image is first analyzed in real time using a real-time target detection algorithm to identify the position, category, and quantity of the target (i.e., the workpiece) in the box image. Then, the unstructured detection results are converted into workpiece state data that can be used for subsequent model calculations.

[0057] Preferably, workpieces are detected from the box image using a YOLO model (such as YOLO8) trained on a large amount of field image data. YOLO is a one-stage object detection algorithm that completes object localization and object classification in one step, so it is extremely fast and can meet the real-time requirements of industrial sites (up to tens of frames per second). For any frame of input box image, the output is the bounding box, class label (such as "20-foot dry cargo box", "40-foot refrigerator box"), and confidence score for all detected workpieces in the box image.

[0058] The YOLOv8 model is trained by collecting over 10,000 images of containers on-site, covering various container types, lighting conditions, and angles. Labeled data is obtained by manually annotating bounding boxes and category labels (20-foot standard containers, 40-foot high cube containers, refrigerated containers, etc.). Transfer learning is performed on the labeled data, with a training cycle of 100 epochs. When the YOLOv8 model achieves a detection accuracy of mAP@0.5:0.92 on the test set, the model training ends, resulting in a YOLOv8 model that can identify and distinguish containers from container images.

[0059] For example, after the box image is processed by the YOLOv8 model, the detection result of unstructured data is output, in the format: [x_center, y_center, width, height, confidence, class_id], where x_center and y_center represent the center coordinates of the bounding box, width and height represent the width and height, confidence is the confidence score, and class_id is the class label.

[0060] Furthermore, the raw output of the YOLO model (such as bounding boxes and category labels) is parsed into structured workpiece state data that can be directly used by the backend control model. For example, based on width and height, and category labels, workpieces are categorized into workpiece types such as "20-foot dry cargo box" and "40-foot refrigerator box," and the number of workpieces with the same workpiece type in the box image is counted to obtain the number of workpieces corresponding to each workpiece type. For example, by performing a structured transformation on the detection results, the workpiece state data is obtained as {workpiece list: [{category: 'Type A', quantity: 1}, {category: 'Type B', quantity: 2}]}.

[0061] For example, assuming the output of the YOLO model is [0.45, 0.52, 0.3, 0.4, 0.95, 1], then this unstructured data is converted to: json { "timestamp": "2026-01-12T14:53:23", "workpiece_list": [ { "category": "20-foot dry goods box", "count": 1, "coating_type": "intermediate coating", "confidence": 0.95 } ] }

[0062] Optionally, the workpiece status data may also include a timestamp to calculate the time when the workpiece enters the drying oven, enabling proactive control.

[0063] Specifically, the estimated entry time can be calculated based on the current timestamp, the conveyor belt speed, and the distance from the container to the drying room.

[0064] In addition, timestamps can establish a complete timeline for the processing of each workpiece, facilitating quality analysis.

[0065] In some embodiments, step S103 may include, but is not limited to, steps S401 to S403.

[0066] Step S401: Obtain future production information for future batches.

[0067] Step S402: Input the current status data of the drying room environment, the workpiece status data and future production information into the pre-trained prediction model for nonlinear mapping to obtain the predicted energy consumption under different control parameters.

[0068] Step S403: With the goal of minimizing energy consumption, the parameters are optimized based on the predicted energy consumption to obtain the optimal control parameters.

[0069] In this embodiment, the acquisition of optimal control parameters is an intelligent decision-making process. It is based on a predictive model trained by machine learning. By receiving the current status data of the drying room and the status data of the workpiece, and taking into account the production plan of future batches, it outputs the optimal control parameters that are best for the current batch and can achieve a smooth transition of working conditions and avoid drastic fluctuations in parameters. This forward-looking planning ensures that the energy consumption of the entire production process is optimal, not just for a single workpiece.

[0070] Specifically, based on the current timestamp, future production information of the workpiece sequence immediately following the current batch is obtained in the future. The current status data of the drying room environment, the workpiece status data, and the future production information are integrated into a feature vector that integrates multi-source information. This feature vector mainly includes F_visual, F_env(t), and F_plan, where F_visual is the workpiece status data, F_env(t) is the current status data of the drying room environment, and F_plan is the future production information.

[0071] The pre-trained prediction model learns a nonlinear mapping relationship between (F_visual, F_env) and future energy consumption E(t+1), E(t+2)... through a large amount of historical data. Using this learned nonlinear mapping relationship, it predicts energy consumption based on the current timestamp's corresponding drying room environment status data, workpiece status data, and future production information. The simplified mathematical expression of the nonlinear mapping relationship is to find a function F such that: [E_pred] = F( [F_visual, F_env(t), F_env(t-1), ..., F_plan]); Where E_pred is the energy consumption predicted by the model.

[0072] The predictive model uses virtual simulation to quickly predict future energy consumption under different control parameters within a feasible control parameter space, obtaining multiple predicted energy consumption values. Then, by comparing these predicted energy consumption values, the control parameter corresponding to the minimum predicted energy consumption is selected as the optimal control parameter, thus completing the optimization process.

[0073] In some embodiments, the training method for the prediction model may include, but is not limited to, steps S501 to S502.

[0074] Step S501: Obtain historical data on workpiece status, historical data on drying oven environment, and historical energy consumption data.

[0075] Step S502: Based on historical data of workpiece status, historical data of drying room environment, and historical energy consumption data, a machine learning algorithm is used to train a neural network model to learn nonlinear mapping relationships. A loss function with the goal of minimizing energy consumption is introduced to optimize the training of the neural network model, resulting in a well-trained prediction model.

[0076] In this embodiment, the neural network model to be trained can use LSTM (Long Short-Term Memory Network, a recurrent neural network that is good at processing time-series data) or XGBoost (Limited Gradient Boosting Tree, a high-performance ensemble learning model) to learn decision-making capabilities through a large amount of historical data.

[0077] Specifically, historical data on workpiece status is acquired, and based on the historical timestamp of each workpiece status historical data, the corresponding historical data on the drying room environment and historical energy consumption are matched.

[0078] By learning the nonlinear mapping relationship between historical workpiece state data, historical drying oven environment data, and historical energy consumption data, and introducing a loss function with the objective of minimizing energy consumption for optimization training, the expression of the loss function is as follows: Loss = argmin( E_total+λ * Penalty(Q) ); Where Loss is the loss function, E_total is the predicted energy consumption, Q represents the drying quality, Penalty(Q)=α×(1-Hardness_score) +β×(1-Adhesion_score) +γ×Surface_defect_score, Hardness_score is the paint film hardness penalty term, Adhesion_score is the adhesion penalty term, Surface_defect_score is the surface defect penalty term, and λ, α, β, and γ are the weights.

[0079] Specifically, the penalty factors are set as follows: The paint film hardness penalty is used to penalize paint film hardness that does not meet the standard. The hardness standard is ≥2H (pencil hardness), and the penalty calculation formula is as follows: Hardness_score = (1 - actual hardness / standard hardness)^2.

[0080] Preferably, the weight α of the paint film hardness penalty term is 0.6.

[0081] The adhesion penalty item is used to penalize adhesion failures. The adhesion standard is grade 0 (no peeling) using the cross-cut adhesion test. The penalty calculation formula is: Adhesion_score = Adhesion area ratio × 10.

[0082] Preferably, the weight β of the adhesion penalty term is 0.3.

[0083] The surface defect penalty item is used to penalize defects such as orange peel, runs, and bubbles. The penalty calculation formula is as follows: Surface_defect_score = defect area / total area × 5.

[0084] Preferably, the weight γ of the surface defect penalty term is 0.1.

[0085] It should be noted that each score must meet the following quality constraints: Hardness score ∈ [0,1], Adhesion score ∈ [0,1], and Surface defect score ∈ [0,1].

[0086] The goal of model training is to find a set of optimal future control parameters [Control_params] (such as temperature curves and wind speed) that minimizes the loss.

[0087] The final model output is a set of optimal, future-oriented control parameter sequences, for example: {Zone 1 temperature setting: [80℃, 85℃, 90℃...], Zone 2 wind speed setting: [5m / s, 5m / s, 4m / s...]}.

[0088] In some embodiments, step S104 may include, but is not limited to, steps S601 to S603.

[0089] Step S601: Perform instruction conversion based on the optimal control parameters to obtain the control instructions to be sent to the programmable logic controller.

[0090] Step S602: Based on the preset safety interlock logic of the programmable logic controller, the upper and lower limit amplitude values ​​and the rate of change of the control command are checked to obtain the safety command.

[0091] Step S603: Drive the actuators inside the drying chamber to operate according to the safety instructions, thereby changing the operating status of the drying chamber.

[0092] In this embodiment, after receiving the optimal control parameters output by the prediction model, the optimal control parameters are converted into an instruction format that the programmable logic controller (PLC) can recognize and execute, thus obtaining control instructions.

[0093] Next, the instructions are controlled by a security gateway, which is responsible for communicating with the PLC (using common protocols such as OPC UA or Modbus TCP / IP). It has built-in strict safety logic, including checking the upper and lower limits of all issued instructions, limiting the rate of change, and binding it with the PLC's inherent safety interlock logic such as emergency stop, over-temperature, and over-pressure, to ensure that no control instruction will endanger the safety of equipment and personnel.

[0094] For example, for instructions that exceed the limits of process parameters, the security gateway can adjust the instructions that originally exceeded the upper or lower limits of process parameters to the corresponding upper or lower limits to obtain a safe instruction through amplitude limiting. The gateway can also issue an alarm declaration for the safe instruction, indicating that the instruction has been limited.

[0095] Finally, the safety commands reliably sent through the security gateway are sent to the corresponding actuators (such as burners and circulating fans) to drive the actions, thus completing the closed-loop control of the entire drying room operation.

[0096] The following is a detailed description and explanation of the solutions in the embodiments of the present invention, using specific application examples: This application's embodiments belong to the interdisciplinary technical field of Industrial Machine Vision, Artificial Intelligence, and Industrial Process Control. It can be applied to purely back-end core control systems, and its overall architecture is as follows: Figure 3 As shown, it mainly includes a visual perception layer, an edge computing layer, a model layer, and a control layer. These layers work together to form a complete "perception-decision-control" closed loop. This system is an integrated hardware and software industrial control system, without a user interface. Its core function is to automatically generate and issue optimal control commands through data analysis and intelligent algorithms.

[0097] A typical application scenario for this system is in the coating production line of a container manufacturing plant, especially in the drying process of intermediate paint (or primer, topcoat). Specifically, the system is deployed at the entrance of the drying chamber and in the nearby control room. It uses industrial cameras to "observe" the container body about to enter the drying chamber and automatically adjusts parameters such as temperature and air speed inside the drying chamber.

[0098] Understandably, this system is applicable to all industrial thermal equipment that requires refined and differentiated energy consumption management based on the specific characteristics of the workpiece being processed (such as size, material, and coating), and has great potential for promotion to industries such as automotive painting, furniture painting, and building material drying.

[0099] Specifically, the visual perception layer contains two icons: an industrial camera array and the workpiece being measured (the container body). The arrows indicate that the camera is acquiring images of the workpiece. This layer consists of one or more high-resolution industrial cameras with high-temperature resistance and dustproof capabilities, deployed in an open area at the entrance of the drying chamber. These cameras can be used to capture high-definition images of the container body in real time, providing raw data for subsequent intelligent analysis.

[0100] Edge computing layer: Located at the center of the architecture, it is represented by an industrial computer icon. It contains four modules, listed in boxes, including: Image acquisition and preprocessing module: Receives images from the camera and performs operations such as "distortion correction" and "illuminance equalization".

[0101] The YOLO object detection module is the "eye" of the system. It receives the preprocessed image, performs object detection on the image, and outputs the detection results.

[0102] The workpiece status parsing module acts as the system's "translator." It translates the unstructured information output by the YOLO model into structured data that the backend control model can directly use. For example, it receives the output from the YOLO module and converts it into structured data such as "workpiece type = A" and "quantity = 2." This information is crucial for triggering dynamic control.

[0103] The energy consumption prediction and optimization module is the brain and decision-making center of the system. As the core decision-making module, it is responsible for receiving workpiece status and environmental data and outputting "optimal control parameters".

[0104] Model Layer and Control Layer: In the diagram, the functions of these two layers are integrated into modules to the right and below the edge computing layer. They receive instructions from the energy consumption prediction module and ultimately connect to the factory's existing programmable logic controller (PLC) via the safety control interface module. The PLC then controls the actuators (burners, fans), forming a closed loop.

[0105] Data flow and control flow: Figure 3 The lines with arrows clearly indicate the direction of data flow (solid lines) and the direction of control command flow (dashed lines).

[0106] Reference Figure 4 , Figure 4 This is a flowchart illustrating the application of the vision-based intelligent control system for container painting drying rooms in a pure back-end core control system.

[0107] First, capture images of the workpiece. Then, use an industrial camera array to capture images of the container body that is about to enter the drying oven, such as... Figure 2 As shown, the raw image captured by the industrial camera contains the complete container body, with an image resolution of 1920×1080 pixels, and includes information such as the container outline and surface coating status.

[0108] Next, the image data is transmitted to the YOLO target detection module, which runs the YOLO model and outputs the target category, quantity, and location in the image. The workpiece state parsing module then parses the original output of the YOLO model (such as bounding boxes and category labels) into structured workpiece state data that can be used for model computation.

[0109] For example, the logic for converting unstructured information to structured information is as follows: Python def parse_workpiece_status(yolo_outputs, confidence_threshold=0.8): structured_data = {"timestamp": get_current_time(), "workpieces":[]} for detection in yolo_outputs: if detection.confidence>confidence_threshold: workpiece_info = { "category": CLASS_NAMES[detection.class_id], "count": 1, "coating_type": infer_coating_type(detection), "position": detection.bbox, "confidence": detection.confidence } structured_data["workpieces"].append(workpiece_info) return structured_data.

[0110] It should be noted that the necessary fields in the structured workpiece status data include workpiece category (20 feet / 40 feet / refrigerated box, etc.), quantity, coating type (identifying primer / intermediate paint / topcoat through image features), timestamp (used for time-series correlation, accurately aligning visual inspection results with historical environmental data of the drying room, and establishing a time correspondence between F_visual and F_env(t),) and confidence level (used for quality control).

[0111] Then, the workpiece status data and the current status data of the drying room environment are input into the energy consumption prediction and optimization model to calculate the optimal control parameter sequence for the next stage. The calculated optimal control parameters are then sent to the field PLC through the safety control interface.

[0112] For example, the actuator can employ time-segmented control: the entire drying process is divided into multiple time segments, each corresponding to a different temperature-wind speed combination. The output of the optimal control parameter sequence for time-segmented control is as follows: Json { "control_sequence": [ { "time_slot": "0-300s", "zone1_temp": 80, "zone2_temp": 85, "fan_speed": 5, "purpose": "Preheating stage, uniform temperature increase" }, { "time_slot": "300-600s", "zone1_temp": 85, "zone2_temp": 90, "fan_speed": 4, Purpose: "Consolidate the main phase, and operate at energy efficiency" }, { "time_slot": "600-900s", "zone1_temp": 75, "zone2_temp": 80, "fan_speed": 3, Purpose: During the cooling stage, to prevent over-baking. } ] }

[0113] It should be noted that each time segment has safety boundary conditions, and exceeding the limits will immediately trigger a safety interlock.

[0114] Finally, the PLC drives the burners, fans, and other actuators to change the operating state of the drying chamber. Then, the system returns to the step of capturing workpiece images, forming a closed loop that continuously and intelligently controls the drying chamber.

[0115] The PLC adjusts the setpoints step by step according to the time sequence to ensure the precise execution of the process curve and achieve real-time adjustment.

[0116] Reference Figure 5 This application also provides a vision-based intelligent control system for container painting drying ovens, which can implement the above-mentioned method. The system includes: The visual perception module is used to acquire images of the containers in the current batch and obtain images of the container bodies.

[0117] The status analysis module is used to analyze the workpiece status based on the box image to obtain structured workpiece status data, which includes workpiece category and statistical quantity.

[0118] The parameter control module is used to acquire current environmental data of the drying chamber and, with the goal of minimizing energy consumption, predict energy consumption and optimize parameters based on the current environmental data and workpiece status data to obtain the optimal control parameters.

[0119] The drying oven drive module is used to drive the actuators inside the drying oven to change the operating state of the drying oven according to the optimal control parameters.

[0120] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0121] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0122] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0123] Reference Figure 6 , Figure 6 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0124] The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the methods described in the embodiments of this application.

[0125] The input / output interface 903 is used to implement information input and output.

[0126] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0127] Bus 905 transmits information between various components of the device, such as processor 901, memory 902, input / output interface 903, and communication interface 904.

[0128] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0129] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0130] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0131] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0132] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0133] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0134] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0135] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0136] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0137] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0138] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0139] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0140] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0141] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0142] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0143] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0144] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for intelligent control of container coating drying ovens based on vision detection, characterized in that, The method includes the following steps: Images of the containers in the current batch are acquired to obtain images of the container bodies; The workpiece status is analyzed based on the box image to obtain structured workpiece status data, which includes workpiece category and statistical quantity. Acquire current environmental data of the drying chamber, and with the goal of minimizing energy consumption, perform energy consumption prediction and parameter optimization based on the current environmental data of the drying chamber and the workpiece status data to obtain the optimal control parameters; The actuators inside the drying chamber are driven to move according to the optimal control parameters, thereby changing the operating state of the drying chamber.

2. The method according to claim 1, characterized in that, Before the step of performing workpiece state analysis processing based on the box image to obtain structured workpiece state data, the method further includes the following steps: The box image is preprocessed to obtain a preprocessed image, wherein the preprocessing includes lens distortion correction, illumination equalization and noise filtering; The step of performing workpiece state analysis processing based on the box image to obtain structured workpiece state data includes the following steps: The preprocessed image is used to perform workpiece state analysis to obtain structured workpiece state data.

3. The method according to claim 1, characterized in that, The step of performing workpiece state analysis processing based on the box image to obtain structured workpiece state data includes the following steps: Target detection is performed based on the box image to obtain detection results, wherein the detection results include category labels, confidence scores, and location information of several targets in the box image; The detection results are processed by structured transformation to obtain workpiece status data, which includes workpiece category and workpiece quantity.

4. The method according to claim 1, characterized in that, The process of minimizing energy consumption, based on current data of the drying chamber environment and workpiece status data, to predict energy consumption and optimize parameters to obtain optimal control parameters, includes the following steps: Obtain future production information for future batches; The current environmental data of the drying room, the status data of the workpiece, and the future production information are input into a pre-trained prediction model for nonlinear mapping to obtain the predicted energy consumption under different control parameters. With the goal of minimizing energy consumption, the parameters are optimized based on the predicted energy consumption to obtain the optimal control parameters.

5. The method according to claim 4, characterized in that, The training method for the prediction model includes the following steps: Acquire historical data on workpiece status, drying oven environment, and historical energy consumption; Based on the historical data of the workpiece state, the historical data of the drying room environment, and the historical energy consumption data, a neural network model is trained using a machine learning algorithm to learn nonlinear mapping relationships. A loss function with the goal of minimizing energy consumption is introduced to optimize the training of the neural network model, resulting in a well-trained prediction model.

6. The method according to claim 5, characterized in that, The loss function includes a predicted energy consumption, a paint film hardness penalty, an adhesion penalty, and a surface defect penalty. The expression for the loss function is: Loss = argmin( E_total+λ * Penalty(Q) ); Where Loss is the loss function, E_total is the predicted energy consumption, Q represents the drying quality, Penalty(Q)=α×(1-Hardness_score) +β×(1-Adhesion_score) +γ×Surface_defect_score, Hardness_score is the paint film hardness penalty term, Adhesion_score is the adhesion penalty term, Surface_defect_score is the surface defect penalty term, and λ, α, β, and γ are the weights.

7. The method according to claim 1, characterized in that, The step of driving the actuators in the drying chamber to operate according to the optimal control parameters and changing the operating state of the drying chamber includes the following steps: The instruction is converted according to the optimal control parameters to obtain the control instructions to be sent to the programmable logic controller. The control command is subjected to upper and lower limit amplitude checks and change rate limits based on the preset safety interlock logic of the programmable logic controller to obtain a safety command; The safety instructions drive the actuators inside the drying chamber to change the operating state of the drying chamber.

8. A visual inspection-based intelligent control system for container coating drying ovens, characterized in that, The system includes: The visual perception module is used to acquire images of the containers in the current batch and obtain images of the container bodies; The status analysis module is used to perform workpiece status analysis processing based on the box image to obtain structured workpiece status data, wherein the workpiece status data includes workpiece category and statistical quantity. The parameter control module is used to acquire the current environmental data of the drying room and, with the goal of minimizing energy consumption, perform energy consumption prediction and parameter optimization based on the current environmental data of the drying room and the workpiece status data to obtain the optimal control parameters. The drying oven drive module is used to drive the actuators inside the drying oven to operate according to the optimal control parameters, thereby changing the operating state of the drying oven.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images