Method and system for detecting and alarming blending of printing ink in oil weighing room

Abstract

The invention provides an ink blending detection and alarm method and system for an oil weighing room, and the method comprises the steps: recording an ink weight change video and a personnel action video in real time in the ink blending process of the oil weighing room; marking the printing ink weight and the personnel action in the printing ink weight change video and the personnel action video respectively; training an ink weight detection model and a personnel action detection model based on the labeling processing result; respectively adopting the trained ink weight detection model and the trained personnel action detection model to carry out ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, setting ink weight and personnel action alarm conditions, and if an event meeting the alarm conditions is detected, giving an alarm; otherwise, recording the ink weight and the personnel action information. The system can detect, alarm and record the work of the oil weighing room in real time, reduces the waste of human resources, identifies the risk, gives an alarm in time, and improves the automation level of a factory.

Description

Technical Field

[0001] The invention relates to the fields of chemical industry and computer vision, and more specifically to a method and system for detecting and alarming ink mixing in an oil weighing room. Background Art

[0002] In the process of ink mixing in modern factories, in order to ensure that the work in the oil weighing room can be completed accurately, two staff members are usually deployed to work together. One staff member is responsible for the ink mixing work, and must strictly follow the production standards and specific process requirements. Before taking the ink, the necessary ink information check must be carried out to ensure the accuracy of the ink mixing, and the amount of crude oil, curing agent, and oil-opening water added must be accurately controlled to ensure that the ink weight ratio meets the standard. Any minor error may seriously affect the final quality of the ink; the other staff member is responsible for recording various data in the ink mixing process. However, the recorder is in a waiting state most of the time and only records when necessary, resulting in a waste of human resources. Summary of the invention

[0003] In view of this, the present invention provides an oil weighing room ink mixing detection and alarm method and system, which aims to detect, alarm, and record the work of the oil weighing room in real time, reduce the waste of human resources, prevent errors caused by human factors, identify risks and issue alarms in time, and improve the factory's automation level.

[0004] On the one hand, the present invention provides a method for detecting and alarming ink mixing in an oil weighing room, comprising:

[0005] Step S1, during the ink mixing process in the oil weighing room, real-time recording of ink weight change video and personnel action video;

[0006] Step S2, marking the ink weight and the personnel action in the ink weight change video and the personnel action video respectively;

[0007] Step S3, training an ink weight detection model and a personnel action detection model based on the result of the labeling process;

[0008] Step S4, respectively using the trained ink weight detection model and the trained personnel action detection model to perform ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, setting ink weight and personnel action alarm conditions, and if an event that meets the alarm conditions is detected, an alarm is triggered; otherwise, the ink weight and personnel action information are recorded.

[0009] In the ink mixing detection and alarm method in the oil weighing room of the present invention, the step S1 comprises:

[0010] During the ink mixing process in the oil weighing room, two cameras are used to record video. One camera records the data on the LED display screen of the electronic scale when the ink is weighed in real time, and the other camera records the movements of the personnel in real time.

[0011] In the ink mixing detection and alarm method for the oil weighing room of the present invention, in step S2, the specific method of marking the ink weight and the personnel action in the ink weight change video and the personnel action video is:

[0012] Extracting a video frame from the ink weight change video, cutting out a weight display area on the LED display screen, and marking the displayed content therein, wherein the marking types include numbers, decimal points, and negative signs;

[0013] Video frames are extracted from the personnel action video, and the personnel actions in the video frames are annotated, and the types of annotations include ink taking action, ink putting back action, and ink information checking action.

[0014] In the ink mixing detection and alarm method for the oil weighing room described in the present invention, the ink weight detection model and the personnel action detection model in step S3 are detection models based on the YOLOv5s algorithm.

[0015] In the ink mixing detection and alarm method in the oil weighing room of the present invention, in the step S4, the alarm condition includes an ink weight alarm condition and a personnel action alarm condition;

[0016] The ink weight alarm condition means that after the ink is mixed, if it is detected that the ink weight ratio does not meet the standard, an alarm will be triggered;

[0017] The personnel action alarm condition means: if the ink information checking action is not detected and the ink picking action is detected, an alarm is triggered; if the ink information checking action is detected first and then the ink picking action is detected, but the time interval between the two actions exceeds the preset time length, an alarm is triggered.

[0018] In the ink mixing detection and alarm method in the oil weighing room of the present invention, the specific method of recording the ink weight and personnel action information is as follows:

[0019] When a negative number is identified, the ink mixing is considered to be finished, and the positive numbers on the LED display of the electronic scale before returning to zero each time during the ink mixing process are recorded as the weights of the crude oil, curing agent, and oil-opening water;

[0020] Record the time, action, and confidence of personnel actions.

[0021] On the other hand, the present invention also provides an ink mixing detection and alarm system for an oil weighing room, comprising: an LED display screen, a first camera, a second camera, a network device, a model training module, a video detection module, an alarm, an information recording module, and a GUI user operation interface;

[0022] LED display screen, used to display the weighing information obtained by the electronic scale;

[0023] The first camera is used to record the ink weight change video displayed on the LED display screen in real time during the ink mixing process in the oil weighing room;

[0024] The second camera is used to record the real-time video of personnel actions during the ink mixing process in the oil weighing room;

[0025] A network device for transmitting the ink weight change video and the personnel action video;

[0026] Model training module, which trains the ink weight detection model and the personnel action detection model based on the results of the annotation processing;

[0027] The video detection module is used to use the trained ink weight detection model and the personnel action detection model to perform ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, and if an event that meets the alarm conditions is detected, an alarm is triggered through an alarm device and a GUI user operation interface;

[0028] An alarm device for sending out an alarm signal;

[0029] An information recording module, used to record the ink weight and personnel action information;

[0030] The GUI user operation interface is used to start the camera to record the ink weight video and the personnel action video, call the detection model to detect the ink weight ratio and personnel action, display the detection results of the ink weight and personnel action in real time, and display alarm information.

[0031] The ink mixing detection and alarm method and system for the oil weighing room provided by the present invention have the following beneficial effects: the scheme of the present invention records the ink weight change video and the personnel action video in real time during the ink mixing process in the oil weighing room; the ink weight and the personnel action in the ink weight change video and the personnel action video are respectively annotated; the ink weight detection model and the personnel action detection model are trained based on the result of the annotating process; the trained ink weight detection model and the personnel action detection model are respectively used to perform ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, and the ink weight and personnel action alarm conditions are set. If an event that meets the alarm conditions is detected, an alarm is triggered; otherwise, the ink weight and personnel action information are recorded. The ink mixing detection and alarm system for the oil weighing room provided by the present invention can detect, alarm, and record the work of the oil weighing room in real time, reduce the waste of human resources, prevent errors caused by human factors, identify risks and give timely alarms, and improve the automation level of the factory. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of the ink mixing detection and alarm method of the oil weighing room provided by the embodiment of the present invention;

[0033] Figure 2 It is a schematic diagram of the flow of the ink mixing detection and alarm system in the oil weighing room provided by an embodiment of the present invention;

[0034] Figure 3 This is a schematic diagram of the entire process of ink mixing in the oil-pounding room provided by an embodiment of the present invention;

[0035] Figure 4 This is a cutout sample of an ink weight detection chart collected in an oil weighing room provided in an embodiment of the present invention;

[0036] Figure 5 This is a digitally labeled classification chart of 12 types of ink weights provided in an embodiment of the present invention;

[0037] Figure 6 It is a personnel action detection diagram collected in an oil weighing room provided by an embodiment of the present invention;

[0038] Figure 7 These are three types of personnel action annotation classification diagrams provided by the embodiments of the present invention;

[0039] Figure 8 This is the YOLOv5s network structure diagram;

[0040] Fig. 9 It is a prediction graph of the ink weight detection model during training provided by an embodiment of the present invention;

[0041] Fig.10 It is a prediction graph of a human action detection model during training provided by an embodiment of the present invention;

[0042] Fig.11 It is a YOLOv5s ink weight detection model training graph provided by an embodiment of the present invention;

[0043] Fig.12 It is a YOLOv5s personnel action detection model training diagram provided by an embodiment of the present invention;

[0044] Fig.13 This is an ink weight detection diagram provided by an embodiment of the present invention;

[0045] Fig.14 is a personnel action detection diagram provided by an embodiment of the present invention;

[0046] Fig.15 It is a diagram of the hardware facilities and the connection relationship of each device provided by the embodiment of the present invention;

[0047] Fig.16 This is a schematic diagram of the initial state of the GUI user operation interface of the ink mixing detection and alarm system for the oil-pound room provided by an embodiment of the present invention;

[0048] Fig.17 It is a schematic diagram of a GUI user operation interface of an ink mixing detection and alarm system for an oil-pounding room provided by an embodiment of the present invention;

[0049] Fig.18 It is an ink weight information storage diagram of the ink mixing detection and alarm system of the oil weighing room provided by the embodiment of the present invention;

[0050] Fig.19 It is a personnel action information storage diagram of the ink mixing detection and alarm system of the oil weighing room provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0051] 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.

[0052] like Figure 1 As shown, an embodiment of the present invention provides a method for detecting and alarming ink mixing in an oil-pounding room. The method of this embodiment may include the following steps:

[0053] Step S1, during the ink mixing process in the oil weighing room, real-time recording of ink weight change video and personnel action video. In this step, during the ink mixing process in the oil weighing room, two cameras are used to record video, one of which records the data on the LED display screen of the electronic scale when the ink is weighed in real time, and the other camera records the personnel action in real time. Specifically, one camera is facing the LED display screen of the electronic scale, and the other camera is facing the shelf where the crude oil barrels are placed. In the embodiment of the present invention, an LED display screen is innovatively connected to the electronic scale, and a camera is facing the LED display screen to shoot the ink weight information, which solves the problem of difficulty in obtaining the ink weight during the ink mixing process in the oil weighing room. The camera facing the shelf where the crude oil barrels are placed can capture the operator's actions of taking the crude oil barrels, putting the crude oil barrels back, and checking the ink information, so as to determine whether the operator's actions are standardized.

[0054] like Figure 2 As shown, an overall flow chart of an ink mixing detection and alarm system for an oil weighing room provided by an embodiment of the present invention can simultaneously call an ink weight detection model and a personnel motion detection model through a multi-threaded manner to perform real-time target detection on the ink weight change video and the personnel motion video respectively.

[0055] like Figure 3 As shown in the figure, the whole process of ink mixing includes the following steps:

[0056] Step S11, selecting the type of ink; selecting the type of ink specifically refers to selecting the ink formula according to the product model. For different ink formulas, the crude oil model and the proportion of crude oil, curing agent, and oil-opening water may be different;

[0057] Step S12, entering the oil weighing room;

[0058] Step S13, checking ink information; checking ink information means searching for corresponding crude oil model information according to the ink formula selected in step S11;

[0059] Step S14, taking ink; taking ink mainly refers to taking a crude oil barrel of a corresponding model from the shelf where the crude oil barrels are placed;

[0060] Step S15, adding crude oil, curing agent, and oil-opening water in proportion; specifically, adding crude oil, curing agent, and oil-opening water in proportion according to the ink formula selected in step S11.

[0061] Step S16, stirring the ink;

[0062] Step S17, ink mixing is completed.

[0063] In the step S15, the following steps are included:

[0064] Step S151, placing the empty oil-mixing cup on the electronic scale, and the data on the LED display screen changes from zero to a positive number, and the positive number is the weight of the empty cup;

[0065] Step S152, pressing the reset button of the electronic scale, the data on the LED display screen is reset to zero;

[0066] Step S153, pouring crude oil into the empty oil blending cup, and the data on the LED display screen turns into a positive number, which is the weight of the crude oil;

[0067] Step S154, after confirming the weight of the crude oil, the reset button of the electronic scale is pressed, and the data on the LED display screen is reset to zero;

[0068] Step S155, pouring curing agent into the oil mixing cup, and the data on the LED display screen turns into a positive number, which is the weight of the curing agent;

[0069] Step S156, after confirming the weight of the curing agent, the reset button of the electronic scale is pressed, and the data on the LED display screen is reset to zero;

[0070] Step S157, pouring the boiling oil water into the oil mixing cup, and the data on the LED display screen turns into a positive number, which is the weight of the boiling oil water;

[0071] Step S158, after confirming the weight of the oil and water, take away the oil mixing cup, and the LED display screen displays a negative number, indicating that the current ink mixing is completed;

[0072] Step S159, press the reset button of the electronic scale, the data on the LED display screen is reset to zero, and prepare for the next ink mixing.

[0073] During the process of ink mixing in the oil weighing room, the mixed ink needs to be used within a certain period of time, otherwise it cannot be used. The ink weight detection model and the personnel action detection model are used to detect personnel actions and ink mixing progress in real time, record the oil mixing time, and identify risks and issue alarms in time. Step S2, respectively mark the ink weight and personnel actions in the ink weight change video and the personnel action video.

[0074] In step S2, a video frame is extracted from the ink weight change video, the weight display area on the LED display screen is cropped, and the display content therein is annotated, and the types of annotations include numbers, decimal points and negative signs.

[0075] In the embodiment of the present invention, the specific method of extraction is to extract one frame from every ten frames of the ink weight change video to obtain an image with a resolution of 1920×1080 as the original data set of the ink mixing detection and alarm system of the oil weighing room. Since there is a lot of interference information in the image containing the ink weight information and the application scenario is fixed, in order to reduce interference and improve the detection accuracy, the image containing the ink weight information is uniformly cropped. The image effect after cropping is as follows Figure 4 shown.

[0076] like Figure 5 As shown, in the embodiment of the present invention, the specific method of the marking processing is to use the rectangular box tool in the LabelImg marking software to mark the ink weight numbers, which are divided into twelve categories, namely "1", "2", "3", "4", "5", "6", "7", "8", "9", "0", "-", ".". During the marking process, try to make the smallest circumscribed rectangle possible, and the more precise the marking box, the better.

[0077] As shown in Table 1, after the annotation process, 196 annotated images were obtained, of which there were a total of 935 ink weight labels. The images were divided into training set, test set and validation set in a ratio of 7:2:1, resulting in 137 training sets, 39 test sets and 20 validation sets.

[0078] Table 1 Ink weight distribution of various labels

[0079]

[0080] In the step S2, video frames are extracted from the personnel action video, and the personnel actions in the video frames are annotated, and the types of annotations include ink taking action, ink putting back action, and ink information checking action.

[0081] like Figure 6 As shown, the specific method of the extraction is to extract one frame from every ten frames of the personnel action video to obtain an image with a resolution of 1920×1080 as the original data set of the ink mixing detection and alarm system of the oil weighing room. Figure 7As shown, the specific method of the labeling process is to use the rectangular box tool in the LabelImg labeling software to label the personnel actions in the image; the main judgment basis for the ink taking action is facing the camera and the hand touching the crude oil barrel, the label name is get_ink, which represents the action of taking the crude oil barrel from the shelf before the ink is mixed; the main judgment basis for the ink putting back action is facing the camera and the hand touching the crude oil barrel, the label name is put_ink, which represents the action of putting the crude oil barrel back after the ink mixing is completed; the main judgment basis for the ink information checking action is whether there is a white ink information table in the hand, the label name is check_ink. In the process of ink mixing in the oil weighing room, check_ink should be before get_ink. If the operation sequence error is detected, the system will issue an alarm signal. Since for different types of ink mixing, only the type of crude oil is different, and the types of curing agent and oil water are the same, the personnel action detection is specifically crude oil-related personnel action detection.

[0082] As shown in Table 2, after the annotation process, 416 annotated images were obtained, of which there were 427 person action labels in total. The images were divided into training set, test set and validation set in a ratio of 7:2:1, and 291 images were obtained for the training set, 83 images for the test set and 42 images for the validation set.

[0083] Table 2 Personnel action label distribution table

[0084]

[0085] Step S3: training an ink weight detection model and a personnel action detection model based on the result of the labeling process. In an embodiment of the present invention, the ink weight detection model and the personnel action detection model are detection models based on the YOLOv5s algorithm. The detection model may be other deep learning models.

[0086] Figure 8 This is the network structure diagram of YOLOv5s, which includes four parts: input end, backbone network, transition layer, and output layer:

[0087] The image is input into the detection model after preprocessing, wherein the preprocessing includes adjusting the image size, normalizing the image, and converting the image data into a tensor format;

[0088] The input end of the YOLOv5s adopts a batch processing method to give full play to the parallel computing capabilities of the GPU or CPU, process multiple input images at the same time, speed up the training speed and reasoning speed of the detection model, and reduce the time consumed by the detection model to process a single image; by inputting a variety of different changes of image data into the detection model, such as rotating, scaling, cropping, adding noise, etc., the model can learn in a rich and diverse data environment, thereby improving its adaptability to different scenes and different changes;

[0089] The YOLOv5s backbone network can efficiently extract high-quality features from the input image, providing strong support for subsequent target detection, classification, and positioning tasks;

[0090] The transition layer of YOLOv5s further processes and fuses the features to improve the performance of target detection;

[0091] The output layer of the YOLOv5s uses anchor boxes to predict the location of the target. Based on the previously extracted and fused features, a fully connected layer or a convolutional layer is used to implement the classification function. The output layer integrates and outputs the prediction results of the target location, category, and confidence. Confidence refers to the model's confidence in the prediction results, which is usually a value between 0 and 1. For object detection models, confidence indicates the possibility that the model believes that a certain bounding box contains an object of a specific category. The higher the confidence, the more confident the model is that its prediction is correct.

[0092] Considering the actions of the personnel in the oil weighing room and the data characteristics of the data set, the target to be detected in the image is large and the detection accuracy requirement is low. Selecting a detection model based on the YOLOv5s algorithm has the following advantages:

[0093] (1) Compared with detection models based on other algorithms, the detection model based on the YOLOv5s algorithm is smaller, which makes it have lower computing and storage requirements when deployed on resource-constrained devices (such as edge devices, embedded systems, etc.);

[0094] (2) Lightweight models also mean faster inference speed, which helps achieve real-time detection.

[0095] In an embodiment of the present invention, the ink weight detection model and the personnel action detection model are constructed by using the deep learning framework Pytorch, and its advantages are mainly reflected in dynamic graph design, ease of use and getting started, rich deep learning libraries, scalability and flexibility, and an active community and ecosystem.

[0096] As shown in Table 3, this is the configuration of the computer when training the network model, including information such as chips, memory, operating system, and development environment.

[0097] Table 3 Detailed parameters of training computer

[0098]

[0099] As shown in Table 4-1 and Table 4-2, these are the configurations of the hyperparameters in the YOLOv5s network before training the ink weight detection model and the personnel action detection model.

[0100] Table 4-1 YOLOv5s ink weight detection model hyperparameter configuration table

[0101]

[0102] Table 4-2 YOLOv5s Personnel Action Detection Model Hyperparameter Configuration Table

[0103]

[0104] As shown in Table 5-1 and Table 5-2, they are the configuration of the training parameters of the ink weight detection model and the human motion detection model. The configuration of the training parameters is based on the computer configuration and mainly depends on the graphics card. Among them, the batch size refers to the number of samples used in one iteration during YOLOv5s network training, and the training round refers to how many times the entire data set will be iterated during the training process.

[0105] Table 5-1 YOLOv5s ink weight detection model training parameter configuration table

[0106]

[0107] Table 5-2 YOLOv5s personnel action detection model training parameter configuration table

[0108]

[0109] like Fig. 9 As shown in FIG. 1 , it is a prediction diagram of the ink weight detection model during training provided by an embodiment of the present invention. Fig.10 As shown, it is a prediction diagram of the personnel action detection model during training provided by an embodiment of the present invention.

[0110] The embodiment of the present invention uses four indicators, namely precision (P), recall (R), mAP50, and detection speed, to evaluate the detection model. mAP50 represents the average accuracy of multi-class detection when IoU is greater than 0.5. The calculation of average precision, recall, and mean average accuracy is shown in formulas 1-1, 1-2, and 1-3.

[0111]

[0112] In the formula, TP means the case where the model predicts a positive sample and it is actually a positive sample; FP means the case where the model predicts a positive sample but it is actually a negative sample; FN means the case where the model predicts a negative sample but it is actually a positive sample. C is the number of detection categories.

[0113] like Fig.11 As shown in FIG. 1 , it is a YOLOv5s ink weight detection model training diagram provided by an embodiment of the present invention, which shows the detailed process of ink weight detection model training and the specific parameters of each detection index; Fig.12 As shown, it is a YOLOv5s personnel action detection model training diagram provided by an embodiment of the present invention, which shows the detailed process of personnel action detection model training and the specific parameters of each detection index. As the number of training rounds increases, the loss function curve of the detection model continues to decrease, and the detection accuracy and recall rate continue to increase until the curve converges, indicating that the detection model training is completed.

[0114] like Fig.13 As shown in FIG. 1 , it is an ink weight detection diagram provided by an embodiment of the present invention; Fig.14 As shown in FIG. 6 , it is a personnel action detection diagram provided by an embodiment of the present invention; the ink weight information and personnel action information can be accurately identified. As shown in Table 6, in the embodiment of the present invention, the accuracy of the personnel action detection model is 86.48%, the recall rate is 92.52%, and the accuracy of the ink weight detection model is 98.87%, and the recall rate is 100%, which can meet the needs of the ink mixing detection and alarm system of the oil weighing room.

[0115] Table 6 Target detection algorithm performance parameters

[0116]

[0117] Step S4, respectively using the trained ink weight detection model and the trained personnel action detection model to perform ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, setting ink weight and personnel action alarm conditions, and if an event that meets the alarm conditions is detected, an alarm is triggered; otherwise, the ink weight and personnel action information are recorded.

[0118] In the embodiment of the present invention, the system is equipped with two special cameras for different detection tasks. One of the cameras is specifically used to collect videos related to ink weight. The camera cooperates with the ink weight detection model to monitor the ink ratio in real time and alarm for abnormalities based on the change information of ink weight. The other camera focuses on collecting videos of personnel operations. The camera combines with the personnel action detection model to monitor the operation standardization in real time and alarm for violations based on the collected personnel action information. In actual applications, users can flexibly configure the working mode of the camera according to specific needs: one of the cameras can be turned on separately for shooting and detection of specific tasks, or both cameras can be started at the same time to achieve dual monitoring and detection functions.

[0119] In step S4, the alarm conditions include ink weight alarm conditions and personnel action alarm conditions; the ink weight alarm condition means that after the ink mixing is completed, if it is detected that the ink weight ratio does not meet the standard, an alarm is triggered; the personnel action alarm condition means: if no ink information check action is detected and the ink picking action is detected, an alarm is triggered; if the ink information check action is detected first and then the ink picking action is detected, but the time interval between the two actions exceeds the preset time length, an alarm is triggered. In the embodiment of the present invention, the preset time length is 30 seconds. If the time interval exceeds 30 seconds, it means that the detected ink information check action is the check action when the ink was picked up last time.

[0120] In this step, the specific method of recording the ink weight and personnel action information is: when a negative number is identified, it is determined that the current ink mixing is completed, and the positive numbers on the electronic scale LED display before each return to zero during the current ink mixing process are recorded as the weight of crude oil, curing agent, and oil-opening water; the time, action, and confidence of the personnel action are recorded. Among them, the confidence is output by the detection model output layer.

[0121] It is understandable that, in the solution of the present invention, the ink weight detection model and the personnel motion detection model may also be detection models based on other deep learning algorithms, as long as their accuracy, recall rate and other parameters meet the requirements.

[0122] like Fig.15 As shown, an embodiment of the present invention also provides an ink mixing detection and alarm system for an oil weighing room, including: an LED display screen, a first camera, a second camera, a network device, a model training module, a video detection module, an alarm, an information recording module, and a GUI user operation interface;

[0123] LED display screen, used to display the weighing information obtained by the electronic scale;

[0124] The first camera is used to record the ink weight change video displayed on the LED display screen in real time during the ink mixing process in the oil weighing room; in the embodiment of the present invention, the first camera is arranged facing the LED display screen.

[0125] The second camera is used to record the real-time video of personnel actions during the ink mixing process in the oil weighing room; in the embodiment of the present invention, the second camera is arranged opposite to the shelf where the crude oil barrels are placed.

[0126] A network device for transmitting the ink weight change video and the personnel action video;

[0127] Model training module, which trains the ink weight detection model and the personnel action detection model based on the results of the annotation processing;

[0128] The video detection module is used to use the trained ink weight detection model and the personnel action detection model to perform ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, and if an event that meets the alarm conditions is detected, an alarm is triggered through an alarm device and a GUI user operation interface;

[0129] An alarm device for sending out an alarm signal;

[0130] An information recording module, used to record the ink weight and personnel action information;

[0131] The GUI user operation interface is used to start the camera to record the ink weight video and the personnel action video, call the detection model to detect the ink weight ratio and personnel action, display the detection results of the ink weight and personnel action in real time, and display alarm information.

[0132] In the embodiment of the present invention, the GUI user operation interface is made by Pyqt5. Pyqt5 has the advantages of cross-platform support, rich GUI elements, modern style, powerful event handling capabilities, excellent performance and stability, easy integration and expansion, etc. It is easy to import the trained model into the GUI user operation interface through Pycharm, which is convenient for development.

[0133] In addition, the ink mixing detection and alarm system of the oil weighing room of the embodiment of the present invention also includes an electronic scale and a computer terminal; the electronic scale is used to weigh the ink, and the ink weight information obtained by weighing will be displayed on the LED display screen, and the computer terminal is used to run the ink mixing detection and alarm system of the oil weighing room.

[0134] like Fig.16 As shown in FIG. 1 , it is the initial state of the GUI user operation interface of the ink mixing detection and alarm system for the oil room provided by the embodiment of the present invention; Fig.17As shown, click "Camera Detection" in the ink weight information section or the personnel action information section to call the corresponding camera, and the video of the corresponding camera is transmitted to the system through the network device for ink weight detection or personnel action detection. The detected real-time ink weight or personnel action will be displayed on the GUI user operation interface; when an event that meets the alarm conditions is detected, the system will transmit the alarm information to the GUI user operation interface. Fig.18 As shown, it is an ink weight information storage diagram of the ink mixing detection and alarm system of the oil weighing room provided by the embodiment of the present invention; Fig.19 As shown in FIG. 1 , it is a diagram of personnel action information storage of the ink mixing detection and alarm system of the oil weighing room provided by an embodiment of the present invention. If no event meeting the alarm condition is detected, the system will save the ink weight information and personnel action information in an Excel table in real time. The recorded data can be synchronized to other systems in real time, such as MES, inventory management system, etc., to realize the recording of information such as inventory and raw material usage.

[0135] The above-mentioned oil weighing room ink mixing detection and alarm method and system can detect, alarm and record the work of the oil weighing room in real time, reduce the waste of human resources, prevent errors caused by human factors, identify risks and alarm in time, and improve the factory's automation level.

[0136] The ink mixing detection and alarm method and system provided by the present invention have the following significant beneficial effects:

[0137] 1. Real-time monitoring and multi-dimensional data collection:

[0138] During the ink mixing process, the system can record the ink weight change video and personnel action video in real time. This not only provides intuitive operation records, but also provides detailed data support for subsequent analysis and tracing.

[0139] 2. Intelligent labeling and model training: The system labels the recorded videos of ink weight changes and personnel movements to ensure the accuracy and consistency of the data. Based on these labeled data, high-precision ink weight detection models and personnel movement detection models are trained to improve the reliability and intelligence level of detection.

[0140] 3. Automated detection and alarm mechanism: Using the trained ink weight detection model and personnel action detection model, the system can intelligently detect the real-time video of the oil weighing room. Through the preset ink weight and personnel action alarm conditions, the system can immediately trigger an alarm when an abnormal event is detected, and promptly notify relevant personnel to take measures to avoid potential risks.

[0141] 4. Efficient recording and data analysis: For normal operation without triggering alarm conditions, the system will automatically record the ink weight and personnel action information. These records can not only be used for daily management and review, but also serve as an important reference for optimizing production processes and training employees.

[0142] 5. Reduce waste of human resources: By introducing automated detection and alarm mechanisms, the system significantly reduces the need for manual monitoring, reduces labor costs, and improves work efficiency. At the same time, the system can prevent operational errors caused by human factors, further improving the stability and safety of production.

[0143] 6. Improve the level of factory automation: This invention realizes the comprehensive intelligent management of the ink mixing process in the oil-pounding room. From data collection, model training to real-time detection and alarm, each link reflects a high degree of automation and intelligence. This not only improves the overall automation level of the factory, but also lays a solid foundation for the realization of intelligent manufacturing.

[0144] To sum up, the ink mixing detection and alarm method and system for the oil weighing room provided in the embodiments of the present invention, by integrating real-time monitoring, automatic detection and alarm functions, significantly improves the efficiency and safety of the oil weighing room, reduces human errors, and promotes the development of the factory towards intelligence and automation.

[0145] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for detecting and alarming ink mixing in an oil weighing room, characterized in that: include: Step S1, during the ink mixing process in the oil weighing room, real-time recording of ink weight change video and personnel action video; Step S2, marking the ink weight and the personnel action in the ink weight change video and the personnel action video respectively; Step S3, training an ink weight detection model and a personnel action detection model based on the result of the labeling process; Step S4, respectively using the trained ink weight detection model and the trained personnel action detection model to perform ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, setting ink weight and personnel action alarm conditions, and if an event that meets the alarm conditions is detected, an alarm is triggered; otherwise, the ink weight and personnel action information are recorded.

2. The ink mixing detection and alarm method for the oil weighing room according to claim 1 is characterized in that: The step S1 comprises: During the ink mixing process in the oil weighing room, two cameras are used to record video. One camera records the data on the LED display screen of the electronic scale when the ink is weighed in real time, and the other camera records the movements of the personnel in real time.

3. The ink mixing detection and alarm method for the oil weighing room according to claim 1 is characterized in that: In the step S2, the specific method for marking the ink weight and the personnel action in the ink weight change video and the personnel action video is: Extracting a video frame from the ink weight change video, cutting out a weight display area on the LED display screen, and marking the displayed content therein, wherein the marking types include numbers, decimal points, and negative signs; Video frames are extracted from the personnel action video, and the personnel actions in the video frames are annotated, and the types of annotations include ink taking action, ink putting back action, and ink information checking action.

4. The ink mixing detection and alarm method for the oil weighing room according to claim 1 is characterized in that: The ink weight detection model and the personnel action detection model in step S3 are detection models based on the YOLOv5s algorithm.

5. The ink mixing detection and alarm method for the oil weighing room according to claim 1 is characterized in that: In the step S4, the alarm conditions include ink weight alarm conditions and personnel action alarm conditions; The ink weight alarm condition means that after the ink is mixed, if it is detected that the ink weight ratio does not meet the standard, an alarm will be triggered; The personnel action alarm condition means: if the ink information checking action is not detected and the ink picking action is detected, an alarm is triggered; if the ink information checking action is detected first and then the ink picking action is detected, but the time interval between the two actions exceeds the preset time length, an alarm is triggered.

6. The ink mixing detection and alarm method for the oil weighing room according to claim 1 is characterized in that: The specific method of recording the ink weight and personnel action information is: When a negative number is identified, the ink mixing is considered to be finished, and the positive numbers on the LED display of the electronic scale before returning to zero each time during the ink mixing process are recorded as the weights of the crude oil, curing agent, and oil-opening water; Record the time, action, and confidence of personnel actions.

7. An oil weighing room ink mixing detection and alarm system, used to implement the oil weighing room ink mixing detection and alarm method according to any one of claims 1 to 6, characterized in that: include: LED display screen, used to display the weighing information obtained by the electronic scale; The first camera is used to record the ink weight change video displayed on the LED display screen in real time during the ink mixing process in the oil weighing room; The second camera is used to record the real-time video of personnel actions during the ink mixing process in the oil weighing room; A network device for transmitting the ink weight change video and the personnel action video; Model training module, which trains the ink weight detection model and the personnel action detection model based on the results of the annotation processing; The video detection module is used to use the trained ink weight detection model and the personnel action detection model to perform ink weight ratio detection and personnel action detection on the real-time video of the oil weighing room, and if an event that meets the alarm conditions is detected, an alarm is triggered through an alarm device and a GUI user operation interface; An alarm device for sending out an alarm signal; An information recording module, used to record the ink weight and personnel action information; The GUI user operation interface is used to start the camera to record the ink weight video and the personnel action video, call the detection model to detect the ink weight ratio and personnel action, display the detection results of the ink weight and personnel action in real time, and display alarm information.

Images