Wagon balance cheating detection method and device and storage medium

By obtaining the weight value of the floor scale in real time and generating the weight change curve, the convolutional neural network model analyzes the curve characteristics, the problem of driver cheat detection inefficient during the weighing process of the car scale is solved, and automated and intelligent cheat detection is realized, which improves the Detection accuracy and efficiency.

CN119023049APending Publication Date: 2024-11-26SUPCON TECH CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202410963938.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The prior art is difficult to effectively detect and prevent driver cheating during the vehicle weighing process, resulting in inefficiency and additional costs.

Method used

By obtaining the weight value of the floor scale in real time, a weight change curve is generated, and the curve characteristics are analyzed using the convolutional neural network model to judge and identify cheating behaviors during the weighing process, and automatic and intelligent cheat detection are achieved.

Benefits of technology

It improves the automation, intelligence and accuracy of cheat detection, reduces interference from human factors, enhances the ability to identify cheating behaviors, and reduces operating costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119023049A_ABST
    Figure CN119023049A_ABST
Patent Text Reader

Abstract

The invention provides a wagon balance cheating detection method and device and a storage medium, and relates to the technical field of wagon balance weighing, and the method comprises the steps: obtaining a real-time weight value of a wagon balance; according to the real-time weight value, a weight change curve of the wagon balance in the weighing process of the vehicle is obtained; and according to the weight change curve in the weighing process, judging whether a cheating behavior exists in the weighing process and identifying a cheating type. According to the invention, the weight value of the wagon balance is obtained in real time, the weight change curve is generated, the abnormal mode in the curve is identified according to the weight change curve, and the abnormal condition is found in time, so that whether cheating behaviors exist in the weighing process is judged; and the cheating type is identified according to the weight change curve, so that the detection automation and intelligence level and the cheating detection precision are improved, human interference is reduced, the cheating behavior identification capability is enhanced, and the cheating detection effect in the automobile weighing process is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of weighing scales, and in particular to a method, device and storage medium for detecting cheating in a weighing scale. Background Art

[0002] During the weighing process of the truck scale, the driver is required to drive the vehicle onto the scale for vehicle weight detection. Since the driver's cheating behavior will cause huge losses to the company, the weighing process needs to be supervised. At present, the detection of cheating behavior during the weighing process is mainly through manual supervision or adding human scale equipment to weigh the driver to prevent the driver from smuggling.

[0003] In the prior art, the traditional weighing process mainly relies on manual supervision, which not only results in low supervision efficiency, but also makes it easy to ignore the driver's cheating behavior due to human negligence or fatigue. At the same time, the method of adding human scale equipment not only requires additional equipment costs, but also causes complicated pre-weighing preparations, resulting in reduced efficiency of car weighing. Summary of the invention

[0004] The problem solved by the present invention is how to improve the detection effect of cheating in automobile weighing.

[0005] In order to solve the above problems, the present invention provides a method, device and storage medium for detecting cheating in a weighbridge.

[0006] In a first aspect, the present invention provides a method for detecting cheating on a weighbridge, comprising:

[0007] Get the real-time weight value of the scale;

[0008] According to the real-time weight value, a weight change curve of the scale during the weighing process of the vehicle is obtained;

[0009] According to the weight change curve of the weighing process, it is determined whether there is cheating in the weighing process and the type of cheating is identified.

[0010] Optionally, obtaining a weight change curve of the scale during the weighing process of the vehicle according to the real-time weight value includes:

[0011] When the real-time weight value starts to change from 0, a weighing curve is generated according to the real-time weight value;

[0012] When the real-time weight value drops to 0, the generation of the weighing curve is stopped and the weighing curve is used as the weight change curve.

[0013] Optionally, the weighing process includes a high-risk stage, and judging whether cheating behavior exists in the weighing process according to the weight change curve of the weighing process includes:

[0014] Obtaining, by a convolutional neural network model, a curve feature of the weight change curve in the high-risk stage according to the weight change curve;

[0015] According to the curve characteristics in the high-risk stage, it is determined whether there is cheating behavior in the high-risk stage.

[0016] Optionally, obtaining, according to the weight change curve, a curve feature of the weight change curve in the high-risk stage by using a convolutional neural network model includes:

[0017] Inputting the weight change curve into the convolutional neural network model, extracting the high-risk stage from the weighing process through the convolutional neural network model, and obtaining the weight change curve of the high-risk stage;

[0018] The curve characteristics in the high-risk stage are obtained according to the peak data and the fluctuation data of the weight change curve in the high-risk stage.

[0019] Optionally, inputting the weight change curve into the convolutional neural network model, and extracting the high-risk stage from the weighing process through the convolutional neural network model, comprises:

[0020] Inputting the weight change curve of the weighing process into the convolutional neural network model, and obtaining trend characteristics of the weight change curve of the weighing process according to the weight change curve of the weighing process, wherein the trend characteristics include the shape and slope change of the weight change curve;

[0021] The trend feature is compared with a preset weight change trend of the weighing process to obtain the high-risk stage in the weighing process.

[0022] Optionally, judging whether there is cheating in the weighing process and identifying the type of cheating according to the weight change curve of the weighing process includes:

[0023] Predicting the curve features in the high-risk stage through the convolutional neural network model, and outputting the predicted probability of the curve features in a plurality of preset feature labels; wherein each of the feature labels corresponds to one of the cheating types or normal weighing types;

[0024] According to the predicted probability, obtaining the feature label corresponding to the curve feature;

[0025] Determining whether the cheating behavior occurs during the weighing process according to the feature label corresponding to the curve feature;

[0026] If yes, the cheating type is identified according to the feature tag.

[0027] Optionally, obtaining the feature label corresponding to the curve feature according to the predicted probability includes:

[0028] Determine the relationship between the predicted probability and a preset probability threshold;

[0029] The feature label whose predicted probability is greater than or equal to the preset probability threshold is regarded as the feature label to which the curve feature belongs.

[0030] Optionally, predicting the curve feature in the high-risk stage by using the convolutional neural network model and outputting a predicted probability of the curve feature in a plurality of preset feature labels includes:

[0031] The weight change curve in the high-risk stage is processed sequentially through multiple convolutional layers, activation functions, pooling layers, and fully connected layers of the convolutional neural network model to obtain a feature vector of the curve feature for each feature label;

[0032] The feature vector is converted into the predicted probability of the curve feature in each feature label by the classifier of the convolutional neural network model.

[0033] In a second aspect, the present invention provides a weighbridge cheating detection device, comprising:

[0034] Weighing module, used to obtain the real-time weight value of the scale;

[0035] An analysis module, used for obtaining a weight change curve of the scale during the weighing process of the vehicle according to the real-time weight value;

[0036] The prediction module is used to determine whether there is cheating in the weighing process and identify the type of cheating according to the weight change curve of the weighing process.

[0037] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the above-mentioned weighbridge cheating detection method is implemented.

[0038] The weighbridge cheating detection method, device and storage medium of the present invention realize dynamic monitoring of the vehicle weighing process by acquiring the weight value of the weighbridge in real time and generating a weight change curve, and accurately identify abnormal patterns in the curve and discover abnormal situations in time according to the weight change curve of the weighing process, so as to judge whether there is cheating behavior in the weighing process, and identify the cheating type according to the weight change curve, thereby realizing automatic detection of cheating behavior and active identification of cheating type, improving the automation and intelligence level of detection and the cheating detection accuracy, reducing the interference of human factors, enhancing the recognition ability of cheating behavior, and thus improving the cheating detection effect in the automobile weighing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a flow chart of a method for detecting cheating on a weighbridge in one embodiment of the present invention;

[0040] Figure 2 A weight curve diagram of a weighing process in another embodiment of the present invention;

[0041] Figure 3 A curve diagram of a high-risk stage of a normal label in another embodiment of the present invention;

[0042] Figure 4 A curve diagram of a high-risk stage of a first abnormal label in another embodiment of the present invention;

[0043] Figure 5 A curve diagram of a high-risk stage of a second abnormal label in another embodiment of the present invention;

[0044] Figure 6 is a curve diagram of a high-risk stage of a third abnormal label in yet another embodiment of the present invention;

[0045] Figure 7 It is a structural schematic diagram of a weighbridge cheating detection device in another embodiment of the present invention;

[0046] Figure 8 The figure is a schematic structural diagram of an electronic device in another embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be interpreted as being limited to the embodiments described herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0048] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0049] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to"; the term "based on" means "based at least in part on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optionally" means "optional embodiments". The relevant definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0050] It should be noted that the modifications of "one" and "plurality" mentioned in the present invention are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0051] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only used for illustrative purposes, and are not used to limit the scope of these messages or information.

[0052] In the relevant technology, during the weighing process of the truck scale, cheating by the driver will bring huge losses to the company. How to monitor weighing cheating at low cost is the focus of the company's attention. At present, the purpose of preventing drivers from smuggling is mainly achieved by human supervision or adding human scale equipment to weigh the driver. The traditional weighing process mainly relies on manual supervision, which is not only inefficient, but also easy to miss cheating due to human negligence or fatigue. If the method of adding human scale equipment is adopted, not only additional equipment costs are required, but also regular maintenance and calibration are required, which increases the operating costs of the company.

[0053] In view of the problems existing in the above-mentioned related technologies, this embodiment provides a method, device and storage medium for detecting cheating in a weighbridge.

[0054] Combination Figure 1 As shown, a method for detecting cheating on a weighbridge provided by an embodiment of the present invention includes:

[0055] S100: Obtain the real-time weight value of the scale.

[0056] Specifically, the real-time weight is detected by the sensors and data acquisition system equipped on the scale, and the weight value can be recorded and transmitted to the weighing management software in real time. Among them, after the front gate of the scale is opened, the vehicle begins to slowly enter the scale. As the vehicle gradually enters, the real-time weight of the scale gradually increases until the vehicle completely enters the scale, at which time the weight value of the scale reaches a peak. In this process, due to the influence of vehicle driving and ground conditions, the reading of the scale may fluctuate slightly, but generally shows a steady upward trend. Once the vehicle stops completely, the driver gets off and leaves the scale area. At this time, the weight value of the scale will gradually decrease after experiencing a small fluctuation, and then tend to stabilize, reflecting that only the weight of the vehicle is left. Next, the driver swipes the card under the scale to perform the weighing operation, and the system records the weight of the vehicle at this time as the reference value for subsequent calculations. After completing the weighing, the driver gets on the car again. As the weight of the vehicle increases, the weight value of the scale rises again and eventually remains stable. At this time, the gate of the lower scale opens, the vehicle starts to leave the scale, and the weight value of the scale gradually decreases as the vehicle leaves until it returns to zero, completing the entire weighing process. This step of the present invention monitors and obtains the real-time weight value of the scale during the entire weighing process.

[0057] S200: Obtaining a weight variation curve of the scale during the weighing process of the vehicle according to the real-time weight value.

[0058] Specifically, the weight change curve is obtained by recording the continuous changes in the real-time weight value of the scale during the weighing process. The real-time weight value is used to generate an accurate weighing curve and conduct in-depth analysis. Before generating the weight change curve, pre-processing operations such as data collection and data cleaning (removing outliers, noise or damaged data) are required, and then these data are used to draw a curve of weight changes over time during the weighing process.

[0059] S300: judging whether there is cheating in the weighing process and identifying the type of cheating according to the weight change curve of the weighing process.

[0060] Specifically, the present invention uses a convolutional neural network method to identify cheating. The ResNet50 model can be used, and the weighing curve image and its corresponding label are used as training data. After the network learning and optimization process, a high-performance model file is successfully trained. First, it is necessary to produce a weight curve data set for various situations, including normal weighing curves and abnormal weighing curves; secondly, the ResNet50 network is used to classify and identify the curve data set for training and test sets, so that in the application process, it can be determined whether there is cheating in the weighing process and identify the type of cheating based on the weight change curve.

[0061] In a preferred embodiment of the present invention, the preparation of a normal weighing curve data set includes: collecting more than 3220 historical normal weighing curves from the customer site, which are capable of representing weighing behaviors under normal circumstances. Check the collected historical weighing curves and remove possible outliers, noise or damaged data. Ensure the accuracy and reliability of the reference template. From the cleaned historical normal curves, select representative curves as reference templates, and the number of templates selected this time is 2800. The preparation of an abnormal weighing curve data set includes: simulating the following abnormal conditions at the customer site: the driver gets off the car, unloads the heavy objects from the car, walks off the scale, the driver gets off the car, walks off the scale, steps on the scale, the driver gets off the car, walks off the scale, and places the heavy objects on the scale. The data set obtained based on the above is divided into training set data and test set data. On this basis, the model is trained. In order to optimize the training effect, the following hyperparameters are set: the number of iterations is set to 20,000 times to ensure that the model has enough opportunities to learn features from the data set and optimize. The momentum is set to 0.9, which is a technique for accelerating the stochastic gradient descent method and suppressing oscillations. The learning rate is set to 0.0001. In the original 3434 training set data set, multiple sub-datasets containing 500, 1000, 1500, 2000, 2500 and the complete 3434 images are constructed by random sampling.

[0062] In this embodiment, by acquiring the weight value of the scale in real time and generating a weight change curve, the weighing process of the vehicle is dynamically monitored, and according to the weight change curve of the weighing process, the abnormal pattern in the curve is accurately identified and the abnormal situation is discovered in time, so as to judge whether there is cheating in the weighing process, and the cheating type is identified according to the weight change curve, so as to realize the automatic detection of cheating and the active identification of cheating type, improve the automation and intelligence level of detection and the accuracy of cheating detection, reduce the interference of human factors, enhance the ability to identify cheating, and thus improve the cheating detection effect in the process of automobile weighing. Since there is no need to install additional software or set up a complex hardware environment, the model can be deployed to the existing system more quickly, effectively identifying the driver's cheating behavior and accelerating the resolution of the problem.

[0063] Optionally, obtaining a weight change curve of the scale during the weighing process of the vehicle according to the real-time weight value includes:

[0064] When the real-time weight value starts to change from 0, a weighing curve is generated according to the real-time weight value;

[0065] When the real-time weight value drops to 0, the generation of the weighing curve is stopped and the weighing curve is used as the weight change curve.

[0066] Specifically, the scale records the weight value in real time through the sensor, combined with Figure 2 As shown, these weight values ​​start to change from 0, and as the vehicle enters and leaves, the weight value will increase and decrease accordingly. When the vehicle begins to enter the weighbridge, the weight value begins to rise, and the system begins to record the change in weight value and generate a weighing curve based on these real-time data. This process is continuous until the vehicle completely leaves the weighbridge and the weight value drops to 0.

[0067] In this optional embodiment, through automated weighing curve generation and analysis, reliance on manual supervision is reduced, efficiency of the weighing process is improved, and continuity and real-time performance of the weighing operation are ensured.

[0068] Optionally, the weighing process includes a high-risk stage, and judging whether cheating behavior exists in the weighing process according to the weight change curve of the weighing process includes:

[0069] Obtaining, by a convolutional neural network model, a curve feature of the weight change curve in the high-risk stage according to the weight change curve;

[0070] According to the curve characteristics in the high-risk stage, it is determined whether there is cheating behavior in the high-risk stage.

[0071] Specifically, according to the weight change curve, the specific curve of the high-risk stage is analyzed, that is, a convolutional neural network, especially a ResNet50 model, is used to analyze the weight change curve of the high-risk stage, and the curve features are automatically extracted. These features may include the fluctuation amplitude, change trend, stable state, etc. of the curve. According to the curve characteristics in the high-risk stage, it is judged whether there is cheating in the high-risk stage. In a preferred embodiment of the present invention, the high-risk stage in the weighing process mainly refers to the driver's off-board weighing section, because this section is the stage where the driver may resort to cheating. The prediction results and probability values ​​output by the model are used to finally determine whether there is cheating in the driver's off-board weighing section and to identify the type of cheating.

[0072] In this optional embodiment, the convolutional neural network is used to conduct an in-depth analysis of the weight change curve in the high-risk stage, which can accurately identify cheating behavior. The intelligent analysis of the convolutional neural network reduces the misjudgment caused by human factors and improves the accuracy of cheating detection. The automated cheating detection process reduces the reliance on manual supervision, improves the efficiency of the weighing process, and reduces the delay caused by waiting for manual judgment.

[0073] Optionally, obtaining, according to the weight change curve, a curve feature of the weight change curve in the high-risk stage by using a convolutional neural network model includes:

[0074] Inputting the weight change curve into the convolutional neural network model, extracting the high-risk stage from the weighing process through the convolutional neural network model, and obtaining the weight change curve of the high-risk stage;

[0075] The curve characteristics in the high-risk stage are obtained according to the peak data and the fluctuation data of the weight change curve in the high-risk stage.

[0076] Specifically, first, the weight change curve is input into the convolutional neural network model, and the model automatically extracts the characteristics of the high-risk stage through its convolution layer, such as multiple peak data and fluctuations of the curve. This process does not rely on manually set rules, but the model automatically identifies possible cheating behavior patterns after learning a large amount of weighing data. Subsequently, the model focuses on the high-risk stage and analyzes subtle changes in the curve, such as abnormal increases or decreases in peak values, and unusual fluctuations in the curve, to obtain the curve characteristics of this stage. Combined with Figure 2 As shown in the figure, the weighing process generally includes: the vehicle weighing section and the driver getting off the vehicle weighing section. The vehicle weighing section is the stage from the front wheel of the vehicle touching the scale to the completion of driving on the scale. The curve in this stage should show a gradual upward trend. If the curve fluctuates abnormally, it may indicate that the driver affects the weighing result by means of sudden braking or other means. The rear wheel of the vehicle is weighed to the stage when the driver is ready to get off the scale. The curve in this stage should reach a relatively stable peak value. The getting off the vehicle weighing section is generally when the driver gets off the scale and swipes the card under the scale to weigh and record the vehicle weight. The curve in this stage should drop slightly, but the amplitude should not be too large. After that, the curve should remain relatively stable. If there is fluctuation, it may be caused by external interference of the scale. Therefore, the driver getting off the vehicle weighing section is a high-risk stage. At this time, the driver gets off the scale and swipes the card under the scale to weigh and record the vehicle weight. The curve in this stage should drop slightly, but the amplitude should not be too large. After that, the curve should remain relatively stable. If there is fluctuation, it may be caused by external interference of the scale.

[0077] In this optional embodiment, by automatically detecting and acquiring the high-risk stage and the weight change curve corresponding to the high-risk stage, the reliance on manual supervision is reduced, thereby reducing labor costs and improving the efficiency of the weighing process. The monitoring of the high-risk stage is utilized to improve the accuracy of identifying cheating behavior.

[0078] Optionally, inputting the weight change curve into the convolutional neural network model, and extracting the high-risk stage from the weighing process through the convolutional neural network model, comprises:

[0079] Inputting the weight change curve of the weighing process into the convolutional neural network model, and obtaining trend characteristics of the weight change curve of the weighing process according to the weight change curve of the weighing process, wherein the trend characteristics include the shape and slope change of the weight change curve;

[0080] The trend feature is compared with a preset weight change trend of the weighing process to obtain the high-risk stage in the weighing process.

[0081] Specifically, first, the weight change curve generated by the weighing process is input into the convolutional neural network model. The model recognizes the shape and slope changes of the curve through its convolutional layer. These trend features are crucial for understanding the dynamic changes in the weighing process. Then, the model compares these trend features with the preset normal weight change trend, and identifies the high-risk stage that is significantly different from the normal trend through the learned pattern, so as to automatically identify possible cheating behavior patterns.

[0082] In this optional embodiment, the shape and slope of the weight change curve are carefully analyzed through a deep learning model. The model can process the input weighing data in real time, quickly identify high-risk stages, and improve the detection accuracy of cheating behavior.

[0083] Optionally, judging whether there is cheating in the weighing process and identifying the type of cheating according to the weight change curve of the weighing process includes:

[0084] Predicting the curve features in the high-risk stage through the convolutional neural network model, and outputting the predicted probability of the curve features in a plurality of preset feature labels; wherein each of the feature labels corresponds to one of the cheating types or normal weighing types;

[0085] According to the predicted probability, obtaining the feature label corresponding to the curve feature;

[0086] Determining whether the cheating behavior occurs during the weighing process according to the feature label corresponding to the curve feature;

[0087] If yes, the cheating type is identified according to the feature tag.

[0088] Specifically, first, the weight change curve during the weighing process is input into the convolutional neural network model, and the model identifies and extracts the curve features of the high-risk stage. Then, the model predicts these features and calculates the prediction probability for each preset feature label (each cheating type). This step involves the model's learning of normal and abnormal curves in the training data, as well as modeling the relationship between features and cheating types. Finally, based on the predicted probability, the model determines the feature label that the curve feature is most likely to correspond to, thereby determining whether there is cheating behavior and identifying the specific cheating type, where if the feature label corresponds to a normal weighing type, it is determined that there is no cheating behavior. In a preferred embodiment of the present invention, the feature label corresponding to the curve feature is determined based on the predicted probability of the curve feature in a plurality of preset feature labels, wherein four feature labels are included, namely: a normal label, a first abnormal label, a second abnormal label, and a third abnormal label.

[0089] Among them, combined Figure 3 As shown in the figure, the weighing curve of the normal label is that the weight of the scale gradually decreases as the driver gets off the scale, and remains stable without fluctuations after the driver leaves the scale. Figure 4 As shown, the abnormal situation in the curve of the first abnormal label is that the weight of the scale gradually drops to the lowest point as the driver gets off the scale, then the weight of the scale increases slightly and fluctuates, and remains stable when the weight of the scale is slightly larger. It can be judged that the driver is cheating by still stepping on the scale after getting off the scale. Figure 5 As shown, the abnormal situation in the curve of the second abnormal label is that the weight of the scale gradually drops to the lowest point as the driver gets off the scale. After a short period of stability, the driver places a heavy object on the scale, and the weight of the scale rises slightly and stabilizes. It can be judged that the driver placed a foreign object on the vehicle. Figure 6 As shown, the abnormal situation in the curve of the third abnormal label is that the weight of the scale gradually decreases and stabilizes as the driver gets off the scale. After a short period of stability, the driver unloads the goods from the scale. The weight of the scale decreases again and then stabilizes. It can be judged that the driver has performed an unloading operation after getting off the scale. Figures 3 to 6 Through the convolutional neural network model, each curve is accurately labeled with its own feature label, and is accompanied by the probability value predicted by the model that the curve belongs to the corresponding label.

[0090] In this optional embodiment, by outputting the prediction probability, the model can quantitatively evaluate the possibility of cheating behavior, thereby improving the accuracy of cheating detection. At the same time, since the automated prediction and identification process reduces the need for manual analysis, it reduces labor costs and improves efficiency.

[0091] Optionally, obtaining the feature label corresponding to the curve feature according to the predicted probability includes:

[0092] Determine the relationship between the predicted probability and a preset probability threshold;

[0093] The feature label whose predicted probability is greater than or equal to the preset probability threshold is regarded as the feature label to which the curve feature belongs.

[0094] Specifically, the curve features of the high-risk stage are analyzed through the convolutional neural network model, and the corresponding predicted probability is output for each possible cheating type. A preset probability threshold is set, which is used to evaluate the confidence of the model prediction to determine whether there is enough evidence to believe that the curve feature corresponds to a certain type of cheating. The predicted probability output by the model is compared with the preset threshold to determine whether the predicted probability reaches or exceeds this threshold. If the predicted probability is greater than or equal to the preset threshold, the feature label is regarded as the attribution of the curve feature, that is, the model's prediction is considered to be credible, and the corresponding cheating behavior is judged accordingly.

[0095] In this optional embodiment, by introducing a probability threshold, uncertain predictions with low prediction probabilities are filtered out, thereby improving the accuracy of cheating detection.

[0096] Optionally, predicting the curve feature in the high-risk stage by using the convolutional neural network model and outputting a predicted probability of the curve feature in a plurality of preset feature labels includes:

[0097] The weight change curve in the high-risk stage is processed sequentially through multiple convolutional layers, activation functions, pooling layers, and fully connected layers of the convolutional neural network model to obtain a feature vector of the curve feature for each feature label;

[0098] The feature vector is converted into the predicted probability of the curve feature in each feature label by the classifier of the convolutional neural network model.

[0099] Specifically, the input data is processed by multiple convolutional layers in the convolutional neural network model, and the spatial features of the curve are extracted through convolution operations. The feature map after the convolution layer introduces nonlinearity through the activation function (such as ReLU) to enhance the expressive power of the model. The feature map is reduced in dimension using a pooling layer (such as maximum pooling) to reduce the number of parameters while retaining important information. The feature map after dimensionality reduction is converted into a feature vector through a fully connected layer to provide input for the classifier. The classifier of the convolutional neural network model calculates and outputs the predicted probability for each preset feature label based on the feature vector.

[0100] In this optional embodiment, through multi-layer convolution and pooling operations, the model can learn multi-level features from shallow to deep layers, enhancing the ability to identify complex cheating behaviors. The design of the fully connected layer and classifier ensures that the model can accurately convert feature vectors into predicted probabilities, improving the accuracy of classification.

[0101] Combination Figure 7 As shown, a weighbridge cheating detection device 700 provided in an embodiment of the present invention includes:

[0102] Weighing module 710, used to obtain the real-time weight value of the scale;

[0103] An analysis module 720, for obtaining a weight change curve of the scale during the weighing process of the vehicle according to the real-time weight value;

[0104] The prediction module 730 is used to determine whether there is cheating in the weighing process and identify the type of cheating according to the weight change curve of the weighing process.

[0105] Optionally, the analysis module 720 is specifically used to generate a weighing curve according to the real-time weight value when the real-time weight value starts to change from 0; when the real-time weight value drops to 0, stop generating the weighing curve and use the weighing curve as the weight change curve.

[0106] Optionally, the prediction module 730 is specifically used to obtain, through a convolutional neural network model, a curve feature of the weight change curve in the high-risk stage according to the weight change curve; and determine whether there is cheating behavior in the high-risk stage according to the curve feature in the high-risk stage.

[0107] Optionally, the prediction module 730 is further specifically used to input the weight change curve into the convolutional neural network model, extract the high-risk stage from the weighing process through the convolutional neural network model, and obtain the weight change curve of the high-risk stage; according to the peak data and fluctuation data of the weight change curve of the high-risk stage, obtain the curve characteristics in the high-risk stage.

[0108] Optionally, the prediction module 730 is further specifically used to predict the curve features in the high-risk stage through the convolutional neural network model, and output the predicted probability of the curve features in the preset multiple feature labels; wherein each of the feature labels corresponds to a cheating type or a normal weighing type; according to the predicted probability, the feature label corresponding to the curve feature is obtained; according to the feature label corresponding to the curve feature, it is determined whether the cheating behavior exists in the weighing process; if so, the cheating type is identified according to the feature label.

[0109] Optionally, the prediction module 730 is further specifically used to determine the size relationship between the predicted probability and a preset probability threshold; and take the feature label whose predicted probability is greater than or equal to the preset probability threshold as the feature label belonging to the curve feature.

[0110] Optionally, the prediction module 730 is further specifically used to process the weight change curve in the high-risk stage through multiple convolutional layers, activation functions, pooling layers and fully connected layers of the convolutional neural network model in sequence to obtain a feature vector of the curve feature for each feature label; and convert the feature vector into the predicted probability of the curve feature in each feature label through the classifier of the convolutional neural network model.

[0111] The weighbridge cheating detection device of the present invention realizes dynamic monitoring of the vehicle weighing process by acquiring the weight value of the weighbridge in real time and generating a weight change curve, and accurately identifies abnormal patterns in the curve and promptly discovers abnormal situations according to the weight change curve of the weighing process, so as to judge whether there is cheating behavior in the weighing process, and identifies the cheating type according to the weight change curve, thereby realizing automatic detection of cheating behavior and active identification of cheating type, improving the automation and intelligence level of detection and cheating detection accuracy, reducing interference from human factors, enhancing the ability to identify cheating behavior, and thus improving the cheating detection effect in the automobile weighing process.

[0112] Combination Figure 8 As shown, an electronic device 800 provided by an embodiment of the present invention includes a memory 810 and a processor 820; the memory 810 is used to store a computer program; the processor 820 is used to implement the model reasoning method as described above when executing the computer program.

[0113] In other words, an electronic device 800 includes a memory 810 and a processor 820 coupled to the memory 810; the memory 810 is configured to store a computer program; and the processor 820 is configured to perform the following operations when executing the computer program:

[0114] Get the real-time weight value of the scale;

[0115] According to the real-time weight value, a weight change curve of the scale during the weighing process of the vehicle is obtained;

[0116] According to the weight change curve of the weighing process, it is determined whether there is cheating in the weighing process and the type of cheating is identified.

[0117] The electronic device of the present invention can dynamically monitor the weighing process of the vehicle by acquiring the weight value of the scale in real time and generating a weight change curve. It can accurately identify abnormal patterns in the curve and timely discover abnormal situations according to the weight change curve of the weighing process, so as to judge whether there is cheating in the weighing process, and identify the cheating type according to the weight change curve, thereby realizing automatic detection of cheating and active identification of cheating types, improving the automation and intelligence level of detection and the cheating detection accuracy, reducing the interference of human factors, enhancing the ability to identify cheating, and thus improving the cheating detection effect in the automobile weighing process.

[0118] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned weighbridge cheating detection method is implemented.

[0119] In other words, a non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor performs the following operations:

[0120] Get the real-time weight value of the scale;

[0121] According to the real-time weight value, a weight change curve of the scale during the weighing process of the vehicle is obtained;

[0122] According to the weight change curve of the weighing process, it is determined whether there is cheating in the weighing process and the type of cheating is identified.

[0123] The computer-readable storage medium of the present invention can dynamically monitor the weighing process of the vehicle by acquiring the weight value of the scale in real time and generating a weight change curve. It can accurately identify abnormal patterns in the curve and timely discover abnormal situations according to the weight change curve of the weighing process, so as to judge whether there is cheating in the weighing process, and identify the cheating type according to the weight change curve, thereby realizing automatic detection of cheating and active identification of cheating types, improving the automation and intelligence level of detection and the cheating detection accuracy, reducing the interference of human factors, enhancing the ability to identify cheating, and thus improving the cheating detection effect in the automobile weighing process.

[0124] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc. In the present application, the unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment of the present invention. In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0125] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A method for detecting cheating on a weighbridge, characterized in that: include: Get the real-time weight value of the scale; According to the real-time weight value, a weight change curve of the scale during the weighing process of the vehicle is obtained; According to the weight change curve of the weighing process, it is determined whether there is cheating in the weighing process and the type of cheating is identified.

2. The method for detecting cheating on a weighbridge according to claim 1, characterized in that: The step of obtaining a weight change curve of the scale during the weighing process of the vehicle according to the real-time weight value includes: When the real-time weight value starts to change from 0, a weighing curve is generated according to the real-time weight value; When the real-time weight value drops to 0, the generation of the weighing curve is stopped and the weighing curve is used as the weight change curve.

3. The method for detecting cheating on a weighbridge according to claim 1, characterized in that: The weighing process includes a high-risk stage, and judging whether there is cheating in the weighing process according to the weight change curve of the weighing process includes: Obtaining, by a convolutional neural network model, a curve feature of the weight change curve in the high-risk stage according to the weight change curve; According to the curve characteristics in the high-risk stage, it is determined whether there is cheating behavior in the high-risk stage.

4. The method for detecting cheating on a weighbridge according to claim 3, characterized in that: The method of obtaining the curve characteristics of the weight change curve in the high-risk stage according to the weight change curve by using a convolutional neural network model includes: Inputting the weight change curve into the convolutional neural network model, extracting the high-risk stage from the weighing process through the convolutional neural network model, and obtaining the weight change curve of the high-risk stage; The curve characteristics in the high-risk stage are obtained according to the peak data and the fluctuation data of the weight change curve in the high-risk stage.

5. The method for detecting cheating on a weighbridge according to claim 4, characterized in that: The step of inputting the weight change curve into the convolutional neural network model and extracting the high-risk stage from the weighing process through the convolutional neural network model comprises: Inputting the weight change curve of the weighing process into the convolutional neural network model, and obtaining trend characteristics of the weight change curve of the weighing process according to the weight change curve of the weighing process, wherein the trend characteristics include the shape and slope change of the weight change curve; The trend feature is compared with a preset weight change trend of the weighing process to obtain the high-risk stage in the weighing process.

6. The method for detecting cheating on a weighbridge according to claim 5, characterized in that: The determining, based on the weight change curve of the weighing process, whether there is cheating behavior in the weighing process and identifying the type of cheating includes: Predicting the curve features in the high-risk stage through the convolutional neural network model, and outputting the predicted probability of the curve features in a plurality of preset feature labels; wherein each of the feature labels corresponds to one of the cheating types or normal weighing types; According to the predicted probability, obtaining the feature label corresponding to the curve feature; Determining whether the cheating behavior occurs during the weighing process according to the feature label corresponding to the curve feature; If yes, the cheating type is identified according to the feature tag.

7. The method for detecting cheating on a weighbridge according to claim 6, characterized in that: The obtaining, according to the predicted probability, the feature label corresponding to the curve feature comprises: Determine the relationship between the predicted probability and a preset probability threshold; The feature label whose predicted probability is greater than or equal to the preset probability threshold is regarded as the feature label to which the curve feature belongs.

8. The method for detecting cheating on a weighbridge according to claim 6, characterized in that: The predicting of the curve features in the high-risk stage by the convolutional neural network model and outputting the predicted probability of the curve features in a plurality of preset feature labels includes: The weight change curve in the high-risk stage is processed sequentially through multiple convolutional layers, activation functions, pooling layers, and fully connected layers of the convolutional neural network model to obtain a feature vector of the curve feature for each feature label; The feature vector is converted into the predicted probability of the curve feature in each feature label by the classifier of the convolutional neural network model.

9. A weighbridge cheating detection device, characterized in that: include: Weighing module, used to obtain the real-time weight value of the scale; An analysis module, used for obtaining a weight change curve of the scale during the weighing process of the vehicle according to the real-time weight value; The prediction module is used to determine whether there is cheating in the weighing process and identify the type of cheating according to the weight change curve of the weighing process.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the weighbridge cheating detection method according to any one of claims 1 to 8 is implemented.

Citation Information

Cited By

  • State analysis method and system for vehicle weighing

    CN120832469A

  • A state analysis method and system for vehicle weighing

    CN120832469B