Abnormality detection method, device and equipment for truck scale and storage medium

Abstract

The application relates to the technical field of data processing, and discloses a truck scale abnormality detection method, device, equipment and storage medium, which comprises the following steps: obtaining running state data and surrounding image data of a truck scale when a vehicle to be weighed drives onto the truck scale; processing the running state data and the surrounding image data by using a preset target detection model, cross-verification and correlation analysis are performed on the processing results, and preliminary abnormality diagnosis results and potential fault prediction results are obtained; and generating a target abnormality detection report according to the preliminary abnormality diagnosis results and the potential fault prediction results. When the vehicle to be weighed drives onto the truck scale, the running state data and the surrounding image data are processed by using the preset target detection model, cross-verification and correlation analysis are performed, and the target abnormality detection report is generated. The abnormality in the running process of the truck scale can be found in a more timely and accurate manner, detailed basis is provided for subsequent maintenance and repair, and the stable running of the truck scale is effectively ensured.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, equipment and storage medium for detecting anomalies in truck scales. Background Technology

[0002] A truck scale is a large measuring instrument whose working principle is mainly based on the combination of mechanics and electronics. With the in-depth development of sensor manufacturing technology and microelectronics technology, truck scales are constantly being infused with new technologies. At the same time, it is also necessary to handle abnormal situations that occur on truck scales. However, existing methods for acquiring abnormal truck scale data are limited in scope, mostly focusing only on acquiring weighing data. The ability to simultaneously acquire truck scale operating status data and surrounding image data is weak, resulting in an insufficient understanding of the overall operating status of the truck scale and making it difficult to detect potential faults in advance. Summary of the Invention

[0003] The main objective of this application is to provide a method, apparatus, equipment, and storage medium for detecting anomalies in truck scales, aiming to solve the technical problem that most existing methods for obtaining abnormal truck scale data only focus on obtaining weighing data and do not provide a comprehensive understanding of the overall operating status of the truck scale.

[0004] To achieve the above objectives, this application proposes a method for detecting anomalies in truck scales, the method comprising:

[0005] When a vehicle to be weighed drives onto the truck scale, the operating status data of the truck scale and surrounding image data are acquired.

[0006] The operating status data and the surrounding image data are processed using a preset target detection model, and the processing results are cross-validated and correlated to obtain preliminary anomaly diagnosis results and potential fault prediction results.

[0007] A target anomaly detection report is generated based on the preliminary anomaly diagnosis results and the potential fault prediction results.

[0008] Optionally, the target detection model includes a first detection model and a second detection model;

[0009] The steps of processing the operating status data and the surrounding image data using a preset target detection model, and performing cross-validation and correlation analysis on the processing results to obtain preliminary anomaly diagnosis results and potential fault prediction results include:

[0010] Based on the first detection model, the operating status data is analyzed to obtain the first abnormal feature and sensor performance indicators;

[0011] Based on the second detection model, target recognition is performed on the surrounding image data to obtain second abnormal features and mechanical structure features;

[0012] Cross-validation is performed on the first abnormal feature and the second abnormal feature to generate preliminary abnormal diagnosis results;

[0013] Correlation analysis is performed on the sensor performance indicators and the mechanical structure characteristics to obtain potential fault prediction results.

[0014] Optionally, the step of cross-validating the first abnormal feature and the second abnormal feature to generate a preliminary abnormal diagnosis result includes:

[0015] Feature matching is performed on the first abnormal feature and the second abnormal feature to obtain the matching result;

[0016] When the matching result is consistent, the first abnormal feature and the second abnormal feature are determined to be deterministic anomalies, and a first-level anomaly alarm is generated.

[0017] When the matching result is inconsistent, the first abnormal feature and the second abnormal feature are weighted according to the fault priority rule to generate a second-level abnormal alarm.

[0018] Based on the first-level and / or second-level anomaly alarms, determine the preliminary anomaly diagnosis results.

[0019] Optionally, the step of performing correlation analysis on the sensor performance indicators and the mechanical structure characteristics to obtain potential fault prediction results includes:

[0020] The time-series data of the sensor performance indicators and the spatial data of the mechanical structure features are compared and processed to generate a set of feature pairs with spatiotemporal correlation.

[0021] When sensor anomalies and / or mechanical deformation areas appear in the feature pair set, a third-level anomaly alarm is generated.

[0022] When the number of third-level abnormal alarms exceeds the preset abnormal threshold, redundant data is removed from the coverage of the sensor abnormal points and the mechanical deformation area, and the target feature pairs in the feature pair set with a confidence level higher than the preset confidence threshold are retained.

[0023] The output potential fault prediction results are based on the target features.

[0024] Optionally, the step of generating a target anomaly detection report based on the preliminary anomaly diagnosis results and the potential fault prediction results includes:

[0025] The abnormal dataset of the truck scale is determined based on the preliminary anomaly diagnosis results and the potential fault prediction results.

[0026] The abnormal dataset is matched with the first historical fault database, and the similarity between the abnormal faults in the abnormal dataset and the historical faults is calculated.

[0027] Based on the similarity, a fault analysis report and repair plan for the abnormal fault are determined;

[0028] Based on the abnormal dataset, the fault analysis report, and the maintenance plan, a target abnormality detection report for the truck scale during the current inspection process is generated.

[0029] Optionally, after the step of generating a target anomaly detection report for the truck scale during the current inspection process based on the anomaly dataset, the fault analysis report, and the maintenance plan, the method further includes:

[0030] Upon receiving the target question input from the target user, the natural language processing module parses the fault description in the target question to obtain the key fault feature vector;

[0031] The key fault feature vectors are semantically matched with the target anomaly detection report to determine the root cause node of the fault.

[0032] Based on the root cause node of the fault, a voice interaction command is generated, and the fault answer corresponding to the target problem is returned to the target user through the voice interaction command.

[0033] Optionally, before the step of acquiring the operating status data and surrounding image data of the truck scale when the vehicle to be weighed enters the truck scale, the method further includes:

[0034] When the weighing sensors and instruments in the truck scale receive the start signal sent by the platform, they use a machine learning model to intelligently detect the core components of the truck scale and obtain component status data.

[0035] The component status data is matched based on the second historical fault database, and the matching results are classified by a deviation parameter within a preset range to obtain the classification results and the corresponding abnormal feature codes.

[0036] If the classification result is the first abnormal result, then the target calibration instruction corresponding to the first abnormal result is matched according to the preset abnormal type mapping table, and after the calibration instruction is executed, a ready signal is sent to the alarm module of the monitoring system through the message bus.

[0037] If the classification result is the second abnormal result, the abnormal feature code is persistently stored based on the self-inspection processing module of the truck scale, and a maintenance request signal with a time limit identifier is sent to the work order module of the monitoring system according to the preset priority rules. After the maintenance request signal disappears, the abnormal feature code is updated.

[0038] Furthermore, to achieve the above objectives, this application also proposes a weighbridge anomaly detection device, which includes:

[0039] The data acquisition module is used to acquire the operating status data and surrounding image data of the truck scale when the vehicle to be weighed enters the truck scale.

[0040] The data processing module is used to process the operating status data and the surrounding image data using a preset target detection model, and to perform cross-validation and correlation analysis on the processing results to obtain preliminary anomaly diagnosis results and potential fault prediction results.

[0041] The report generation module is used to generate a target anomaly detection report based on the preliminary anomaly diagnosis results and the potential fault prediction results.

[0042] In addition, to achieve the above objectives, this application also proposes a weighbridge anomaly detection device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the weighbridge anomaly detection method as described above.

[0043] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the truck scale anomaly detection method described above.

[0044] This application discloses a method for acquiring operational status data and surrounding image data of a truck scale when a vehicle enters it; processing the operational status data and surrounding image data using a preset target detection model; and performing cross-validation and correlation analysis on the processing results to obtain preliminary anomaly diagnosis results and potential fault prediction results; and generating a target anomaly detection report based on the preliminary anomaly diagnosis results and the potential fault prediction results. By simultaneously acquiring operational status data and surrounding image data when a vehicle enters the truck scale, and then processing and cross-validating and analyzing the two types of data using a preset target detection model, a target anomaly detection report is finally generated. Compared to traditional detection methods based on a single data source, this method can more timely and accurately detect anomalies during the operation of the truck scale, providing detailed information for subsequent maintenance and repair, and effectively ensuring the stable operation of the truck scale. Attached Figure Description

[0045] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating the first embodiment of the truck scale anomaly detection method of this application;

[0048] Figure 2 This is a structural diagram of the data processing for the truck scale anomaly detection method of this application;

[0049] Figure 3 This is a flowchart illustrating the second embodiment of the truck scale anomaly detection method of this application;

[0050] Figure 4 This is a schematic diagram of the structure of the first detection model in this application;

[0051] Figure 5 This is a schematic diagram of the structure of the second detection model in this application;

[0052] Figure 6 This is a flowchart illustrating the third embodiment of the truck scale anomaly detection method of this application;

[0053] Figure 7 This is a schematic diagram of the module structure of the truck scale anomaly detection device according to an embodiment of this application;

[0054] Figure 8 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the abnormal detection method of the truck scale in this application embodiment.

[0055] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0056] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0057] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0058] The main solution of this application embodiment is as follows: when a vehicle to be weighed enters the truck scale, the operating status data and surrounding image data of the truck scale are acquired; the operating status data and surrounding image data are processed using a preset target detection model, and the processing results are cross-validated and correlated to obtain preliminary anomaly diagnosis results and potential fault prediction results; a target anomaly detection report is generated based on the preliminary anomaly diagnosis results and the potential fault prediction results.

[0059] A truck scale is a large weighing instrument whose working principle is mainly based on a combination of mechanics and electronics. When vehicles and goods are parked on the platform of the truck scale, the resulting gravity is transmitted to the load cells. With the in-depth development of sensor manufacturing technology and microelectronics technology, truck scales are constantly being infused with new technologies. At the same time, abnormal situations have also occurred during the weighing process, causing significant economic losses to enterprises and users.

[0060] Extensive investigation and analysis were conducted on various forms of cheating on truck scales. The results show that cheating in the truck scale weighing process can be categorized into two types: human cheating and technical cheating. Human cheating refers to methods that exploit management loopholes in the truck scale weighing process. This includes vehicle following, incomplete weighing, vehicle concealment, repeated weighing, data tampering, and collusion. Technical cheating refers to methods that involve installing cheating devices on key weighing modules such as weighing sensors, weighing instruments, junction boxes, and communication cables to alter weighing data. This type of cheating is characterized by its high technical sophistication, concealment, and difficulty in detection.

[0061] Existing anti-cheating detection methods acquire abnormal data from truck scales in a single dimension, with most focusing only on acquiring weighing data. They are weak in simultaneously acquiring truck scale operating status data and surrounding image data, resulting in an insufficient understanding of the overall operating status of the truck scale and making it difficult to detect potential faults in advance.

[0062] This application provides an anomaly data detection method based on a large model. In terms of alarm statistics, it integrates alarm information generated by instruments, sensors, weighing equipment and other devices in various areas of the warehouse, classifies and statistically analyzes them according to alarm type, time, location and other dimensions, and generates intuitive alarm trend charts and percentage charts to help decision-makers quickly grasp alarm dynamics and allocate resources in a timely manner to deal with potential risks.

[0063] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, anomaly detection, and intelligent response capabilities, such as an instrument equipped with intelligent AI, or an electronic device capable of performing the aforementioned functions. The following description uses a truck scale anti-cheating anomaly detection platform as an example to illustrate this embodiment and the subsequent embodiments.

[0064] Based on this, embodiments of this application provide a method for detecting anomalies in a truck scale, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the truck scale anomaly detection method of this application.

[0065] In this embodiment, the abnormal detection method for the truck scale includes:

[0066] Step S10: When the vehicle to be weighed drives onto the truck scale, acquire the operating status data of the truck scale and the surrounding image data.

[0067] It should be noted that the operational status data is a collection of physical quantity data reflecting the real-time working status of the truck scale and its sensors, such as sensor data: sensor temperature, humidity, tilt angle, voltage, operating status of each sensor, connection status, and casing damage, etc.; instrument data: whether there is off-center load, communication status (normal / offline), power supply voltage, filtering parameters, etc. Surrounding image data is visual information about the surrounding environment of the truck scale obtained through devices such as cameras and LiDAR, such as smoke and fire detection, personnel at the weighbridge, and safety helmet detection, etc. This embodiment does not impose any limitations on this.

[0068] It should be understood that the vehicle to be weighed can be a motor vehicle that requires weighing, condition detection, or other compliance checks, such as a truck or van. When the vehicle enters, the detection platform automatically identifies the vehicle type (such as axle number matching) based on features like weight distribution and wheel position, and initiates the corresponding data acquisition and detection procedures.

[0069] Furthermore, in order to detect potential problems with core components in advance and ensure they are in good working order during the start-up phase of the truck scale, the following steps are included before step S10:

[0070] When the weighing sensors and instruments in the truck scale receive the start signal sent by the platform, they intelligently detect the core components of the truck scale using a machine learning model to obtain component status data. The component status data is matched against a second historical fault database, and the matching results are classified using a preset range of deviation parameters to obtain classification results and corresponding anomaly feature codes. If the classification result is a first anomaly, a target calibration instruction corresponding to the first anomaly is matched according to a preset anomaly type mapping table. After the calibration instruction is executed, a ready signal is sent to the alarm module of the monitoring system via a message bus. If the classification result is a second anomaly, the anomaly feature code is persistently stored based on the truck scale's self-test processing module. A maintenance request signal with a timeliness identifier is sent to the work order module of the monitoring system according to a preset priority rule. After the maintenance request signal disappears, the anomaly feature code is updated.

[0071] It should be noted that before the load cells and instruments receive the start signal from the platform, all core components of the truck scale are in a powered-off state. The receipt of the start signal indicates that the truck scale is initializing. Core components refer to those that play a crucial role in the weighing function, including load cells (measuring weight), instruments (data processing and display), and mechanical connection structures (support and fixing components). The second historical fault database stores historical fault cases of the truck scale during startup, along with corresponding component status data, used to match anomalies detected during startup.

[0072] Additionally, it should be noted that the first abnormal result represents a fault that the truck scale itself can handle. Based on the preset abnormality type mapping table, the corresponding target calibration command can be selected for the first abnormal result, and automatic correction and clearing of the abnormality can be performed. The second abnormal result represents a fault that the truck scale itself cannot handle, requiring an alarm to be triggered and the corresponding maintenance personnel to be notified for repair.

[0073] Understandably, the machine learning model is used to analyze the status data of the core components of the truck scale during startup. It can be a time-series-based long short-term memory neural network. Input data includes initial values ​​of the weighing sensors, power supply voltage fluctuation curves of the instrument module, and vibration spectra of the mechanical structure. Outputs include component status labels (normal / abnormal), abnormal feature codes, and various signals. The training data consists of historical startup data of the truck scale.

[0074] Specifically, when the weighing sensors and instruments in the truck scale receive the start signal sent by the platform, a machine learning model is triggered to intelligently detect the core components and acquire component status data, such as sensor bridge voltage and instrument response time. The detected component status data is matched with cases in a second historical fault database to determine the degree to which the data deviates from the normal threshold. The matching results are then categorized into "first abnormal result" (faults that can be handled automatically) or "second abnormal result" (faults that cannot be handled automatically), and corresponding abnormal feature codes are generated. For the first abnormal result, according to a preset abnormality type mapping table, the corresponding target calibration instruction (such as zero-point calibration) is matched. After calibration, a "ready signal" is sent to the alarm module of the monitoring system via the message bus, indicating that the fault has been resolved. For the second abnormal result, the truck scale's self-test processing module stores the abnormal feature code in local storage and simultaneously sends a maintenance request signal with a "timeliness identifier" to the work order module according to preset priority rules (such as sorting by fault impact). When maintenance is completed and the signal disappears, the abnormal feature code status is updated (e.g., marked as "processed").

[0075] In one example, the weighbridge operator starts the weighbridge with a single click on the platform each time they start work, and the truck scale automatically triggers a self-test program. After communication between the load cell, instrument, and platform is restored, the machine learning model detects that the load cell's output voltage continuously exceeds the preset range, triggering a load cell alarm. This status data is then matched against the historical database, finding a match for a "sensor zero-point drift" fault, triggering an alarm, and the deviation parameter is determined as the first abnormal result. A zero-point calibration command is automatically executed, and a ready signal is sent to the alarm module after calibration, clearing the fault. If the self-test detects occasional slight delays in the instrument's response time, but these delays do not exceed the severe fault threshold, this is determined as the second abnormal result. The abnormality signature is stored in the self-test module, and a maintenance request with a "process within 24 hours" timeframe is sent to the work order module. After the maintenance personnel repair the instrument within the time limit, the system detects the disappearance of the maintenance request signal and updates the abnormality signature to "repaired."

[0076] Step S20: The operating status data and the surrounding image data are processed using a preset target detection model, and the processing results are cross-validated and correlated to obtain preliminary anomaly diagnosis results and potential fault prediction results.

[0077] It should be noted that the target detection model includes a first detection model and a second detection model. The first detection model is an algorithmic model used to analyze equipment operating status data. Based on deep learning, it extracts features from time-series data (such as temperature, pressure, and current) collected by sensors to identify abnormal patterns or evaluate sensor performance. The second detection model is a computer vision model designed for surrounding image data, used to identify objects, abnormal shapes, or mechanical structural features in the images.

[0078] Understandably, preliminary anomaly diagnosis results are based on data processing and analysis, making a preliminary judgment on whether the truck scale currently has any anomalies and the type of such anomalies. Potential fault prediction results are predictions of future faults that may occur on the truck scale by analyzing data trends or historical patterns.

[0079] Step S30: Generate a target anomaly detection report based on the preliminary anomaly diagnosis results and the potential fault prediction results.

[0080] It should be noted that the target anomaly detection report can be presented in a structured format, a data statistics dashboard, or a 3D-style truck scale interface, where each fault is displayed in the corresponding area of ​​the truck scale interface where the anomaly occurred. This embodiment does not impose any limitations on this.

[0081] In one example, reference Figure 2 , Figure 2This diagram illustrates the data processing structure of the truck scale anomaly detection method proposed in this application. The data sources include ambient environmental data and real-time operational data. Ambient environmental data, after acquisition, is stored in a database via an image collection module, then drives the fault analysis and target recognition modules. The recognition results are sent to a data warehouse for further processing. Real-time operational data is captured via a real-time data acquisition interface, sorted and categorized through a data queue, and stored in real-time in storage modules such as the data warehouse, heap operational data, and latest operational data values. Both types of data are further aggregated into the "anti-cheating detection platform," where they are combined with structured data from the business platform for data mining or model calculation. The processed data, through a data centralization module, supports model optimization work such as "model or type training, model data preparation management, and statistical analysis," while also serving historical detection data analysis. Simultaneously, it triggers real-time alarms based on anomaly models and provides data support for real-time monitoring or alarm modules, ultimately forming a complete closed loop from data acquisition, processing, analysis to application, ensuring the accuracy and real-time nature of truck scale anomaly detection. This solution was validated using data from 500 truck scales, achieving an anomaly detection accuracy of 95.2%, a 16.8% improvement over traditional single-model methods. The false alarm rate was reduced to 2.3%, a 9.8% decrease, meeting real-time monitoring requirements. Through multimodal collaborative analysis of sensors and images, the model successfully detected complex anomalies such as shell breakage (92.3% confidence level) and mechanical deformation (bounding box error less than 5 pixels). Potential fault warning coverage was increased to 89.7%, and maintenance suggestions were generated based on a dynamic weighting strategy, with a historical fault matching success rate of 91%. Experiments show that the fusion of sensor temporal features and mechanical image spatial features significantly improves diagnostic robustness, maintaining over 85% detection accuracy even in complex scenarios with lighting interference and data noise. It also supports voice interaction for generating diagnostic reports with a response latency of less than 500ms, providing efficient guidance for on-site maintenance.

[0082] In this embodiment, when a vehicle to be weighed enters the truck scale, the operating status data of the truck scale and surrounding image data are acquired. A preset target detection model is used to process the operating status data and surrounding image data, and the processing results are cross-validated and correlated to obtain preliminary anomaly diagnosis results and potential fault prediction results. A target anomaly detection report is generated based on the preliminary anomaly diagnosis results and the potential fault prediction results. By simultaneously acquiring operating status data and surrounding image data when a vehicle enters the truck scale, and using a preset target detection model to process the two types of data followed by cross-validation and correlation analysis, a target anomaly detection report is finally generated. Compared to traditional detection methods based on a single data source, this method can more timely and accurately detect anomalies during the operation of the truck scale, providing detailed information for subsequent maintenance and repair, and effectively ensuring the stable operation of the truck scale.

[0083] Reference Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the truck scale anomaly detection method of this application. Based on the first embodiment described above, a second embodiment of the truck scale anomaly detection method of this application is proposed.

[0084] In the second embodiment, step S20 includes:

[0085] Step S201: Based on the first detection model, perform data analysis on the operating status data to obtain the first abnormal feature and sensor performance indicators.

[0086] It should be noted that the first abnormal feature is a characteristic that deviates from the normal operating range, identified by the first detection model after analyzing the operating status data. Examples include instantaneous voltage exceeding limits, regular fluctuations in vibration frequency, and continuous temperature increases. Sensor performance indicators are key parameters used to quantitatively evaluate the sensor's operating status, including sensitivity, response time, stability, and drift rate. This embodiment does not impose any limitations on these parameters.

[0087] Specifically, after collecting the operational status data generated by the sensors during the operation of the truck scale, the data needs to be preprocessed, including cleaning and normalization. Then, the preprocessed data is input into the first detection model. The model scans the data using algorithmic logic (such as time series analysis and anomaly detection algorithms) to identify the first abnormal feature that does not conform to the normal operating mode. For example, if the voltage value of the weighing sensor is found to be consistently below the standard threshold, it is determined as a "signal abnormal feature." Simultaneously, the model calculates sensor performance indicators based on preset rules or a trained evaluation system. For instance, by comparing historical sensor measurement data with the actual weight value, it assesses whether the accuracy has decreased; by monitoring the sensor data update frequency, it determines whether the response time meets the standard. Finally, the model outputs the first abnormal feature reflecting the current operational anomaly, as well as sensor performance indicators that quantitatively evaluate the sensor's working status.

[0088] In one example, reference Figure 4 , Figure 4This is a schematic diagram of the structure of the first detection model in this application. The first detection model is a hybrid deep learning architecture designed based on multimodal sensor temporal data analysis. This model integrates the temporal modeling capabilities of Long Short-Term Memory (LSTM) networks and the spatial feature extraction advantages of Convolutional Neural Networks (CNNs). It also uses a Residual Network (ResNet) module to solve the gradient vanishing problem in deep network training, achieving accurate monitoring and anomaly diagnosis of the sensor's operating status. The first half is a CNN module, where input data sequentially passes through convolutional layers (extracting local spatial patterns), pooling layers (compressing feature dimensions), and compression layers (dimensionality reduction and feature fusion), abstracting spatial information layer by layer using different weights and filters. The second half is a bidirectional LSTM module, where h... t-1 h t h t+1 Representing the hidden states at different times, conveying temporal information, x t-1 x t x t+1 This represents the input data at different times. An LSTM unit processes time-series data and captures long-term dependencies. The double-headed arrows indicate a bidirectional LSTM structure, moving from the positive direction (from x...) to the negative direction (from x...). t-1 To x t+1 ) and the reverse (from x) t+1 To x t-1 The system processes data and fuses bidirectional information to better extract temporal features. The bidirectional LSTM module consists of two sets of LSTM units, one forward and one backward. It captures long-term dependencies in sensor data (such as voltage trend drift or temperature periodic fluctuations) through bidirectional temporal modeling. The spatial features extracted by CNN interact and fuse with the temporal state of LSTM, ultimately outputting multidimensional diagnostic results that combine sudden anomaly detection (such as voltage drop) and gradual pattern analysis (such as angle shift), realizing comprehensive dynamic monitoring of sensor operating status.

[0089] The model uses six-dimensional time-series data as training data and input (including voltage, temperature, and XYZ axis angles of a gyroscope). It captures long-range temporal dependencies in sensor data through a two-layer LSTM network, such as gradually changing features like continuous voltage decreases or angle drift. The high-order time-series features output by the LSTM are transformed by spatial coordinates and then input into a one-dimensional convolutional layer to extract local spatial patterns, such as features of sudden events like voltage drops and abnormal temperature fluctuations. The core residual connection module, through cross-layer feature fusion, learns complex abstract patterns while preserving effective information from lower layers, significantly improving the model's ability to model multi-scale features.

[0090] Meanwhile, the model adopts a multi-task learning framework, outputting results in parallel through two branches: 1. Anomaly classification branch, based on fully connected layers and the Softmax function, identifies 8 predefined anomalies (such as shell breaking cheating, data tampering, etc.) and outputs confidence probabilities; 2. Performance evaluation branch maps features to a health score of 0-100 through the Sigmoid activation function. During training, the focus loss function is used to alleviate the class imbalance problem, while the Huber loss function is used to improve the regression robustness of the health score. A dynamic weight adjustment mechanism is used to balance the optimization objectives of classification and regression tasks.

[0091] Step S202: Based on the second detection model, target recognition is performed on the surrounding image data to obtain the second abnormal feature and mechanical structure feature.

[0092] It should be noted that the second type of anomaly feature is abnormal visual information extracted from the image through target recognition, such as vehicles following each other, incomplete weighing, vehicles carrying items, and repeated weighing. The mechanical structure feature is information related to the mechanical structure of the truck scale identified from the image, including component cracks, loose screws, structural deformation, surface wear, and liquid leakage.

[0093] Specifically, when a vehicle enters the truck scale, the anomaly detection platform uses cameras to collect real-time or periodic image data of the surrounding area, covering key mechanical components, installation areas, and personnel. The collected images undergo preprocessing, including noise reduction, brightness adjustment, and resolution optimization. The preprocessed images are then input into a second detection model. This model uses techniques such as convolutional neural networks to analyze the images pixel-by-pixel, identifying target objects such as weighing sensors and instrument modules. During the identification process, the model ultimately outputs second anomaly features (e.g., vehicle not fully on the scale) and mechanical structural features (e.g., weighing sensor installation angle deviation of 5°), providing intuitive visual analysis for subsequent truck scale fault diagnosis and structural assessment.

[0094] In one example, reference Figure 5 , Figure 5This is a schematic diagram of the structure of the second detection model in this application. The second detection model is a computer vision residual attention network. It integrates the feature extraction capabilities of residual networks with the context-aware advantages of spatial attention mechanisms. Its core achieves multi-scale feature fusion through 1×1 convolutional layers and cross-layer feature concatenation. The input image data first passes through the concatenation module to integrate contextual information, and then passes through 1×1 convolutional layers for channel-dimensional feature compression and spatial enhancement. The feature map dimensions are dynamically adjusted through reshaping and transposition operations, and non-linear expressive power is introduced by combining activation functions (such as ReLU). At the same time, residual skip connections are used to retain the original effective information to avoid gradient degradation. The spatial / channel dual attention mechanism in the network further calibrates the weights of the feature map—spatial attention focuses on abnormal regions (such as structural deformation), and channel attention suppresses background interference. Finally, the output is a multi-task diagnostic result that combines pixel-level localization (damage bounding boxes) and semantic classification (five types of image anomalies), which significantly improves the detection robustness in complex scenes.

[0095] This model takes 640×480 resolution RGB images as input and, through multi-stage joint optimization, outputs dual results: anomaly type identification and mechanical structure quantitative assessment, providing high-precision visual analysis for equipment condition monitoring in industrial scenarios. Based on a residual network backbone architecture, the model removes the traditional global pooling layer and introduces a hollow spatial pyramid pooling module. It utilizes convolutional kernels with different dilation rates to capture multi-scale contextual information in the image in parallel, effectively identifying diverse features ranging from minute cracks (local scale) to overall structural deformation (global scale). To further enhance its ability to focus on key regions, the model integrates a channel-spatial dual attention mechanism: the channel attention module dynamically calibrates the importance weights of feature channels through global average pooling, suppressing irrelevant background interference; the spatial attention module generates a spatial weight heatmap through 1×1 convolution, highlighting the pixel response of abnormal areas (such as foreign object intrusion locations or deformed parts). The two work together, enabling the model to accurately locate targets even in complex environments.

[0096] To address the need for multi-task output, the model adopts a bifurcated design: 1. The anomaly classification branch outputs the probability distribution of five preset anomalies (such as foreign object intrusion and structural deformation) through a four-layer fully connected network, and uses a focal loss function to optimize training, alleviating the problem of imbalanced sample distribution; 2. The structural regression branch outputs component integrity scores (Sigmoid activation) and key component bounding box coordinates (normalized format, Smooth L1 loss supervision) in parallel. The scoring mechanism combines a nonlinear decay formula to quantify the health status of the equipment, while the bounding box accurately defines the damaged area, providing spatial guidance for manual review.

[0097] During training, the model achieves stable convergence through the AdamW optimizer and cosine annealing learning rate scheduling strategy, and employs data augmentation techniques such as random erasure and multi-scale brightness and contrast perturbation to improve generalization ability. In actual deployment, the model can complete single-frame image processing in just 22 milliseconds on embedded devices, supports real-time analysis at 45 frames per second at 4K resolution, achieves an average detection accuracy of 91.4%, and has an intersection-over-union ratio (IoU) of over 0.89 for structural deformation regions. Its spatial attention heatmap visualization function can intuitively demonstrate the model's decision-making basis. For example, in the detection of truck scale load-bearing platforms, the highlighted areas in the heatmap highly match the manually labeled deformation locations, significantly improving the reliability of the results.

[0098] The second detection model, together with the first detection model, constructs a multimodal anomaly verification system. When both models have an anomaly confidence level of over 85% for the same event, the system automatically triggers the highest-level handling protocol, suppressing the overall false alarm rate to below 0.3%, thereby achieving comprehensive intelligent monitoring and early warning of mechanical structure anomalies.

[0099] Step S203: Cross-validate the first abnormal feature and the second abnormal feature to generate a preliminary abnormal diagnosis result.

[0100] Understandably, cross-validation aligns information from a first anomalous feature (such as abnormal voltage in a weighing sensor) with a second anomalous feature (such as a sensor's external image showing damaged wiring), establishing a spatial or logical relationship between the two. For example, it confirms whether a sensor with abnormal operating data is the same object as a component with detected physical damage in an image. It utilizes different features to supplement the limitations of single data points. If the operating data only shows parameter fluctuations, but the image data further reveals that the fluctuations are caused by component wear, complementary analysis can be used to clarify the root cause of the anomaly.

[0101] Furthermore, to avoid potential misjudgments from judging a single data anomaly and to improve the reliability of anomaly diagnosis, step S203 may include:

[0102] Feature matching is performed on the first abnormal feature and the second abnormal feature to obtain a matching result; when the matching result is consistent, the first abnormal feature and the second abnormal feature are determined as deterministic anomalies, and a first-level anomaly alarm is generated; when the matching result is inconsistent, the first abnormal feature and the second abnormal feature are weighted according to the fault priority rule to generate a second-level anomaly alarm; based on the first-level anomaly alarm and / or the second-level anomaly alarm, a preliminary anomaly diagnosis result is determined.

[0103] It should be noted that feature matching is a process of comparing and analyzing the first and second abnormal features to determine whether they point to the same source of the fault. When the two types of abnormal features match, the type of the anomaly can be clearly determined, and the same fault can be confirmed by multiple sources of data. The first-level anomaly alarm indicates a high-priority alarm generated for deterministic anomalies, prompting maintenance personnel to handle it first.

[0104] Understandably, fault priority rules are pre-defined fault classification standards that categorize faults based on factors such as the scope of their impact and their urgency, and are used for weight allocation. When matching results are inconsistent, different weights need to be assigned to mismatched anomaly features according to fault priority to quantify the overall severity of the anomaly. The second-level anomaly alarm is generated based on the weight allocation results, reflecting the overall risk level of the non-deterministic anomaly.

[0105] In one example, sensor data fluctuations and sensor surface cracks are matched. If both are associated with the same fault—that is, both the data anomaly and the hardware crack point to sensor damage—the matching result is consistent, classified as a deterministic anomaly, and a Level 1 anomaly alarm is generated, indicating an urgent need for handling. If the matching results are inconsistent, based on the fault priority rules, the weight of anomalies affecting the core weighing function is set to 0.7, and the weight of minor wear on secondary components is set to 0.3. The overall risk is calculated using these weights, generating a Level 2 anomaly alarm that reflects the overall severity of the anomaly. After one round of anomaly detection (i.e., one vehicle weighing), combined with multiple Level 1 anomaly alarms and / or multiple Level 2 anomaly alarms, a preliminary anomaly diagnosis result is output.

[0106] Step S204: Perform correlation analysis on the sensor performance indicators and the mechanical structure characteristics to obtain potential fault prediction results.

[0107] Understandably, correlation analysis examines the relationship between sensor performance metrics and mechanical structural features, such as whether a decrease in sensor accuracy is caused by wear of mechanical components. Correlation analysis between sensor performance metrics and mechanical structural features can be performed by establishing a mapping relationship between them, determining the correlation through machine learning, or based on a set of spatiotemporally correlated feature pairs between sensor performance metrics and mechanical structural features.

[0108] Furthermore, in order to predict failures more proactively, reduce the probability of failures, and lower downtime and maintenance costs caused by failures, step S204 may include:

[0109] The time-series data of the sensor performance indicators and the spatial data of the mechanical structure features are compared and processed to generate a set of feature pairs with spatiotemporal correlation. When sensor anomalies and / or mechanical deformation areas appear in the feature pair set, a third-level anomaly alarm is generated. When the number of third-level anomaly alarms exceeds a preset anomaly threshold, redundant data is removed from the coverage areas of the sensor anomalies and mechanical deformation areas, and target feature pairs with confidence levels higher than a preset confidence threshold are retained in the feature pair set. Based on the target feature pairs, potential fault prediction results are output.

[0110] It should be noted that time-series data is a sequence of performance parameters collected by the sensor at different points in time. Spatial data describes the spatial attributes of mechanical components, such as the three-dimensional dimensions of the component, the location and shape of the deformation area, and the loosening position of screws. Feature pair sets are sets of "time-space-feature value" triplets formed by spatiotemporally associating the time-series data of sensor performance indicators with the spatial data of mechanical structural features. For example, "On XX year XX month XX day at 10:00, the accuracy of weighing sensor A decreased by 15%, corresponding to a 2mm deformation at point Y of the mechanical support structure."

[0111] Understandably, the Level 3 anomaly alarm is triggered when a sensor anomaly or mechanical deformation area is detected in the feature pair set. The preset anomaly threshold is a pre-defined critical value for the number of alarms, used to determine whether redundant data needs to be processed. This removes duplicate, low-value, or low-confidence information from the feature pair set, retaining core, valid data to improve analysis efficiency and accuracy. The preset confidence threshold is a confidence standard used to filter feature pairs; for example, only feature pairs with a confidence level higher than 80% are retained to ensure the reliability of the analysis results. Target feature pairs are high-confidence spatiotemporal correlation feature pairs retained after redundancy removal and confidence screening, which can serve as the basis for generating potential fault prediction results.

[0112] In one example, when comparing sensor accuracy fluctuations over time with data on the location and extent of component deformation, the process first associates the time points of sensor accuracy degradation with the corresponding deformation areas of mechanical components, forming a "time-location-outlier" combination. Next, it iterates through the feature pair set, checking for feature pairs where sensor accuracy is below a threshold or structural crack length exceeds a preset value. If such anomalies are detected, a feature pair is generated displaying "Sensor B's response time at 10:30 exceeds the standard by 20%, and a 3mm bend appears at the corresponding mechanical link C," triggering an alarm. When the number of such anomalies triggers 50 alarms within one hour, only feature pairs with a confidence level greater than or equal to 90% are retained, excluding random anomalies. Finally, based on the current sensor drift rate and mechanical deformation rate, it is predicted that "Sensor 3 will break within 24 hours due to excessive deformation and requires immediate repair."

[0113] In this embodiment, the operating status data is analyzed based on a first detection model to obtain first anomaly features and sensor performance indicators; target recognition is performed on the surrounding image data based on a second detection model to obtain second anomaly features and mechanical structure features; cross-validation is performed on the first and second anomaly features to generate preliminary anomaly diagnosis results; correlation analysis is performed on the sensor performance indicators and the mechanical structure features to obtain potential fault prediction results. By subdividing the target detection model into a first detection model and a second detection model, and processing the operating status data and surrounding image data respectively, different dimensions of feature information are obtained, making the analysis of different types of data more professional and in-depth, and further improving the accuracy of anomaly detection.

[0114] Reference Figure 6 , Figure 6 This is a flowchart illustrating the third embodiment of the truck scale anomaly detection method of this application. Based on the second embodiment described above, the third embodiment of the truck scale anomaly detection method of this application is proposed.

[0115] In the third embodiment, step S30 includes:

[0116] Step S301: Determine the abnormal dataset of the truck scale based on the preliminary abnormal diagnosis results and the potential fault prediction results.

[0117] Specifically, when determining the abnormal dataset of the truck scale, the currently identified abnormal information is first extracted through preliminary abnormality diagnosis results, including but not limited to abnormality type, abnormality location, abnormality characteristics, and time. Next, potential fault information, such as the predicted fault type and the possible time range, is extracted through potential fault prediction results. Finally, the information from the preliminary diagnosis and prediction results is integrated into a unified data table according to the aforementioned fields.

[0118] Step S302: Match the abnormal dataset with the first historical fault database and calculate the similarity between the abnormal faults in the abnormal dataset and the historical faults.

[0119] It should be noted that the first historical fault database is a database that stores the faults that have occurred in the truck scale system in history and their related characteristics.

[0120] It should be understood that calculating the similarity between abnormal faults and historical faults requires selecting an appropriate algorithm based on the data type, such as using Euclidean distance or cosine similarity.

[0121] Understandably, by quantifying feature differences, a similarity score can be generated, with higher scores indicating greater similarity between the two faults. The type of the current anomaly can be determined based on the similarity score. Matching strategies between anomalies and historical faults can be one-to-one matching: for each record in the anomaly dataset, traversing the historical fault database, calculating its similarity to all historical faults, and retaining the top N records with the highest similarity; or rule-based filtering optimization: first filtering a subset of historical faults based on key features (such as fault type and location), and then calculating similarity within the subset to reduce computational load; or time window matching: matching records in historical faults whose timestamps fall within a certain range before and after the anomaly time to determine the anomaly type.

[0122] In one example, for each record in the anomaly dataset, key feature vectors, such as anomaly type, location, and time window, are extracted. The matching degree with records in the first historical fault database is calculated using an improved cosine similarity formula.

[0123]

[0124] Where S represents similarity, w i The feature weights for sensor malfunctions and mechanical structure malfunctions can be manually set during the installation of the truck scale according to the operating environment, or they can be derived through continuous optimization during subsequent use. i and y i These are the feature values ​​for abnormal faults and historical faults, respectively. Historical faults with a similarity greater than 80% are retained as a reference after calculation. If an abnormal fault matches multiple historical faults, the solution with the highest success rate in the repair plan is selected first.

[0125] The following are some of the historical faults in the first historical fault database:

[0126] 1. Sensor related

[0127] Cheating by breaking the shell: A shell-breaking fault was detected;

[0128] Communication failure: Unable to communicate normally;

[0129] Data tampering: Data between sensors and instruments, and between instruments and sensors, cannot be decrypted normally according to the encryption method;

[0130] Reading fault: There is a fault in the sensor data;

[0131] Data acquisition interruption: Data acquisition is suddenly interrupted;

[0132] Shell breakage fault: Shell breakage exists;

[0133] Voltage change: Sudden change in sensor voltage;

[0134] Abnormal attitude: The sensor tilt angle is abnormal;

[0135] Temperature anomaly: The internal temperature of the sensor is abnormal.

[0136] 2. Instrument related

[0137] Zero-point anomaly: The data at midnight contains an anomaly;

[0138] Parameter backup: Compared with historical data, there is evidence of improper scale calibration;

[0139] Instrument disconnection: The device disconnects during use;

[0140] Instrument casing opening: If the instrument casing is opened, cheating will occur.

[0141] 3. Scanning fault

[0142] Health abnormality: When the scale is empty, analysis using the coaxial sensor AD in real time indicates that the scale body is offset.

[0143] Non-empty scale weighing: The scale body is not zeroed and is being weighed;

[0144] Off-center load detection: When the scale is not empty, the coaxial sensor is subjected to an off-center load.

[0145] By employing multimodal data fusion (sensors and images), fine-grained model design (LSTM and residual attention network), and dynamic weight allocation strategies, this approach reduces the false alarm rate to 2.3%, a 5-fold improvement over traditional methods, while also supporting 24-hour early warning for 83% of potential faults.

[0146] Step S303: Determine the fault analysis report and repair plan for the abnormal fault based on the similarity.

[0147] Understandably, after identifying the corresponding anomaly, a preliminary maintenance strategy can be formulated based on the solutions to the most similar historical faults. Simultaneously, the differences between the current anomaly and historical faults can be analyzed to optimize maintenance procedures.

[0148] Step S304: Generate a target anomaly detection report for the truck scale during the current detection process based on the anomaly dataset, the fault analysis report, and the maintenance plan.

[0149] Furthermore, in order to facilitate users in obtaining accurate fault solutions through intelligent user interaction, improve problem-solving efficiency, and enhance user experience, the process after step S304 also includes:

[0150] Upon receiving a target question input from a target user, the system parses the fault description in the target question using a natural language processing module to obtain a key fault feature vector; it then performs semantic matching between the key fault feature vector and the target anomaly detection report to determine the root cause node of the fault; based on the root cause node, it generates a voice interaction command and returns the fault response corresponding to the target question to the target user through the voice interaction command.

[0151] It should be noted that the natural language processing module is responsible for parsing the target question input by the user, extracting its semantic information (such as fault type, location, phenomenon, etc.), and converting it into structured data that can be processed by a computer. The format of the target question input by the user can be text, images, speech, etc., and this embodiment does not impose any restrictions on this.

[0152] Understandably, the key fault feature vector is the core fault information extracted from the target problem by the natural language processing module. Semantic matching compares the semantic similarity of two texts (such as a user question and the content of a target anomaly detection report) to determine whether they describe the same or related faults. The fault root cause node is the fault description or root cause node in the target anomaly detection report that has the highest semantic match with the user question. For example, if the user question points to "abnormal sensor voltage," the fault root cause node could be the anomaly record in the report such as "sensor casing damaged, causing low voltage."

[0153] It should be understood that after identifying the root cause node of the fault through semantic matching, the text information can be converted into speech using text-to-speech technology based on the audio response command generated by the root cause node, and returned to the target user to achieve human-computer voice interaction.

[0154] In this embodiment, an abnormal dataset for the truck scale is determined based on the preliminary anomaly diagnosis results and the potential fault prediction results. The abnormal dataset is then matched with a first historical fault database to calculate the similarity between the abnormal faults in the abnormal dataset and historical faults. Based on the similarity, a fault analysis report and repair plan for the abnormal fault are determined. A target anomaly detection report for the truck scale during the current detection process is generated based on the abnormal dataset, the fault analysis report, and the repair plan. When an anomaly is detected, a machine learning-driven solution generation process is executed, allowing maintenance personnel to quickly perform repairs based on the detailed fault analysis and repair plan in the report, improving maintenance efficiency, shortening equipment downtime, and reducing enterprise operating costs.

[0155] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the abnormal detection method of the truck scale in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0156] This application also provides a weighbridge anomaly detection device; please refer to [reference needed]. Figure 7 The abnormal detection device for the truck scale includes:

[0157] The data acquisition module 10 is used to acquire the operating status data and surrounding image data of the truck scale when the vehicle to be weighed drives into the truck scale.

[0158] The data processing module 20 is used to process the running status data and the surrounding image data using a preset target detection model, and to perform cross-validation and correlation analysis on the processing results to obtain preliminary anomaly diagnosis results and potential fault prediction results.

[0159] The report generation module 30 is used to generate a target anomaly detection report based on the preliminary anomaly diagnosis results and the potential fault prediction results.

[0160] The truck scale anomaly detection device provided in this application, employing the truck scale anomaly detection method described in the above embodiments, can solve the technical problem that existing methods for acquiring truck scale anomaly data mostly focus only on acquiring truck scale weighing data, resulting in an insufficient understanding of the overall operating status of the truck scale. Compared with the prior art, the beneficial effects of the truck scale anomaly detection device provided in this application are the same as those of the truck scale anomaly detection method provided in the above embodiments, and other technical features in the truck scale anomaly detection device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0161] This application provides a truck scale anomaly detection device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the truck scale anomaly detection method in Embodiment 1 above.

[0162] The following is for reference. Figure 8 The diagram illustrates a structural schematic suitable for implementing the truck scale anomaly detection device in the embodiments of this application. The truck scale anomaly detection device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (e.g., vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 8The illustrated truck scale anomaly detection device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0163] like Figure 8 As shown, the truck scale anomaly detection device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the truck scale anomaly detection device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the weighbridge anomaly detection equipment to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows weighbridge anomaly detection equipment with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0164] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0165] The truck scale anomaly detection device provided in this application, employing the truck scale anomaly detection method described in the above embodiments, can solve the technical problem that existing methods for acquiring truck scale anomaly data mostly focus only on acquiring weighing data, resulting in an insufficient understanding of the overall operating status of the truck scale. Compared with the prior art, the beneficial effects of the truck scale anomaly detection device provided in this application are the same as those of the truck scale anomaly detection method provided in the above embodiments, and other technical features of this truck scale anomaly detection device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0166] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0167] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0168] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the truck scale anomaly detection method in the above embodiments.

[0169] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0170] The aforementioned computer-readable storage medium may be included in the truck scale anomaly detection equipment; or it may exist independently and not be assembled into the truck scale anomaly detection equipment.

[0171] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the truck scale anomaly detection device, cause the truck scale anomaly detection device to perform the truck scale anomaly detection method described above.

[0172] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0173] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0174] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0175] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described truck scale anomaly detection method. This addresses the technical problem that existing methods for acquiring truck scale anomaly data mostly focus on acquiring weighing data, resulting in insufficient understanding of the overall operational status of the truck scale. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the truck scale anomaly detection method provided in the above embodiments, and will not be elaborated upon here.

[0176] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for detecting anomalies in a truck scale, characterized in that, The method for detecting anomalies in a truck scale includes: When a vehicle to be weighed enters the truck scale, the operating status data of the truck scale and surrounding image data are acquired. The operating status data includes sensor data and instrument data. The operating status data and the surrounding image data are processed using a preset target detection model, and the processing results are cross-validated and correlated to obtain preliminary anomaly diagnosis results and potential fault prediction results. The target detection model includes a first detection model and a second detection model. A target anomaly detection report is generated based on the preliminary anomaly diagnosis results and the potential fault prediction results. The steps of processing the operating status data and the surrounding image data using a preset target detection model, and performing cross-validation and correlation analysis on the processing results to obtain preliminary anomaly diagnosis results and potential fault prediction results include: Based on the first detection model, the operating status data is analyzed to obtain the first abnormal feature and sensor performance indicators; Based on the second detection model, target recognition is performed on the surrounding image data to obtain second abnormal features and mechanical structure features; Cross-validation is performed on the first abnormal feature and the second abnormal feature to generate preliminary abnormal diagnosis results; Correlation analysis is performed on the sensor performance indicators and the mechanical structure features to obtain potential fault prediction results; The step of cross-validating the first abnormal feature and the second abnormal feature to generate a preliminary abnormal diagnosis result includes: Feature matching is performed on the first abnormal feature and the second abnormal feature to obtain the matching result; When the matching result is consistent, the first abnormal feature and the second abnormal feature are determined to be deterministic anomalies, and a first-level anomaly alarm is generated. When the matching result is inconsistent, the first abnormal feature and the second abnormal feature are weighted according to the fault priority rule to generate a second-level abnormal alarm. Based on the first-level and / or second-level anomaly alarms, determine the preliminary anomaly diagnosis results.

2. The method for detecting anomalies in a truck scale as described in claim 1, characterized in that, The step of performing correlation analysis on the sensor performance indicators and the mechanical structure characteristics to obtain potential fault prediction results includes: The time-series data of the sensor performance indicators and the spatial data of the mechanical structure features are compared and processed to generate a set of feature pairs with spatiotemporal correlation. When sensor anomalies and / or mechanical deformation areas appear in the feature pair set, a third-level anomaly alarm is generated. When the number of third-level abnormal alarms exceeds the preset abnormal threshold, redundant data is removed from the coverage of the sensor abnormal points and the mechanical deformation area, and the target feature pairs in the feature pair set with a confidence level higher than the preset confidence threshold are retained. The output potential fault prediction results are based on the target features.

3. The method for detecting anomalies in a truck scale as described in claim 1, characterized in that, The step of generating a target anomaly detection report based on the preliminary anomaly diagnosis results and the potential fault prediction results includes: The abnormal dataset of the truck scale is determined based on the preliminary anomaly diagnosis results and the potential fault prediction results. The abnormal dataset is matched with the first historical fault database, and the similarity between the abnormal faults in the abnormal dataset and the historical faults is calculated. Based on the similarity, a fault analysis report and repair plan for the abnormal fault are determined; Based on the abnormal dataset, the fault analysis report, and the maintenance plan, a target abnormality detection report for the truck scale during the current inspection process is generated.

4. The method for detecting anomalies in a truck scale as described in claim 3, characterized in that, After the step of generating the target anomaly detection report of the truck scale in the current inspection process based on the anomaly dataset, the fault analysis report, and the maintenance plan, the method further includes: Upon receiving the target question input from the target user, the natural language processing module parses the fault description in the target question to obtain the key fault feature vector; The key fault feature vectors are semantically matched with the target anomaly detection report to determine the root cause node of the fault. Based on the root cause node of the fault, a voice interaction command is generated, and the fault answer corresponding to the target problem is returned to the target user through the voice interaction command.

5. The method for detecting anomalies in a truck scale as described in any one of claims 1 to 4, characterized in that, Before the step of acquiring the operational status data and surrounding image data of the truck scale when the vehicle to be weighed enters the truck scale, the method further includes: When the weighing sensors and instruments in the truck scale receive the start signal sent by the platform, they use a machine learning model to intelligently detect the core components of the truck scale and obtain component status data. The component status data is matched based on the second historical fault database, and the matching results are classified by a deviation parameter within a preset range to obtain the classification results and the corresponding abnormal feature codes. If the classification result is the first abnormal result, then the target calibration instruction corresponding to the first abnormal result is matched according to the preset abnormal type mapping table, and after the calibration instruction is executed, a ready signal is sent to the alarm module of the monitoring system through the message bus. If the classification result is the second abnormal result, the abnormal feature code is persistently stored based on the self-inspection processing module of the truck scale, and a maintenance request signal with a time limit identifier is sent to the work order module of the monitoring system according to the preset priority rules. After the maintenance request signal disappears, the abnormal feature code is updated.

6. An abnormality detection device for a truck scale, characterized in that, The device includes: The data acquisition module is used to acquire the operating status data and surrounding image data of the truck scale when the vehicle to be weighed enters the truck scale. The operating status data includes sensor data and instrument data. The data processing module is used to process the running status data and the surrounding image data using a preset target detection model, and to perform cross-validation and correlation analysis on the processing results to obtain preliminary anomaly diagnosis results and potential fault prediction results. The target detection model includes a first detection model and a second detection model. The report generation module is used to generate a target anomaly detection report based on the preliminary anomaly diagnosis results and the potential fault prediction results; The data processing module is further configured to perform data analysis on the operating status data based on the first detection model to obtain first abnormal features and sensor performance indicators; perform target recognition on the surrounding image data based on the second detection model to obtain second abnormal features and mechanical structure features; perform cross-validation on the first abnormal features and the second abnormal features to generate preliminary abnormal diagnosis results; and perform correlation analysis on the sensor performance indicators and the mechanical structure features to obtain potential fault prediction results. The data processing module is further configured to perform feature matching on the first abnormal feature and the second abnormal feature to obtain a matching result; when the matching result is consistent, the first abnormal feature and the second abnormal feature are determined as deterministic anomalies, and a first-level anomaly alarm is generated; when the matching result is inconsistent, the first abnormal feature and the second abnormal feature are weighted according to the fault priority rule to generate a second-level anomaly alarm; and a preliminary anomaly diagnosis result is determined based on the first-level anomaly alarm and / or the second-level anomaly alarm.

7. An abnormality detection device for truck scales, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the weighbridge anomaly detection method as described in any one of claims 1 to 5.

8. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the truck scale anomaly detection method as described in any one of claims 1 to 5.

Images