Load cell data evaluation method, apparatus, device, and storage medium

Abstract

The application relates to the technical field of anti-cheating detection, and discloses a weighing sensor data evaluation method, device, equipment and storage medium, which comprises the following steps: generating associated data associated with current weighing sensor output data; selecting key data from the output data of the current weighing sensor; and evaluating the data of the weighing sensor based on the associated data and the key data. By generating the associated data associated with the current weighing sensor output data first, then extracting the key data, and finally performing health evaluation or anti-cheating early warning based on the two, the active sensing and early warning of the sensor state can be realized. Compared with the traditional single threshold judgment mode, the data correlation analysis and key feature extraction are combined, the accuracy of sensor fault diagnosis and the sensitivity of anti-cheating detection are significantly improved, and the measurement distortion caused by sensor damage or artificial installation of cheating devices is effectively prevented.

Description

Technical Field

[0001] This application relates to the field of anti-cheating detection technology, and in particular to a method, apparatus, device and storage medium for evaluating weighing sensor data. Background Technology

[0002] In existing technologies, load cells, as core components of large weighing instruments such as electronic truck scales and rail scales, are primarily monitored for their health status using single threshold judgments or periodic manual inspections. For example, faults are often diagnosed simply by checking if the output exceeds the rated range, or maintenance personnel periodically inspect the sensor's appearance and wiring. Anti-cheating measures are relatively passive, typically only conducting post-incident investigations after measuring anomalies are discovered, failing to proactively identify and warn of cheating during the process. Summary of the Invention

[0003] The main purpose of this application is to provide a method, apparatus, device and storage medium for evaluating weighing sensor data, which aims to solve the technical problem that the existing anti-cheating measures for weighing sensors are relatively passive and cannot actively identify and warn of cheating behavior.

[0004] To achieve the above objectives, this application proposes a method for evaluating weighing sensor data, the method comprising: Generate associated data that is related to the current weighing sensor output data; Select key data from the output data of the current weighing sensor; The data of the weighing sensor is evaluated based on the associated data and the key data.

[0005] Optionally, selecting key data from the output data of the current weighing sensor includes: Obtain the output data sequence of the current weighing sensor; The output data sequence is subjected to feature recognition to obtain recognition results, wherein the recognition results are at least one of data change rate, data consistency deviation, and data time validity value. Based on the recognition results, the importance score of the output data is determined, and key data is selected from the output data sequence according to the importance score.

[0006] Optionally, the identification result is the data change rate, and the step of performing feature identification on the output data sequence to obtain the identification result includes: Acquire subsequent output data acquired after the current output data. The value of the current output data is compared with the value of the subsequent output data to obtain the data change rate of the current output data.

[0007] Optionally, the identification result is a data consistency deviation, and the step of performing feature identification on the output data sequence to obtain the identification result includes: Acquire the preceding output data whose acquisition time was before the current output data; The value of the current output data is compared with the value of the previous output data to obtain the data consistency deviation of the current output data.

[0008] Optionally, the identification result is a valid data time value, and the step of performing feature identification on the output data sequence to obtain the identification result includes: Obtain the first time difference between the acquisition time of the current output data and the start time of the current weighing sensor; And / or, obtain a second time difference between the acquisition time of the current output data and the previous calibration time; The effective time value of the current output data is determined based on the first time difference and / or the second time difference.

[0009] Optionally, generating associated data related to the current weighing sensor output data includes: Obtain the target historical data waveform corresponding to the current weighing sensor; Determine the probability of occurrence of each feature pattern in the target historical data waveform; Based on the probability of occurrence, associated data is generated that is related to the current weighing sensor output data. The associated data includes a predicted waveform or a standard reference template.

[0010] Optionally, determining the probability of occurrence of each feature pattern in the target historical data waveform includes: The target historical data waveform is segmented to obtain data segments; The data segments are labeled with features, and target feature fragments are selected from the data segments based on the labeling results; The target feature segment is extended by working condition to obtain similar working condition features corresponding to the target feature segment; The probability of occurrence of each feature pattern in the target historical data waveform is determined based on the target feature fragment and the similar working condition features.

[0011] Furthermore, to achieve the above objectives, this application also proposes a weighing sensor data evaluation device, which includes: The generation module is used to generate associated data that is related to the current weighing sensor output data; The selection module is used to select key data from the output data of the current weighing sensor; An evaluation module is used to evaluate the data of the weighing sensor based on the associated data and the key data.

[0012] In addition, to achieve the above objectives, this application also proposes a weighing sensor data evaluation 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 weighing sensor data evaluation method as described above.

[0013] 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 weighing sensor data evaluation method described above.

[0014] This application first generates correlation data that is related to the current weighing sensor output data, then extracts key data, and finally performs health assessments or anti-cheating warnings based on both, enabling proactive sensing and early warning of sensor status. Compared to traditional methods that rely solely on a single threshold, this approach integrates data correlation analysis and key feature extraction, significantly improving the accuracy of sensor fault diagnosis and the sensitivity of anti-cheating detection. It is particularly suitable for high-load, high-frequency usage scenarios such as ports and truck scales, effectively preventing measurement distortion caused by sensor damage or the addition of cheating devices. Attached Figure Description

[0015] 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.

[0016] 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.

[0017] Figure 1 This is a flowchart illustrating the first embodiment of the weighing sensor data evaluation method of this application; Figure 2 This is a schematic diagram illustrating the evaluation of weighing sensor data according to an embodiment of this application; Figure 3 This is a flowchart illustrating the second embodiment of the weighing sensor data evaluation method of this application; Figure 4 This is a flowchart illustrating the third embodiment of the weighing sensor data evaluation method of this application; Figure 5This is a schematic diagram of the module structure of the weighing sensor data evaluation device according to an embodiment of this application; Figure 6 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the weighing sensor data evaluation method in this application embodiment.

[0018] 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

[0019] 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.

[0020] 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.

[0021] Weighing sensors are the core sensing components of large weighing instruments such as electronic truck scales, rail scales, and hopper scales. In practical applications, especially in large ports such as Qingdao Port, truck scales operate under high load, high frequency, and harsh environments (humidity, salt spray, large temperature differences) for a long time. Sensors are prone to progressive failures such as zero drift, creep deviation, strain gauge fatigue, and seal damage. They also face the risk of measurement fraud by adding cheating devices, damaging the casing, and tampering with signals.

[0022] Currently, existing weighing sensor health monitoring and anti-cheating technologies have the following shortcomings: First, monitoring methods mostly rely on single-point threshold judgment or periodic manual inspections. For example, an alarm is only triggered when the sensor output exceeds the rated range, or maintenance personnel measure the output resistance with a multimeter quarterly. This approach cannot capture subtle abnormal trends in sensor output, causing early faults (such as strain gauge micro-cracks or temperature compensation failure) to often only be discovered after they have developed into severe drift or even complete failure, leading to metrological disputes or production interruptions. Second, anti-cheating methods mostly employ passive protection measures such as physical seals and locking the instrument housing, or retrospectively investigate cheating behavior through video playback, lacking proactive, real-time intelligent early warning capabilities. When cheating occurs (such as remotely installing cheating devices or vehicles not fully on the scale), the system cannot identify and alarm in a timely manner during the weighing process. Third, existing technologies fail to fully utilize the inherent correlations between sensor output data, such as the force ratio relationship of coaxial sensors, the load distribution patterns of sensors at different positions, and the time series characteristics of historical data. These related information contain rich clues about health status and characteristics of cheating behavior, but most current methods treat each sensor as an independent unit and only focus on single-point values, resulting in a waste of a lot of valuable information and one-sided and unreliable judgment results.

[0023] Therefore, this application provides a method for health status assessment and anti-cheating warning based on reference data that is related to the current sensor output and intelligently filtering key features from real-time data, so as to make up for the deficiencies of the prior art in active diagnosis, early warning and correlation analysis.

[0024] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a computer, or an electronic device capable of performing the above functions. The following description uses a weighing sensor data evaluation system as an example to illustrate this embodiment and the subsequent embodiments.

[0025] Based on this, embodiments of this application provide a method for evaluating weighing sensor data, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the weighing sensor data evaluation method of this application.

[0026] In this embodiment, the weighing sensor data evaluation method includes: Step S10: Generate associated data that is related to the current weighing sensor output data.

[0027] It should be noted that the load cell output data is the load signal collected and uploaded in real time by digital load cells installed under the truck scale loader. This data may include the AD value (analog-to-digital conversion code value), temperature value, attitude angle value, and shell integrity flag for each sensor. Correlated data refers to reference data that has an inherent logical relationship or historical regularity with the current output data, including predicted values ​​from theoretical force models, statistical characteristic values ​​of the same group of sensors, and historical trend fitting values. Optionally, related data can also be coupling constraint relationship data between multiple sensors, such as the force ratio of coaxial sensors and the load deviation of symmetrically positioned sensors.

[0028] In this embodiment, the associated data can be divided into three categories: predictive associated data based on physical models, statistical associated data based on data-driven methods, and feature associated data based on historical patterns. Predictive associated data can be the theoretical output value of each sensor under ideal conditions, obtained through a preset deep learning model based on the mechanical structure of the truck scale, the sensor placement locations, and the vehicle axle load distribution. Statistical associated data based on data-driven methods can be constructed by using statistical measures such as the mean, median, variance, and skewness of the outputs of multiple sensors in the same group to construct the population distribution characteristics that the current output data should follow. Feature associated data based on historical patterns is the typical waveforms, trend features, or state cluster centers extracted from the historical data of the sensor or similar sensors during their normal operating cycle.

[0029] Specifically, as an optional implementation, when generating associated data related to the current weighing sensor output data, the real-time output vectors of all sensors on the truck scale are first obtained. For example, for eight column-type digital sensors, each output includes a load code value, temperature, attitude angle, and a shell breakage indicator. Then, a mechanical model pre-installed in the weighing indicator or monitoring server is invoked. This model, based on the U-beam structure of the load-bearing structure, the coordinates of the sensor support points, and the vehicle axle limit positions, calculates the theoretical load distribution ratio of each sensor under the same total load. This ratio is multiplied by the current total load to obtain the theoretical predicted value of each sensor, which serves as the associated data. For example, when a truck is fully on the scale, if the axle position calculates that sensor number 1 should bear 12.5% ​​of the total load, and the actual total load is 80t, then its theoretical predicted value is 10t, which is the associated data.

[0030] In another optional implementation, when the correlated data is statistical correlated data of the same group of sensors, the system collects the load outputs of the eight sensors in real time and calculates their average value. and standard deviation Then, for each sensor's current output value x, statistical bias correlation data relative to the population is generated, such as the z-score standardized value. or generate This serves as a correlation constraint. When the output of a sensor exceeds this range, an anomaly is determined. This method does not require a pre-established mechanical model and relies entirely on the statistical regularities of the data itself. It is suitable for truck scale scenarios with symmetrical sensor layouts and uniform load distribution.

[0031] Optionally, when the associated data is feature-based data based on historical patterns, the system collects output data under normal operating conditions within the first 30 days after the sensor is first installed and calibrated, constructing a standard output waveform library for each load range and temperature band. When data associated with the current output data needs to be generated, the system matches the most similar operating condition from the historical waveform library based on the current load value, ambient temperature, and timestamp, extracting the historical output waveform or feature value (such as creep curve, zero-point recovery trend) under that operating condition as the associated data. For example, if the current load is 40t and the ambient temperature is 30℃, the system retrieves the output waveform of the same load and temperature from history, smooths and filters it, and uses it as the expected reference waveform for the current output.

[0032] Understandably, if valid correlation data cannot be generated (e.g., insufficient historical data from newly installed sensors, or the mechanical model failing due to deformation of the load-bearing structure), the system will automatically downgrade to single-sensor threshold judgment mode and output a "correlation data unavailable" message, requiring manual intervention to calibrate or supplement modeling data.

[0033] Step S20: Select key data from the output data of the current weighing sensor.

[0034] It should be noted that key data refers to characteristic data points or data segments selected from the current weighing sensor output data sequence that have high characterization value for health status assessment and anti-cheating diagnosis. The purpose of selecting key data is to reduce data redundancy and focus on sensitive information that can reflect sensor performance degradation, sudden anomalies, or cheating behavior. The selection process may include steps such as data sequence acquisition, multi-dimensional feature identification, importance score calculation, and threshold screening. Optionally, key data can be an abnormal jump point of a single sensor or a point of inconsistency deviation of multiple sensors.

[0035] In this embodiment, the output data sequence refers to the sensor output dataset collected within a continuous time window (such as the most recent 1 minute, the most recent 10 weighing cycles, or the most recent 100 vehicle passage records). Each data point includes at least a timestamp, load value, temperature value, attitude angle, and shell flag. Feature recognition involves mathematical transformation and pattern extraction of the output data sequence. The resulting recognition includes at least one of the following: data change rate, data consistency deviation, and data time validity value. Specifically, the data change rate reflects the transient fluctuation of the sensor output; the data consistency deviation reflects the degree of deviation between the sensor and other sensors in the same group or its own historical output; and the data time validity value reflects the timeliness weight of the data.

[0036] Step S30: Evaluate the data of the weighing sensor based on the associated data and the key data.

[0037] It should be noted that evaluating the data from the weighing sensor can involve comparing, fusing, or logically reasoning with the generated correlated data and selected key data to determine the current health status of the weighing sensor and whether cheating is occurring. The evaluation result can be a quantitative score (e.g., health score 0-100), a qualitative rating (e.g., excellent / good / poor), or triggering specific alarm events (e.g., sensor damage alarm, abnormal posture alarm, data consistency error alarm). Optionally, the evaluation process can be executed in real-time on the local weighing indicator, or the correlated and key data can be uploaded to a monitoring server for comprehensive big data analysis.

[0038] In this embodiment, the evaluation can include three dimensions: deviation evaluation, trend evaluation, and logical consistency evaluation. Deviation evaluation refers to calculating the absolute or relative deviation of key data relative to associated data to determine whether it exceeds the allowable error range; trend evaluation refers to analyzing the changing trend of key data over multiple consecutive time windows to determine whether there is gradual degradation; logical consistency evaluation refers to verifying whether the key data conforms to the physical constraints or statistical laws defined by the associated data, such as the force ratio relationship of multiple sensors, the coupling relationship between load and attitude, etc.

[0039] Specifically, as an optional implementation, when evaluating deviation based on correlated data and key data, the system compares the key data (e.g., the real-time load output value x of a sensor) with the correlated data (e.g., the theoretical prediction value μ calculated based on a mechanical model) to calculate the relative deviation. Then, the evaluation level is divided according to the preset deviation threshold: when At that time, the assessment was normal; At that time, the assessment was to be taken seriously; At that time, the assessment was raised to a warning level; When an abnormality is detected, an alarm is triggered. For example, if a column-type digital sensor has a theoretical predicted value of 39.95t under a 40t load, but the actual key data output is 40.80t, the relative deviation is 2.13%, exceeding the 2.0% threshold. The system immediately outputs a "sensor is severely abnormal" assessment result and recommends immediate shutdown for repair. This method is suitable for scenarios with a mechanical model that provides support and has high model accuracy.

[0040] In another optional implementation, when performing trend assessment based on associated and key data, the system acquires key data sequences extracted from multiple consecutive vehicle passage cycles (e.g., the most recent 30 vehicle passages), along with corresponding associated data sequences (e.g., standard reference curves in a historical waveform library). Time series analysis methods (e.g., moving average, exponential smoothing, or linear regression) are used to calculate the cumulative offset trend of the key data relative to the associated data. Specifically, the deviation value for each vehicle passage is calculated, and the slope k of the deviation change over time is fitted. If k > 0 and the deviation is positive for 10 consecutive vehicle passages, it is determined to be a "positive drift trend"; if k < 0 and the deviation is negative for 10 consecutive vehicle passages, it is determined to be a "negative drift trend". When the cumulative offset exceeds 1.5 times the maximum permissible error of the sensor (e.g., ±0.02%FS for a C3-level sensor), a warning is output, prompting for angle difference calibration or zero-point tracking. This allows for the early detection of signs of slow aging before the sensor completely fails, enabling predictive maintenance.

[0041] Optionally, when performing logical consistency assessment based on associated data and key data, the associated data represents the coupling constraints between multiple sensors. For example, two sensors on the same axis (on the same side) should output similar load values ​​when the vehicle is centered, with a theoretical ratio of 1; or, for a bridge sensor in a horizontal orientation, the load output should satisfy the constraint that the attitude angle is linearly independent (i.e., angle changes should not cause sudden load changes). The system substitutes key data (such as the real-time load and real-time attitude angle of a sensor) into these constraints to determine whether logical violations occur. For example, a truck scale's left front sensor outputs 15t, and the left rear sensor outputs 5t, while the associated data (theoretical ratio calculated based on wheelbase and track width) requires the ratio to be between 0.8 and 1.2, but the actual ratio is as high as 3.0, clearly violating logical consistency. In this case, the system assesses it as "suspected cheating," specifically, the vehicle may be severely unbalanced or not fully on the scale, and outputs corresponding alarm information. For example, if the shell breakage flag of a certain column-type digital sensor is triggered (this flag is a state latch and is not lost when the power is off), the shell integrity requirement in the associated data is "intact", but the actual shell breakage flag in the critical data is "damaged". The logic consistency check fails, and the system evaluates it as "sensor shell breakage alarm", indicating that it may have been disassembled and a cheating device has been installed.

[0042] Furthermore, the system can integrate the results of deviation assessment, trend assessment, and logical consistency assessment to generate a comprehensive health index. As an optional implementation, a weighted scoring method is used. When the comprehensive score is ≥90, the system is assessed as healthy; 80 ≤ score <90, it is assessed as sub-healthy; 60 ≤ score <80, it is assessed as warning; and a score <60, it is assessed as dangerous. For example, if a sensor has a deviation of 0.8% (deduct 10 points), a positive trend slope with a cumulative offset of 0.015% (deduct 5 points), and normal logical consistency (deduct 0 points), and uses α=0.5, β=0.3, and γ=0.2, the comprehensive score is 93.5, and the system is assessed as healthy. This comprehensive score can be directly displayed on the equipment health dashboard of the monitoring platform for reference by port dispatchers and maintenance personnel.

[0043] Understandably, if related or critical data is missing (e.g., due to communication interruption preventing the acquisition of related data, or failure to extract critical data), the system cannot complete the normal assessment. It will then output a "Insufficient assessment conditions" message and automatically downgrade to a single-sensor threshold alarm mode. Simultaneously, a log of this assessment failure will be recorded for subsequent troubleshooting of network or sensor malfunctions. For sensors with assessment results of "Warning" or "Danger," the system should automatically push alarm messages to the monitoring platform's PC and mobile app. Based on preset strategies (e.g., three consecutive "serious anomalies" occurring), it can trigger an audible and visual alarm or automatically lock the weighbridge, suspending weighing operations until manual confirmation.

[0044] In one example, reference Figure 2 , Figure 2 This is a schematic diagram illustrating the evaluation of weighing sensor data according to an embodiment of this application. (Based on rated load) The graph, with the x-axis representing the sensor's output readings and the y-axis representing its input and output characteristics during the loading and return processes—the key data curves—is displayed. The difference between the loading and return curves on the y-axis represents the hysteresis error. Below the loading and return curves are the correlated data curves, and the dashed line parallel to the rated load represents the zero-point output curve. The difference between the loading and correlated data curves represents the nonlinearity error. The difference between the return curve and the correlated data curves represents the combined error. This graph provides a visual assessment of the sensor's output linearity, return consistency, and zero-point stability under different loads.

[0045] In this embodiment, by first generating correlation data that is related to the current weighing sensor output data, then extracting key data, and finally performing health assessments or anti-cheating warnings based on both, proactive sensing and early warning of sensor status can be achieved. Compared with the traditional method that relies solely on a single threshold judgment, this approach integrates data correlation analysis and key feature extraction, significantly improving the accuracy of sensor fault diagnosis and the sensitivity of anti-cheating detection, effectively preventing measurement distortion caused by sensor damage or the addition of cheating devices.

[0046] Reference Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the weighing sensor data evaluation method of this application. Based on the first embodiment described above, a second embodiment of the weighing sensor data evaluation method of this application is proposed.

[0047] In the second embodiment, step S20 includes: Step S201: Obtain the output data sequence of the current weighing sensor.

[0048] It should be noted that the output data sequence refers to a time-ordered collection of data acquired and stored from one or more weighing sensors within a continuous time window or event window. The output data sequence can be acquired based on fixed time intervals (e.g., sampling every 100 milliseconds) or event triggers (e.g., each time a vehicle is fully on the scale, or after the weight stabilizes). The sequence length can be dynamically configured according to the application scenario; for example, in a health monitoring scenario, the most recent 5 minutes of continuous data can be used, while in an anti-cheating scenario, the data from the most recent complete vehicle passage can be used. Optionally, the output data sequence may also include sensor status flags (e.g., temperature, attitude angle, shell breakage indicator) as verification data.

[0049] In this embodiment, the weighing sensor is a digital weighing sensor, such as a column-type sensor. Its output data is uploaded to the instrument or anti-cheating platform in the form of data frames via a CAN bus or interface. Each data frame includes at least the sensor ID, load code value (AD value), temperature value, attitude angle value, shell integrity flag, check code, and timestamp. Multiple consecutive frames arranged chronologically constitute the output data sequence.

[0050] Specifically, as an optional implementation, when the truck scale is in idle standby mode (no vehicle is on the scale), the system acquires the output data sequence at a low sampling frequency, for example, acquiring data from all sensors once per second. This low-frequency sequence is mainly used to monitor the zero-point drift, temperature changes, and attitude stability of the sensors. When a vehicle is detected to be entering the scale (triggered by an infrared grating or weight threshold), the system automatically switches to a high-frequency sampling mode, continuously acquiring the output data sequence at a frequency of 50 times per second until the vehicle is completely off the scale, thus fully recording the sensor response waveforms of the entire process from the vehicle entering, stopping, weighing, to leaving the scale.

[0051] In another optional implementation, the output data sequence is acquired using vehicle passage events as a window. Whenever a vehicle completes weighing (i.e., the weight stabilizes and the vehicle has left), the system encapsulates all sensor data collected during this passage into a data sequence and assigns a unique event ID. The sequence begins at the moment the vehicle's front axle touches the leading edge of the load-bearing platform (determined by a sudden change in load from the first pair of sensors) and ends at the moment the vehicle's rear axle leaves the trailing edge of the load-bearing platform (determined by the load returning to zero from the last pair of sensors).

[0052] Optionally, the output data sequence can also be a stored sequence read from a historical database or local cache. For example, when offline analysis of sensor data from a historical vehicle passage is required, the system queries and loads the corresponding output data sequence from the anti-fraud platform's database based on the user-specified time period or event ID. The format of this sequence is completely consistent with the real-time acquired sequence, including fields such as timestamp, sensor ID, and payload value. Offline sequences are typically used for post-event auditing, fault backtracking, or model training. When reading historical sequences, the system also loads the corresponding tag information (such as whether an alarm occurred, manual review results, etc.) to facilitate supervised model optimization.

[0053] Step S202: Perform feature recognition on the output data sequence to obtain recognition results. The recognition results are at least one of data change rate, data consistency deviation, and data time validity value.

[0054] Understandably, if there is a significant change between the current output data and subsequent data, it indicates that the sensor may have been subjected to a sudden shock, strain gauge breakage, or abrupt signal injection from an external cheating device. This abrupt change is localized and drastic in time, providing crucial clues for anti-cheating and fault diagnosis. Therefore, in this embodiment, the output data sequence can be characterized from the dimension of data change rate. The subsequent data is acquired after the current data, and the acquisition time can be a timestamp in the sensor's output data frame.

[0055] It should be understood that if the current output data is significantly inconsistent with its previous data or data from other sensors in the same group, it indicates that the sensor may have experienced zero-point drift, aging, damage, or local interference (such as a wheel pressing against the edge, or a sensor being tampered with by a cheating device). This inconsistency reflects a deviation in the sensor's behavior relative to its historical state or the group's state. Therefore, in this embodiment, feature identification can also be performed on the output data sequence from the dimension of data consistency deviation. Specifically, the previous data was collected before the current data, and the data in the same group comes from the same time window output of other sensors on the same truck scale.

[0056] Understandably, if the sensor corresponding to the current output data has been in service for a long time or has been out of service for too long since the last calibration, the reliability of the sensor data may be reduced, and its reference value for health assessment will also decrease accordingly. Data from newer sensors or those that have just been calibrated carries more weight in decision-making. Therefore, in this embodiment, feature identification can also be performed on the output data sequence from the dimension of the data time validity value. The time validity value is calculated based on the sensor's start-up time and the last calibration time; the higher the value, the more reliable the data.

[0057] In the specific implementation process, the transient jump trend of the sensor output can be detected in real time by acquiring subsequent data after the current output data and comparing the rate of change of the values. This time series-based differential analysis method is extremely sensitive to sudden load changes, instantaneous sensor failure, or external shock interference, and can identify abnormal fluctuations within milliseconds, providing a rapid response basis for the anti-cheating system. For example, when the vehicle is not fully on the scale or the sensor output is momentarily tampered with, the rate of change of the data will deviate significantly from the normal range, and the system can immediately trigger an alarm, effectively curbing cheating behavior. At this time, the identification result is the rate of change of data, and step S202 may include: Acquire subsequent output data collected after the current output data; compare the value of the current output data with the value of the subsequent output data to obtain the data change rate of the current output data.

[0058] It should be noted that the rate of change of data is used to measure the degree of fluctuation in the output value of the weighing sensor over a short period of time. Subsequent output data refers to one or more sampling points on the timeline that are later than the current output data, and their acquisition time is determined by the timestamp in the sensor's output data frame or the local clock at the time of system reception.

[0059] In this embodiment, the number of subsequent output data points can be dynamically set according to the sampling frequency and application scenario. At high-frequency sampling (e.g., 50Hz), 1 to 5 sampling points immediately following the current data can be taken; at low-frequency sampling (e.g., 1Hz), only one subsequent sampling point can be taken. The calculation method for the data change rate includes: absolute change rate. ,in This is the current output data value. For subsequent output data values, This represents the sampling interval. For the load output, a relative rate of change can be used because the absolute rate of change varies significantly between sensors with different ranges.

[0060] Specifically, as an optional implementation, when the truck scale is in high-frequency sampling mode (50 times per second, i.e., Δt = 0.02 seconds), the relative rate of change is calculated. If the rate of change far exceeds the normal fluctuation range caused by uneven road surfaces during normal vehicle operation (typically <10% / second), the system immediately determines that there is a drastic jump at that moment, which may be due to sensor impact or high-frequency interference signals injected by a cheating device. This data point will be assigned a high rate of change score and thus selected as key data.

[0061] In another alternative implementation, to suppress misjudgments caused by single-point noise, instead of comparing adjacent single points, the mean of the N consecutive sampling points following the current output data can be used as the subsequent data value. For example, if N=3, the average of the next three points is calculated, and then the rate of change is calculated. After this processing, for isolated glitch noise, the mean of the next three points will not deviate significantly due to a single glitch, and the rate of change will not increase abnormally; however, for true step jumps, the mean will also deviate significantly, and the rate of change remains sensitive.

[0062] Alternatively, the data change rate can also employ a dynamic threshold comparison method based on historical data. For example, the data change rate distribution of the sensor under normal operating conditions (no impact, no cheating, and smooth vehicle driving) can be learned in advance, and its mean can be calculated. and standard deviation For the current data point, calculate the rate of change between it and subsequent data points. ,like If the change exceeds three times the normal range (i.e., more than three times the standard deviation), it is considered an abnormal change. This method can adapt to the baseline of the rate of change for different sensors and different load ranges, avoiding the fixed threshold being too sensitive under light load (because the same absolute change under light load will lead to an extremely high relative rate of change) or not sensitive enough under heavy load.

[0063] In the specific implementation process, by comparing the current output data with previous historical data and calculating the data consistency deviation, cumulative anomalies such as long-term sensor drift, zero-point offset, or periodic interference can be effectively identified. Utilizing the temporal continuity and spatial symmetry that sensors should possess under normal operating conditions, when an unexplained deviation occurs between the output of a sensor and previous data, it can be determined that it may be due to aging, damage, or external interference. This is particularly suitable for multi-sensor parallel systems in truck scales, enabling rapid location of faulty sensors and reducing maintenance and troubleshooting time. The identification result at this point is the data consistency deviation. Step S202 may include: Acquire the preceding output data whose acquisition time is before the current output data; compare the value of the current output data with the value of the preceding output data to obtain the data consistency deviation of the current output data.

[0064] It should be noted that data consistency deviation measures the degree of deviation between the current load cell output data and its own historical data or data from other sensors in the same group. It is a key feature for identifying sensor zero drift, aging, damage, or local anomalies (such as off-center loading or partial tampering by a cheating device). Previous output data refers to one or more sampling points on the timeline that are earlier than the current output data. The acquisition time is determined by the timestamp in the sensor output data frame or the local clock at the time of system reception.

[0065] In this embodiment, data consistency deviation is divided into two types: self-consistency deviation and mutual consistency deviation. Self-consistency deviation refers to the degree of deviation between the current output data and the sensor's previous output data, used to detect the sensor's temporal stability, such as gradual anomalies like zero-point drift and creep accumulation. Mutual consistency deviation refers to the degree of deviation between the current output data and the output data of other sensors on the same truck scale within the same time window, used to detect whether the force ratio between sensors is normal, for example, if a sensor outputs abnormally due to damage or cheating. The two types of deviation can be used individually or weighted and combined into a single comprehensive consistency deviation index.

[0066] Specifically, as an optional implementation, when the identification result shows a self-consistency deviation, the system acquires M consecutive preceding output data points (M is typically 10~100, depending on the sampling frequency and the required statistical stability) whose acquisition time precedes the current output data. The mean of these M preceding data points is then calculated. and standard deviation Current output data The formula for calculating the self-consistency deviation relative to historical data is: Self-deviation When the self-deviation exceeds a preset threshold, it is considered a self-consistency deviation. For example, if a sensor's self-deviation far exceeds the threshold in an unloaded state, it indicates that the sensor may have experienced sudden zero-point drift or been subjected to external interference. This prompts the need to check whether the sensor has been touched, whether there are foreign objects obstructing it, or whether it has been injected with a cheating signal.

[0067] In another optional implementation, when the identification result shows a mutual consistency deviation, the system acquires the output data of the other K sensors in the same group within the same time window (i.e., the same sampling time or the same vehicle passage event). Assume the truck scale has 8 sensors, numbered 1 to 8. For the current output of sensor i... Calculate the average output of the remaining 7 sensors. and standard deviation The formula for calculating mutual consistency deviation is: Mutual Deviation Furthermore, theoretical proportional deviations can be calculated based on mechanical models: for sensor pairs in symmetrical positions (such as the left front end and the right front end), the theoretical output ratio should be 1:1, and the degree of deviation of the actual ratio can also be used as a mutual consistency deviation. When the system determines that there is a serious inconsistency deviation in the sensor, it may be due to sensor damage, signal wire breakage, or the vehicle only partially pressing on the load-bearing structure (such as one wheel being suspended in the air). The data points are marked as critical data, triggering a "severely uneven sensor force" alarm.

[0068] Optionally, data consistency deviation can be further extended by incorporating temperature data. The zero-point output of a load cell is affected by temperature; typically, a temperature sensor is integrated within the sensor for temperature compensation. The system acquires the temperature value corresponding to the current output data. It calculates the temperature consistency deviation by taking the output values ​​near the same temperature point in the preceding data, and is suitable for environments with large temperature variations, such as ports.

[0069] Furthermore, the system can also fuse self-consistency deviations and mutual consistency deviations to obtain a comprehensive consistency deviation score. The fusion method can be weighted summation. ,in and This represents the weighting coefficient. The higher the overall deviation score, the worse the consistency between the data point and its historical system or group system, and the more likely it is abnormal data caused by malfunction or cheating.

[0070] In the specific implementation process, the time difference between the current output data acquisition time and the start working time or the last calibration time can be calculated, and the time validity value of the data can be determined accordingly, which can quantify the weight of the timeliness of sensor data. For sensors that have not been calibrated for a long time or have exceeded their service life, their data confidence naturally decreases, and the system can reduce the decision weight of the data or prompt for mandatory calibration accordingly. This mechanism helps to establish a data quality control system for the entire life cycle of sensors, prevents misjudgment of health status due to sensor aging, and also provides a quantitative basis for preventive maintenance. At this time, the identification result is the data time validity value, and step S202 may include: Obtain a first time difference between the acquisition time of the current output data and the current start time of the weighing sensor; and / or, obtain a second time difference between the acquisition time of the current output data and the last calibration time; determine the effective time value of the current output data based on the first time difference and / or the second time difference.

[0071] It should be noted that the effective time value is used to quantify the timeliness weight of the current weighing sensor output data, reflecting the reliability of the data in the time dimension. As the service life of the sensor increases, performance degradation phenomena such as elastomer fatigue, strain gauge creep, and seal aging will gradually intensify, and the reliability of the output data will show a downward trend. At the same time, the longer the time since the last calibration, the more significant the cumulative error of the sensor due to environmental changes, load impacts, etc. The first time difference reflects the service life of the sensor, and the second time difference reflects the time since the last calibration. Both can be used alone or in combination.

[0072] In this embodiment, the sensor's initial operating time refers to the moment when the sensor is first powered on and completes its initial calibration, which is recorded in the device file of the monitoring platform. The acquisition time of the current output data is provided by the timestamp in the data frame. First time difference. This is equal to the difference between the current acquisition time and the start time. The sensor's design life is typically 10 years, but its actual performance declines with service life. Generally, an exponential decay function or a piecewise linear function can be used to mitigate this. Time effective value components mapped to the [0,1] interval For example, the first three years are a stable period. The decay period is 1 year; it lasts from 3 to 8 years. It decreased linearly to 0.7; after 8 years, the decay accelerated. The index fell below 0.5.

[0073] In another alternative implementation, the valid time value can be determined based on a second time difference (the time elapsed since the last calibration). The system records the date of the sensor's most recent calibration and calculates the difference between the current acquisition time and the last calibration time. (Unit: days). Valid time value. The calculation method is within the valid calibration period. , It decreases linearly from 1 to 0.8; after exceeding the calibration period. , This accelerates the degradation. The above solution can effectively remind maintenance personnel to calibrate in a timely manner, avoiding measurement inaccuracies caused by exceeding the service life.

[0074] Optionally, the effective time value can be determined simultaneously based on the first and second time differences, and then fused using a weighted product or weighted summation method. Furthermore, other time-related factors can be incorporated into the effective time value calculation, such as the time of the sensor's most recent maintenance, the time of the most recent strain gauge replacement, and the time of the most recent seal inspection. These extended factors can be added to the effective time value calculation as product correction terms.

[0075] Step S203: Determine the importance score of the output data based on the recognition result, and select key data from the output data sequence according to the importance score.

[0076] It should be noted that the importance score is used to evaluate the value of each data point in the output data sequence for sensor health status diagnosis and anti-cheating warning. A higher importance score indicates that the data point is more likely to contain fault characteristics, cheating traces, or performance degradation information. The importance score is determined based on the identification results obtained in the preceding steps, namely at least one of data change rate, data consistency deviation, and valid data time value. The importance score for each data point is obtained by normalizing and weighting the identification results from multiple dimensions. Then, based on the importance score, the most representative data points are selected from the output data sequence as key data for subsequent evaluation steps. Optionally, the importance score can be calculated using linear weighting, product fusion, fuzzy logic, or machine learning models.

[0077] Specifically, firstly, the data change rate and data consistency deviation (including self-consistency deviation and mutual consistency deviation) are normalized and mapped to the [0,1] interval. The higher the score, the more significant the abnormality. The effective value of the data time is already in the [0,1] interval, so no additional normalization is required. Then, weight coefficients are assigned to each indicator according to different application scenarios. Finally, the weighted summation or product is used to obtain the comprehensive importance score.

[0078] Furthermore, after determining the importance scores, the system selects key data from the output data sequence according to a preset strategy. As an optional implementation, a fixed-quantity selection method is used: the number of key data points to be extracted each time is set to K (e.g., K=5 or K=10), all data points in the data sequence are sorted from highest to lowest importance score, and the top K points are selected as key data. This method is suitable for scenarios with stable data volume and where the computational load of subsequent evaluation needs to be controlled. For example, if a single vehicle passage generates 650 frames of data, with 5200 data points from 8 sensors per frame, and K=20 is chosen, then only the 20 points with the highest importance scores are selected and sent to the evaluation module, significantly reducing the computational burden.

[0079] In another optional implementation, a dynamic threshold selection method can be used. A lower threshold for the importance score is set, and all data points with importance scores greater than this threshold are selected as key data. When sensor data is generally normal, very few or no key data points may be selected; however, when an abnormal event occurs, multiple consecutive abnormal points will be selected to facilitate subsequent analysis of the abnormal waveform. For example, if a sensor suddenly experiences a step jump, and the rate of change and consistency deviation of 10 consecutive data points are both very high, with importance scores exceeding 0.7, then all 10 points are selected, and the evaluation module can extract complete features such as the jump amplitude and duration from them.

[0080] Optionally, when the important score rises continuously to a peak and then begins to decline, data points near the peak can be selected as key data to identify transient impact events. The system maintains a sliding window (e.g., containing 10 data points) and calculates the local maximum value of the important score within the window. When a local maximum value is detected and exceeds a threshold, the peak point and two points before and after it are selected as key data, thus fully capturing the rising edge, peak value, and falling edge of the event. For example, when a sensor suffers a transient impact, the important score reaches a peak of 0.95 at the moment of impact, and five points are selected (two before and two after). The evaluation module can analyze the impact waveform characteristics (maximum impact value, impact duration, etc.) to help determine whether it is a cheat injection or a real physical impact.

[0081] In this embodiment, feature recognition is performed on the output data sequence of the symmetric sensors to obtain multi-dimensional recognition results such as data change rate, consistency deviation, and time validity value. Key data is then filtered based on importance scores, compressing massive amounts of raw data into a small number of high-value features and reducing the computational complexity of subsequent analysis. This multi-indicator comprehensive evaluation avoids the risk of misjudgment by a single indicator. For example, the data change rate can capture abrupt changes, consistency deviation can detect abnormal forces between sensors, and time validity value can eliminate outdated data, thereby improving the robustness and representativeness of key data selection and providing reliable input for accurate health assessment.

[0082] Reference Figure 4, Figure 4 This is a flowchart illustrating the third embodiment of the weighing sensor data evaluation method of this application. Based on the above embodiments, the third embodiment of the weighing sensor data evaluation method of this application is proposed.

[0083] In the third embodiment, step S10 includes: Step S101: Obtain the target historical data waveform corresponding to the current weighing sensor.

[0084] It should be noted that the target historical data waveform refers to the output data waveform collected by a sensor of the same model, range, and installation method as the current weighing sensor during its historical normal operation. It is typically stored in time series format. The historical data waveform serves as the basis for generating subsequent correlated data (such as predicted waveforms or standard reference templates). The purpose of obtaining the target historical data waveform is to extract statistical regularities or characteristic patterns from the historical normal operation data for comparison and evaluation with the current real-time output data. Optionally, the target historical data waveform can come from the current sensor's own historical records, from other sensors of the same model and operating conditions, or from factory standard test waveforms.

[0085] In this embodiment, the target historical data waveform can be the historical record of the current sensor itself, that is, the output data collected by the sensor under normal operating conditions over a period of time. This type of waveform best reflects the individual characteristics of the sensor (such as installation posture, base zero point, temperature response curve), and is suitable for long-term stability monitoring and personalized fault diagnosis. It can also be the record of similar sensors under the same or similar operating conditions, that is, the historical data of multiple sensors with the same model, application scenario (such as port vehicle scale digital sensors), and similar environmental conditions (such as temperature range, humidity range) selected from the regulatory platform database, which is suitable for determining whether the current sensor has deviated from normal behavior. Alternatively, it can be the factory standard test waveform, that is, the ideal output waveform recorded when the sensor completes calibration and performance testing at the manufacturer, which can include sensitivity curves, creep curves, temperature compensation curves, etc., under different load points, suitable for assessing whether the sensor has deviated from its factory performance due to aging or damage.

[0086] Step S102: Determine the probability of occurrence of each feature pattern in the target historical data waveform.

[0087] It should be noted that a feature pattern refers to a signal segment in the target's historical data waveform that has a specific shape, statistical characteristics, or physical meaning, such as a typical load change waveform, creep curve, temperature zero-point drift trajectory, or impact response waveform. Each feature pattern can be viewed as a template for the sensor's output behavior under specific operating conditions. The occurrence probability refers to the proportion of a certain feature pattern in all valid data segments in the sensor's historical operating records, or the conditional probability of observing that pattern under given operating conditions. Determining the occurrence probability of a feature pattern is used to quantify the prevalence of different behavioral patterns, thereby generating the most probable predicted waveform or standard reference template in subsequent steps based on the probability. Optionally, the occurrence probability can be calculated using frequency statistics, kernel density estimation, hidden Markov models, or deep learning classifiers.

[0088] In the specific implementation process, by segmenting the target historical data waveform, labeling features, expanding similar working conditions, and calculating probabilities, a more refined and generalized feature pattern library can be constructed. Improving the adaptability of associated data generation, enabling the system to maintain stable early warning performance under different seasons, load ranges, and port operation conditions, is the core supporting technology for realizing intelligent health management of sensors. At this point, step S102 may include: The target historical data waveform is segmented to obtain data segments; the data segments are labeled with features, and target feature segments are selected from the data segments based on the labeling results; the target feature segments are extended with operating conditions to obtain similar operating condition features corresponding to the target feature segments; the occurrence probability of each feature pattern in the target historical data waveform is determined based on the target feature segments and the similar operating condition features.

[0089] It should be noted that segmentation processing divides continuous historical target data waveforms into multiple independent data segments according to certain rules. Each segment corresponds to a complete event or stable operating condition, such as a complete vehicle passage process, a period of no-load stability, or a temperature change cycle. Data segments are the basic units for subsequent feature annotation and pattern statistics. Feature annotation assigns one or more labels to each data segment, describing the physical meaning or behavioral characteristics of the segment, such as "smooth vehicle passage," "rapid weighing," "zero-point drift," and "temperature compensation response." Annotation results can be manually annotated based on expert knowledge or automatically annotated based on rules or classification models. Target feature segments refer to segments selected from all segments that possess representative or high-value features. Segments with high confidence levels, good signal-to-noise ratios, and the ability to typically reflect a certain pattern can be selected. Optionally, target feature segments may also include rare segments that, although occurring infrequently, have significant diagnostic value.

[0090] In this embodiment, the segmented processing method can be based on event detection, fixed time windows, or state change detection. For truck scale applications, the most commonly used segmentation method is based on vehicle passage events: by monitoring changes in the total load of the sensors, when the load rises from below a threshold to above a threshold and then stabilizes before falling back below the threshold, this process is considered a complete segment. This segment begins when the front axle of the vehicle is placed on the scale and ends when the rear axle is completely removed from the scale, encompassing the output waveforms of the entire process of loading, stabilizing the weighing, and unloading.

[0091] Optionally, the purpose of operating condition extension is to overcome the problem of sparse samples under certain operating conditions in historical data. By using similarity metrics, features of sparse regions are extended from the operating condition regions with abundant samples, making the probability estimation of feature patterns more robust. The basic assumption of operating condition extension is that under similar operating conditions (such as load difference not exceeding 10%, temperature difference not exceeding 5℃, and vehicle speed difference not exceeding 20%), the feature patterns of sensor output waveforms are similar.

[0092] Specifically, firstly, for each target feature segment, its operating condition parameter vector is extracted, including average load, load change rate, temperature, attitude angle, vehicle speed, etc. Then, an operating condition similarity metric function is defined, such as Euclidean distance or Mahalanobis distance. For the current target feature segment, all other segments in historical data whose operating condition parameters have a similarity to this segment less than a threshold (including ordinary segments not selected as target feature segments) are searched, and the feature patterns of these segments are appended to the current target feature segment as similar operating condition features.

[0093] It should be understood that determining the probability of occurrence of each feature pattern in the target historical data waveform based on the target feature segment and the similar working condition feature can be based on the proportion of the segments corresponding to the target feature segment and the similar working condition feature in the data segmentation to determine the probability of occurrence of each feature pattern.

[0094] Step S103: Generate associated data that is related to the current weighing sensor output data based on the occurrence probability. The associated data includes a predicted waveform or a standard reference template.

[0095] It should be noted that the correlated data mainly includes predicted waveforms and standard reference templates. The predicted waveform, given the current operating conditions (such as current load, temperature, and timestamp), is the most likely output change trajectory of the sensor over a future period, predicted based on historical probability models. The standard reference template is a static baseline waveform or eigenvalue vector (such as load-output mapping curves, creep curves, and temperature compensation curves) representing the sensor's "normal behavior" under specific operating conditions. The closer the current real-time output data is to the correlated data, the healthier the sensor; the greater the deviation, the more likely there is a malfunction or cheating.

[0096] In this embodiment, the specific process for generating associated data depends on the application scenario and data type. When it is necessary to predict the complete output waveform of the sensor during the current vehicle passage in real time, a predicted waveform generation method is adopted. The system first acquires the partially observed output data (e.g., the vehicle has been weighed but has not yet stabilized), combines it with the current operating parameters (load change rate, temperature, vehicle axle type, etc.), matches the most probable feature pattern from the historical probability model, and then generates the predicted waveform from the current moment to the end of the vehicle passage based on the typical waveform template of the pattern. When a static benchmark is needed for offline comparison or calibration verification, a standard reference template generation method is adopted. According to the given operating conditions (e.g., fixed load point, fixed temperature), the system extracts the template waveform corresponding to the feature pattern with the highest probability of occurrence under the given operating conditions from the historical probability model, or extracts the weighted average waveform of multiple patterns as the standard reference template.

[0097] Specifically, as an optional implementation, assume that a vehicle is currently passing through the weighbridge, and the system has collected the output data (partial waveform) for the first 3 seconds after the vehicle's front axle is on the scale. The system extracts the characteristics of the current operating condition: load increases from 0 to 15t, rate of change 5t / s, temperature 28℃. According to the historical probability model, under the current operating condition, the conditional probabilities of each characteristic mode are: smooth passage 65%, rapid weighing 20%, significant creep 10%, and overshoot 5%. The system selects the historical waveform template corresponding to the "smooth passage" mode with the highest probability. Then, the first 3 seconds of the template's waveform are time-aligned and amplitude-matched with the collected 3 seconds of actual waveform. If the two match well (correlation coefficient > 0.9), the last 5 seconds of the template are used as the predicted waveform to generate a sequence of predicted sensor output values ​​for the next 5 seconds starting from the current moment.

[0098] Optionally, the standard reference template does not need to match some observed waveforms in real time, but instead directly outputs a static reference waveform or numerical characteristic for specific operating conditions. For example, a standard reference template used for daily zero-point calibration: under no-load and 20°C conditions, the system extracts the zero-point pattern with the highest probability of occurrence under this condition from the historical probability model. Based on the most typical response waveform under this condition in historical data, the system generates the theoretical output ratio of each sensor. The standard reference template can also be presented in the form of a function, such as a load-output fitting curve: based on the load-output relationship under all "smooth vehicle passage" modes in historical data, a linear or polynomial curve is fitted using the least squares method, serving as the standard characteristic curve of the sensor. When the real-time output deviates from this curve beyond the allowable range, it is judged as abnormal.

[0099] In this embodiment, by acquiring historical sensor data waveforms and analyzing the probability of occurrence of each characteristic pattern, a predicted waveform or standard reference template associated with the current output is generated, achieving an upgrade from "passive monitoring" to "proactive prediction." When the current output waveform matches a high-probability associated template, the system is considered healthy; if there is a significant deviation, an early warning is issued. This approach can uncover the regular patterns of sensors under different loads, temperatures, and time conditions, significantly improving the ability to identify early, minor faults (such as strain gauge fatigue and seal aging) and reducing the risk of sudden failures.

[0100] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the weighing sensor data evaluation method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0101] This application also provides a weighing sensor data evaluation device; please refer to... Figure 5 The weighing sensor data evaluation device includes: The generation module 10 is used to generate associated data that is related to the current weighing sensor output data; Selection module 20 is used to select key data from the output data of the current weighing sensor; Evaluation module 30 is used to evaluate the data of the weighing sensor based on the associated data and the key data.

[0102] The weighing sensor data evaluation device provided in this application, employing the weighing sensor data evaluation method in the above embodiments, can solve the technical problem that existing weighing sensor anti-cheating measures are relatively passive and unable to actively identify and warn of cheating behavior during its occurrence. Compared with the prior art, the beneficial effects of the weighing sensor data evaluation device provided in this application are the same as those of the weighing sensor data evaluation method provided in the above embodiments, and other technical features in the weighing sensor data evaluation device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0103] This application provides a weighing sensor data evaluation 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 weighing sensor data evaluation method in Embodiment 1 above.

[0104] The following is for reference. Figure 6The diagram illustrates a structural schematic suitable for implementing the weighing sensor data evaluation device in the embodiments of this application. The weighing sensor data evaluation 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), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The weighing sensor data evaluation device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0105] like Figure 6 As shown, the weighing sensor data evaluation 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 weighing sensor data evaluation 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 load cell data evaluation device to communicate wirelessly or wiredly with other devices to exchange data. While the figure shows load cell data evaluation devices 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.

[0106] 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.

[0107] The weighing sensor data evaluation device provided in this application, employing the weighing sensor data evaluation method in the above embodiments, can solve the technical problem that existing weighing sensor anti-cheating measures are relatively passive and unable to actively identify and warn of cheating behavior during its occurrence. Compared with the prior art, the beneficial effects of the weighing sensor data evaluation device provided in this application are the same as those of the weighing sensor data evaluation method provided in the above embodiments, and other technical features of this weighing sensor data evaluation device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0108] 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.

[0109] 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.

[0110] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the weighing sensor data evaluation method in the above embodiments.

[0111] 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.

[0112] The aforementioned computer-readable storage medium may be included in the weighing sensor data evaluation device; or it may exist independently and not assembled into the weighing sensor data evaluation device.

[0113] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the weighing sensor data evaluation device, cause the weighing sensor data evaluation device to perform the weighing sensor data evaluation method described above.

[0114] 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).

[0115] 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.

[0116] 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.

[0117] 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 weighing sensor data evaluation method. This solves the technical problem that existing weighing sensor anti-cheating measures are relatively passive and cannot actively identify and warn of cheating during the cheating process. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the weighing sensor data evaluation method provided in the above embodiments, and will not be repeated here.

[0118] The above description is only a part of the embodiments of this application and does not limit the scope of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included within the protection scope of this application.

[0119] It should be noted that the data collection, tag management, rule setting, and push decision-making processes involved in this application are designed to work with other technical features to solve technical problems. They do not involve or support any illegal activities. Any data processing that may violate laws and regulations (such as unauthorized collection of privacy data, generation of discriminatory tags, setting unfair rules, or pushing illegal information) is not within the scope of protection of this application's technical solution. Of course, the user data in this application will be encrypted, anonymized, or de-identified before storage to ensure user data security.

Claims

1. A method for evaluating weighing sensor data, characterized in that, The evaluation of the weighing sensor data includes: Generate associated data that is related to the current weighing sensor output data; Select key data from the output data of the current weighing sensor; The data of the weighing sensor is evaluated based on the associated data and the key data.

2. The weighing sensor data evaluation method as described in claim 1, characterized in that, The step of selecting key data from the output data of the current weighing sensor includes: Obtain the output data sequence of the current weighing sensor; The output data sequence is subjected to feature recognition to obtain recognition results, wherein the recognition results are at least one of data change rate, data consistency deviation, and data time validity value. Based on the recognition results, the importance score of the output data is determined, and key data is selected from the output data sequence according to the importance score.

3. The weighing sensor data evaluation method as described in claim 2, characterized in that, The identification result is the data change rate. The step of performing feature identification on the output data sequence to obtain the identification result includes: Acquire subsequent output data acquired after the current output data. The value of the current output data is compared with the value of the subsequent output data to obtain the data change rate of the current output data.

4. The weighing sensor data evaluation method as described in claim 2, characterized in that, The identification result is a data consistency deviation. The step of performing feature identification on the output data sequence to obtain the identification result includes: Acquire the preceding output data whose acquisition time was before the current output data; The value of the current output data is compared with the value of the previous output data to obtain the data consistency deviation of the current output data.

5. The weighing sensor data evaluation method as described in claim 2, characterized in that, The identification result is a valid data time value. The step of performing feature identification on the output data sequence to obtain the identification result includes: Obtain the first time difference between the acquisition time of the current output data and the start time of the current weighing sensor; And / or, obtain a second time difference between the acquisition time of the current output data and the previous calibration time; The effective time value of the current output data is determined based on the first time difference and / or the second time difference.

6. The weighing sensor data evaluation method according to any one of claims 1 to 5, characterized in that, The generation of associated data related to the current weighing sensor output data includes: Obtain the target historical data waveform corresponding to the current weighing sensor; Determine the probability of occurrence of each feature pattern in the target historical data waveform; Based on the probability of occurrence, associated data is generated that is related to the current weighing sensor output data. The associated data includes a predicted waveform or a standard reference template.

7. The weighing sensor data evaluation method as described in claim 6, characterized in that, Determining the probability of occurrence of each feature pattern in the target historical data waveform includes: The target historical data waveform is segmented to obtain data segments; The data segments are labeled with features, and target feature fragments are selected from the data segments based on the labeling results; The target feature segment is extended by working condition to obtain similar working condition features corresponding to the target feature segment; The probability of occurrence of each feature pattern in the target historical data waveform is determined based on the target feature fragment and the similar working condition features.

8. A weighing sensor data evaluation device, characterized in that, The device includes: The generation module is used to generate associated data that is related to the current weighing sensor output data; The selection module is used to select key data from the output data of the current weighing sensor; An evaluation module is used to evaluate the data of the weighing sensor based on the associated data and the key data.

9. A weighing sensor data evaluation device, 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 weighing sensor data evaluation method as described in any one of claims 1 to 7.

10. 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 weighing sensor data evaluation method as described in any one of claims 1 to 7.

Images