Intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overweight detection equipment

Abstract

The application discloses a vehicle-mounted overload detection equipment intelligent operation and maintenance and dynamic calibration management system, and relates to the technical field of vehicle-mounted weighing detection equipment operation and maintenance and calibration. The system adopts a four-layer architecture of a perception layer, a transmission layer, an algorithm layer and an application layer, the perception layer collects multi-source data such as weight and temperature, the transmission layer realizes efficient data transmission, the algorithm layer integrates various algorithms such as trend prediction, fault diagnosis and dynamic calibration, and the application layer completes operation and maintenance scheduling and calibration control. The system realizes fine management of the whole life cycle of equipment by quantifying the health index of equipment, performing Bayesian fault diagnosis, self-adaptive recursive least square dynamic calibration and operation and maintenance priority sorting, solves the problems of low precision and slow response of traditional operation and maintenance, improves detection precision and operation and maintenance efficiency, and is suitable for various types of vehicle-mounted overload detection equipment of motor vehicles and has high practicability.

Description

Technical Field

[0001] This invention relates to the field of operation, maintenance and calibration technology for vehicle-mounted weighing and testing equipment, and specifically to an intelligent operation, maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment. Background Technology

[0002] Vehicle-mounted overload detection equipment, such as load cells and data acquisition terminals, operates under complex conditions including vibration, high and low temperatures, and dust. Prolonged operation under these conditions can lead to problems such as sensor drift, circuit aging, and data transmission loss / delay, resulting in decreased weighing accuracy and frequent equipment failures. Furthermore, existing maintenance methods cannot provide real-time monitoring of equipment status. Current technologies rely heavily on manual periodic inspections and static calibration with fixed cycles, which suffers from slow response times, insufficient calibration accuracy, and low maintenance efficiency. Therefore, there is an urgent need for an integrated system combining intelligent maintenance and dynamic calibration, using algorithms to achieve real-time monitoring of equipment status, accurate fault diagnosis, and dynamic adaptive calibration. Summary of the Invention

[0003] To address the aforementioned problems, this invention introduces an intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment, comprising:

[0004] The perception layer is used to collect the operating parameters and detection data of the vehicle-mounted overload detection equipment. The operating parameters include ambient temperature, vibration amplitude and power supply voltage, and the detection data includes weighing data. The transmission layer is used to transmit the collected data to the cloud server in real time.

[0005] The algorithm layer assesses the health status of equipment through an equipment status assessment algorithm, identifies the type of fault through a fault diagnosis algorithm, calibrates the coefficients through a dynamic calibration algorithm, and obtains the operation and maintenance priority results through an operation and maintenance priority ranking algorithm, providing quantitative decision support for the system's operation and maintenance scheduling and dynamic calibration.

[0006] The application layer is used to implement functions such as device status visualization, fault alarm, and calibration command issuance.

[0007] Preferably, the sensing layer includes: a weighing sensor for collecting weighing data; a temperature sensor for collecting ambient temperature; a vibration sensor for collecting vibration amplitude; and a voltage monitoring module for collecting power supply voltage.

[0008] Preferably, the transmission layer includes a 5G communication module and a BeiDou positioning communication module;

[0009] The algorithm layer also includes:

[0010] The detection data trend prediction algorithm is based on a linear regression model to perform time series analysis on historical weighing detection data, fit the data change trend and predict future detection data, and triggers an operation and maintenance warning when the predicted value exceeds a preset deviation threshold.

[0011] A multi-source data consistency verification algorithm is provided, which is based on linear fitting error analysis to perform consistency verification on relevant data collected by the perception layer and identify and mark abnormal data.

[0012] Preferably, the equipment condition assessment algorithm is used to calculate the equipment's health index. :

[0013] Select , , and The normalization process is performed using the following formula:

[0014] ( ),

[0015] in, The stability index of weighing data is the coefficient of variation of the most recent N weighing data, which is the ratio of the standard deviation to the mean. N is a positive integer and is preset by the system. This is a sensor drift index, representing the absolute value of the relative deviation between the current detected value and the standard value. This is an environmental adaptability index, representing a comprehensive coefficient of the influence of temperature and vibration on measurement accuracy; This is an energy consumption indicator for the equipment, representing the ratio of actual power consumption to rated power consumption. For the first Minimum threshold for class indicators For the first The maximum threshold of the class indicator, after normalization ;

[0016] The weights of the indicators are determined using the Analytic Hierarchy Process (AHP). The weights are allocated as follows:

[0017] And satisfy ;

[0018] according to:

[0019] Calculate the health index ;

[0020] The health index Standards for classifying health levels:

[0021] This is the normal state;

[0022] This is a mildly abnormal condition;

[0023] This is a moderately abnormal condition.

[0024] This is a severely abnormal condition.

[0025] Preferably, the fault diagnosis algorithm is used in the health index When the value is less than the third preset threshold, the fault type is located:

[0026] Define the root node of the Bayesian network as the cause of failure, including sensor drift. Circuit fault Data transmission failure Power supply failure ;

[0027] The leaf nodes of a Bayesian network are defined as fault characteristics, including: The weighing deviation is greater than 5%, i.e., |detected value - standard value| / standard value > 5%; Data transmission packet loss rate greater than 10%; The power supply voltage fluctuation exceeds ±10%, i.e., |actual voltage - rated voltage| / rated voltage > 10%; The vibration amplitude is greater than 0.5g; Temperatures outside the normal operating range of -20°C to 60°C.

[0028] Prior probabilities are obtained based on historical failure data statistics:

[0029] Set the conditional probability of each fault characteristic under the corresponding fault cause. Where j = 1, 2, 3, 4, 5; when at least one fault characteristic is detected At that time, according to the formula:

[0030]

[0031] Calculate the posterior probability of each cause of failure, where The cause of failure with the highest posterior probability; As the output of the equipment's fault diagnosis results, corresponding fault handling suggestions are generated simultaneously.

[0032] Preferably, the dynamic calibration algorithm is used in health index When the detection deviation is between the first preset threshold and the second preset threshold, or between the second preset threshold and the third preset threshold, the correction steps include:

[0033] Establish a weighing and testing calibration model that takes into account the influence of ambient temperature:

[0034] ,

[0035] in; for The actual weight input by the time sensor; for The detected weight output by the time sensor. This is the zero-bias calibration coefficient, used to correct zero-point drift; This is the sensitivity calibration coefficient, used to correct for gain error; This is the temperature compensation coefficient, used to correct for the effects of temperature. for Ambient temperature at all times; For random error and ;

[0036] Define parameter vector

[0037] Input vector ,

[0038] Rewrite the calibration model as follows

[0039] ;

[0040] Parameter estimation is performed using adaptive recursive least squares, and the recursive formula includes:

[0041] 1. Coefficient estimation update: ;

[0042] 2. Gain matrix: ;

[0043] 3. Covariance matrix update: ;

[0044] in, for Time calibration coefficient estimate; This is the gain matrix; To estimate the error covariance matrix, the initial values ​​of the estimated error covariance matrix are... , It is a 3x3 identity matrix and a parameter vector. Dimensional matching, The initial value is a large positive number to ensure fast convergence; Forgetting factor and , used to control the weight of historical data;

[0045] Calculate the calibrated weighing detection value according to the real-time estimated calibration coefficient :

[0046] ,

[0047] where is the real-time estimated value of the zero-bias calibration coefficient, which is updated in real time through an adaptive recursive least squares algorithm; is the real-time estimated value of the sensitivity calibration coefficient, which is updated in real time through an adaptive recursive least squares algorithm; is the real-time estimated value of the temperature compensation coefficient, which is updated in real time through an adaptive recursive least squares algorithm;

[0048] When it is determined that the calibration is qualified, is the weight of the standard weight.

[0049] Preferably, the operation and maintenance priority sorting algorithm is used to determine the allocation order of operation and maintenance resources when there are multiple abnormal devices. The steps include:

[0050] Define three objective functions:

[0051] Urgency objective , the larger the value, the more urgent the device status;

[0052] Influence range objective , where is the average daily transportation volume of the service vehicle of the th device, is the total average daily transportation volume of the service vehicles of all devices;

[0053] Operation and maintenance cost objective is the estimated cost for the operation and maintenance of a single abnormal device, in thousands of yuan, calibrated by the system according to historical operation and maintenance data. The smaller the value, the higher the priority;

[0054] Perform normalization processing on each objective function in the [0,1] interval to eliminate the dimension difference, and obtain ;

[0055] Assign weights to the objective functions: ,

[0056] Calculate the priority index: , and determine the operation and maintenance priority in ascending order of the priority index .

[0057] Preferably, the detection data trend prediction algorithm is based on linear regression. The implementation process includes:

[0058] Establish a linear regression model:

[0059] ,

[0060] in This is a time series, with units in hours. The detection data corresponds to the time point. For the intercept term, The slope For the random error term, satisfying , ;

[0061] Model parameters are estimated using the least squares method. and This makes the sum of squared residuals The minimum parameter estimation formula is:

[0062] ;

[0063] ;

[0064] in, , Based on the model obtained through fitting Predict detection data at the next m time points Set the prediction deviation threshold for weighing test data. If the predicted data satisfies , If this is detected, the equipment is deemed to have a potential risk of detection deviation, triggering an operation and maintenance warning.

[0065] Preferably, the multi-source data consistency verification algorithm is based on linear fitting error analysis, and its specific implementation process includes:

[0066] A linear correlation model is established by selecting two types of related data: ,in and For the relevant detection data collected by the perception layer, For constant terms, The linear correlation coefficient, For fitting residuals;

[0067] The least squares method was used to collect historical synchronous data. Group data Perform fitting to obtain parameter estimates. The fitting model is ;

[0068] Calculate the fit bias for each set of historical data:

[0069] , These are the fitted values;

[0070] Calculate the standard deviation of the deviation Set a data consistency verification threshold , Confidence coefficient;

[0071] For real-time collected synchronous data Calculate its fitting deviation ,like If the data is consistent, then the data is considered to be consistent; if If the data is inconsistent, a data anomaly alarm will be triggered.

[0072] Preferably, the workflow of the intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment includes the following steps:

[0073] S1: The sensing layer collects equipment operating parameters and detection data at a preset frequency and uploads them to the cloud server through the transmission layer;

[0074] S2: The algorithm layer calls the device status assessment algorithm to calculate the health index. Determine the device status;

[0075] S3: If The system remains in standby monitoring mode; if or The dynamic calibration algorithm is activated to correct the detection deviation; if The fault diagnosis algorithm is activated to locate the fault type and generate an operation and maintenance plan;

[0076] S4: For equipment requiring maintenance or calibration, allocate maintenance resources using a maintenance priority ranking algorithm, and execute dynamic calibration operations or fault maintenance handling operations according to maintenance priority. After the operation is completed, re-collect equipment data and calculate the health index. After verifying that the device status has returned to normal, update the device's full lifecycle status file.

[0077] Beneficial effects

[0078] Enhance the intelligence level of equipment operation and maintenance and calibration: Through algorithm integration, realize real-time perception of equipment status, accurate fault diagnosis and dynamic calibration, replace the traditional manual inspection or static calibration mode, reduce manual intervention and reduce operation and maintenance costs.

[0079] Ensuring stable weighing and detection accuracy: To address issues such as sensor drift and environmental interference under complex working conditions, environmental adaptability indicators are optimized and dynamic calibration models are improved, resulting in a post-calibration detection error of ≤0.25%, which is 58.3% higher than the accuracy of traditional static calibration, ensuring accurate and reliable overweight detection data.

[0080] Achieve refined management of the entire equipment lifecycle: Through quantitative assessment of health indices and prioritization of maintenance, enable precise scheduling and closed-loop handling of abnormal equipment, thereby extending equipment lifespan. Attached Figure Description

[0081] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0082] Figure 1 This is an overall architecture diagram of the intelligent operation and maintenance and dynamic calibration management system for the vehicle-mounted overload detection equipment described in this invention. Detailed Implementation

[0083] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0084] like Figure 1 As shown, this invention introduces an intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment, including a perception layer, a transmission layer, an algorithm layer, and an application layer. Deep collaboration is achieved through a closed-loop logic of data flow-processing-decision-feedback, with each layer closely connected and its responsibilities seamlessly integrated, as detailed below:

[0085] Sensing Layer and Transmission Layer: The sensing layer consists of a resistance strain gauge load cell, a PT100 temperature sensor, a MEMS vibration sensor, and a voltage monitoring module. Each sensor / module synchronously collects data at a preset frequency to acquire equipment operating parameters, including ambient temperature, vibration amplitude, and power supply voltage, as well as detection data, including weighing data. The sensing layer is the data source of the system. The equipment operating parameters it collects, such as temperature, vibration, and voltage, and detection data, such as weighing data, are uploaded to the cloud server in real time through the 5G combined with Beidou dual-mode communication module of the transmission layer, providing raw input for subsequent data processing. The communication stability of the transmission layer directly determines the timeliness and completeness of the sensing data and serves as a data bridge connecting the sensing layer and the upper-layer modules.

[0086] Transport layer and algorithm layer: The transport layer pushes the initially packaged sensing data to the algorithm layer. The algorithm layer uses this data to perform various calculations such as data consistency verification, health index assessment, and trend prediction. If data loss or delay occurs during transmission, the algorithm layer will identify the anomaly through a multi-source data consistency verification algorithm and provide feedback on possible communication faults in the transport layer to assist in optimizing transmission quality.

[0087] Algorithm Layer and Application Layer: The algorithm layer outputs processing results, such as device health status, fault type, calibration instructions, and maintenance priorities, to the application layer. The application layer visualizes the results, such as status dashboards and early warning prompts, through the maintenance management platform and user interface. It also issues operation instructions, such as starting dynamic calibration and scheduling maintenance resources, based on the algorithm results. At the same time, the application layer receives user operation feedback, such as calibration parameter adjustments and maintenance task confirmations, and synchronizes it back to the algorithm layer to optimize algorithm parameters, such as calibration coefficients and weight allocation, and improve algorithm adaptability.

[0088] The entire process is a closed-loop system: the perception layer collects data, the transmission layer transmits data, the algorithm layer analyzes and makes decisions, and the application layer executes and provides feedback, forming a complete closed loop. For example, if the perception layer collects abnormal weighing data, it is uploaded by the transmission layer. The algorithm layer identifies data anomalies through consistency checks and determines equipment malfunctions through status assessment. The application layer issues an alert and triggers operation and maintenance scheduling. After operation and maintenance are completed, the data is collected again by the perception layer and circulated through each layer to verify the equipment's recovery status, achieving intelligent management throughout its entire lifecycle.

[0089] In the sensing layer: the weighing sensor is a resistance strain gauge type with a range of 0~200t and an accuracy of ±0.2%; the temperature sensor is a PT100 type with a measurement range of -50℃~100℃; the vibration sensor is a MEMS type with a range of 0~2g; and the voltage monitoring module has a measurement range of 10V~15V.

[0090] The transmission layer uses a combination of Huawei 5G modules and Beidou positioning and communication modules; the cloud server is Alibaba Cloud ECS, configured with 8 cores, 16GB of memory and 1TB of storage; the operation and maintenance management platform in the application layer is developed based on Java Spring Boot, and the vehicle calibration control terminal uses an embedded Linux system.

[0091] The equipment status assessment algorithm is based on weighted multi-index fusion, used to quantitatively assess the health status of the testing equipment and output a health index ( This provides a basis for triggering operations and maintenance; the specific implementation process includes:

[0092] Selecting stability indexes for weighing data Sensor drift index Environmental adaptability indicators and equipment energy consumption indicators Normalization is performed using the following formula:

[0093] ( ),

[0094] in, For the first Minimum threshold for class indicators For the first The maximum threshold of the class indicator, after normalization ;

[0095] The weights of the indicators are determined using the Analytic Hierarchy Process (AHP). The weight allocation is as follows And satisfy ;

[0096] Calculate the health index The calculation formula is: and according to Values ​​are used to classify health levels: For normal For mild abnormalities, For moderate abnormality, This indicates a severe abnormality.

[0097] The fault diagnosis algorithm is based on an improved Bayesian network and accurately locates fault types such as sensor drift, circuit faults, data transmission faults, and power supply faults for severely malfunctioning equipment. The specific implementation process includes:

[0098] Define a Bayesian network node: root node These correspond to four types of fault causes: sensor drift, circuit failure, data transmission failure, and power supply failure; leaf nodes ,in For weighing deviation >5%, For data transmission packet loss rate >10%, For power supply voltage fluctuations > ±10%, For vibration amplitude > 0.5g, The temperature is out of range (-20℃~60℃).

[0099] Prior probabilities are obtained based on historical failure data statistics:

[0100] ;

[0101] Set conditional probability , indicating a fault Characteristics of occurrence The probability of occurrence;

[0102] Calculating the posterior probability of a fault based on Bayes' theorem:

[0103] ,

[0104] in Take the one with the highest posterior probability. As a diagnostic result.

[0105] The dynamic calibration algorithm is based on adaptive recursive least squares. For mildly and moderately abnormal devices, it can correct the sensor output deviation in real time to ensure the detection accuracy. The specific implementation process includes:

[0106] Establish a weighing detection calibration model considering the influence of environmental temperature: ,

[0107] where is the actual weight input by the sensor at time is the detected weight output by the sensor at time is the calibration coefficient, is the environmental temperature at time is a random error and ;

[0108] Define the parameter vector , the input vector , and rewrite the calibration model as ; Use adaptive recursive least squares for parameter estimation. The recursive formulas include:

[0109] 1. Coefficient estimation update: ;

[0110] 2. Gain matrix: ;

[0111] 3. Covariance matrix update: ;

[0112] where is the estimated value of the calibration coefficient at time is the estimated error covariance matrix, and the initial value of the estimated error covariance matrix , is the 3rd-order identity matrix, which matches the dimension of the parameter vector , is the forgetting factor;

[0113] The weighed detection value after calibration :

[0114] ,

[0115] When it is determined that the calibration is qualified, is the weight of the standard weight.

[0116] The operation and maintenance priority sorting algorithm is based on multi-objective optimization. When there are multiple abnormal devices, operation and maintenance resources are allocated according to the priority. The specific implementation process includes:

[0117] Define three objective functions:

[0118] Urgency level target A larger value indicates a more urgent device status.

[0119] Scope of impact target ,in For the first The average daily transport volume of the equipment service vehicles. The average daily transport volume of vehicles serving all equipment;

[0120] Operation and maintenance cost target The estimated cost for maintenance and repair of a single malfunctioning device is expressed in thousands of yuan. It is determined by the system based on historical maintenance data, and the smaller the value, the higher the priority.

[0121] Normalize each objective function in the interval [0,1] to eliminate dimensional differences, and obtain... ;

[0122] Assign weights to the objective function: Calculate the priority index: ,

[0123] The larger the value, the higher the maintenance priority.

[0124] The detection data trend prediction algorithm is based on linear regression. It analyzes the linear variation pattern of historical detection data, predicts the trend of equipment detection data in the future, identifies potential detection deviation risks in advance, and provides forward-looking support for operation and maintenance decisions. The specific implementation process includes:

[0125] Establish a linear regression model: ,in This is a time series, with units in hours. The detection data corresponds to the time point. For the intercept term, The slope For the random error term, satisfying , The least squares method is used to estimate the model parameters. and This makes the sum of squared residuals The minimum parameter estimation formula is:

[0126] ;

[0127] ;

[0128] in, , Based on the model obtained through fitting Predict detection data at the next m time points Set a prediction deviation threshold Δy for the weighing test data. If the prediction data meets the following conditions... , If this is detected, the equipment is deemed to have a potential risk of detection deviation, triggering an operation and maintenance warning.

[0129] The multi-source data consistency verification algorithm is based on linear fitting error analysis. It verifies the consistency of data collected from multiple sensors in the sensing layer, such as multiple weighing sensors of the same device and different types of auxiliary sensors, identifying abnormal data and ensuring data reliability. This provides high-quality data input for subsequent algorithms. The specific implementation process includes:

[0130] A linear correlation model is established by selecting two types of related data: ,in and For the relevant detection data collected by the perception layer, For constant terms, The linear correlation coefficient, For fitting residuals;

[0131] The least squares method was used to analyze n sets of historically synchronized data. Perform fitting to obtain parameter estimates. The fitting model is ;

[0132] Calculate the fit bias for each set of historical data: ;

[0133] Calculate the standard deviation of the deviation Set a data consistency verification threshold , Confidence coefficient;

[0134] For real-time collected synchronous data Calculate its fitting deviation ,like If the data is consistent, then the data is considered to be consistent; if If the data is inconsistent, a data anomaly alarm will be triggered.

[0135] The system's workflow is characterized by the following steps:

[0136] S1: The sensing layer collects equipment operating parameters and detection data at a preset frequency and uploads them to the cloud server through the transmission layer;

[0137] S2: The algorithm layer calls the device status assessment algorithm to calculate the health index. Determine the device status;

[0138] S3: If The system remains in standby monitoring mode; if or The dynamic calibration algorithm is activated to correct the detection deviation; if The fault diagnosis algorithm is activated to locate the fault type and generate an operation and maintenance plan;

[0139] S4: For equipment requiring maintenance or calibration, allocate maintenance resources using a maintenance priority ranking algorithm, and execute dynamic calibration operations or fault maintenance handling operations according to maintenance priority. After the operation is completed, re-collect equipment data and calculate the health index. After verifying that the device status has returned to normal, update the device's full lifecycle status file.

[0140] Implementation environment description:

[0141] This embodiment is for a heavy-duty truck on-board overload detection device, model: ZCS-2000. The application scenario is a highway freight logistics fleet consisting of 50 trucks, with one detection device for each truck. The ambient temperature range is -15℃ to 45℃, the vibration amplitude is 0.1g to 0.8g, and the power supply voltage is 12V±1V.

[0142] System hardware configuration:

[0143] Sensing layer: resistance strain gauge load cell, range: 0~200t, accuracy ±0.2%; PT100 temperature sensor, measurement range: -50℃~100℃; MEMS vibration sensor, range: 0~2g; voltage monitoring module, measurement range: 10V~15V.

[0144] Transport layer: Huawei 5G module + Beidou positioning and communication module;

[0145] Cloud server: Alibaba Cloud ECS;

[0146] Application layer: Operation and maintenance management platform, developed based on Java Spring Boot; vehicle calibration control terminal, embedded Linux system.

[0147] Detailed explanation of system operation process

[0148] 1. Data Acquisition Phase

[0149] The sensing layer device collects data at a frequency of 1Hz. An example of the collected data is shown below:

[0150] The last 10 weighing data are: 120.5t, 120.7t, 120.3t, 120.6t, 120.4t, 120.8t, 120.5t, 120.6t, 120.7t, and 120.4t.

[0151] Ambient temperature ;

[0152] Vibration amplitude: 0.35g;

[0153] Power supply voltage: 12.3V;

[0154] Data transmission packet loss rate: 2%.

[0155] 2. Equipment condition assessment

[0156] (1) Calculation and normalization of indicators

[0157] X1, the stability index of weighing data, is the coefficient of variation of the most recent 100 weighing data points.

[0158] Mean = 120.6t, Standard deviation = 0.21t, X1 = 0.21 / 120.6 ≈ 0.00174;

[0159] Sensor drift index X2: Standard weight 150t, current detection value 150.8t.

[0160] X2=|150.8-150| / 150≈0.00533;

[0161] Environmental adaptability index X3: The influence coefficient of temperature / vibration on accuracy was obtained through experimental fitting, X3=0.08;

[0162] Equipment energy consumption index X4: Rated power consumption 10W, actual power consumption 10.5W, X4=10.5 / 10=1.05.

[0163] Normalization processing;

[0164] Threshold values: ; ; ; ;

[0165]

[0166]

[0167]

[0168]

[0169] (2) Health Index Calculation

[0170]

[0171] According to the health level classification, 0.7 < 0.802185 is judged as a mild abnormality, triggering the dynamic calibration process.

[0172] 3. Implementation of dynamic calibration

[0173] (1) Calibration model parameter estimation

[0174] Known: Standard weight Detect weight ,temperature Initial values ​​of parameter vector The initial values ​​for the zero-bias calibration coefficient are 0, the initial values ​​for the sensitivity calibration coefficient are 1, and the initial values ​​for the temperature compensation coefficient are 0. The initial covariance matrix is... Forgetting factor .

[0175] First recursive calculation:

[0176] Input vector

[0177] ;

[0178] Prediction error:

[0179] ;

[0180] Gain matrix:

[0181] ;

[0182] Coefficient update:

[0183] ;

[0184] Covariance matrix: .

[0185] After 10 adaptive recursive least squares iterations, the calibration coefficients converged to a stable value:

[0186] .

[0187] (2) Verification of calibration results

[0188] Detected values ​​after calibration:

[0189]

[0190] Standard weight ,error The calibration was successful.

[0191] 4. Example of Fault Diagnosis and Maintenance Priority

[0192] Fault diagnosis scenarios

[0193] Data collected by a truck inspection device: Weighing deviation 12%. Power supply voltage fluctuation of 15% Calculate the health index The problem was identified as a severe anomaly, and fault diagnosis was initiated.

[0194] Calculating posterior probabilities based on Bayesian networks:

[0195]

[0196] ;

[0197] Posterior probability The fault type was determined to be a power supply fault.

[0198] Implementation effect verification

[0199] This embodiment verifies the system's effectiveness through a three-month trial run using 50 trucks:

[0200] 1. Fault Response Time: The average system response time after equipment malfunction is 3.2 minutes, which is 98.6% shorter than the 4-hour fault response time of traditional manual inspection.

[0201] 2. Calibration accuracy: After dynamic calibration, the weighing detection error of the equipment is ≤0.25%, which is 58.3% higher than the 0.6% detection accuracy of traditional fixed-period static calibration;

[0202] 3. Equipment failure rate: reduced from 12% before trial operation to 4.2%, a decrease of 65%.

[0203] 4. Data anomaly detection accuracy: The newly added multi-source data consistency verification algorithm improves the data anomaly detection accuracy to 99.2%;

[0204] 5. Accuracy of potential risk warning: The newly added detection data trend prediction algorithm has improved the accuracy of potential risk warning to 92%, effectively reducing the occurrence of sudden failures.

[0205] The above technical solutions only embody the preferred technical solutions of the present invention. Any modifications that may be made by those skilled in the art to certain parts thereof embody the principles of the present invention and fall within the protection scope of the present invention.

Claims

1. An intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment, characterized in that, include: The sensing layer is used to collect the operating parameters and detection data of the vehicle-mounted overload detection equipment. The operating parameters include ambient temperature, vibration amplitude and power supply voltage, and the detection data includes weighing data. The transport layer is used to transmit the collected data to the cloud server in real time; The algorithm layer assesses the health status of equipment through an equipment status assessment algorithm, identifies the type of fault through a fault diagnosis algorithm, calibrates the coefficients through a dynamic calibration algorithm, and obtains the operation and maintenance priority results through an operation and maintenance priority ranking algorithm, providing quantitative decision support for the system's operation and maintenance scheduling and dynamic calibration. The application layer is used to implement functions such as device status visualization, fault alarm, and calibration command issuance; The transmission layer includes a 5G communication module and a BeiDou positioning communication module; The algorithm layer also includes: a detection data trend prediction algorithm, which performs time series analysis on historical weighing detection data based on a linear regression model, fits the data change trend and predicts future detection data, and triggers an operation and maintenance warning when the predicted value exceeds a preset deviation threshold. The multi-source data consistency verification algorithm is based on linear fitting error analysis. It performs consistency verification on relevant data collected by multiple sensors in the sensing layer, such as multiple weighing sensors of the same device and different types of auxiliary sensors, to identify abnormal data, ensure data reliability, and provide high-quality data input for subsequent algorithms. The dynamic calibration algorithm is used to adjust health index. When the detection deviation is between the first preset threshold and the second preset threshold, or between the second preset threshold and the third preset threshold, the correction steps include: Establish a weighing and testing calibration model that takes into account the influence of ambient temperature: ， in; for The actual weight input by the time sensor; for The detected weight output by the time sensor. This is the zero-bias calibration coefficient, used to correct zero-point drift; This is the sensitivity calibration coefficient, used to correct for gain error; This is the temperature compensation coefficient, used to correct for the effects of temperature. for Ambient temperature at all times; For random error and ; Define parameter vector : Input vector , Rewrite the calibration model as follows ； Parameter estimation is performed using adaptive recursive least squares, and the recursive formula includes: Coefficient estimation update: ; Gain matrix: ; Covariance matrix update: ; in, for Time calibration coefficient estimate; This is the gain matrix; To estimate the error covariance matrix, the initial values ​​of the estimated error covariance matrix are... , It is a 3x3 identity matrix and a parameter vector. Dimensional matching, The initial value is a large positive number to ensure fast convergence; Forgetting factor and , used to control the weight of historical data; Calculate the calibrated weighing value based on the real-time estimated calibration coefficient. : ， in, The real-time estimated value of the zero-bias calibration coefficient is obtained by updating it in real time using an adaptive recursive least squares algorithm; The real-time estimated value of the sensitivity calibration coefficient is obtained by updating it in real time through an adaptive recursive least squares algorithm; The real-time estimated value of the temperature compensation coefficient is obtained by updating it in real time through an adaptive recursive least squares algorithm; When it is determined that the calibration is qualified, is the weight of the standard weight.

2. The intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment according to claim 1, characterized in that, The sensing layer includes: a weighing sensor for collecting weighing data; a temperature sensor for collecting ambient temperature; a vibration sensor for collecting vibration amplitude; and a voltage monitoring module for collecting power supply voltage.

3. The intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment according to claim 1, characterized in that, The equipment status assessment algorithm is used to calculate the equipment's health index. : Select , , and The normalization process is performed using the following formula: （ ）， in, The stability index of weighing data is the coefficient of variation of the most recent N weighing data, which is the ratio of the standard deviation to the mean. N is a positive integer and is preset by the system. This is a sensor drift index, representing the absolute value of the relative deviation between the current detected value and the standard value. This is an environmental adaptability index, representing a comprehensive coefficient of the influence of temperature and vibration on measurement accuracy; This is an energy consumption indicator for the equipment, representing the ratio of actual power consumption to rated power consumption. For the first Minimum threshold for class indicators For the first The maximum threshold of the class indicator, after normalization ; Use AHP (Analogous Hierarchical Analysis) to determine the weights of the indicators. The weights are allocated as follows: And satisfy ; according to: Calculate the health index ; The health index Standards for classifying health levels: This is the normal state; This is a mildly abnormal condition; This is a moderately abnormal state. This is a severely abnormal condition.

4. The intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment according to claim 1, characterized in that, The fault diagnosis algorithm is used in the health index When the value is less than the third preset threshold, the fault type is located: Define the root node of the Bayesian network as the cause of failure, including sensor drift. Circuit fault Data transmission failure Power supply failure ; The leaf nodes of a Bayesian network are defined as fault characteristics, including: The weighing deviation is greater than 5%, i.e., |detected value - standard value| / standard value > 5%; Data transmission packet loss rate greater than 10%; The power supply voltage fluctuation exceeds ±10%, i.e., |actual voltage - rated voltage| / rated voltage > 10%; The vibration amplitude is greater than 0.5g; Temperatures outside the normal operating range of -20℃ to 60℃; Prior probabilities are obtained based on historical failure data statistics: Set the conditional probability of each fault characteristic under the corresponding fault cause. Where j = 1, 2, 3, 4, 5; the cause of failure with the highest posterior probability. When at least one fault characteristic is detected At that time, according to the formula: ； Calculate the posterior probability of each fault cause, where The cause of failure with the highest posterior probability; As the output of the equipment's fault diagnosis results, corresponding fault handling suggestions are generated simultaneously.

5. The intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment according to claim 1, characterized in that, The operation and maintenance priority ranking algorithm is used to determine the allocation order of operation and maintenance resources when there are multiple abnormal devices. The steps include: Define three objective functions: Urgency level target A larger value indicates a more urgent device status. Scope of impact ,in For the first The average daily transport volume of the equipment service vehicles. The average daily transport volume of vehicles serving all equipment; Operation and maintenance cost target The estimated cost for maintenance and repair of a single malfunctioning device is expressed in thousands of yuan. It is determined by the system based on historical maintenance data, and the smaller the value, the higher the priority. Normalize each objective function in the interval [0,1] to eliminate dimensional differences, and obtain... ; Assign weights to the objective function: , Calculate the priority index: According to priority index Determine maintenance priorities in ascending order of size.

6. The intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment according to claim 1, characterized in that, The multi-source data consistency verification algorithm is based on linear fitting error analysis, and its specific implementation process includes: A linear correlation model is established by selecting two types of related data: ,in and For the relevant detection data collected by the perception layer, For constant terms, The linear correlation coefficient, For fitting residuals; The least squares method was used to collect historical synchronous data. Group data Perform fitting to obtain parameter estimates. The fitting model is ; Calculate the fit bias for each set of historical data: , These are the fitted values; Calculate the standard deviation of the deviation Set a data consistency verification threshold , Confidence coefficient; For real-time collected synchronous data Calculate its fitting deviation ,like If the data is consistent, then the data is considered to be consistent; if If the data is inconsistent, a data anomaly alarm will be triggered.

7. The intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment according to claim 1, characterized in that, The workflow of the intelligent operation and maintenance and dynamic calibration management system for vehicle-mounted overload detection equipment includes the following steps: S1: The sensing layer collects equipment operating parameters and detection data at a preset frequency and uploads them to the cloud server through the transmission layer; S2: The algorithm layer calls the device status assessment algorithm to calculate the health index. Determine the device status; S3: If The system remains in standby monitoring mode; if or The dynamic calibration algorithm is activated to correct the detection deviation; if The fault diagnosis algorithm is activated to locate the fault type and generate an operation and maintenance plan; S4: For equipment requiring maintenance or calibration, allocate maintenance resources using a maintenance priority ranking algorithm, and execute dynamic calibration operations or fault maintenance handling operations according to maintenance priority. After the operation is completed, re-collect equipment data and calculate the health index. After verifying that the device status has returned to normal, update the device's full lifecycle status file.

Images