Vehicle type identification method based on shaftless identification device

Abstract

The invention discloses a vehicle type recognition method based on a shaftless recognition device, and belongs to the technical field of vehicle type recognition, and the method comprises the following steps: S1, collecting on-site vehicle passing waveform data of different vehicle types, and carrying out the preprocessing of the data through a big data analysis technology; s2, analyzing features and laws of waveforms of all vehicle types, extracting key features by adopting a machine learning algorithm, and establishing a vehicle type recognition model; s3, weighing shafts are identified through the weighing channel data, and shaft number statistics and vehicle type discrimination are carried out in combination with shaft spacing and shaft width parameters; s4, identifying weighing shafts by using the weighing channel data, and judging the number of the shafts staying on the weighing platform by combining the shaft distance and the shaft width parameters; and S5, calculating the weight of the vehicle, querying the average axle weight of the corresponding vehicle model through the vehicle model database in combination with the vehicle model recognition result and the axle number of the vehicle, and further calculating the total weight of the vehicle. The method has the characteristics of high recognition accuracy, low transformation cost, low maintenance cost and the like.

Description

Technical Field

[0001] The invention belongs to the technical field of vehicle type identification, and in particular relates to a vehicle type identification method based on an axleless identification device. Background Art

[0002] In current weighing systems, in order to ensure the accuracy of vehicle axle count and vehicle model identification, system designers usually use an upper axle identifier to identify the upper axle of the vehicle. The function of this identifier is to accurately detect the number of axles of the vehicle, thereby providing necessary data support for subsequent vehicle model identification. At the same time, the system also needs to rely on a lower axle identifier to identify the lower axle resting on the scale. In this way, the system can accurately determine the number of axles of the vehicle and then calculate the weight of the vehicle. This process is crucial to ensure the accuracy and reliability of the weighing results.

[0003] However, the entire weighing system is highly dependent on the upper and lower axle identifiers, which means that the performance and stability of the axle identifier will directly affect the operating efficiency and accuracy of the entire system. However, due to the limitations of its design and manufacturing, the axle identifier often has some inherent defects, such as possible jamming, failure to rebound normally, and easy damage to the sensor. The existence of these problems not only increases the maintenance cost of the system, but also may cause a series of problems in actual use, such as vehicle model identification errors and abnormal vehicle gross weight display. These problems not only affect the accuracy of the weighing data, but may also have a negative impact on the vehicle's traffic efficiency and may even cause safety accidents. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a vehicle type identification method based on an axleless identification device, which has the advantage of being able to perform vehicle type identification without axle identification, and solves the problems of the prior art.

[0005] The present invention is implemented as follows: a vehicle type identification method based on an axleless identification device comprises the following steps: Step S1, collecting on-site vehicle waveform data of different vehicle models, and pre-processing the data using big data analysis technology; Step S2, analyzing the characteristics and rules of the waveforms of each vehicle model, using machine learning algorithms to extract key features, and establishing a vehicle model recognition model; Step S3, identifying the upper scale axle through the upper scale channel data, and performing axle number statistics and vehicle model discrimination in combination with the axle spacing and axle width parameters; Step S4, using the lower scale channel data to identify the lower scale axle, and combining the axle spacing and axle width parameters to determine the number of axles remaining on the scale platform; Step S5, calculate the vehicle weight, combine the vehicle model identification result and the number of vehicle axles, query the average axle weight of the corresponding vehicle model through the vehicle model database, and then calculate the total vehicle weight.

[0006] As preferred embodiment of the present invention, step S1 includes the following steps: S1.1 Set data collection points: Set sensors on the vehicle's path to ensure that waveform data of different vehicle models can be collected; S1.2 Real-time data acquisition: Start the data acquisition system to record the waveform data of passing vehicles in real time; S1.3 Data storage: Store the collected waveform data in the database for subsequent processing and analysis; S1.4 Data preprocessing: Filter the stored data to remove high-frequency noise and interference signals; S1.5 Data denoising: further denoise the filtered data to ensure data purity; S1.6 Data quality check: Ensure the quality of processed data through data verification and outlier detection.

[0007] As preferred embodiment of the present invention, step S2 includes the following steps: S2.1 Feature extraction: Analyze the characteristics of waveform data of different vehicle models; S2.2 Sample preparation: Select representative waveform data from the database as training samples and test samples; S2.3 Machine learning algorithm selection: Select a machine learning algorithm based on the vehicle model recognition requirements; S2.4 Model training: Use training samples to train the machine learning algorithm, extract key features and establish a vehicle model recognition model; S2.5 Model verification: Use test samples to verify the trained vehicle model recognition model and evaluate the accuracy and stability of the model.

[0008] As preferred embodiment of the present invention, step S3 includes the following steps: S3.1 Weighing axle identification: Analyze the weighing channel data and identify the axle where the vehicle starts to weigh; S3.2 Axis parameter measurement: measure the axis spacing and axis width parameters of the upper scale axis; S3.3 Axle count: Count the axle count of the vehicle based on the identification of the axle on the scale and the measurement of axle parameters; S3.4 Preliminary vehicle model identification: Based on the number of axles and axle parameters, a preliminary judgment is made on the possible vehicle model.

[0009] As preferred embodiment of the present invention, step S4 includes the following steps: S4.1 Identification of the weighing axle: Analyze the weighing channel data and identify the axle where the vehicle starts to weigh; S4.2 Reconfirm the axis parameters: Confirm the axis spacing and axis width of the lower scale axis and compare them with the data when the upper scale was used; S4.3 Determination of the number of axes staying on the scale: Determine the number of axes staying on the scale platform based on the identification of the lower scale axis and the confirmation of the axis parameters.

[0010] As preferred embodiment of the present invention, step S5 comprises the following steps: S5.1 Preliminary calculation of vehicle weight: Based on the number of axles resting on the scale and the scale data, a preliminary calculation of the partial weight of the vehicle is made; S5.2 Vehicle model database query: Combine the vehicle model recognition result and the number of vehicle axles to query the average axle weight of the corresponding vehicle model in the vehicle model database; S5.3 Gross vehicle weight calculation: Calculate the gross vehicle weight based on the preliminary calculated vehicle partial weight and the average axle weight in the vehicle model database; S5.4 Result output: Output the vehicle type identification result and the total vehicle weight to the user or related system.

[0011] As a preferred embodiment of the present invention, step S6 is also included to construct a real-time dynamic vehicle model identification and weight estimation system: Combined with steps S1 and S2, the vehicle waveform data is collected and processed in real time, and the vehicle type is quickly identified using a machine learning model. Then, through step S5, the system combines the vehicle type identification results and axle number statistics, queries the average axle weight in the vehicle type database, and dynamically estimates the total vehicle weight.

[0012] As a preferred embodiment of the present invention, step S7 is also included to optimize the classification and weighing of axle number adaptive vehicle models: On the basis of steps S3 and S4, the vehicle classification standard is automatically adjusted dynamically according to the change in the number of axles when the vehicle is on and off the scale. Combined with the weight calculation in step S5, the weighing process is optimized to improve the accuracy and efficiency of weighing.

[0013] As a preferred embodiment of the present invention, step S8 is also included, and the anti-cheating system for jointly verifying waveform characteristics and the number of axes is constructed, which includes the following steps: Step S81: Analyze waveform data and extract characteristic values; Step S82: combining the vehicle type recognition result, analyzing whether the waveform feature matches the vehicle type; Step S83: monitor the change of the number of axes to ensure that the waveform data is consistent with the axis number statistics; Step S84: Develop an anomaly detection algorithm to identify situations where waveform features do not match axis number information; Step S85: triggering the anti-cheating mechanism.

[0014] As a preferred embodiment of the present invention, if step 4 is damaged, the following settings are performed: In step S3, the positions of all axes are identified through the scale channel data, and the position information of each axis is recorded; According to the vehicle type recognition model established in step S1 and step S2, combined with the position information of the axis obtained in step S3, the vehicle type recognition is performed on each axis; Compare the identified vehicle model information with the average axle weight of the corresponding vehicle model to calculate the weight of each axle; Add the weights of all axles to get the total weight of the vehicle; The calculation result of the total vehicle weight can be further calibrated by querying the average axle weight of the corresponding vehicle model in the vehicle model database and combining it with the weight of each axle calculated in step 3.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This method collects a large number of on-site vehicle waveform data of different vehicle models, analyzes the waveform characteristics and rules of each vehicle model, and proposes an innovative axle identification method. It does not rely on a separate axle identification device, but only uses the weighing sensor data of the scale platform itself to accurately identify the vehicle model and weight calculation. The upper and lower axles are identified respectively through the data of the upper and lower scale channels, realizing axle count, vehicle model identification and vehicle weight calculation. This method has the characteristics of high recognition accuracy, low transformation cost, and low maintenance cost, and meets the requirements of on-site use.

[0016] Improved the accuracy of vehicle model recognition. By analyzing a large amount of on-site vehicle waveform data and extracting key features using machine learning algorithms, accurate vehicle model recognition is achieved. Reduced transformation costs. No separate axle recognition device is required. Vehicle model recognition and weight calculation can be completed only through the weighing sensor data of the scale platform itself. Reduced maintenance costs. Since it does not rely on additional axle recognition devices, maintenance work is simpler and less costly. Enhanced the adaptability of the system to meet the use requirements of different sites and improved the flexibility and reliability of the weighing system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of a vehicle type identification method based on an axleless identification device provided by an embodiment of the present invention; Figure 2 is a waveform diagram provided by an embodiment of the present invention; Figure 3 is a flow chart of step S1 provided in an embodiment of the present invention; Figure 4 is a schematic flow chart of step S2 provided in an embodiment of the present invention; Figure 5 is a flow chart of step S3 provided in an embodiment of the present invention; Figure 6 is a schematic flow chart of step S4 provided in an embodiment of the present invention; Figure 7 It is a flowchart of step S5 provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to further understand the content, features and effects of the present invention, the following embodiments are given as examples and described in detail with reference to the accompanying drawings.

[0019] The structure of the present invention is described in detail below in conjunction with the accompanying drawings.

[0020] like Figures 1 to 7 As shown, a vehicle type identification method based on an axleless identification device provided by an embodiment of the present invention comprises the following steps: Step S1, collecting on-site vehicle waveform data of different vehicle models, and using big data analysis technology to pre-process the data, including filtering, denoising, etc., to ensure data quality; Step S2, analyzing the characteristics and regularities of the waveforms of each vehicle model, using machine learning algorithms, such as support vector machine (SVM), neural network, etc., to extract key features and establish a vehicle model recognition model; Step S3, identifying the upper scale axle through the upper scale channel data, and performing axle number statistics and vehicle model discrimination in combination with the axle spacing and axle width parameters; Step S4, using the lower scale channel data to identify the lower scale axle, and combining the axle spacing and axle width parameters to determine the number of axles remaining on the scale platform; Step S5, calculate the vehicle weight, combine the vehicle model identification result and the number of vehicle axles, query the average axle weight of the corresponding vehicle model through the vehicle model database, and then calculate the total vehicle weight.

[0021] By collecting a large number of on-site vehicle waveforms of different vehicle models, analyzing the waveform characteristics and rules of each vehicle model, an innovative axle identification method is proposed, which can accurately identify the vehicle model and calculate the weight only through the weighing sensor data of the scale platform itself without relying on a separate axle identification device. On the one hand, this method uses the upper scale channel data to identify the upper scale axle, thereby realizing axle count and vehicle model discrimination. On the other hand, it uses the lower scale channel data to identify the lower scale axle and determine the number of axles staying on the scale platform, thereby realizing the calculation of vehicle weight. This method fundamentally eliminates the dependence of traditional weighing systems on axle identification devices, and has the characteristics of high identification accuracy, low transformation cost, and low maintenance cost, meeting on-site use requirements.

[0022] Specifically, step S1 includes the following steps: S1.1 Set data collection points: Set sensors on the vehicle's travel path to ensure that waveform data of different vehicle models can be collected.

[0023] S1.2 Real-time data acquisition: Start the data acquisition system to record the waveform data of passing vehicles in real time.

[0024] S1.3 Data storage: The collected waveform data is stored in the database for subsequent processing and analysis.

[0025] S1.4 Data preprocessing: Filter the stored data to remove high-frequency noise and interference signals.

[0026] S1.5 Data denoising: Further denoising is performed on the filtered data, such as using median filtering, mean filtering and other methods to ensure the purity of the data.

[0027] S1.6 Data quality check: Ensure the quality of processed data through data verification and outlier detection.

[0028] Specific: S1.1 Set data collection points: Set sensors on the vehicle's path to ensure that waveform data of different vehicle models can be collected. For example, speed monitoring sensors can be installed at the entrances and exits of expressways and important nodes of urban roads.

[0029] S1.2 Real-time data collection: Start the data collection system to record the waveform data of passing vehicles in real time. For example, use radar sensors or LiDAR systems to monitor the speed, distance and other parameters of each vehicle in real time.

[0030] S1.3 Data storage: Store the acquired waveform data in a database for subsequent processing and analysis. For example, use a relational database (such as MySQL) or a NoSQL database (such as MongoDB) to classify and store the data.

[0031] S1.4 Data preprocessing: Filter the stored data to remove high-frequency noise and interference signals. For example, use a low-pass filter to remove high-frequency noise and improve data quality.

[0032] S1.5 Data denoising: Further denoise the filtered data, such as using median filtering, mean filtering, etc., to ensure the purity of the data. For example, median filtering is performed on vehicle speed data to remove outliers.

[0033] S1.6 Data quality check: Ensure the quality of processed data through data verification and outlier detection. For example, set a speed threshold and mark and remove abnormal data that exceeds the threshold.

[0034] Specifically, step S2 includes the following steps: S2.1 Feature extraction: Analyze the characteristics of waveform data of different vehicle models, such as waveform shape, amplitude, frequency, etc.

[0035] S2.2 Sample preparation: Select representative waveform data from the database as training samples and test samples.

[0036] S2.3 Machine learning algorithm selection: Select an appropriate machine learning algorithm, such as SVM or neural network, based on the vehicle model recognition requirements.

[0037] S2.4 Model training: Use training samples to train the machine learning algorithm, extract key features and establish a vehicle model recognition model.

[0038] S2.5 Model verification: Use test samples to verify the trained vehicle model recognition model and evaluate the accuracy and stability of the model.

[0039] The specific steps are as follows: S2.1 Feature extraction: Preprocess the waveform data, including denoising and smoothing.

[0040] Extract waveform features, such as peak value, valley value, average value, variance, frequency distribution, etc.

[0041] For example, for a car engine sound waveform, its harmonic components and resonant frequencies can be extracted.

[0042] S2.2 Sample preparation: Filter out waveform data of different models under different working conditions from the database.

[0043] Split the dataset to ensure a balanced distribution of training and test sets.

[0044] For example, we select 1000 samples, 500 of which are used as training sets and 500 as test sets.

[0045] S2.3 Machine Learning Algorithm Selection: According to the characteristics of waveform data, SVM or neural network is selected as the main algorithm.

[0046] For nonlinear problems, neural networks are used.

[0047] For example, a neural network model based on a multilayer perceptron is selected.

[0048] S2.4 Model training: Use the training set to train the selected machine learning algorithm.

[0049] Adjust model parameters such as learning rate, number of iterations, etc. to optimize model performance.

[0050] For example, set the learning rate to 0.01 and the number of iterations to 100.

[0051] S2.5 Model Validation: Use the test set to validate the trained model.

[0052] Calculate the model's accuracy, recall, F1 score and other indicators.

[0053] For example, the model achieved 90% precision and 85% recall on the test set.

[0054] Specifically, step S3 includes the following steps: S3.1 Weighing axle identification: Analyze the weighing channel data and identify the axle where the vehicle starts to be weighed.

[0055] S3.2 Axis parameter measurement: measure the axis spacing and axis width of the upper scale axis.

[0056] S3.3 Axle number statistics: Count the axle number of the vehicle based on the identification of the axles on the scale and the measurement of axle parameters.

[0057] S3.4 Preliminary vehicle model identification: Based on the number of axles and axle parameters, a preliminary judgment is made on the possible vehicle model.

[0058] Analyze the weighing channel data, identify the moment when the front wheel of the vehicle presses onto the weighing platform, and determine the axis from which the vehicle starts to be weighed. For example, when the data shows that the front wheel presses onto the weighing platform, it is marked as axis 1.

[0059] The distance between axle 1 and the front axle is measured through sensor data and recorded as the axle spacing parameter; at the same time, the width of axle 1 is measured as the axle width parameter, such as the axle spacing is 3.5 meters and the axle width is 1.2 meters.

[0060] Based on the identified axis 1 and the measured axis spacing and axis width, continue to search backward, identify subsequent axes, and record their respective axis spacing and axis width. For example, identify axis 2 and axis 3, with axis spacings of 4.0 meters and 4.5 meters respectively, and axis widths of both 1.2 meters.

[0061] Count the number of identified axles. If there are 3 axles in total, it means it is a three-axle vehicle.

[0062] Combine the number of axles, wheel spacing, and axle width parameters to make a preliminary judgment on the vehicle type. For example, if the wheel spacing is large and the number of axles is large, it may be a large truck; if the wheel spacing is small and there are only two axles, it may be a small car.

[0063] Specifically, step S4 includes the following steps: S4.1 Identification of the weighing axle: Analyze the weighing channel data and identify the axle where the vehicle starts to weigh.

[0064] S4.2 Reconfirm the axis parameters: Confirm the axis spacing and axis width of the lower scale axis and compare them with the data when the upper scale was used.

[0065] S4.3 Determination of the number of axes staying on the scale: Determine the number of axes staying on the scale platform based on the identification of the lower scale axis and the confirmation of the axis parameters.

[0066] The specific implementation is as follows: Analyze the data of the weighing channel to determine the first axle of the vehicle starting to weigh, such as recording the time point T1 when the axle passes through the weighing platform and the time point T2 when the axle leaves the weighing platform.

[0067] Confirm the axis spacing and axis width of the lower scale axis, such as the axis spacing is D1 and the axis width is W1.

[0068] Compare the axis spacing D1 and axis width W1 of the lower scale axis with the data recorded when the upper scale was used to verify consistency.

[0069] According to the identified lower scale axis and its parameters, combined with the length of the scale platform, the number of axes staying on the scale platform is calculated. For example, if the length of the scale platform is L and the axis spacing D1 is known, the number of axes staying N = L / D1 (rounded).

[0070] example: Assume that the time when the first axle of the vehicle starts to leave the scale is T1=10:00:00, the time of leaving is T2=10:00:05, the axle spacing D1=1.5 meters, and the axle width W1=0.6 meters. The axle spacing recorded when the vehicle is on the scale is 1.5 meters, and the axle width is 0.6 meters. The length of the scale platform is 9 meters. Based on these data, it can be determined that the number of axles staying on the scale platform is N=9 / 1.5=6 axles.

[0071] Specifically, step S5, calculating the vehicle weight, combining the vehicle model recognition result and the number of vehicle axles, querying the average axle weight of the corresponding vehicle model through the vehicle model database, and then calculating the total vehicle weight; S5.1 Preliminary calculation of vehicle weight: Based on the number of axles resting on the scale and the scale data, a preliminary calculation of the partial weight of the vehicle is made.

[0072] S5.2 Vehicle model database query: Based on the vehicle model recognition result and the number of vehicle axles, query the average axle weight of the corresponding vehicle model in the vehicle model database.

[0073] S5.3 Gross vehicle weight calculation: Calculate the gross vehicle weight based on the preliminary calculated vehicle partial weight and the average axle weight in the vehicle model database.

[0074] S5.4 Result output: Output the vehicle type identification result and the total vehicle weight to the user or related system.

[0075] The specific implementation of the above scheme is as follows: S5.1 Preliminary calculation of vehicle weight: Assuming the scale data shows that the number of axles of a vehicle resting on the scale is 3, and the scale data shows that the weight borne by each axle is 5 tons, then the partial weight of the vehicle is 3 axles * 5 tons / axle = 15 tons.

[0076] S5.2 Vehicle type database query: Combined with the vehicle type identification result of "medium-sized truck" and the number of vehicle axles being 3, the vehicle type database query shows that the average axle weight of medium-sized trucks is 4 tons / axle.

[0077] S5.3 Calculation of gross vehicle weight: Based on the preliminary calculated vehicle partial weight of 15 tons and the average axle weight of 4 tons / axle, the gross vehicle weight is calculated to be 3 axles * 4 tons / axle * 3 = 36 tons.

[0078] S5.4 Result output: The vehicle type identification result "medium-sized truck" and the gross vehicle weight of 36 tons are output to the user or related system.

[0079] Specifically, the method further includes step S6, constructing a real-time dynamic vehicle model recognition and weight estimation system: Combined with steps S1 and S2, the system can collect and process the waveform data of passing vehicles in real time, and use the machine learning model to quickly identify the vehicle type. Then, through step S5, the system combines the vehicle model identification results and the axle number statistics (from S3), queries the average axle weight in the vehicle model database, and dynamically estimates the total vehicle weight. This function has important application value in the fields of traffic monitoring, logistics management, and vehicle charging.

[0080] Real-time dynamic vehicle model recognition and weight estimation system refinement: Step S61: Install a high-precision sensor array to collect waveform data generated when a vehicle passes by in real time.

[0081] Step S62: Using a deep learning algorithm, such as a convolutional neural network (CNN), to train a vehicle model recognition model to recognize different vehicle models.

[0082] Step S63: Develop an axle number statistical algorithm to identify the vehicle axle number through changes in waveform data.

[0083] Step S64: Establish a vehicle model database, including average axle weight information of different vehicle models.

[0084] Step S65: Based on the vehicle model recognition result and the axle number statistics, query the vehicle model database and estimate the total vehicle weight.

[0085] For example, when a heavy truck passes the sensor, the system collects waveform data in real time, identifies it as a heavy truck through the CNN model, and counts that the vehicle has 6 axles. Based on the database information, the system estimates the total weight of the vehicle to be 30 tons.

[0086] Specifically, the method further includes step S7, optimizing the classification and weighing of the axle number adaptive vehicle models: On the basis of steps S3 and S4, the vehicle classification standard is automatically adjusted dynamically according to the change in the number of axles when the vehicle is on and off the scale. Combined with the weight calculation in step S5, the weighing process is optimized to improve the accuracy and efficiency of weighing. It is especially suitable for handling complex and changeable vehicle types and load conditions.

[0087] Specifically, the method further includes step S8, constructing an anti-cheating system for jointly verifying waveform characteristics and the number of axes: By comprehensively analyzing the waveform data of step S1, the vehicle model recognition results of S2, and the axle count statistics of S3 and S4, the system can achieve comprehensive monitoring of the vehicle weighing process. Once the waveform characteristics and axle count information are found to be mismatched or abnormal, the system will automatically trigger the anti-cheating mechanism to ensure the fairness and accuracy of the weighing process.

[0088] The construction of an anti-cheating system for joint verification of waveform characteristics and axis number includes the following steps: Step S81: Analyze the waveform data and extract characteristic values, such as peaks, troughs, frequency, etc.

[0089] Step S82: Analyze whether the waveform features match the vehicle model based on the vehicle model recognition result.

[0090] Step S83: Monitor the change in the number of axes to ensure that the waveform data is consistent with the axis number statistics.

[0091] Step S84: Develop an anomaly detection algorithm to identify situations where waveform features do not match axis number information.

[0092] Step S85: triggering anti-cheating mechanisms, such as alarming, recording abnormal events, etc.

[0093] Example: When a small truck is being weighed, the waveform data is abnormal, showing features that do not match the actual number of axles. The system detects the anomaly and triggers an alarm to prevent cheating.

[0094] Furthermore, an intelligent vehicle dispatching and load management system can be constructed. For example, in the field of transportation and logistics, the system can integrate the functions of steps S1 to S5 to form an intelligent vehicle dispatching and load management system. The system can grasp the vehicle type, load condition and driving trajectory in real time, provide accurate dispatching decision support for logistics companies, optimize transportation routes and resource allocation, and improve transportation efficiency and safety.

[0095] Specifically, steps S1 to S5: Integrate all the above steps to form a complete vehicle dispatch and load management system. It also includes the following steps: Develop an intelligent dispatch algorithm to optimize dispatch according to vehicle type, load condition and driving trajectory. Provide a real-time monitoring interface for dispatchers to view vehicle status and route information. Analyze historical data, optimize transportation routes and resource allocation, and improve transportation efficiency. For example: A logistics company monitors the load and driving trajectory of multiple trucks in real time through the system, and the system automatically recommends the optimal route to avoid congestion and reduce transportation time.

[0096] Furthermore, a vehicle health monitoring and early warning system is constructed: Although the focus is mainly on vehicle model identification and weight calculation, the waveform data collected in step S1 also contains rich information about the vehicle's driving status. Through further data mining and analysis, the system can identify potential problems such as abnormal vibration and imbalance of the vehicle, and then construct a vehicle health monitoring and early warning system. This function is of great significance for ensuring vehicle safety and preventing traffic accidents.

[0097] Specifically, the vehicle health monitoring and early warning system includes the following steps: In-depth analysis of waveform data to extract vehicle vibration, balance and other features. Develop data analysis models to identify potential vehicle problems, such as tire wear and suspension system failure. Establish an early warning mechanism to automatically notify vehicle maintenance personnel or drivers when an abnormality is detected. Record historical data for long-term health monitoring and maintenance decision support. For example: The system analyzes the waveform data of a truck and finds abnormal vibration, prompting maintenance personnel to check the vehicle suspension system, promptly detect and replace damaged parts, and prevent accidents.

[0098] Specifically, if step 4 is damaged, perform the following settings: In step S3, the positions of all axes are identified through the scale channel data, and the position information of each axis is recorded.

[0099] According to the vehicle type recognition model established in step S1 and step S2, combined with the position information of the axis obtained in step S3, the vehicle type recognition is performed on each axis; Compare the identified vehicle model information with the average axle weight of the corresponding vehicle model to calculate the weight of each axle; Add the weights of all axles to get the total weight of the vehicle; The calculation result of the total vehicle weight can be further calibrated by querying the average axle weight of the corresponding vehicle model in the vehicle model database and combining it with the weight of each axle calculated in step 3.

[0100] Through this method, although step 4 is omitted, the total weight of the vehicle can still be estimated based on the vehicle type identification information obtained in steps S3 and S2, and the axle position information collected in step 1. It should be noted that this estimation method may have a certain error, and the specific error size depends on the accuracy of vehicle type identification and the accuracy of axle weight estimation.

[0101] Working principle of the present invention: Step S1: First, we start to collect on-site vehicle waveform data of different models. These data are obtained through high-precision sensors and include various physical signals generated when the vehicle passes through the scale. In order to ensure the quality and accuracy of the data, we use big data analysis technology to pre-process the data. The pre-processing process includes filtering and denoising, which can effectively remove interference signals and noise in the data to ensure the accuracy of subsequent analysis.

[0102] Step S2: Next, we analyze the characteristics and patterns of the waveforms of each vehicle model. This step is the core of the entire recognition method. We use advanced machine learning algorithms, such as support vector machines (SVMs) and neural networks, to extract key features. These algorithms can identify the most helpful features for vehicle model recognition from complex waveform data, thereby establishing an accurate vehicle model recognition model.

[0103] Step S3: Then, we identify the axle on the scale through the data of the scale channel. This step involves counting the number of axles of the vehicle and preliminarily determining the vehicle model. We combine the parameters such as the axle spacing and axle width, and through precise algorithm calculation, we can accurately identify the number of axles of the vehicle and preliminarily determine the vehicle model.

[0104] Step S4: Use the data of the lower scale channel to identify the lower scale axle. This step is to further confirm the number of axles that stay on the scale. We also combine the parameters such as axle spacing and axle width, and through careful analysis and calculation, we can accurately determine the number of axles of the vehicle.

[0105] Step S5: Finally, we calculate the weight of the vehicle. This step combines the vehicle model recognition result and the number of vehicle axles, queries the average axle weight of the corresponding vehicle model through the vehicle model database, and then calculates the total weight of the vehicle. This step is the ultimate goal of the entire recognition method, which can provide users with accurate vehicle weight information.

[0106] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0107] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A vehicle type identification method based on an axleless identification device, characterized in that: The following steps are involved: Step S1, collecting on-site vehicle waveform data of different vehicle models, and pre-processing the data using big data analysis technology; Step S2, analyzing the characteristics and rules of the waveforms of each vehicle model, using machine learning algorithms to extract key features, and establishing a vehicle model recognition model; Step S3, identifying the upper scale axle through the upper scale channel data, and performing axle number statistics and vehicle model discrimination in combination with the axle spacing and axle width parameters; Step S4, using the lower scale channel data to identify the lower scale axle, and combining the axle spacing and axle width parameters to determine the number of axles remaining on the scale platform; Step S5, calculate the vehicle weight, combine the vehicle model identification result and the number of vehicle axles, query the average axle weight of the corresponding vehicle model through the vehicle model database, and then calculate the total vehicle weight.

2. A vehicle type identification method based on an axleless identification device as claimed in claim 1, characterized in that: Step S1 includes the following steps: S1.1 Set data collection points: Set sensors on the vehicle's path to ensure that waveform data of different vehicle models can be collected; S1.2 Real-time data acquisition: Start the data acquisition system to record the waveform data of passing vehicles in real time; S1.3 Data storage: Store the collected waveform data in the database for subsequent processing and analysis; S1.4 Data preprocessing: Filter the stored data to remove high-frequency noise and interference signals; S1.5 Data denoising: further denoise the filtered data to ensure data purity; S1.6 Data quality check: Ensure the quality of processed data through data verification and outlier detection.

3. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: Step S2 includes the following steps: S2.1 Feature extraction: Analyze the characteristics of waveform data of different vehicle models; S2.2 Sample preparation: Select representative waveform data from the database as training samples and test samples; S2.3 Machine learning algorithm selection: Select a machine learning algorithm based on the vehicle model recognition requirements; S2.4 Model training: Use training samples to train the machine learning algorithm, extract key features and establish a vehicle model recognition model; S2.5 Model verification: Use test samples to verify the trained vehicle model recognition model and evaluate the accuracy and stability of the model.

4. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: Step S3 includes the following steps: S3.1 Weighing axle identification: Analyze the weighing channel data and identify the axle where the vehicle starts to weigh; S3.2 Axis parameter measurement: measure the axis spacing and axis width parameters of the upper scale axis; S3.3 Axle count: Count the axle count of the vehicle based on the identification of the axle on the scale and the measurement of axle parameters; S3.4 Preliminary vehicle model identification: Based on the number of axles and axle parameters, a preliminary judgment is made on the possible vehicle model.

5. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: Step S4 includes the following steps: S4.1 Identification of the weighing axle: Analyze the weighing channel data and identify the axle where the vehicle starts to weigh; S4.2 Reconfirm the axis parameters: Confirm the axis spacing and axis width of the lower scale axis and compare them with the data when the upper scale was used; S4.3 Determination of the number of axes staying on the scale: Determine the number of axes staying on the scale platform based on the identification of the lower scale axis and the confirmation of the axis parameters.

6. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: Step S5 includes the following steps: S5.1 Preliminary calculation of vehicle weight: Based on the number of axles resting on the scale and the scale data, a preliminary calculation of the partial weight of the vehicle is made; S5.2 Vehicle model database query: Combine the vehicle model recognition result and the number of vehicle axles to query the average axle weight of the corresponding vehicle model in the vehicle model database; S5.3 Gross vehicle weight calculation: Calculate the gross vehicle weight based on the preliminary calculated vehicle partial weight and the average axle weight in the vehicle model database; S5.4 Result output: Output the vehicle type identification result and the total vehicle weight to the user or related system.

7. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: The system also includes step S6, which is to construct a real-time dynamic vehicle model recognition and weight estimation system: Combined with steps S1 and S2, the vehicle waveform data is collected and processed in real time, and the vehicle type is quickly identified using a machine learning model. Then, through step S5, the system combines the vehicle type identification results and axle number statistics, queries the average axle weight in the vehicle type database, and dynamically estimates the total vehicle weight.

8. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: The step S7 is also included, which is to optimize the classification and weighing of the axle number adaptive vehicle models: On the basis of steps S3 and S4, the vehicle classification standard is automatically adjusted dynamically according to the change in the number of axles when the vehicle is on and off the scale. Combined with the weight calculation in step S5, the weighing process is optimized to improve the accuracy and efficiency of weighing.

9. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: The method further includes step S8, wherein the anti-cheating system for jointly verifying waveform characteristics and the number of axes is constructed, including the following steps: Step S81: Analyze waveform data and extract characteristic values; Step S82: combining the vehicle type recognition result, analyzing whether the waveform feature matches the vehicle type; Step S83: monitor the change of the number of axes to ensure that the waveform data is consistent with the axis number statistics; Step S84: Develop an anomaly detection algorithm to identify situations where waveform features do not match axis number information; Step S85: triggering the anti-cheating mechanism.

10. The vehicle type identification method based on the shaftless identification device according to claim 1, characterized in that: If step 4 is damaged, perform the following settings: In step S3, the positions of all axes are identified through the scale channel data, and the position information of each axis is recorded; According to the vehicle type recognition model established in step S1 and step S2, combined with the position information of the axis obtained in step S3, the vehicle type recognition is performed on each axis; Compare the identified vehicle model information with the average axle weight of the corresponding vehicle model to calculate the weight of each axle; Add the weights of all axles to get the total weight of the vehicle; The calculation result of the total vehicle weight can be further calibrated by querying the average axle weight of the corresponding vehicle model in the vehicle model database and combining it with the weight of each axle calculated in step 3.

Images