Self-adaptive acquisition, identification and classification method based on driving dynamic load frequency

Abstract

The invention discloses a self-adaptive acquisition, identification and classification method based on driving dynamic load frequency, which relates to the technical field of traffic transportation engineering, and comprises the following steps of: acquiring vibration and acceleration signals generated in a driving process in real time through a plurality of high-sensitivity sensors integrated in equipment; through the integrated design and the self-adaptive acquisition, identification and classification method, the accuracy and efficiency of traffic infrastructure dynamic load monitoring are remarkably improved. The device automatically adjusts the sampling frequency to adapt to different vehicles, data inaccuracy or redundancy is avoided, signals are intelligently classified through an artificial intelligence algorithm, visual images are generated, and reliable support is provided for infrastructure design and maintenance. The system achieves intelligent and automatic monitoring, reduces manual intervention, improves the data processing speed and accuracy, helps a traffic management department to make scientific decisions based on big data prediction analysis, and effectively avoids infrastructure damage and accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of transportation engineering, and particularly to an adaptive acquisition, recognition and classification method based on the driving dynamic load frequency of a vehicle. Background Art

[0002] With the continuous development of transportation facilities, especially the acceleration of the urbanization process, the dynamic load effects of vehicles on infrastructure such as roads, bridges and tunnels are becoming more and more significant. The driving dynamic load refers to the dynamic load generated by vehicles (such as cars, trains, trucks, etc.) on the ground or other structures (such as bridges, tracks, etc.). These dynamic loads will affect the structural safety, durability of roads and bridges, and even the scheduling and management of traffic flow. Therefore, accurately measuring, identifying and analyzing the frequency characteristics of driving dynamic loads has become one of the core issues in modern traffic engineering research. Most traditional dynamic load measurement methods rely on vibration sensors and data acquisition instruments to analyze vibration signals to judge the impact of vehicles on infrastructure. However, these methods usually have problems such as scattered equipment, complex operation and insufficient analysis accuracy. Especially in complex traffic environments, it is impossible to achieve precise classification of various vehicles and in-depth analysis of dynamic load characteristics.

[0003] The existing technologies have the following deficiencies:

[0004] Currently, most traditional driving dynamic load measurement technologies adopt independent modular equipment systems. These devices usually include multiple parts such as sensors, data acquisition devices and signal processing units. The connection and debugging between devices are cumbersome, which is not conducive to rapid deployment and on-site operation. Especially in a dynamic environment, the assembly and maintenance of the measurement system require a lot of time and manpower. In addition, most existing technologies adopt a fixed-frequency sampling method in data acquisition. This method cannot flexibly adjust according to the differences in dynamic load characteristics generated by different types of vehicles, resulting in inaccurate capture of frequency response and load characteristics. For example, the dynamic load frequencies generated by cars and heavy trucks are quite different, and the fixed sampling frequency of the traditional system cannot meet this differential requirement, thus affecting the accuracy and reliability of the data. In addition, the existing technologies lack the adaptive analysis ability based on artificial intelligence and deep learning, resulting in the inability to effectively classify and identify multiple dynamic load signals in complex traffic environments, restricting the depth and breadth of subsequent data processing. These technical limitations make the existing driving dynamic load measurement methods difficult to meet the needs of modern traffic infrastructure construction and management in terms of efficient monitoring, intelligent evaluation and precise analysis.

[0005] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The object of the present invention is to provide an adaptive acquisition, recognition and classification method based on the dynamic load frequency of vehicles. Through integrated design and adaptive acquisition, recognition and classification methods, the present invention significantly improves the accuracy and efficiency of dynamic load monitoring of traffic infrastructure. The device automatically adjusts the sampling frequency to adapt to different means of transportation, avoiding inaccurate or redundant data. At the same time, intelligent classification of signals is carried out through artificial intelligence algorithms to generate visual images, providing reliable support for infrastructure design and maintenance. The system realizes intelligent and automatic monitoring, reduces manual intervention, improves the speed and accuracy of data processing, and based on big data prediction analysis, helps traffic management departments make scientific decisions, effectively avoiding infrastructure damage and accidents, so as to solve the problems in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions: An adaptive acquisition, recognition and classification method based on the dynamic load frequency of vehicles, comprising the following steps:

[0008] Real-time acquisition of vibration and acceleration signals generated during vehicle driving through multiple high-sensitivity sensors integrated in the device;

[0009] According to the frequency characteristics of the acquired signals, through the built-in adaptive frequency adjustment module, automatically adjust the sampling frequency and data processing method to ensure that the dynamic load characteristics generated by different types of means of transportation can be accurately captured;

[0010] Preprocess the acquired original signals, and extract the key features of the signals through methods such as spectrum analysis for subsequent classification and recognition;

[0011] Use the trained deep learning algorithm to analyze the processed signal features, identify the types of means of transportation based on the extracted feature data, and classify the dynamic load characteristics;

[0012] According to the classification results, generate multi-dimensional visual images through algorithms, real-time display the dynamic load characteristics of means of transportation and their influence range on infrastructure, and transmit the analysis results to external devices through a wireless transmission module.

[0013] Preferably, the specific steps of real-time acquisition of vibration and acceleration signals generated during vehicle driving through multiple high-sensitivity sensors integrated in the device are as follows:

[0014] The device needs to be installed at key positions on the path passed by the means of transportation, and the height and angle can be adjusted to ensure that the sensors can accurately capture the dynamic load signals;

[0015] Start the device and perform self-check to ensure that all components are operating normally and ready for data acquisition;

[0016] The sensor collects vibration and acceleration signals generated when vehicles pass by in real time, and records the amplitude, duration and frequency characteristics of dynamic loads;

[0017] The collected data is uploaded to external devices in real time through a wireless transmission module for engineers to monitor and analyze.

[0018] Preferably, according to the frequency characteristics of the collected signals, through the built-in adaptive frequency adjustment module, the sampling frequency and data processing method are automatically adjusted to ensure that the dynamic load characteristics generated by different types of vehicles can be accurately captured. The specific steps are as follows:

[0019] Identify the signal frequency characteristics generated by vehicles through spectrum analysis to provide basic data for adaptive sampling;

[0020] Automatically adjust the sampling frequency according to the dynamic load frequency characteristics of different vehicles to ensure data accuracy;

[0021] Dynamically optimize the data processing method according to the collected signal characteristics to ensure effective signal capture and noise removal;

[0022] Continuously adjust the sampling frequency and data processing method through a real-time feedback mechanism to ensure adaptation to the dynamic load characteristics of different types of vehicles.

[0023] Preferably, the following are the specific steps for preprocessing the collected original signals and extracting the key characteristics of the signals through methods such as spectrum analysis for subsequent classification and identification:

[0024] Remove noise and interference through a filter, and retain useful dynamic load signals to improve the quality and accuracy of the data;

[0025] Enhance the signal amplitude through a gain amplifier to ensure the clarity and resolvability of the signal in subsequent processing;

[0026] Use spectrum analysis methods to extract the main frequency components in the signal to provide key characteristics for subsequent classification and identification;

[0027] Calibrate the extracted features with vehicle types and prepare data input for subsequent classification and identification.

[0028] Preferably, use the trained deep learning algorithm to analyze the processed signal characteristics, and identify the types of vehicles based on the extracted feature data and classify the dynamic load characteristics. The specific steps are as follows:

[0029] First, input the processed signal characteristics into the trained deep learning model for feature mapping. Use a convolutional neural network to extract and map features. The model gradually extracts higher-level features through multiple convolutional and pooling operations. Let the input feature matrix be X = [xi =[x 1 , x 2 , ……, x n , where x i represents the i-th signal feature, n is the total number, and the signal features have been processed by filtering, denoising, and spectral analysis. In the model, the convolutional layer operates through the following formula:

[0030] f conv (X) = σ(W·X + b)

[0031] , where f conv (X) represents the output result of the convolutional operation, σ is the activation function, W is the convolutional kernel, X is the input signal feature matrix, and b is the bias term;

[0032] Through multi-layer convolution and pooling operations, a high-dimensional feature vector Z is finally obtained, where each element represents a certain complex dynamic load characteristic;

[0033] After the deep learning model extracts the high-dimensional feature vector Z, the high-dimensional feature vector Z is input into the fully connected layer for classification, and then the type of the transportation vehicle is identified and the dynamic load characteristics are classified. Through the trained classification network, the model will output the classification result, and this result obtains the classification probability of each transportation vehicle through the softmax function. The softmax function is used for normalization processing, and the calculation formula is as follows:

[0034]

[0035] , where y p is the prediction probability of the model for the p-th type of transportation vehicle, m is the total number of transportation vehicle categories, is the exponentialized eigenvalue of each category.

[0036] Preferably, according to the classification result, the specific steps of generating a multi-dimensional visualization image through an algorithm to display the dynamic load characteristics of the transportation vehicle and its influence range on the infrastructure in real time, and transmitting the analysis result to an external device through a wireless transmission module are as follows:

[0037] After signal preprocessing and feature extraction, first generate a time history graph to display the dynamic response of the dynamic load signal of the transportation vehicle changing with time. Calculate the displacement response by integrating the collected acceleration signal to reflect the influence of the transportation vehicle on the infrastructure. The displacement response is calculated through the following formula:

[0038] x(t) = ∫ 0 t ∫ 0 t′ a(t″)dt″dt′

[0039] , where \(x(t)\) is the displacement response, \(a(t)\) is the acceleration signal, and \(t\) is the time variable;

[0040] After obtaining the time history diagram and calculating the displacement response, the next step is to generate a spectrogram to analyze the distribution of the signal at different frequencies. The time-domain signal is converted into a frequency-domain signal using the fast Fourier transform, and the spectrogram is calculated using the following spectral formula:

[0041]

[0042] , where \(X(f)\) is the spectral component at frequency \(f\), \(e\) is the natural base, \(x(t)\) is the displacement signal in the time history diagram, \(j\) is the imaginary unit, \(f\) is the frequency variable, and \(2\pi\) is a mathematical constant;

[0043] Finally, based on the results of the spectrogram, a power spectral density diagram is generated. The power spectral density describes the power distribution within a unit frequency bandwidth, and the frequency range where the energy of the dynamic load is concentrated is determined. The power spectral density is calculated using the following formula:

[0044]

[0045] , where \(P(f)\) is the power spectral density and \(T\) is the total duration of the signal.

[0046] In the above technical solution, the technical effects and advantages provided by the present invention are as follows:

[0047] Through the integrated design and adaptive acquisition, recognition, and classification methods, the present invention effectively improves the accuracy and efficiency of dynamic load monitoring of traffic infrastructure. The device can automatically adjust the sampling frequency to adapt to the dynamic load characteristics of different types of vehicles, avoiding the problems of inaccurate or redundant data caused by improper frequency sampling in traditional systems. By using artificial intelligence algorithms to intelligently classify the collected signals, the system can accurately distinguish the dynamic loads generated by different vehicles and generate visual images in real time, providing reliable data support for infrastructure design, maintenance, and safety assessment. In addition, the modular design and wireless data transmission function improve the deployment flexibility and data processing efficiency of the system, enabling traffic management departments to conduct real-time monitoring and decision-making more efficiently.

[0048] The present invention realizes the intelligence and automation of traffic load monitoring. By using deep learning algorithms, it can automatically identify the types of vehicles and evaluate their impacts on infrastructure, significantly improving the data processing speed and accuracy. The intelligent classification and analysis system makes the monitoring process fully automated, reduces the need for manual intervention, and enhances the operational convenience. The system can not only monitor the current traffic load status in real time but also analyze the future load change trends based on big data and prediction models, helping traffic management departments conduct preventive maintenance and make scientific decisions, thereby improving traffic safety, management efficiency, and effectively avoiding infrastructure damage and traffic accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0050] Figure 1 It is a flowchart of the adaptive acquisition, recognition, and classification method based on the driving dynamic load frequency of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.

[0052] The present invention provides an adaptive acquisition, recognition, and classification method based on the driving dynamic load frequency as shown in Figure 1 the following, including the following steps:

[0053] Real-time acquisition of vibration and acceleration signals generated during vehicle driving through multiple high-sensitivity sensors integrated in the device. The sensors are set on the passing path of the vehicle and can capture and record the dynamic load characteristics generated during vehicle driving;

[0054] The specific steps of real-time acquisition of vibration and acceleration signals generated during vehicle driving through multiple high-sensitivity sensors integrated in the device. The sensors are set on the passing path of the vehicle and can capture and record the dynamic load characteristics generated during vehicle driving are as follows:

[0055] The device needs to be installed at key positions on the passing path of the vehicle and the height and angle can be adjusted to ensure that the sensors can accurately capture the dynamic load signals;

[0056] First, the device needs to be correctly installed on the path where the vehicle passes. To ensure the accuracy and reliability of the data, the choice of installation location is crucial. Usually, the sensor should be installed at key positions on bridges, tunnels or roads to ensure comprehensive coverage of the vehicle passing path. For structures such as bridges or tunnels, it is usually selected to install in areas such as the middle of the lane or the center line of the track, because these positions can usually most directly reflect the dynamic impact of the vehicle on the infrastructure. In addition, to adapt to different traffic environments, the installation method of the device should be flexible and able to adjust the height and angle according to specific measurement requirements to ensure that the sensor is parallel to the road surface or the structure surface, thereby minimizing the impact of external interference on the data.

[0057] Start the device and perform a self-check to ensure that all components are operating normally and ready for data collection;

[0058] After installation, the device should be started and initialized to ensure that all components are operating normally. In this step, first, the power status of the device needs to be checked to ensure that the device has sufficient power, and then start the main control system of the device. Through interfaces such as the liquid crystal display screen, the operator can confirm the operating status of the device, including the working status of the sensor and the configuration parameters of the data acquisition module. During initialization, the device will automatically perform a self-check to ensure that there are no problems with the coordinated operation of various components such as the sensor, signal acquisition module, and data processing module. After the system initialization is completed, the device enters the standby state and is ready to start collecting data.

[0059] The sensor continuously collects vibration and acceleration signals generated when the vehicle passes by, and records the amplitude, duration, and frequency characteristics of the dynamic load;

[0060] After the device is ready, it enters the signal acquisition stage. At this time, the sensor will start to monitor and continuously record the vibration and acceleration signals generated when the vehicle passes by. The sensor captures the dynamic load of the vehicle's driving in a timely manner by sensing the vibration or acceleration changes. Specifically, when a vehicle or other vehicle passes by, the contact between the wheel and the road surface or track will cause vibrations in structures such as the ground or bridge. These vibration signals will be converted into electrical signals by the sensor and transmitted to the data acquisition module for processing. By continuously collecting these signals, the device can record various dynamic load characteristics at different time points, reflecting the impact of the vehicle on the infrastructure, including the amplitude, duration, and frequency characteristics of the load.

[0061] The collected data is uploaded to an external device in real time through a wireless transmission module for engineers to monitor and analyze;

[0062] After signal acquisition, the device will transmit the real-time acquired data to the central processing module or external devices. The acquired data is transmitted to mobile devices (such as smartphones, tablets) or remote servers through a wireless transmission module (such as Wi-Fi or Bluetooth) for real-time monitoring and analysis by engineers or traffic management personnel. In this stage, the device can not only transmit the original data, but also transmit real-time dynamic load information, including vibration waveforms and acceleration time series, etc. Through wireless transmission, data can be remotely accessed, making the management of the data acquisition process more convenient. In some application scenarios, the data can also be synchronously uploaded to the cloud for further analysis, forming a more refined dynamic monitoring system and improving the efficiency of data sharing and collaborative work.

[0063] According to the frequency characteristics of the acquired signals, through the built-in adaptive frequency adjustment module, automatically adjust the sampling frequency and data processing method to ensure that the dynamic load characteristics generated by different types of transportation tools (such as cars, trucks, trains, etc.) can be accurately captured;

[0064] The specific steps to automatically adjust the sampling frequency and data processing method according to the frequency characteristics of the acquired signals to ensure that the dynamic load characteristics generated by different types of transportation tools (such as cars, trucks, trains, etc.) can be accurately captured are as follows:

[0065] Identify the frequency characteristics of the signals generated by transportation tools through spectrum analysis to provide basic data for adaptive sampling;

[0066] After the sensor real-time acquires the vibration and acceleration signals generated when a transportation tool passes by, the first step is to conduct a preliminary analysis of these signals to identify their frequency characteristics. The signals captured by the sensor contain a large amount of dynamic load information, including vibration components with different frequencies, amplitudes, and durations. By conducting spectrum analysis on these signals, the system can extract the frequency distribution characteristics of the signals. Different types of transportation tools, such as cars, trucks, and trains, tend to have different dynamic load characteristics in terms of frequency. For example, heavy vehicles (such as trucks) generate more significant low-frequency dynamic loads, while light vehicles (such as cars) generate more high-frequency dynamic loads. The signal analysis module identifies the frequency range of the signals in real time based on these characteristics, providing basic data for subsequent adaptive sampling.

[0067] Automatically adjust the sampling frequency according to the dynamic load frequency characteristics of different transportation tools to ensure data accuracy;

[0068] Once the frequency characteristics of the signal are identified, the built-in adaptive frequency adjustment module automatically adjusts the sampling frequency according to the dynamic load frequency characteristics of different vehicles. In traditional systems, the sampling frequency is fixed, which may lead to under-sampling or over-sampling for different types of vehicles, affecting the accuracy of data. Through adaptive adjustment, the system can dynamically adjust the sampling rate according to the specific frequency range generated by the vehicle. For example, for trucks or trains with lower frequencies, the system will reduce the sampling frequency to avoid excessive redundant data; while for small cars with higher frequencies, the system will increase the sampling frequency to ensure that high-frequency components can be captured. Adaptive frequency adjustment ensures that data can always be collected at the optimal frequency under different types of vehicles and road conditions, guaranteeing the accuracy and integrity of the data.

[0069] Dynamically optimize the data processing method according to the collected signal characteristics to ensure effective signal capture and noise removal;

[0070] During the frequency adjustment, the data processing module also optimizes the data processing method according to different types of signal characteristics. Traditional signal processing methods often rely on fixed algorithms for processing, but these algorithms may not be able to adapt to all types of dynamic load signals. By combining the results of frequency adjustment, the system will automatically select the most appropriate data processing method. For example, for low-frequency signals, a band-pass filter may be used for filtering to retain the core frequency information; while for high-frequency signals, higher-order filtering methods or spectral analysis methods may be required to ensure that the high-frequency components of the signal are not lost. Through this strategy of dynamically optimizing the data processing method, the system can effectively eliminate noise and maintain the effective information of the signal while capturing the signal, thereby improving the quality of the signal and the accuracy of subsequent analysis.

[0071] Continuously adjust the sampling frequency and data processing method through a real-time feedback mechanism to ensure adaptation to the dynamic load characteristics of different types of vehicles;

[0072] With the continuous optimization of adaptive frequency adjustment and data processing methods, the system will continuously perform real-time feedback and adjustment according to the changes of the vehicle. This means that when different vehicles pass by, the device will dynamically adjust the frequency and processing method according to the changes in the signals collected in real time. For example, when the system detects the passing of a heavier vehicle (such as a train or a large truck), it may automatically reduce the sampling frequency and change the filtering strategy; while when the system detects the passing of a light car or a motorcycle, it will increase the sampling frequency to ensure that more high-frequency components are captured. This dynamic feedback mechanism not only ensures the adaptability of the device in different traffic environments, but also makes the entire data collection and processing system more flexible and efficient, thus providing accurate data support for subsequent dynamic load analysis, vehicle identification and classification.

[0073] Preprocess the collected original signals through filtering, denoising, amplification, etc., and extract the key features of the signals through methods such as spectral analysis for subsequent classification and recognition;

[0074] The specific steps for preprocessing the collected original signals through filtering, denoising, amplification, etc., and extracting the key features of the signals through methods such as spectral analysis for subsequent classification and recognition are as follows:

[0075] Remove noise and interference through a filter, and retain the useful dynamic load signals to improve the quality and accuracy of the data;

[0076] In the collected original signals, there usually contain some noise and interference components. These noises may come from environmental vibrations, electronic device interferences, or other external factors. To ensure the accuracy of subsequent analysis, the first step is to filter and denoise the original signals. The selection of the filter is customized according to the frequency characteristics of the signals. Common filtering methods include low-pass filtering, high-pass filtering, and band-pass filtering. Through these filtering methods, high-frequency noise or low-frequency drift can be effectively removed, and the useful dynamic load signals can be retained. For example, when the system detects low-frequency background noise, the low-pass filter can remove signals with frequencies higher than a certain threshold. For low-frequency signals caused by heavy vehicles such as trucks, the irrelevant frequency parts can be filtered out through the band-pass filter, leaving the key dynamic load characteristics. In this way, the subsequent analysis can ensure the data quality and is not affected by noise interference.

[0077] Enhance the signal amplitude through a gain amplifier to ensure the clarity and distinguishability of the signal in subsequent processing;

[0078] After filtering, the signal may still have a small amplitude or signal attenuation. In this case, signal amplification processing is required. Through the gain amplifier, the system can enhance the amplitude of the signal, making the signal clearer and more distinguishable in the subsequent processing stage. The enhancement of the signal can not only improve the signal-to-noise ratio but also ensure that low-amplitude signals will not lose important information due to insufficient amplification. This step is usually closely combined with adaptive frequency adjustment. Based on the sampling frequency adjustment, the amplitude of the signal is amplified in real time to ensure that the characteristics of the signal will not be weakened under the dynamic loads of different vehicles. For example, when a light vehicle (such as a car) passes by, the amplitude of the signal may be low and needs to be appropriately amplified; for heavy vehicles (such as trucks), although the signal amplitude is large, the amplification factor may also need to be adjusted to adapt to different environmental conditions.

[0079] Use spectral analysis methods to extract the main frequency components in the signal and provide key features for subsequent classification and recognition;

[0080] After the signal preprocessing is completed, the next step is to extract the key features of the signal through methods such as spectral analysis. Spectral analysis can use mathematical tools such as the Fourier transform (FFT) to convert the time-domain signal into a frequency-domain signal, thereby identifying the main frequency components in the signal. The dynamic load signals generated by different types of transportation vehicles have significant differences in frequency. The signal frequency ranges of transportation vehicles such as cars, trucks, and trains are different, and even the signals of the same transportation vehicle may vary under different road conditions. Therefore, spectral analysis can effectively help the system identify different frequency components in the signal and provide necessary features for subsequent classification and recognition. By extracting features such as the main frequency, amplitude, and phase, the system can comprehensively describe the time-frequency characteristics of the dynamic load and provide basic data for automated identification and classification. For example, the low-frequency dynamic load components of trucks may be more prominent, while cars may contain more high-frequency components, and spectral analysis can help the system distinguish these differences.

[0081] Calibrate the extracted features with the transportation vehicle type and prepare the data input for subsequent classification and recognition;

[0082] After the spectral analysis and feature extraction of the signal are completed, the last step is to calibrate the extracted features and prepare the data for subsequent classification and recognition. In this process, the system will calibrate the extracted features through machine learning or deep learning algorithms, that is, correspond different frequency components, amplitudes, and time-domain features to the corresponding transportation vehicle types (such as cars, trucks, trains, etc.). By training the model, the system can automatically determine the source of the dynamic load based on the extracted signal features. The key to this step is to enable the system to accurately identify the dynamic load signals generated by various transportation vehicles and classify them through a large amount of training data. In addition, the calibrated data will be used as input for the artificial intelligence algorithm for further learning and recognition, thereby providing support for the accurate classification and recognition of transportation vehicles and ensuring that effective and operable results can be generated for subsequent analysis.

[0083] Use the trained deep learning algorithm to analyze the processed signal features, identify the types of transportation vehicles based on the extracted feature data, and classify the dynamic load characteristics;

[0084] The specific steps of using the trained deep learning algorithm to analyze the processed signal features, identify the types of transportation vehicles based on the extracted feature data, and classify the dynamic load characteristics are as follows:

[0085] First, input the processed signal features (such as frequency components, amplitudes, time-domain features, etc.) into the trained deep learning model for feature mapping. Use a convolutional neural network (CNN) to extract and map features. The model gradually extracts higher-level features through multiple convolutional and pooling operations. Let the input feature matrix be X = [xi = [x 1 , x 2 , ……, x n , where x i represents the i-th signal feature, n is the total number, and the signal features have been processed by filtering, denoising, and spectral analysis. In the model, the convolutional layer operates through the following formula:

[0086] f conv (X) = σ(W·X + b)

[0087] , where f conv (X) represents the output result of the convolution operation, σ is the activation function (such as ReLU), W is the convolutional kernel (filter), X is the input signal feature matrix containing the dynamic load characteristics obtained through filtering and spectral analysis, and b is the bias term;

[0088] Through multiple convolutional and pooling operations, a high-dimensional feature vector Z is finally obtained, where each element represents a certain complex dynamic load characteristic;

[0089] This operation helps the model extract more complex feature mappings from the original signal features, such as the feature patterns generated by different transportation tools. The high-dimensional feature vector Z will be used as the input for the subsequent classification step.

[0090] After the deep learning model extracts the high-dimensional feature vector Z, the high-dimensional feature vector Z is input into the fully connected layer for classification, and then the type of transportation tool is identified and the dynamic load characteristics are classified. Through the trained classification network, the model will output the classification result, and this result obtains the classification probability of each transportation tool through the softmax function. The softmax function is used for normalization processing, and the calculation formula is as follows:

[0091]

[0092] , where y p is the predicted probability of the model for the p-th type of transportation tool, m is the total number of transportation tool categories, is the exponentiated eigenvalue of each category, representing the relative importance of that category.

[0093] By comparing the probability values of each category, the model can identify which transportation tool the signal comes from and classify it according to the dynamic load characteristics. If y p corresponding to a specific transportation tool category has the highest probability value, then the signal is classified as that transportation tool type and the dynamic load characteristics are classified according to the category.

[0094] According to the classification results, multi-dimensional visualization images are generated through algorithms, including time history diagrams, spectrograms, power spectral density diagrams, etc., to display the dynamic load characteristics of transportation vehicles and their influence ranges on infrastructure in real time, and the analysis results are transmitted to external devices through a wireless transmission module;

[0095] The specific steps for generating multi-dimensional visualization images according to the classification results, including time history diagrams, spectrograms, power spectral density diagrams, etc., to display the dynamic load characteristics of transportation vehicles and their influence ranges on infrastructure in real time, and transmitting the analysis results to external devices through a wireless transmission module are as follows:

[0096] After signal preprocessing and feature extraction, a time history diagram is first generated to show the dynamic response of the dynamic load signal of the transportation vehicle changing with time. The time history diagram can clearly show the vibration and acceleration changes generated when the transportation vehicle passes by. The displacement response is calculated by integrating the collected acceleration signal to reflect the impact of the transportation vehicle on the infrastructure. The displacement response is calculated by the following formula:

[0097] x(t)=∫ 0 t ∫ 0 t′ a(t″)dt″dt′

[0098] , where x(t) is the displacement response, a(t) is the acceleration signal, and t is the time variable;

[0099] This step converts the acceleration signal into a displacement signal through two integrations, reflecting the actual impact of the dynamic load generated by the transportation vehicle on the road or bridge. The generation step of the time history diagram provides basic data for the subsequent calculation of the spectrogram and power spectral density diagram because the displacement signal is the input of the spectral analysis.

[0100] After obtaining the time history diagram and calculating the displacement response, the next step is to generate a spectrogram to analyze the distribution of the signal at different frequencies. The spectrogram can reveal the different frequency components contained in the dynamic load signal generated by the transportation vehicle and help analyze the impact of various frequency components on the infrastructure. The fast Fourier transform (FFT) is used to convert the time-domain signal into a frequency-domain signal to calculate the spectrogram. The spectral calculation formula is:

[0101]

[0102] , where X(f) is the spectral component at frequency f, e is the natural base, x(t) is the displacement signal in the time history diagram, j is the imaginary unit, f is the frequency variable, and 2π is a mathematical constant representing twice the value of the mathematical constant π;

[0103] The result of the FFT, X(f), describes the intensity distribution of the signal at different frequencies, which helps to analyze the dynamic load characteristics generated by different vehicles. For example, lower frequencies may correspond to the vibrations of trucks, while the high-frequency components may be related to the dynamic loads of light vehicles. This spectrogram provides frequency information for the subsequent calculation of the power spectral density diagram.

[0104] Finally, based on the results of the spectrogram, a power spectral density diagram is generated. This diagram can reveal the signal energy distribution corresponding to different frequency components and further analyze the actual impact of the dynamic load signal on the infrastructure. The power spectral density describes the power distribution within a unit frequency bandwidth and determines the frequency range where the energy of the dynamic load is mainly concentrated. The power spectral density is calculated by the following formula:

[0105]

[0106] , where P(f) is the power spectral density and T is the total duration of the signal.

[0107] This step describes the power distribution of the signal at each frequency and can reveal which frequency components have the greatest impact on the infrastructure. Through the power spectral density diagram, the system can accurately identify the frequency ranges where the dynamic load signals of vehicles are mainly concentrated, providing precise data support for subsequent infrastructure monitoring and maintenance.

[0108] Specific implementation method 1: The core of this implementation method is to integrate multiple functional modules into a single device through a highly integrated design to achieve efficient and convenient dynamic load acquisition and analysis. The device is embedded with multiple high-sensitivity sensors, signal acquisition modules, data processing modules, and intelligent analysis modules, forming a compact and integrated system. The device can be flexibly deployed on the passing paths of vehicles, such as key positions like bridges, tunnels, and road surfaces. The acquisition module of the sensor can continuously monitor the vibration and acceleration signals generated when a vehicle passes, covering different frequency components generated when a vehicle passes, from the low-frequency heavy trucks to the high-frequency small cars, and the device can accurately capture their dynamic load characteristics.

[0109] Since the device integrates data acquisition, processing, and analysis modules internally, users do not need to connect multiple external devices, reducing connection complexity and operation difficulty, making the device more convenient and having efficient deployment capabilities. The device can directly complete data acquisition, preprocessing, signal analysis, and classification and identification on-site. At the same time, through the adaptive frequency adjustment module, the device can automatically adjust the sampling frequency according to the characteristics of different vehicles to ensure that both low-frequency heavy vehicles and high-frequency small vehicles can accurately capture their dynamic load signals.

[0110] During the signal acquisition process, high-sensitivity sensors in the device continuously monitor the vibrations and accelerations when a vehicle passes by. The collected raw signals are filtered, denoised, and amplified through an integrated signal processing module to ensure signal clarity and accuracy. On this basis, the system performs frequency-domain conversion on the signals through a built-in spectrum analysis module, extracting key features such as the frequency and amplitude of the dynamic load. These features enable the system to accurately identify the types of vehicles and classify them according to their characteristics.

[0111] Data classification uses artificial intelligence algorithms, especially deep learning models (such as Convolutional Neural Network CNN or Recurrent Neural Network RNN). By learning the characteristics of large datasets, these algorithms enable the system to automatically identify different types of vehicles from complex signals and conduct subsequent analysis based on these identification results. For example, the dynamic load characteristics of vehicles are not only related to their weight but also closely related to factors such as driving speed, wheelbase, and road conditions. Therefore, the system can accurately classify the dynamic loads of vehicles based on these complex variables.

[0112] The system supports real-time generation of images from the analysis results, including time history diagrams, spectrograms, power spectral density diagrams, etc., to enable users to more intuitively understand the dynamic load characteristics of vehicles. These visualized data can not only help traffic management departments evaluate the dynamic load conditions of infrastructure such as roads and bridges but also serve as a basis for decision-making to optimize infrastructure design and maintenance strategies.

[0113] In addition, this device is equipped with a built-in wireless transmission module (such as Wi-Fi or Bluetooth), which can transmit the collected raw data and analysis results to external devices in real time, such as mobile devices or cloud platforms. Through remote devices, users can view the data analysis results at any time and place, and conduct data sharing and further processing. For traffic management departments, this means being able to monitor the data of multiple monitoring points in real time without geographical restrictions and make timely decisions and responses.

[0114] The device has high adaptability during data acquisition and analysis, and can provide real-time feedback and adjustment according to changes in the traffic environment and types of vehicles. For example, when the system detects different types of vehicles, it automatically adjusts the signal sampling frequency and data processing strategy to ensure complete capture of signal characteristics. At the same time, the system can identify and analyze complex dynamic load signals, classify the dynamic load characteristics of different vehicles through artificial intelligence algorithms, thereby improving the accuracy and efficiency of data processing. This adaptive ability enables the device to maintain high-efficiency and precise working performance in the face of complex traffic environments.

[0115] Specific Embodiment 2: This embodiment adopts a modular design. The system separates the functions of signal acquisition, data processing, and intelligent analysis, providing great flexibility and scalability. Each module can be deployed independently according to specific measurement requirements, facilitating installation and debugging in different environments. The signal acquisition module can transmit data to the data processing module via wireless or wired means. The latter is responsible for preprocessing the signal and extracting features. Finally, the intelligent analysis module identifies and classifies the collected data. This modular design enables each functional module to be customized according to different requirements, meeting the needs of diverse traffic monitoring scenarios.

[0116] For example, at specific locations such as bridges and tunnels, the signal acquisition module can be flexibly installed in different positions to ensure that the dynamic load signals of vehicles can be captured to the maximum extent. The signal processing module can be remotely set to uniformly manage the data processing tasks of multiple acquisition points. This design not only improves the flexibility of the equipment but also simplifies the system maintenance and upgrade process. The independence and adjustability between modules enable the equipment to adapt to changing environmental requirements, optimizing the convenience and efficiency of the installation process.

[0117] Different from traditional fixed-frequency sampling systems, the equipment in this embodiment can automatically adjust the sampling frequency according to the dynamic load characteristics of vehicles through a built-in adaptive frequency adjustment module. For signals with different frequency components, the system automatically selects an appropriate frequency range for acquisition to ensure that all important frequency components can be fully captured. For example, for heavy trucks, the system will reduce the sampling frequency to avoid excessive data redundancy; for light vehicles, the system will increase the sampling frequency to ensure the capture of high-frequency dynamic load characteristics.

[0118] After signal acquisition, the data is filtered, denoised, and signal-amplified through the preprocessing module. During this process, the system uses an adaptive filtering method to select the most suitable filter according to the signal characteristics, removing noise and enhancing the effective part of the signal. For the high-frequency part of the dynamic load signal, the system will use high-precision spectrum analysis technology to extract key features. These features include not only frequency information but also amplitude, phase, and other information. By extracting these features, the system can assist subsequent classification and identification tasks.

[0119] One of the greatest advantages of modular equipment is its support for wireless data transmission. The signal acquisition module transmits the collected data to the remote data processing module in real time via a wireless communication protocol (such as Wi-Fi, Bluetooth, etc.) and generates analysis results. These analysis results can not only be stored on local devices but also be uploaded to the cloud platform wirelessly. The cloud platform supports real-time synchronization of multi-point data, allowing centralized management of monitoring data from multiple monitoring points, greatly improving the efficiency of data processing.

[0120] The traffic management department can access this data in real time through mobile devices or computers for remote monitoring and management. Users can obtain multi-point data without having to operate on-site, effectively reducing the workload of maintenance personnel. The real-time data transmission also enables managers to respond quickly in case of emergencies. For example, if abnormal loads are detected on a certain section of the road, the system will immediately send an alarm signal, and managers can make quick decisions based on the real-time data.

[0121] The modular design of the device enables it to adapt to different traffic environments. Whether it is a bridge, tunnel, road or railway, the system can be flexibly configured according to different scenarios to optimize the position and method of signal acquisition. The interconnection and interoperability between modules ensure the real-time transmission and analysis of data. The system also has an adaptive ability, which can automatically adjust the sampling frequency and data processing method according to the signal changes and different traffic environments, thus improving the intelligence and automation level of the monitoring system.

[0122] Specific implementation method 3: Based on artificial intelligence and big data analysis technologies, this implementation method constructs an intelligent traffic load monitoring system. The system arranges multiple sensors on different traffic facilities, such as key positions of bridges, tunnels, railways, etc. The dynamic load signals collected by the sensors are transmitted to the cloud platform in real time. The cloud platform not only has powerful data processing capabilities, but also can conduct joint analysis with external factors such as historical data, traffic flow, and weather, thus forming an all-round traffic load analysis platform. Through artificial intelligence algorithms, the system can extract valuable information from the massive data, identify the influence patterns of different transportation tools on the infrastructure, and then conduct more accurate load analysis and prediction.

[0123] The core application of artificial intelligence in this implementation method is the intelligent

[0124] recognition and classification of signals. The system can efficiently analyze and classify the collected signals through deep learning models, especially convolutional neural networks (CNNs) and recurrent neural networks (RNNs). After the signal data is transmitted to the cloud platform, the system will first perform spectral analysis on the signal to extract features such as frequency and amplitude, and then input these features into the trained deep learning model. The learned model can automatically identify different types of transportation tools (such as cars, trucks, trains, etc.) from the complex signals and accurately classify them. Based on these recognition results, the system can further analyze the impact of the dynamic loads of transportation tools on facilities such as roads, bridges, and tunnels, helping relevant departments to evaluate the operating status of the infrastructure in real time.

[0125] Through the big data platform, the system can not only analyze the dynamic load conditions of current transportation vehicles in real time, but also make predictions based on historical data and trends. The cloud platform integrates multiple information sources such as historical load data, traffic flow, road maintenance records, and weather data. Through data mining and prediction models, the system can predict the load change trend within a certain period in the future and identify potential infrastructure damage risks in advance. This prediction function provides more accurate decision-making support for traffic management departments, helping them allocate loads and maintain facilities in advance during traffic peaks, special weather conditions, etc.

[0126] The system has a real-time feedback function, which can feedback the analysis results and prediction data to traffic management personnel in real time. Through terminals such as mobile devices and computers, users can view real-time data, prediction results, and analysis charts at any time. The system also has an intelligent alarm function. When the dynamic load of a certain road or bridge is detected to exceed the set threshold, the system will automatically send an alarm to the management personnel, prompting them to take measures immediately. This real-time feedback mechanism greatly improves the efficiency of traffic management, helping the management department adjust traffic flow or arrange road maintenance in a timely manner to ensure the safe and stable operation of the infrastructure.

[0127] Through integrated design and adaptive acquisition, identification, and classification methods, the present invention effectively improves the accuracy and efficiency of dynamic load monitoring of transportation infrastructure. The device can automatically adjust the sampling frequency to adapt to the dynamic load characteristics of different types of transportation vehicles, avoiding data inaccuracy or redundancy problems caused by improper frequency sampling in traditional systems. Through artificial intelligence algorithms to intelligently classify the collected signals, the system can accurately distinguish the dynamic loads generated by different transportation vehicles and generate visual images in real time, providing reliable data support for infrastructure design, maintenance, and safety assessment. In addition, the modular design and wireless data transmission function improve the deployment flexibility and data processing efficiency of the system, enabling traffic management departments to conduct real-time monitoring and decision-making more efficiently.

[0128] The present invention realizes the intelligence and automation of traffic load monitoring. By using deep learning algorithms to automatically identify the types of transportation vehicles and evaluate their impacts on the infrastructure, it significantly improves the data processing speed and accuracy. The intelligent classification and analysis system makes the monitoring process completely automated, reducing the need for manual intervention and improving the operation convenience. The system can not only monitor the current traffic load conditions in real time, but also analyze the future load change trend based on big data and prediction models, helping traffic management departments conduct preventive maintenance and make scientific decisions, thereby improving traffic safety, management efficiency, and effectively avoiding infrastructure damage and traffic accidents.

[0129] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0130] Only some exemplary embodiments of the present invention have been described above by way of illustration. Without doubt, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0131] It should be noted that in this text, if there are relational terms such as first and second, they 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 term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of another identical element in the process, method, article or device comprising the element.

[0132] It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not mean the order of execution is prior or subsequent. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0133] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this text can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0134] Those skilled in the art can clearly understand that for the sake of convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0135] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed across multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] In addition, in each embodiment of the present application, each functional unit may be integrated in a processing unit, may exist physically separately for each unit, or two or more units may be integrated in one unit.

[0137] As mentioned above, only the specific implementation manners of the present application are described, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all of them should be covered by the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

[0138] Only some exemplary embodiments of the present invention have been described above by way of illustration. Undoubtedly, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

Claims

1. The adaptive acquisition, identification and classification method based on vehicle dynamic load frequency is characterized by: The following steps are involved: Vibration and acceleration signals generated during driving are collected in real time through multiple high-sensitivity sensors integrated in the equipment; According to the frequency characteristics of the collected signals, the built-in adaptive frequency adjustment module automatically adjusts the sampling frequency and data processing method to ensure that the dynamic load characteristics generated by different types of vehicles can be accurately captured; Preprocess the collected original signals and extract the key features of the signals through spectrum analysis and other methods for subsequent classification and identification; Using the trained deep learning algorithm, the processed signal features are analyzed, the type of vehicle is identified based on the extracted feature data, and the dynamic load characteristics are classified; Based on the classification results, the algorithm generates multi-dimensional visualization images to display the dynamic load characteristics of the vehicle and its impact on the infrastructure in real time, and transmits the analysis results to external devices through the wireless transmission module.

2. The adaptive acquisition, identification and classification method based on vehicle dynamic load frequency according to claim 1 is characterized in that: The specific steps for collecting vibration and acceleration signals generated during driving in real time through multiple high-sensitivity sensors integrated in the equipment are as follows: The equipment needs to be installed at key locations along the path of vehicles, and the height and angle should be adjustable to ensure that the sensor can accurately capture dynamic load signals; Start the equipment and perform a self-test to ensure that all components are functioning properly and ready for data collection; The sensor collects the vibration and acceleration signals generated by the passing vehicles in real time, and records the amplitude, duration and frequency characteristics of the dynamic load; The collected data is uploaded to external devices in real time via the wireless transmission module for engineers to monitor and analyze.

3. The adaptive acquisition, identification and classification method based on vehicle dynamic load frequency according to claim 1 is characterized in that: According to the frequency characteristics of the collected signals, the sampling frequency and data processing method are automatically adjusted through the built-in adaptive frequency adjustment module to ensure that the dynamic load characteristics generated by different types of vehicles can be accurately captured. The specific steps are as follows: Identify the frequency characteristics of signals generated by vehicles through spectrum analysis to provide basic data for adaptive sampling; According to the dynamic load frequency characteristics of different vehicles, the sampling frequency is automatically adjusted to ensure data accuracy; Dynamically optimize data processing methods based on collected signal characteristics to ensure effective signal capture and noise removal; The sampling frequency and data processing method are continuously adjusted through a real-time feedback mechanism to ensure that they adapt to the dynamic load characteristics of different types of vehicles.

4. The adaptive acquisition, identification and classification method based on vehicle dynamic load frequency according to claim 1 is characterized in that: The specific steps for preprocessing the collected original signal and extracting the key features of the signal through methods such as spectrum analysis for subsequent classification and identification are as follows: Remove noise and interference through filters to retain useful dynamic load signals to improve data quality and accuracy; The signal amplitude is enhanced through the gain amplifier to ensure the clarity and distinguishability of the signal in subsequent processing; Use spectrum analysis methods to extract the main frequency components in the signal to provide key features for subsequent classification and identification; The extracted features are calibrated with the vehicle type and data input is prepared for subsequent classification and recognition.

5. The adaptive acquisition, identification and classification method based on vehicle dynamic load frequency according to claim 1 is characterized in that: The specific steps of analyzing the processed signal features using the trained deep learning algorithm, identifying the type of vehicle based on the extracted feature data, and classifying the dynamic load characteristics are as follows: First, the processed signal features are input into the trained deep learning model for feature mapping. A convolutional neural network is used to extract and map features. The model gradually extracts higher-level features through multi-layer convolution and pooling operations. Let the input feature matrix be X = [x i ]=[x1,x2,……,x n ], where x i represents the i-th signal feature, n is the total number, and the signal feature has been filtered, denoised and spectral analyzed. The convolution layer in the model is operated by the following formula: f conv (X)=σ(W·X+b) In the formula, f conv (X) represents the output result of the convolution operation, σ is the activation function, W is the convolution kernel, X is the input signal feature matrix, and b is the bias term; Through multi-layer convolution and pooling operations, a high-dimensional feature vector Z is finally obtained, in which each element represents a complex dynamic load characteristic; After the deep learning model extracts the high-dimensional feature vector Z, the high-dimensional feature vector Z is input into the fully connected layer for classification, and then the type of vehicle is identified and the dynamic load characteristics are classified. Through the trained classification network, the model will output the classification result. The result is obtained through the softmax function to obtain the classification probability of each vehicle, and the softmax function is used for normalization. The calculation formula is as follows: In the formula, y p is the model's predicted probability for the pth type of transportation, m is the total number of types of transportation, is the exponential eigenvalue of each category.

6. The adaptive acquisition, identification and classification method based on vehicle dynamic load frequency according to claim 1 is characterized in that: Based on the classification results, the algorithm generates a multi-dimensional visualization image to display the dynamic load characteristics of the vehicle and its impact on the infrastructure in real time, and transmits the analysis results to external devices through the wireless transmission module. The specific steps are as follows: After signal preprocessing and feature extraction, a time history diagram is first generated to show the dynamic response of the dynamic load signal of the vehicle over time. The displacement response is calculated by integrating the collected acceleration signal to reflect the impact of the vehicle on the infrastructure. The displacement response is calculated using the following formula: Where x(t) is the displacement response, a(t) is the acceleration signal, and t is the time variable; After obtaining the time history diagram and calculating the displacement response, the next step is to generate a spectrum diagram to analyze the distribution of the signal at different frequencies. The time domain signal is converted into a frequency domain signal using a fast Fourier transform and the spectrum diagram is calculated. The spectrum calculation formula is: Where X(f) is the frequency spectrum component with frequency f, e is the natural base, x(t) is the displacement signal in the time history diagram, j is the imaginary unit, f is the frequency variable, and 2π is a mathematical constant; Finally, based on the results of the spectrum diagram, a power spectrum density diagram is generated. The power spectrum density describes the power distribution within a unit frequency bandwidth and determines the frequency range where the energy of the dynamic load is concentrated. The power spectrum density is calculated using the following formula: Where P(f) is the power spectral density and T is the total duration of the signal.

Images