Traffic signal machine flow analysis device and analysis method based on artificial intelligence, and storage medium

Abstract

The embodiment of the invention relates to the technical field of traffic, and discloses a traffic signal machine flow analysis device and analysis method based on artificial intelligence, and a storage medium, and the method comprises the steps: firstly obtaining an initial vibration curve formed by the collection data of a vibration membrane sensor of a target traffic road section, and carrying out the compensation and correction of the initial vibration curve based on environment parameters, and obtaining a target vibration curve. And then key features of a target vibration curve are identified and extracted, a plurality of candidate vibration data sets containing information such as vibration amplitude and corresponding duration are formed, real-time traffic flow is preliminarily estimated according to the candidate vibration data sets, and a traffic flow estimated value is obtained. If the estimated value is greater than or equal to a preset flow threshold value, triggering a correction instruction, controlling a cloud camera communicating with a cloud server to start, and correcting the estimated value to obtain a target value; and if the estimated value is smaller than the threshold value, directly taking the estimated value as a target value. According to the scheme, vibration data and visual correction are combined, so that the method can adapt to various traffic flow analysis scenes, and accurate analysis results can be obtained in different scenes.

Description

Technical Field

[0001] The present invention relates to the field of traffic technology, and in particular to an artificial intelligence-based traffic signal flow analysis device, analysis method and storage medium. Background Art

[0002] With the rapid development of intelligent transportation systems, the flow analysis accuracy of traffic signals directly affects road traffic efficiency and congestion management level.

[0003] In related technologies, traffic signal systems mainly analyze traffic flow through cameras or ground sensing coils. However, using ground sensing coils for traffic flow analysis has the problem of difficulty and high replacement cost of coil sensors; and using cameras for traffic flow analysis is prone to technical problems such as a sudden drop in recognition accuracy under complex conditions such as network instability. Summary of the Invention

[0004] The main purpose of the present invention is to provide a traffic signal flow analysis device, analysis method and storage medium based on artificial intelligence, aiming to solve the technical problem that traffic flow analysis devices in the prior art are difficult to balance maintenance costs and recognition accuracy.

[0005] To achieve the above objectives, in a first aspect, embodiments of the present application provide an artificial intelligence-based traffic signal flow analysis device, comprising a memory and a processor, wherein the memory stores executable program code, and the processor is configured to implement the following steps of a traffic flow analysis method when calling and running the executable program code: obtaining an initial vibration curve formed by collecting data from a vibrating membrane sensor, the vibrating membrane sensor being located at a target traffic section; Performing environmental factor compensation correction on the initial vibration curve based on environmental parameters to obtain a target vibration curve; Identifying and extracting key features of the target vibration curve to obtain a plurality of candidate vibration data sets, wherein the candidate vibration data sets include at least vibration amplitudes and corresponding vibration durations; Performing a preliminary estimation of the real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimation value; When the traffic flow estimate is greater than or equal to the traffic threshold, a traffic flow correction instruction is triggered, and the cloud camera is controlled to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value, wherein the cloud camera is communicatively connected to the cloud server.

[0006] In one possible implementation, the vibration membrane sensor includes a vehicle crushing and lifting module arranged on the road surface of the target traffic section, a force transmission rod arranged below the vehicle crushing and lifting module, and an elastic body arranged on the side of the road, wherein a vibration membrane is provided in the elastic body; wherein, when the vehicle is located above the vehicle crushing and lifting module, the vehicle crushing and lifting module sinks and drives the force transmission rod to swing, and when the force transmission rod swings, it squeezes the elastic body so that the vibration membrane generates a vibration signal due to air compression.

[0007] In a possible implementation, the vehicle crushing lifting module includes a liftable pressure plate and a return spring, wherein the return spring is used to restore the pressure plate to its initial position after the vehicle leaves, and the initial position of the pressure plate is flush with the road surface or higher than the ground.

[0008] In a possible implementation, performing environmental factor compensation correction on the initial vibration curve based on the environmental parameters to obtain the target vibration curve includes: Acquiring real-time meteorological data collected by a meteorological sensor, wherein the meteorological sensor includes at least one of a wind sensor and a rain sensor; When the rainfall meteorological parameter and / or the wind speed meteorological parameter in the meteorological data exceeds a preset meteorological parameter threshold, a pre-trained environmental compensation neural network is called to perform environmental factor compensation correction on the initial vibration curve to obtain a target vibration curve.

[0009] In a possible implementation, identifying and extracting key features of the target vibration curve to obtain multiple candidate vibration data sets includes: Extracting closed areas from the target vibration curve based on image analysis technology to obtain multiple closed graphs, wherein each closed graph corresponds to a vibration event; The maximum vibration amplitude, amplitude integral area, and vibration duration in each closed graph are extracted to obtain multiple candidate vibration data sets.

[0010] In a possible implementation, performing a preliminary estimation of the real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimate value includes: Inputting the multiple candidate vibration data sets into a pre-trained lightweight convolutional network for feature screening, retaining key features related to target vehicle traffic flow to obtain multiple target vibration data sets; A preliminary estimate of the real-time traffic flow is performed based on the multiple target vibration data sets to obtain a traffic flow estimate value.

[0011] In a possible implementation, performing a preliminary estimation of the real-time traffic flow according to the multiple target vibration data sets to obtain a traffic flow estimation value includes: When the maximum vibration amplitude is less than the amplitude threshold, performing a preliminary traffic flow estimation in a first estimation method to obtain a traffic flow estimation value; When the maximum vibration amplitude is greater than or equal to the amplitude threshold, a second estimation method is used to perform a preliminary estimation of traffic flow to obtain a traffic flow estimation value, wherein the second estimation method is to perform a preliminary estimation of traffic flow based on the amplitude integral area and the vibration duration, the amplitude integral area is positively correlated with the traffic flow, and / or the vibration duration is positively correlated with the traffic flow.

[0012] In one possible implementation, controlling the cloud camera to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value includes: Controlling the cloud camera to start according to the traffic flow correction instruction; After the cloud camera is started, video data collected by the cloud camera is obtained in real time; Extracting a target video frame corresponding to the time of the target vibration dataset; Performing vehicle feature recognition on the target video frame to obtain a vehicle quantity value; The traffic flow estimate is corrected according to the vehicle quantity value to obtain a traffic flow target value.

[0013] In a second aspect, an embodiment of the present application further provides a traffic flow analysis method, which is applied to a traffic signal flow analysis device, and the method includes: obtaining an initial vibration curve formed by collecting data from a vibrating membrane sensor, the vibrating membrane sensor being located at a target traffic section; Performing environmental factor compensation correction on the initial vibration curve based on environmental parameters to obtain a target vibration curve; Identifying and extracting key features of the target vibration curve to obtain a plurality of candidate vibration data sets, wherein the candidate vibration data sets include at least vibration amplitudes and corresponding vibration durations; Performing a preliminary estimation of the real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimation value; When the traffic flow estimate is greater than or equal to the traffic threshold, a traffic flow correction instruction is triggered, and the cloud camera is controlled to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value, wherein the cloud camera is communicatively connected to the cloud server.

[0014] In a third aspect, an embodiment of the present application further provides a storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in the second aspect are implemented.

[0015] Different from the prior art, the artificial intelligence-based traffic signal flow analysis device provided in the embodiment of the present application is composed of a memory and a processor. The memory stores executable program code. When the processor calls and runs the code, it first obtains the initial vibration curve formed by the data collected by the vibration membrane sensor of the target traffic section, and then compensates the initial vibration curve for environmental factors based on environmental parameters to obtain the target vibration curve. Then, the key features of the target vibration curve are identified and extracted to form multiple candidate vibration data sets containing information such as vibration amplitude and corresponding vibration duration. Based on these candidate vibration data sets, the real-time traffic flow is preliminarily estimated to obtain a traffic flow estimate value. If the estimate value is greater than or equal to a preset traffic threshold, a traffic flow correction instruction is triggered, and the cloud camera connected to the cloud server is controlled to start according to this correction instruction to correct the traffic flow estimate value, and finally the traffic flow target value is obtained.

[0016] In scenarios with low traffic volume, diaphragm sensors can independently and accurately perform flow estimation tasks. Since traffic conditions are relatively simple, the vibration characteristics generated by passing vehicles are relatively simple and easy to identify. With their high sensitivity and stability, diaphragm sensors can accurately capture vehicle vibration information, thereby achieving highly accurate flow estimation. Furthermore, diaphragm sensors have a relatively simple structure and low installation and maintenance costs. They do not require large-scale infrastructure construction and complex post-maintenance work, and can meet traffic flow monitoring needs at a low cost.

[0017] In scenarios with heavy traffic, the vibration membrane sensor is used to estimate traffic flow, and the traffic estimate is then corrected again using cloud camera video data. The vibration data and image data verify each other and complement each other, greatly improving the accuracy of traffic flow analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0019] Figure 1 This is a schematic structural diagram of a traffic signal flow analysis device in some embodiments of the present application; Figure 2 This is a flow chart of a traffic flow analysis method in some embodiments of the present application; Figure 3 This is a graphical diagram of a target vibration curve in some embodiments of the present application; Figure 4This is a flow chart of step S200 of the traffic flow analysis method in some embodiments of the present application; Figure 5 This is a flow chart of step S300 of the traffic flow analysis method in some embodiments of the present application; Figure 6 This is a flow chart of step S400 of the traffic flow analysis method in some embodiments of the present application.

[0020] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0023] In addition, the descriptions of "first", "second", etc. in the present invention are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, "and / or" in the full text includes three solutions. Taking A and / or B as an example, it includes technical solution A, technical solution B, and technical solution that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in this field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0024] With the rapid development of intelligent transportation systems, the flow analysis accuracy of traffic signals directly affects road traffic efficiency and congestion management level.

[0025] In related technologies, traffic signal systems mainly analyze traffic flow through cameras or ground sensing coils. However, using ground sensing coils for traffic flow analysis has the problem of difficulty and high replacement cost of coil sensors; and using cameras for traffic flow analysis is prone to technical problems such as a sudden drop in recognition accuracy under complex conditions such as network instability.

[0026] like Figure 1 As shown, the traffic signal flow analysis device of the present application includes a diaphragm sensor 100, a cloud camera 200, a memory 300, and a processor 400. The diaphragm sensor 100 includes a vehicle-crushing and lifting module 110 located on the road surface of the target traffic section, a force transmission rod 120 located below the vehicle-crushing and lifting module 110 (the lower middle portion of the force transmission rod 120 is rotatably connected to a support surface), and an elastic body 130 located on the roadside. A diaphragm 140 is disposed within the elastic body 130. When a vehicle is located above the vehicle-crushing and lifting module 110, the module 110 sinks, causing the force transmission rod 120 to swing. The swinging force transmission rod compresses the elastic body, causing the diaphragm 140 to generate a vibration signal due to air compression. The processor 400 processes the vibration signal from the diaphragm 140 to generate a vibration curve. Furthermore, the greater the vehicle mass, the greater the vibration amplitude sensed by the diaphragm 140.

[0027] The vehicle crushing lifting module 110 includes a liftable pressure plate 111 and a return spring 112 . The return spring 112 is used to restore the pressure plate to its initial position after the vehicle leaves. The initial position of the pressure plate is flush with the road surface or higher than the ground.

[0028] It should be noted that Figure 1 Region A is the road surface area of ​​the target traffic section, and region B is the roadside area of ​​the target traffic section. Therefore, when the relatively fragile elastic body 130 of the diaphragm sensor 100 fails or reaches the maintenance cycle and requires maintenance, workers do not need to perform destructive operations on the road surface; they only need to replace the elastic body 130 in the roadside area. The entire maintenance process does not interfere with normal traffic on the target traffic section, thereby reducing maintenance costs.

[0029] The cloud camera 200 of the present application is located above the road surface, for example, above the traffic signal rack of the target traffic section, and is used to capture video or images of the target traffic section to assist in traffic flow analysis. The cloud camera 200 is connected to a cloud server in communication, and uses the cloud server's high-performance computing resources to achieve rapid feature recognition of video images, thereby improving the accuracy and timeliness of traffic flow analysis. In one scenario, the cloud server can send the recognition results to a local processor for comprehensive analysis of traffic flow.

[0030] In scenarios with low traffic flow, since the traffic conditions are relatively simple at this time, the vibration characteristics generated by the passing of vehicles are relatively simple and easy to identify. The vibrating membrane sensor, with its high sensitivity and stability, can accurately capture the vehicle vibration information, thereby achieving high-precision traffic flow estimation. In scenarios with large traffic flow, the traffic conditions are relatively complex, and it is impossible to accurately analyze the traffic flow by relying solely on the vibrating membrane sensor. Based on this, the present application only uses the vibrating membrane sensor to analyze the traffic flow in scenarios with low traffic flow, which not only ensures the accuracy of the flow analysis, but also reduces maintenance costs and avoids traffic interference. In scenarios with large traffic flow, while using the vibrating membrane sensor to estimate the flow, the cloud camera video data is combined to perform a secondary correction on the flow estimate value. The vibration data and image data verify each other and complement each other's advantages, which greatly improves the accuracy of the traffic flow analysis. In addition, the traffic flow analysis device of the present application does not need to use the camera for traffic analysis throughout the entire process, which can reduce the frequency of camera use, thereby improving the overall accuracy of traffic flow analysis.

[0031] like Figures 1-6 As shown, the following uses an artificial intelligence-based traffic signal flow analysis device to perform the traffic flow analysis method as an example. It should be noted that although the logical order is shown in the flow chart, in some cases, the steps shown or described can be performed in a different order than here. Figure 2 The method includes the following steps S100 to S500: Step S100: obtaining an initial vibration curve formed by collecting data from a vibration membrane sensor, wherein the vibration membrane sensor is located at a target traffic section; Specifically, the appropriate vibrating membrane sensor can be selected based on the actual conditions of the target traffic section, such as traffic volume, vehicle type distribution, and road surface material. For example, on a road section with a high volume of large trucks, a sensor with high sensitivity and a large measurement range should be selected to ensure that the strong vibrations generated by the passing vehicles can be accurately captured.

[0032] In an embodiment of the present application, when a vehicle passes through the vehicle crushing lifting module, the processor can process the vibration signal of the vibrating diaphragm to obtain an initial vibration curve. This vibration signal contains multi-dimensional information such as the vibration amplitude, frequency, and duration generated when the vehicle passes. The processor uses a specific signal processing algorithm, such as the Fourier transform algorithm for analyzing the frequency components of the signal, the wavelet transform algorithm for extracting the characteristics of the signal at different time scales, etc., to perform a series of processing operations such as filtering and feature extraction on these vibration signals, and finally plots the processed data into an initial vibration curve with time as the horizontal coordinate and vibration amplitude as the vertical coordinate. The initial vibration curve can intuitively and accurately reflect the dynamic changes in the vibration of the vibrating diaphragm when the vehicle passes, and provide basic data support for subsequent traffic flow analysis based on vibration data.

[0033] Step S200: performing environmental factor compensation correction on the initial vibration curve based on environmental parameters to obtain a target vibration curve; It is understandable that during the generation of the initial vibration curve, only basic operations such as filtering are performed, and its accuracy is insufficient to fully meet the high data accuracy requirements of subsequent traffic flow analysis. In view of this, in the embodiment of the present application, after obtaining the initial vibration curve, it is necessary to further combine environmental parameters to perform environmental factor compensation correction on the initial vibration curve.

[0034] For example, we can comprehensively consider the impact of various environmental factors such as temperature, humidity, wind speed, and road conditions on the vibration signal, use corresponding mathematical models and algorithms, and accurately adjust the initial vibration curve to obtain a target vibration curve that can more realistically reflect the vibration conditions when the vehicle passes, providing a reliable data basis for subsequent accurate traffic flow analysis.

[0035] In one embodiment, the step S200 of performing environmental factor compensation correction on the initial vibration curve based on environmental parameters to obtain a target vibration curve includes: S210: Acquire real-time meteorological data collected by a meteorological sensor, where the meteorological sensor includes at least one of a wind sensor and a rain sensor; S220: When the rainfall meteorological parameter and / or the wind speed meteorological parameter in the meteorological data exceeds a preset meteorological parameter threshold, calling a pre-trained environmental compensation neural network to perform environmental factor compensation correction on the initial vibration curve.

[0036] In actual traffic scenarios, meteorological conditions (such as rainfall, wind speed, etc.) can interfere with the vibration signals collected by the diaphragm sensor to varying degrees. Therefore, the present embodiment evaluates whether the initial vibration curve needs to be corrected based on the meteorological conditions.

[0037] In the embodiments of this application, when rainfall or wind speed is within a relatively low range, the external environment's interference with the vibration signal is relatively limited. Under these relatively stable environmental conditions, the initial vibration curve acquired by the diaphragm sensor and processed through simple filtering and other operations is accurate enough to meet basic traffic analysis requirements. Therefore, to improve processing efficiency and reduce unnecessary computing resource consumption, the initial vibration curve is directly used as the target vibration curve for subsequent traffic flow analysis and other tasks.

[0038] However, when rainfall or wind speed exceeds a certain threshold and reaches a significant level, the external environment significantly increases interference with the vibration signal. For example, heavy rainfall increases road surface moisture and changes friction, while raindrops directly impact the road surface, generating additional vibration interference. Strong winds can cause the sensor itself or its mounting structure to shake, or air flow can affect the propagation of vibrations generated by the vehicle's movement. This can lead to a significant deviation between the initial vibration curve and the actual vehicle vibration, significantly reducing its accuracy. In this case, directly using the initial vibration curve for analysis will seriously affect the reliability of results such as traffic flow analysis.

[0039] To address this issue, an embodiment of the present application pre-trains an environmental compensation neural network. This neural network is trained based on a large amount of historical meteorological data, initial vibration curves, and real target vibration curves that have been manually calibrated. It can learn the changing patterns of vibration signals under different meteorological conditions and the corresponding compensation amounts. When high rainfall or wind speed is detected, the system automatically calls the pre-trained environmental compensation neural network. The current meteorological data (including rainfall and wind speed, etc.) and the corresponding initial vibration curve are used as input. The neural network outputs the environmental factor compensation correction amount after a series of complex calculations and analyses. Subsequently, the compensation correction amount is applied to the initial vibration curve, and the initial vibration curve is corrected through specific mathematical operations (such as addition, multiplication, etc., the specific operation method is determined according to the training objectives of the neural network and the actual situation), thereby obtaining a target vibration curve that can more accurately reflect the actual vibration conditions when the vehicle passes.

[0040] For example, Figure 3The figure shows the target vibration curve obtained after environmental compensation and correction. In this graph, the horizontal axis is defined as the time axis, which accurately quantifies the evolution of the vibration signal over time; the vertical axis is defined as the vibration amplitude, which quantifies the intensity of the vibration signal. Analysis of the vibration waveform characteristics shows that the vibration amplitude graphs S1, S3, and S4 exhibit a single-peak waveform. A single-peak waveform means that the vibration signal has only one significant maximum point within the corresponding time interval, which generally indicates that only one vehicle passed through the monitoring area during that time period (considering only the number of peaks, not the amplitude). The vibration amplitude graph S2, on the other hand, exhibits a multi-peak waveform. A multi-peak waveform indicates that the vibration signal has multiple significant maximum points within the corresponding time interval, indicating that multiple vehicles passed through the monitoring area in sequence. The passage of each vehicle will stimulate an independent vibration response peak in the vibration signal. The passage of multiple vehicles causes these independent vibration response peaks to superimpose on each other on the time axis, ultimately forming a complex multi-peak waveform.

[0041] Specifically, meteorological sensors can be installed at appropriate locations around the target traffic section. These locations should accurately reflect the microclimate of the road section and avoid obstruction by buildings, trees, and other structures. For example, wind sensors can be mounted on top of roadside light poles in open areas, at a certain height (e.g., 3-5 meters) above the ground to ensure accurate wind speed measurement. Rain sensors can be installed in areas that can adequately receive rainfall, such as on top of streetlights or on specially constructed rainfall measurement brackets. Meteorological sensors establish a communication connection with the processor via a wired (e.g., RS485 bus) or wireless (e.g., LoRa or ZigBee) connection. During the connection process, parameter configuration, including sensor address and communication baud rate, is required to ensure stable and accurate data transmission.

[0042] Meteorological sensors collect meteorological data in real time at a preset sampling frequency (e.g., once per minute). Wind sensors measure wind speed, while rain sensors measure rainfall. For wind speed data, a sliding average filter algorithm can be used to average the wind speed value over several consecutive sampling points to eliminate the effects of random noise. For rainfall data, the data can be checked to ensure it is within a reasonable range (e.g., excluding negative or excessively large values). If any abnormal data is present, interpolation correction can be performed using data from adjacent time points.

[0043] During the pre-training process of the environmental compensation neural network, a large amount of historical meteorological data, the corresponding initial vibration curves, and the manually calibrated target vibration curves are collected as a training dataset. The meteorological data serves as input features, and the difference between the initial and target vibration curves serves as labels. An appropriate neural network architecture is then selected, such as a multi-layer perceptron (MLP), convolutional neural network (CNN), or recurrent neural network (RNN) and its variants (such as LSTM and GRU). The neural network is trained using the training dataset, employing a backpropagation algorithm and an optimizer (such as Adam) to adjust the weights and biases of the neural network so that the output is as close as possible to the actual compensation value. During the training process, the dataset is divided into training, validation, and test sets. The validation set is used to monitor the performance of the neural network to prevent overfitting. Finally, the accuracy and generalization ability of the neural network are evaluated on the test set.

[0044] Step S300: identifying and extracting key features of the target vibration curve to obtain multiple candidate vibration data sets, wherein the candidate vibration data sets at least include vibration amplitudes and corresponding vibration durations; In one embodiment, the step S300 of identifying and extracting key features of the target vibration curve to obtain multiple candidate vibration data sets includes: S310, extracting closed areas from the target vibration curve based on image analysis technology to obtain multiple closed graphs, wherein each closed graph corresponds to a vibration event; S320 , extracting the maximum vibration amplitude, amplitude integral area, and vibration duration in each closed graph to obtain multiple candidate vibration data sets.

[0045] Specifically, the embodiment of the present application can first extract the closed area in the target vibration curve according to the image analysis technology to obtain multiple closed graphics, such as Figure 3 The closed graphs S1, S2, S3, and S4 in the image are then extracted to obtain multiple candidate vibration data sets.

[0046] For example, a candidate vibration data set (H1, S H 1, T1), a candidate vibration data set (H2, S H 2, T2), a candidate vibration data set (H3, S H 3, T3), a candidate vibration data set (Hn, S H n, Tn). Among them, Hn is the vibration amplitude, S H n is the amplitude integrated area, and Tn is the vibration duration.

[0047] Step S400: Preliminarily estimating the real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimation value; In traffic flow monitoring scenarios, by extracting key features from the target vibration curve, multiple candidate vibration datasets can be obtained, each corresponding to a vibration event. However, not all vibration events meet the specific analysis requirements of the system or management personnel. For example, if the system's primary task is to count the traffic flow of small vehicles passing through the target traffic section, these candidate vibration datasets may contain a large amount of vibration event data unrelated to small vehicles, such as vibration data caused by large vehicles, non-motorized vehicles, and even pedestrians.

[0048] To accurately obtain the information that meets the requirements, these candidate vibration datasets must be screened and filtered. Specifically, a set of targeted screening rules must be established based on prior knowledge of the vibration characteristics of small vehicles. These rules may involve the range of vibration amplitude, the interval of vibration duration, and the threshold of the amplitude-integrated area. For example, due to the weight and structural characteristics of small vehicles, the vibration amplitude and duration generated are generally relatively small, and the amplitude-integrated area also has a specific range.

[0049] Based on this, in one embodiment, step S400: performing a preliminary estimation of the real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimation value, includes: S410, inputting the multiple candidate vibration data sets into a pre-trained lightweight convolutional network for feature screening, retaining key features related to target vehicle traffic flow to obtain multiple target vibration data sets; S420 : Preliminarily estimating the real-time traffic flow according to the multiple target vibration data sets to obtain a traffic flow estimation value.

[0050] Specifically, multiple candidate vibration datasets are first fed into a pre-trained lightweight convolutional network for feature screening. Key features relevant to target vehicle traffic flow are retained to generate multiple target vibration datasets. For example, target vibration datasets corresponding to cars under 1.5 tons are selected and retained to analyze traffic flow for cars under 1.5 tons. A preliminary estimate of real-time traffic flow is then generated based on these multiple target vibration datasets to generate traffic flow estimates.

[0051] In the process of conducting a preliminary estimation of real-time traffic flow based on the target vibration data set to obtain the traffic flow estimate, in order to improve the accuracy and adaptability of the estimation, a variety of estimation methods can be comprehensively adopted and flexibly selected according to different vibration characteristic conditions.

[0052] Specifically, an amplitude threshold can be set first. This threshold is determined based on an analysis of the vibration characteristics of different types of vehicles. When the maximum vibration amplitude recorded in the target vibration data set is less than this amplitude threshold, it means that the vibration event may have been generated by a single vehicle, and its vibration characteristics are relatively weak. In this case, the first estimation method is used to make a preliminary estimate of traffic flow. The first estimation method is relatively simple and direct, that is, the vibration event is counted as only 1, indicating that one traffic unit has passed, and this count value is used as the estimated traffic flow value at the current moment.

[0053] If the maximum vibration amplitude in the target vibration dataset is greater than or equal to the amplitude threshold, it indicates that the vibration event was likely generated by multiple vehicles and has significant vibration characteristics. In this case, the second estimation method is used to make a preliminary estimate of traffic flow. This second estimation method comprehensively considers two key characteristics: the amplitude integral area and the vibration duration.

[0054] The amplitude-integrated area reflects the total energy of a vibration event and has a positive correlation with traffic flow. Generally speaking, a larger amplitude-integrated area may indicate more vehicles passing simultaneously or heavier vehicles, leading to increased vibration energy. Therefore, the amplitude-integrated area can be used as an important factor in estimating traffic flow: a larger amplitude-integrated area indicates a higher estimated traffic flow.

[0055] Vibration duration is also positively correlated with traffic flow. Longer vibration durations may indicate slower vehicles passing through the monitoring area, longer vehicles, or multiple vehicles passing through the area simultaneously, resulting in longer vibration duration. In these cases, the number of traffic units passing through the monitoring area is likely to be relatively high, thus increasing the estimated traffic flow.

[0056] In practical applications, depending on the specific scenario and needs, the positive correlation between amplitude-integrated area or vibration duration and traffic flow can be used independently for estimation, or the two can be combined for comprehensive estimation. For example, by establishing a simple linear model with amplitude-integrated area and vibration duration as input variables and traffic flow as output variable, and using historical data for model training, a preliminary estimate of real-time traffic flow can be achieved.

[0057] For example, using the traffic scenario shown in Figure 3 as an example, since the maximum vibration amplitudes of graphs S3 and S4 do not exceed the amplitude threshold, graphs S3 and S4 perform traffic statistics using the first method, meaning that one vehicle passed through time period T3 and one vehicle passed through time period T4. Since graph S1 has only one peak and its maximum vibration amplitude exceeds the amplitude threshold, it is possible that two vehicles passed through the test road surface simultaneously. Since graph S2 has two peaks and its maximum vibration amplitude exceeds the amplitude threshold, it is possible that two vehicles passed through the test road surface sequentially. Furthermore, by summing the number of vehicles in these time periods, the traffic flow for the corresponding time period can be obtained.

[0058] In this way, the embodiment of the present application uses the maximum vibration amplitude, amplitude integral area and vibration duration of the closed figure to comprehensively judge and analyze traffic flow, which greatly improves the accuracy of traffic flow estimation.

[0059] Step S500: When the traffic flow estimate is greater than or equal to the traffic flow threshold, a traffic flow correction instruction is triggered, and the cloud camera is controlled to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value, wherein the cloud camera is communicatively connected to the cloud server.

[0060] When the traffic flow estimate based on data collected by the diaphragm sensor reaches or exceeds a pre-set flow threshold, it typically indicates high traffic flow on that road section. Under these high-flow conditions, the likelihood of multiple vehicles passing through the detection area simultaneously or at closely spaced intervals increases significantly. Due to the overlapping and interfering vibration signals from multiple vehicles, traffic flow data estimated solely by the diaphragm sensor will exhibit significant errors. This error can adversely impact traffic management decisions, such as adjusting traffic light timing and issuing congestion warnings.

[0061] Based on this, the embodiments of the present application propose an effective solution. When the traffic flow estimate is determined to be greater than or equal to the flow threshold, the system automatically triggers a traffic flow correction instruction. This instruction serves as a control signal, guiding the cloud camera to start operation. The cloud camera can obtain traffic condition information in the detection area from a more intuitive visual perspective. Combined with the estimated data from the diaphragm sensor, it corrects the initial traffic flow estimate, thereby obtaining a more accurate and reliable traffic flow target value.

[0062] In one embodiment, controlling the cloud camera to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value includes: controlling the cloud camera to start according to the traffic flow correction instruction; after the cloud camera is started, obtaining video data collected by the cloud camera in real time; extracting a target video frame corresponding to the time of the target vibration data set; performing vehicle feature recognition on the target video frame to obtain a vehicle quantity value; and correcting the traffic flow estimate according to the vehicle quantity value to obtain a traffic flow target value.

[0063] Specifically, the system continuously monitors the traffic flow estimate output by the diaphragm sensor. When the traffic flow estimate exceeds or equals a preset flow threshold, the system automatically triggers a traffic flow correction command. This traffic flow correction command is sent to the cloud camera's control module. Upon receiving the command, the cloud camera immediately activates and enters operation, preparing to collect video data from the detection area. Once activated, the cloud camera begins collecting real-time video stream data from the detection area. This video data is transmitted in real time to the cloud data processing center via a communication network. The data processing center determines the time period requiring correction based on the timestamp information of the target vibration dataset collected by the diaphragm sensor. From the real-time video stream, the video frame corresponding to the time of the target vibration dataset is extracted as the target video frame. This process is achieved through a time synchronization mechanism, ensuring precise temporal correspondence between the video frame and the vibration data. Vehicle feature recognition is performed on the extracted target video frame using image processing and computer vision techniques. Common vehicle feature recognition methods include object detection algorithms (such as YOLO and SSD) and vehicle outline recognition technology. By analyzing vehicle features such as position, shape, and color in the video frames, the number of vehicles within the detection area is accurately counted, resulting in a vehicle count value. The vehicle count value is compared and analyzed with the traffic flow estimate from the diaphragm sensor. If there is a significant discrepancy between the two, the system will adjust the estimate based on the vehicle count. For example, if the discrepancy between the two is less than 10%, the diaphragm sensor's result is used as the final recognition result. If the discrepancy reaches 10%, the average of the two results is used as the final recognition result to complete the correction process. After correction, a more accurate and reliable traffic flow target value is obtained. This target value will be used in subsequent traffic management decisions, such as adjusting traffic light timing and issuing congestion warnings.

[0064] In summary, the traffic flow analysis method provided by the embodiment of the present application first obtains the initial vibration curve formed by the data collected by the vibration membrane sensor of the target traffic section, then performs environmental factor compensation correction on the initial vibration curve based on environmental parameters to obtain the target vibration curve, and then identifies and extracts the key features of the target vibration curve to form multiple candidate vibration data sets containing information such as vibration amplitude and corresponding vibration duration. Based on these candidate vibration data sets, a preliminary estimate of the real-time traffic flow is made to obtain a traffic flow estimate value. If the estimate value is greater than or equal to a preset traffic threshold, a traffic flow correction instruction is triggered, and the cloud camera connected to the cloud server is controlled to start according to the correction instruction, and the traffic flow estimate value is corrected to finally obtain a traffic flow target value. If the estimate value is less than the preset traffic threshold, the estimate value is used as the traffic flow target value. In this way, the technical solution of the present application can adapt to a variety of different traffic flow analysis scenarios, and can obtain accurate analysis results in different scenarios.

[0065] In the traffic signal flow analysis device provided in the embodiment of the present application, the memory 300 is used to store computer-readable instructions, and the processor 400 is used to call the computer-readable instructions to execute the traffic flow analysis method as described above.

[0066] Among them, the processor 400 is used to provide computing and control capabilities to control the traffic signal flow analysis device to perform corresponding tasks, for example, controlling the traffic signal flow analysis device to perform the traffic flow analysis method in any of the above-mentioned method embodiments, the method including: obtaining a vibration curve formed by collecting data from a vibration membrane sensor, the vibration membrane sensor being arranged on a target traffic section; identifying and extracting key features of the vibration curve to obtain multiple candidate vibration data sets, the candidate vibration data sets at least including vibration amplitudes and corresponding vibration durations; making a preliminary estimate of real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimate value; triggering a traffic flow correction instruction when the traffic flow estimate value is greater than or equal to a flow threshold; controlling the cloud camera to start according to the traffic flow correction instruction to correct the traffic flow estimate value to obtain a traffic flow target value, wherein the cloud camera is communicatively connected to the cloud server.

[0067] Processor 400 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a hardware chip, or any combination thereof; it may also be a digital signal processing (DSP), an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or any combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0068] Memory 300, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the traffic flow analysis method in the embodiments of the present application. Processor 400 can implement the traffic flow analysis method in any of the above-described method embodiments by executing the non-transitory software programs, instructions, and modules stored in memory 300.

[0069] Specifically, the memory 300 may include a volatile memory (VM), such as a random access memory (RAM); the memory 300 may also include a non-volatile memory (NVM), such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD) or other non-transitory solid-state storage device; the memory 300 may also include a combination of the above types of memory.

[0070] In summary, the traffic signal flow analysis device of the present application adopts the technical solution of any one of the above-mentioned traffic flow analysis method embodiments, and therefore, has at least the beneficial effects brought by the technical solution of the above-mentioned embodiments, which will not be described one by one here.

[0071] The present application also provides a computer-readable storage medium, such as a memory device containing program code. The program code can be executed by a processor to implement the traffic flow analysis method described in the above embodiment. For example, the computer-readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CDROM), a magnetic tape, a floppy disk, or an optical data storage device.

[0072] The present application also provides a computer program product comprising one or more program codes stored in a computer-readable storage medium. A processor of the early warning system reads the program code from the computer-readable storage medium and executes the program code to perform the steps of the traffic flow analysis method provided in the above embodiment.

[0073] Those skilled in the art will understand that all or part of the steps of implementing the above embodiments can be completed by hardware, or by hardware related to program code, and the program can be stored in a computer-readable storage medium. The above-mentioned storage medium can be a read-only memory, a disk or an optical disk, etc.

[0074] It should be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0075] Through the description of the above embodiments, it is clear to those skilled in the art that each embodiment can be implemented by means of software plus a general hardware platform, or of course by hardware. It is understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0076] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made by using the contents of the present invention description and drawings under the inventive concept of the present invention, or direct / indirect application in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A traffic signal flow analysis device based on artificial intelligence, characterized in that: The system comprises a memory and a processor, wherein the memory stores an executable program code, and the processor is configured to implement the following steps of the traffic flow analysis method when calling and running the executable program code: obtaining an initial vibration curve formed by collecting data from a vibrating membrane sensor, the vibrating membrane sensor being located at a target traffic section; Performing environmental factor compensation correction on the initial vibration curve based on environmental parameters to obtain a target vibration curve; Identifying and extracting key features of the target vibration curve to obtain a plurality of candidate vibration data sets, wherein the candidate vibration data sets include at least vibration amplitudes and corresponding vibration durations; Performing a preliminary estimation of the real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimation value; When the traffic flow estimate is greater than or equal to the traffic threshold, a traffic flow correction instruction is triggered, and the cloud camera is controlled to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value, wherein the cloud camera is communicatively connected to the cloud server.

2. The traffic signal flow analysis device based on artificial intelligence according to claim 1, characterized in that: The vibration membrane sensor includes a vehicle crushing and lifting module arranged on the road surface of the target traffic section, a force transmission rod arranged below the vehicle crushing and lifting module, and an elastic body arranged on the side of the road, wherein a vibration membrane is provided in the elastic body; wherein, when a vehicle is located above the vehicle crushing and lifting module, the vehicle crushing and lifting module sinks and drives the force transmission rod to swing, and when the force transmission rod swings, it squeezes the elastic body so that the vibration membrane generates a vibration signal due to air compression.

3. The traffic signal flow analysis device based on artificial intelligence according to claim 2, characterized in that: The vehicle crushing lifting module includes a liftable pressure plate and a return spring, wherein the return spring is used to restore the pressure plate to its initial position after the vehicle leaves. The initial position of the pressure plate is flush with the road surface or higher than the ground.

4. The traffic signal flow analysis device based on artificial intelligence according to claim 1, characterized in that: The step of performing environmental factor compensation correction on the initial vibration curve based on the environmental parameters to obtain a target vibration curve includes: Acquiring real-time meteorological data collected by a meteorological sensor, wherein the meteorological sensor includes at least one of a wind sensor and a rain sensor; When the rainfall meteorological parameter and / or the wind speed meteorological parameter in the meteorological data exceeds a preset meteorological parameter threshold, a pre-trained environmental compensation neural network is called to perform environmental factor compensation correction on the initial vibration curve to obtain a target vibration curve.

5. The traffic signal flow analysis device based on artificial intelligence according to claim 1, characterized in that: The identifying and extracting key features of the target vibration curve to obtain multiple candidate vibration data sets includes: Extracting closed areas from the target vibration curve based on image analysis technology to obtain multiple closed graphs, wherein each closed graph corresponds to a vibration event; The maximum vibration amplitude, amplitude integral area, and vibration duration in each closed graph are extracted to obtain multiple candidate vibration data sets.

6. The traffic signal flow analysis device based on artificial intelligence according to claim 5, characterized in that: The preliminarily estimating the real-time traffic flow according to the plurality of candidate vibration data sets to obtain a traffic flow estimation value includes: Inputting the multiple candidate vibration data sets into a pre-trained lightweight convolutional network for feature screening, retaining key features related to target vehicle traffic flow to obtain multiple target vibration data sets; A preliminary estimate of the real-time traffic flow is performed based on the multiple target vibration data sets to obtain a traffic flow estimate value.

7. The traffic signal flow analysis device based on artificial intelligence according to claim 6, characterized in that: The preliminarily estimating the real-time traffic flow according to the plurality of target vibration data sets to obtain a traffic flow estimation value includes: When the maximum vibration amplitude is less than the amplitude threshold, a preliminary traffic flow estimation is performed using a first estimation method to obtain a traffic flow estimation value; When the maximum vibration amplitude is greater than or equal to the amplitude threshold, a second estimation method is used to perform a preliminary estimation of traffic flow to obtain a traffic flow estimation value, wherein the second estimation method is to perform a preliminary estimation of traffic flow based on the amplitude integral area and the vibration duration, the amplitude integral area is positively correlated with the traffic flow, and / or the vibration duration is positively correlated with the traffic flow.

8. The traffic signal flow analysis device based on artificial intelligence according to claim 6, characterized in that: The controlling the cloud camera to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value includes: Controlling the cloud camera to start according to the traffic flow correction instruction; After the cloud camera is started, video data collected by the cloud camera is obtained in real time; Extracting a target video frame corresponding to the time of the target vibration dataset; Performing vehicle feature recognition on the target video frame to obtain a vehicle quantity value; The traffic flow estimate is corrected according to the vehicle quantity value to obtain a traffic flow target value.

9. A traffic flow analysis method, applied to a traffic signal flow analysis device, characterized in that: The method comprises: obtaining an initial vibration curve formed by collecting data from a vibrating membrane sensor, the vibrating membrane sensor being located at a target traffic section; Performing environmental factor compensation correction on the initial vibration curve based on environmental parameters to obtain a target vibration curve; Identifying and extracting key features of the target vibration curve to obtain a plurality of candidate vibration data sets, wherein the candidate vibration data sets include at least vibration amplitudes and corresponding vibration durations; Performing a preliminary estimation of the real-time traffic flow based on the multiple candidate vibration data sets to obtain a traffic flow estimation value; When the traffic flow estimate is greater than or equal to the traffic threshold, a traffic flow correction instruction is triggered, and the cloud camera is controlled to start according to the traffic flow correction instruction to correct the traffic flow estimate to obtain a traffic flow target value, wherein the cloud camera is communicatively connected to the cloud server.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to claim 9 are implemented.

Images