Sensor-based traffic flow detection method and system

Abstract

The invention relates to the field of traffic flow detection, and provides a traffic flow detection method and system based on a sensor, the sensor is provided with a protection cover, the sensor is arranged on the first side of the protection cover, the first side of the protection cover is also provided with a millimeter wave radar, and the second side of the protection cover is also provided with a millimeter wave radar signal reflection device. The method comprises the following steps: acquiring a target signal quality index of a sensor and initial sensing data of the sensor based on a millimeter-wave radar and a millimeter-wave radar signal reflection device; the initial sensing data comprises a plurality of sub-initial sensing data; correcting the plurality of sub-initial sensing data according to the target signal quality index to obtain target sensing data; the target sensing data comprises a plurality of sub-target sensing data; extracting data features of the target sensing data; the data features comprise the duration and frequency features of each sub-target sensing data; and identifying the vehicle according to the data features and the target signal quality index. The accuracy of traffic flow detection can be improved.

Description

Technical Field

[0001] This application relates to the field of traffic flow detection, and in particular to a sensor-based traffic flow detection method and system. Background Technology

[0002] In modern urban traffic management, accurately obtaining road traffic flow data is crucial for optimizing traffic signals, providing early warnings of congestion, and planning traffic resources.

[0003] Existing solutions typically rely on various sensors deployed on roads, such as millimeter-wave radar, lidar, or high-resolution vision sensors, which acquire traffic data by sensing vehicle signals. However, these sensors face signal attenuation due to environmental pollution and weak signals from new micro-vehicles (such as electric motorcycles and electric scooters) due to their physical characteristics during long-term operation. The combination of these two factors makes it difficult for traditional detection methods to accurately identify and count traffic flow, and may cause the system to transmit distorted data even when detection accuracy is partially lost, thus affecting the effectiveness of traffic management decisions. Therefore, existing solutions have low accuracy in traffic flow detection. Summary of the Invention

[0004] This application provides a sensor-based traffic flow detection method and system, which can improve the accuracy of traffic flow detection.

[0005] To achieve the above objectives, this application adopts the following technical solution:

[0006] In a first aspect, a sensor-based traffic flow detection method is provided, wherein the sensor is equipped with a protective cover, the sensor is disposed on the first side of the protective cover, a millimeter-wave radar is disposed on the first side of the protective cover, and a millimeter-wave radar signal reflection device is disposed on the second side of the protective cover. The method includes: acquiring the target signal quality index of the sensor and the initial sensing data of the sensor based on the millimeter-wave radar and the millimeter-wave radar signal reflection device; the initial sensing data includes multiple sub-initial sensing data; correcting the multiple sub-initial sensing data according to the target signal quality index to obtain target sensing data; the target sensing data includes multiple sub-target sensing data; extracting data features of the target sensing data; the data features include the duration and frequency features of each sub-target sensing data; and identifying vehicles based on the data features and the target signal quality index.

[0007] Furthermore, the target signal quality index of the sensor is obtained based on the millimeter-wave radar and the millimeter-wave radar signal reflection device, including: controlling the transmitting antenna of the millimeter-wave radar to transmit the original millimeter-wave radar signal; obtaining the target millimeter-wave radar signal received by the receiving antenna of the millimeter-wave radar; the target millimeter-wave radar signal is the reflected signal of the original millimeter-wave radar signal reflected by the millimeter-wave radar signal reflection device; determining the contamination index of the protective cover based on the target millimeter-wave radar signal; and determining the target signal quality index of the sensor based on the contamination index.

[0008] More specifically, in some implementations, the contamination index of the protective shield is determined based on the target millimeter-wave radar signal, including: acquiring the reference peak intensity and reference waveform characteristics of the reflected signal of the original millimeter-wave radar signal; using the ratio of the difference between the reference peak intensity and the target peak intensity of the target millimeter-wave radar signal to the reference peak intensity as the peak intensity deviation index of the protective shield; normalizing the mean square error between the target waveform characteristics and the reference waveform characteristics of the target millimeter-wave radar signal to obtain the waveform characteristic deviation index of the protective shield; and using the weighted sum of the peak intensity deviation index and the waveform characteristic deviation index as the contamination index of the protective shield.

[0009] Based on the above, this application further proposes a method for determining the target signal quality index of a sensor based on a pollution index, including: acquiring test sensing data of the sensor on a target area; the target area being a preset area within the sensor's observation area; determining the background reflection feature drift index of the sensor based on the test sensing data; the background reflection feature drift index having a value range of 0-1; and using the weighted sum of the background reflection feature drift index and the pollution index of the protective cover as the target signal quality index of the sensor.

[0010] Preferably, determining the background reflection feature drift index of the sensor based on the test sensing data includes: acquiring the reference average echo intensity and reference spectral width of the sensor's sensing data for the target area; using the ratio of the difference between the reference average echo intensity and the test average echo intensity of the test sensing data to the reference average echo intensity as the background feature drift index of the sensor; using the ratio of the difference between the reference spectral width and the test spectral width of the test sensing data to the reference spectral width as the spectral width drift index of the sensor; and using the weighted sum of the background feature drift index and the spectral width drift index as the background reflection feature drift index of the sensor.

[0011] In one embodiment, correcting initial sensing data based on a target signal quality index to obtain target sensing data includes: acquiring a first correspondence; the first correspondence includes a one-to-one correspondence between multiple signal quality index ranges and multiple signal compensation gains; using the signal compensation gain corresponding to the target signal quality index in the first correspondence as the target signal compensation gain; and correcting the initial sensing data based on the target signal compensation gain to obtain the target sensing data.

[0012] In another implementation, vehicle identification is performed based on data features and target signal quality index, including: for each sub-target sensing data in a plurality of sub-target sensing data, determining the moving speed corresponding to the sub-target sensing data based on the frequency characteristics of the sub-target sensing data; and identifying the vehicle based on the moving speed corresponding to the sub-target sensing data and the duration of the sub-target sensing data.

[0013] As an optional approach, the moving speed corresponding to the sub-target sensing data is determined based on the frequency characteristics of the sub-target sensing data, including: performing a fast Fourier transform on the frequency characteristics of the sub-target sensing data to obtain the Doppler frequency shift of the sub-target sensing data; and taking half of the ratio of the Doppler frequency shift to the wavelength of the sensor's sensing signal as the moving speed corresponding to the sub-target sensing data.

[0014] To improve the solution, vehicles are identified based on the moving speed corresponding to the sub-target sensing data and the duration of the sub-target sensing data. This includes: obtaining a second correspondence; the second correspondence includes a one-to-one correspondence between multiple first pieces of information and multiple pieces of second information; the first information includes a range of moving speed and a range of duration, and the second information includes the vehicle type; the first information into which the moving speed and duration of the sub-target sensing data in the second correspondence fall is taken as the target first information; and the vehicle type corresponding to the target first information in the second correspondence is taken as the target vehicle type corresponding to the sub-target sensing data.

[0015] Secondly, this application also discloses a sensor-based traffic flow detection system, wherein the sensor is equipped with a protective cover, the sensor is disposed on the first side of the protective cover, a millimeter-wave radar is disposed on the first side of the protective cover, and a millimeter-wave radar signal reflection device is disposed on the second side of the protective cover. The system includes: an acquisition device and a processing device; the acquisition device is used to acquire the target signal quality index of the sensor and the initial sensing data of the sensor based on the millimeter-wave radar and the millimeter-wave radar signal reflection device; the initial sensing data includes multiple sub-initial sensing data; the processing device is used to correct the multiple sub-initial sensing data according to the target signal quality index to obtain target sensing data; the target sensing data includes multiple sub-target sensing data; the processing device is used to extract data features of the target sensing data; the data features include the duration and frequency features of each sub-target sensing data; the processing device is used to identify vehicles according to the data features and the target signal quality index.

[0016] Through the above technical solution, this application effectively solves the problems of signal attenuation caused by protective cover contamination and decreased detection accuracy and data distortion caused by weak signals from new micro-vehicles in existing technologies. This method corrects the original data by introducing a target signal quality index, ensuring the accuracy of data processing and avoiding the transmission of distorted data even when some accuracy is lost. Therefore, this application can provide more realistic and reliable traffic flow data, providing accurate basis for traffic management departments to make decisions such as road congestion warnings, traffic light timing optimization, and traffic flow analysis, significantly improving the overall operational efficiency of urban traffic and user experience, and overcoming the shortcomings of existing technologies where the effectiveness of traffic management decisions is greatly reduced. Attached Figure Description

[0017] Figure 1 A schematic flowchart of a sensor-based traffic flow detection method provided in this application;

[0018] Figure 2 A flowchart illustrating yet another sensor-based traffic flow detection method provided in this application;

[0019] Figure 3 A flowchart illustrating yet another sensor-based traffic flow detection method provided in this application;

[0020] Figure 4 This application provides a schematic diagram of the architecture of a sensor-based traffic flow detection system. Detailed Implementation

[0021] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0022] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] In modern urban traffic management, accurately obtaining road traffic flow data is crucial for optimizing traffic signals, providing early warnings of congestion, and planning traffic resources.

[0024] Existing solutions typically rely on various sensors deployed on roads, such as millimeter-wave radar, lidar, or high-resolution vision sensors, which acquire traffic data by sensing vehicle signals. However, these sensors face signal attenuation due to environmental pollution and weak signals from new micro-vehicles (such as electric motorcycles and electric scooters) due to their physical characteristics during long-term operation. The combination of these two factors makes it difficult for traditional detection methods to accurately identify and count traffic flow, and may cause the system to transmit distorted data even when detection accuracy is partially lost, thus affecting the effectiveness of traffic management decisions. Therefore, existing solutions have low accuracy in traffic flow detection.

[0025] To address this, this application proposes a sensor-based traffic flow detection method. The sensor is equipped with a protective cover, and is positioned on the first side of the cover. A millimeter-wave radar is also mounted on the first side of the cover, and a millimeter-wave radar signal reflection device is mounted on the second side of the cover. The method includes: acquiring the target signal quality index and initial sensing data of the sensor based on the millimeter-wave radar and the millimeter-wave radar signal reflection device; the initial sensing data includes multiple sub-initial sensing data; correcting the multiple sub-initial sensing data according to the target signal quality index to obtain target sensing data; the target sensing data includes multiple sub-target sensing data; extracting data features from the target sensing data; the data features include the duration and frequency characteristics of each sub-target sensing data; and identifying vehicles based on the data features and the target signal quality index.

[0026] This application assesses the sensor's operating status by introducing millimeter-wave radar and a millimeter-wave radar signal reflection device, and corrects the sensing data accordingly. This effectively addresses the problems of sensor contamination and difficulty in identifying vehicles with weak signals, significantly improving the accuracy and reliability of traffic flow detection.

[0027] To better understand the technical solution proposed in this application, some key terms involved will be explained first.

[0028] In this application, "sensor" refers to a device used to sense vehicle signals in a traffic environment, such as an ultrasonic sensor, an infrared sensor, or a geomagnetic sensor, whose main function is to collect raw data when a vehicle passes by.

[0029] A "protective cover" is a protective structure placed on the outside of a sensor to protect it from environmental factors such as dust and rain.

[0030] "Millimeter-wave radar" is a radar system that uses millimeter-wave electromagnetic waves for detection. It has the ability to penetrate non-metallic obstacles such as smoke and dust. In this application, it is used to assist in assessing the contamination level of the protective cover and the background reflection characteristics of the sensor.

[0031] A "millimeter-wave radar signal reflection device" is a reflector used in conjunction with a millimeter-wave radar to provide a stable source of millimeter-wave signal reflection so that the millimeter-wave radar can accurately measure signal attenuation. It can be a metal sheet.

[0032] The "Target Signal Quality Index" is a comprehensive indicator that measures the signal quality of a sensor under its current operating conditions. The higher the value, the better the signal quality.

[0033] "Initial sensing data" is the raw data that the sensor collects directly before any corrections are made.

[0034] "Sub-initial sensing data" is a data segment of a single signal within the initial sensing data.

[0035] "Target sensing data" is sensing data that has been corrected by the target signal quality index, and its accuracy is higher.

[0036] "Sub-target sensing data" refers to a data segment of a single signal within the target sensing data.

[0037] "Data features" refer to key information extracted from sub-target sensing data for vehicle identification, such as the duration of sub-target sensing data and the frequency characteristics of its signals.

[0038] "Duration" refers to the time that the sub-target sensing data remains within the sensor's detection area.

[0039] "Frequency characteristics" refer to the performance of sub-target sensing data in the frequency domain, such as Doppler shift, which can reflect the movement speed corresponding to the sub-target sensing data.

[0040] The core of the sensor-based traffic flow detection method proposed in this application lies in obtaining the target signal quality index of the sensor by introducing millimeter-wave radar and a millimeter-wave radar signal reflection device, and using this index to correct the initial sensing data, ultimately achieving accurate vehicle identification.

[0041] Specifically, the sensor is equipped with a protective cover. The sensor is positioned on the first side of the cover, which also houses a millimeter-wave radar. A millimeter-wave radar signal reflector is located on the second side of the cover. This configuration allows the millimeter-wave radar to detect the radar signal reflector through the protective cover, thereby assessing the contamination level of the cover. For example, the sensor can be mounted above or to the side of a road, with the protective cover protecting it from harsh weather and dust. The millimeter-wave radar can be mounted outside the protective cover, while the radar signal reflector is mounted on the other side, creating a path for the millimeter-wave signal. Figure 1 As shown, the method includes:

[0042] S101. Obtain the target signal quality index of the sensor and the initial sensing data of the sensor based on millimeter-wave radar and millimeter-wave radar signal reflection device.

[0043] Initial sensing data comprises multiple sub-initial sensing data. The target signal quality index of the sensor can be obtained in various ways. For example, millimeter-wave radar can periodically emit millimeter-wave signals and receive signals reflected back from millimeter-wave radar signal reflectors. By analyzing the intensity, waveform, and other characteristics of the reflected signals, the degree of contamination of the protective shield can be assessed, and thus the sensor's target signal quality index can be calculated. Initial sensing data is directly acquired by the sensor; for example, when a vehicle passes through the sensor's detection area, the sensor generates a series of raw signals, which constitute the initial sensing data.

[0044] S102. Correct multiple initial sub-sensing data according to the target signal quality index to obtain target sensing data.

[0045] The target sensing data includes multiple sub-target sensing data.

[0046] The correction process can perform gain compensation or filtering on the initial sensing data based on the magnitude of the target signal quality index. For example, when the target signal quality index is low, it indicates that the sensor signal may be attenuated. In this case, a larger compensation gain can be applied to the initial sensing data to enhance the signal strength; conversely, when the target signal quality index is high, a smaller compensation gain is applied. This correction can be linear or nonlinear, depending on the preset correction model.

[0047] S103. Extract the data features of the target sensing data.

[0048] Data characteristics include the duration and frequency of each sub-target sensing data.

[0049] Data feature extraction can be achieved through signal processing techniques. For example, for each sub-target sensing data, the duration of the signal can be calculated by analyzing its time series. Frequency features can be obtained by performing spectral analysis (e.g., Fourier transform) on the sub-target sensing data, thereby revealing the energy distribution of the signal at different frequencies.

[0050] S104. Identify vehicles based on data characteristics and target signal quality index.

[0051] Vehicle recognition can be based on pre-defined rules or machine learning models. For example, a database containing vehicle types, duration ranges, and frequency characteristic ranges can be established. After acquiring the data features of the target sensor data, it is matched with the information in the database. Simultaneously, the target signal quality index can also serve as an important parameter in the recognition process; for instance, in cases of poor signal quality, the matching threshold can be appropriately relaxed to improve the robustness of the recognition.

[0052] The sensor-based traffic flow detection method proposed in this application effectively solves the data distortion problem caused by sensor contamination and difficulty in identifying weak signal vehicles in traditional methods by introducing millimeter-wave radar and millimeter-wave radar signal reflection device to evaluate the working status of the sensor in real time and correct the original sensor data according to the evaluation results.

[0053] Specifically, traditional methods often fail to detect signal attenuation when sensor protective covers are contaminated, leading to continuous output of distorted data. This application utilizes millimeter-wave radar to detect the reflection of millimeter-wave radar signals, accurately obtaining the contamination index of the protective cover. Combined with the sensor's background reflection characteristic drift index, a comprehensive determination of the sensor's target signal quality index is achieved. This innovation allows the system to monitor its own "health status" in real time, providing a reliable basis for subsequent data correction.

[0054] Furthermore, addressing the challenge of identifying weak signals generated by novel micro-vehicles, this application utilizes the acquired target signal quality index to correct the initial sensing data. This correction mechanism effectively compensates for signal attenuation caused by pollution and weak signals, enabling the effective identification of vehicles with weak signals that might otherwise be ignored. For example, when the target signal quality index is low, the system applies a corresponding signal compensation gain to enhance the strength of the weak signal, bringing it to a recognizable threshold. This contrasts sharply with traditional methods that use fixed discrimination thresholds, leading to weak-signal vehicles being misclassified as background noise.

[0055] By extracting data features (including duration and frequency features) from the corrected target sensing data and combining them with the target signal quality index for vehicle identification, this application can more accurately determine the type and quantity of vehicles. For example, frequency features can be used to determine the vehicle's speed, while duration can help determine the vehicle's size or type. In cases of poor signal quality, the target signal quality index can serve as a basis for adjusting the identification algorithm parameters, further improving the accuracy and robustness of the identification.

[0056] In summary, this application presents a complete and adaptive traffic flow detection scheme by introducing millimeter-wave radar to assist in evaluating sensor operating status, correcting sensor data based on the evaluation results, and combining multi-dimensional data features for vehicle identification. Compared to existing technologies, this application effectively overcomes the challenges of sensor contamination and weak signal vehicle identification, significantly improving the accuracy and reliability of traffic flow data. This provides more precise data support for urban traffic management, thereby optimizing traffic signals, providing congestion warnings, and planning traffic resources, ultimately improving the overall operational efficiency of urban traffic and the user experience.

[0057] like Figure 2 As shown, this application further proposes a method for obtaining the target signal quality index of a sensor based on millimeter-wave radar and a millimeter-wave radar signal reflection device, the specific steps of which include:

[0058] S201, Control the transmitting antenna of the millimeter-wave radar to transmit the raw millimeter-wave radar signal.

[0059] Controlling the transmitting antenna of a millimeter-wave radar to emit raw millimeter-wave radar signals refers to programming or commanding the millimeter-wave radar to cause its transmitting antenna to emit millimeter-wave signals of a specific frequency and power at preset times or periodically towards the direction of the millimeter-wave radar signal reflecting device. This raw millimeter-wave radar signal is used to detect the current state of the protective shield.

[0060] S202. Obtain the target millimeter-wave radar signal received by the receiving antenna of the millimeter-wave radar.

[0061] The target millimeter-wave radar signal is the reflected signal of the original millimeter-wave radar signal reflected by the millimeter-wave radar signal reflecting device.

[0062] Acquiring the target millimeter-wave radar signal received by the receiving antenna of a millimeter-wave radar can be understood as the receiving antenna capturing the signal that was originally a millimeter-wave radar signal, passed through a protective cover, and reflected back by a millimeter-wave radar signal reflection device. This target millimeter-wave radar signal carries the signal characteristics after passing through the protective cover and reflection device; its intensity and waveform are affected by the degree of contamination on the surface of the protective cover.

[0063] S203. Determine the contamination index of the protective shield based on the target millimeter-wave radar signal.

[0064] Determining the contamination index of a protective shield based on the target millimeter-wave radar signal involves analyzing the characteristics of the target millimeter-wave radar signal, such as its peak intensity and waveform features, and comparing it with a pre-defined reference signal under uncontaminated conditions to quantify the degree of contamination of the protective shield. The contamination index can be a numerical value used to characterize the degree of attenuation or distortion of the millimeter-wave signal by the protective shield.

[0065] S204. Determine the target signal quality index of the sensor based on the pollution index.

[0066] Determining the target signal quality index of a sensor based on its contamination index means taking the contamination index of the protective cover obtained above as one of the key factors affecting the sensor's signal quality. By comprehensively considering its impact on sensor performance, a more accurate target signal quality index that better reflects the sensor's actual operating state can be calculated. For example, a higher contamination index may result in a lower target signal quality index.

[0067] This application's solution utilizes millimeter-wave radar for non-contact detection of the contamination status of a protective shield, solving the problem of traditional methods' difficulty in accurately assessing the impact of the protective shield on sensor signal quality. Specifically, the millimeter-wave radar's transmitting antenna emits a raw millimeter-wave radar signal. This signal passes through the protective shield and is reflected by a millimeter-wave radar signal reflector before being received by the receiving antenna. Since the millimeter-wave signal attenuates and distorts when penetrating the contaminated protective shield, the degree of contamination of the protective shield can be accurately quantified by analyzing the characteristics of the received target millimeter-wave radar signal, i.e., determining the shield's contamination index. Subsequently, this contamination index is used as a key parameter to correct or directly determine the sensor's target signal quality index. This approach allows the target signal quality index to reflect the actual condition of the protective shield on sensor performance in real time and accurately, thus providing a more reliable basis for subsequent sensor data correction.

[0068] The above technical solution effectively addresses the problem of inaccurate target signal quality index caused by protective cover contamination. This solution utilizes millimeter-wave radar to accurately assess the degree of protective cover contamination, enabling the sensor's target signal quality index to more accurately reflect the sensor's operating status in the current environment, thereby improving the accuracy of initial sensor data correction. Therefore, this application significantly enhances the overall reliability and data accuracy of sensor-based traffic flow detection methods, especially under harsh environments or long-term operating conditions, ensuring the accuracy of traffic flow detection results.

[0069] like Figure 3 As shown, in some embodiments of this application, a method for determining the contamination index of a protective shield based on a target millimeter-wave radar signal is proposed. Specifically, the step of determining the contamination index of the protective shield includes:

[0070] S301. Obtain the reference peak intensity and reference waveform characteristics of the reflected signal of the original millimeter-wave radar signal.

[0071] The reference peak intensity and reference waveform characteristics of the reflected signal from the original millimeter-wave radar signal refer to the reflection characteristics of the millimeter-wave radar signal when the protective enclosure is in a clean state or with a known level of contamination. These reference values ​​can be measured and stored during system deployment or after periodic maintenance, serving as a reference standard for subsequent contamination level assessments. For example, the reference peak intensity and reference waveform characteristics can be obtained by controlling the transmitting antenna of the millimeter-wave radar to transmit the original millimeter-wave radar signal in a contamination-free environment and acquiring the reflected signal received by the receiving antenna of the millimeter-wave radar.

[0072] S302. The ratio of the difference between the reference peak intensity and the target peak intensity of the target millimeter-wave radar signal to the reference peak intensity is used as the peak intensity deviation index of the protective cover.

[0073] The target peak intensity and target waveform characteristics of a millimeter-wave radar signal refer to the characteristics of the signal reflected by the millimeter-wave radar signal reflecting device and received by the receiving antenna of the millimeter-wave radar during actual operation. These characteristics can be affected by contamination on the surface of the protective cover (such as dust, water droplets, ice, snow, etc.).

[0074] The peak intensity deviation index is used to quantify the degree of change in the intensity of a target millimeter-wave radar signal relative to a reference intensity. When the protective cover surface is contaminated, the reflection intensity of the millimeter-wave radar signal is typically weakened, resulting in the target peak intensity being lower than the reference peak intensity, thus generating a positive deviation index. The larger the index, the greater the impact of contamination on the signal strength.

[0075] S303. The mean square error between the target waveform characteristics and the reference waveform characteristics of the target millimeter-wave radar signal is normalized to obtain the waveform characteristic deviation index of the protective cover.

[0076] The waveform characteristic deviation index is used to quantify the degree of change in the waveform shape of a target millimeter-wave radar signal relative to a reference waveform shape. Contamination not only affects signal strength but can also cause signal scattering, attenuation, and distortion, thereby altering its waveform characteristics. By calculating and normalizing the mean square error between the target waveform characteristics and the reference waveform characteristics, an index reflecting the degree of waveform distortion can be obtained. The larger the mean square error, the larger the normalized waveform characteristic deviation index, indicating more severe waveform distortion and a higher degree of contamination.

[0077] S304. The weighted sum of the peak intensity deviation index and the waveform characteristic deviation index is used as the pollution index of the protective cover.

[0078] The pollution index is a comprehensive indicator obtained by weighting the peak intensity deviation index and the waveform characteristic deviation index. The weighting coefficients can be adjusted based on practical application scenarios and experience to reflect the importance of different deviation indices in contributing to the degree of pollution. For example, different weights can be assigned to the peak intensity deviation index and the waveform characteristic deviation index based on historical data or experimental results to more accurately assess the pollution level of the protective shield.

[0079] This application's solution determines the contamination index of the protective shield by simultaneously considering the peak intensity variation and waveform characteristic variation of the millimeter-wave radar signal. When contamination exists on the surface of the protective shield, the millimeter-wave radar signal is affected by attenuation and scattering during penetration or reflection. Specifically, contamination leads to energy loss in the reflected signal, causing a deviation in the target peak intensity relative to the reference peak intensity; simultaneously, contamination also causes phase and amplitude distortion of the signal, resulting in differences between the target waveform characteristics and the reference waveform characteristics. By calculating the peak intensity deviation index and the waveform characteristic deviation index and weighting them together, the degree of contamination of the protective shield can be comprehensively and accurately reflected, avoiding the misjudgment or inaccuracy that may result from relying on a single indicator.

[0080] The above technical solution enables a more comprehensive and accurate assessment of the contamination status of the sensor protective cover. Traditional contamination detection methods may only focus on signal strength attenuation while ignoring signal waveform distortion, leading to inaccurate judgments of the degree of contamination. This application introduces a peak intensity deviation index and a waveform characteristic deviation index, and performs weighted fusion to comprehensively consider the dual impact of contamination on the signal strength and waveform characteristics of millimeter-wave radar, thereby obtaining a more reliable contamination index. This provides a more accurate input for subsequent correction of initial sensor data and determination of the target signal quality index, thus improving the overall accuracy and reliability of traffic flow detection.

[0081] In one design, this application further proposes a more complete method for determining the target signal quality index of a sensor based on a pollution index, by introducing the background reflection characteristic drift index of the sensor to comprehensively evaluate the signal quality of the sensor.

[0082] Specifically, determining the target signal quality index of the sensor based on the aforementioned pollution index includes: acquiring test sensing data of the sensor on the target area; the target area being a preset area within the observation area of ​​the sensor; determining the background reflection feature drift index of the sensor based on the test sensing data; the background reflection feature drift index having a value range of 0-1; and using the weighted sum of the background reflection feature drift index and the pollution index of the protective cover as the target signal quality index of the sensor.

[0083] Acquiring test sensing data of the aforementioned sensors for a target area refers to continuously collecting sensing data from a preset target area under specific conditions, such as during periods of low traffic or when no vehicles are passing through. This target area is a fixed sub-region within the sensor's observation area that possesses stable background reflection characteristics, such as a fixed structure at the edge of a road or the ground. Collecting sensing data from this area reflects the sensor's performance under relatively stable conditions.

[0084] Furthermore, determining the background reflectance feature drift index of the aforementioned sensor based on the test sensing data refers to assessing whether the sensor's ability to perceive background reflectance features has drifted by analyzing the test sensing data. The background reflectance feature drift index aims to quantify the performance of the sensor itself or the degree of change in the non-traffic-related background environment within its observation area. The index value is set to a range of 0 to 1, where 0 represents no drift and 1 represents maximum drift, to facilitate subsequent quantization and weighted calculations.

[0085] Specifically, the target signal quality index of the sensor is obtained by weighting the background reflection feature drift index and the contamination index of the protective cover. This means that the external contamination of the protective cover and the internal performance of the sensor itself or changes in the background environment are comprehensively considered. By weighting and summing these two indices, a more comprehensive and accurate sensor signal quality assessment result can be obtained. The weighting coefficients can be adjusted according to actual application scenarios and experience to reflect the relative importance of different factors on signal quality.

[0086] This application's solution introduces the background reflection characteristic drift index of the sensor and weights it together with the contamination index of the protective cover to more comprehensively evaluate the sensor's target signal quality index. Specifically, the contamination index of the protective cover mainly reflects the attenuation and interference of external environmental factors (such as dust and moisture) on the sensor signal, while the background reflection characteristic drift index reflects the long-term drift of the sensor's own performance (such as component aging and calibration deviation) as well as subtle changes in the non-traffic-related background environment within the sensor's observation area (such as vegetation growth and road wear). It is precisely because these external and internal, direct and indirect influencing factors are considered simultaneously that the determined target signal quality index can more accurately characterize the sensor's signal quality status in actual operation.

[0087] Through the above technical solution, this application overcomes the limitations of relying solely on the protective cover contamination index to evaluate sensor signal quality. By introducing a background reflection characteristic drift index, it can more precisely capture the degradation of sensor performance or changes in the background environment, thereby making the determination of the target signal quality index more accurate and reliable. This comprehensive evaluation method helps improve the correction accuracy of subsequent initial sensing data, thereby enhancing the overall accuracy and robustness of traffic flow detection. Especially under conditions of long-term sensor operation or complex and changing environments, it can effectively avoid detection errors caused by insufficient evaluation of a single factor.

[0088] In some preferred embodiments, this application is implemented as follows:

[0089] Assuming that during nighttime or periods of low traffic, sensors are controlled to continuously collect test data from a fixed guardrail area alongside the road. This guardrail area is preset as the target area. The test data collected by the sensor includes the echo intensity and spectral characteristics of that area. The system periodically acquires this test data and compares it with reference data collected by the sensor under ideal conditions (e.g., immediately after installation or calibration) from the same target area. For example, by calculating the relative deviation between the average echo intensity of the test data and the reference average echo intensity, and the relative deviation between the spectral width of the test data and the reference spectral width, the sensor's background feature drift index and spectral width drift index can be obtained. The background reflection feature drift index is obtained by weighted summing of these two drift indices.

[0090] Assuming the protective shield contamination index detected by millimeter-wave radar is 0.2 (indicating mild contamination), and the background reflection feature drift index calculated from the test sensor data is 0.1 (indicating slight drift), if the preset weighting coefficients are 0.6 and 0.4 respectively, the final sensor target signal quality index will be calculated as: 0.1*0.6 + 0.2*0.4 = 0.06 + 0.08 = 0.14. This comprehensive index of 0.14 will be used to subsequently correct the initial sensor data, thereby ensuring that even with mild contamination of the protective shield and slight sensor drift, the sensor data can be effectively compensated, guaranteeing the accuracy of traffic flow detection.

[0091] In one design, this application further proposes a step for determining the background reflection feature drift index of the sensor based on the aforementioned test sensing data, comprising: obtaining the reference average echo intensity and reference spectral width of the sensor's sensing data for the target area; using the ratio of the difference between the reference average echo intensity and the test average echo intensity of the test sensing data to the reference average echo intensity as the background feature drift index of the sensor; using the ratio of the difference between the reference spectral width and the test spectral width of the test sensing data to the reference spectral width as the spectral width drift index of the sensor; and using the weighted sum of the background feature drift index and the spectral width drift index as the background reflection feature drift index of the sensor.

[0092] Specifically, acquiring the baseline average echo intensity and baseline spectral width of the sensor's sensing data for the target area refers to collecting sensing data from a predetermined area (i.e., the target area) within the sensor's observation area during the initial installation phase or under known favorable environmental conditions, and extracting the average echo intensity and spectral width of that area as baseline values. These baseline values ​​represent the background reflection characteristics of the sensor under ideal or stable conditions, and their purpose is to provide a reference standard for subsequent drift detection.

[0093] The background feature drift index of the sensor is defined as the ratio of the difference between the reference average echo intensity and the test average echo intensity of the test sensor data to the reference average echo intensity. This can be understood as quantifying the degree of background reflection intensity drift by calculating the relative change between the average echo intensity under the current test condition and the reference average echo intensity. The average echo intensity reflects the overall energy level of the reflected signal within the target area, and its changes may indicate changes in the environmental background, such as water accumulation, snow accumulation, or vegetation growth.

[0094] In practical applications, the ratio of the difference between the reference spectral width and the test spectral width of the test sensor data to the reference spectral width is used as the sensor's spectral width drift index. This refers to evaluating the stability of the background reflected signal frequency distribution by comparing the relative change between the spectral width under the current test state and the reference spectral width. The spectral width reflects the degree of frequency spread of the signal, and its changes may be related to the dynamic characteristics of background clutter or small changes in the sensor's own performance.

[0095] Furthermore, the weighted sum of the background feature drift index and the spectral width drift index is used as the sensor's background reflection feature drift index. This aims to comprehensively consider changes in background reflection intensity and frequency distribution, thereby obtaining a more comprehensive and accurate quantitative indicator of background reflection feature drift. By weighted summing these two indices, different types of drift can be assigned different importance based on actual needs or experience, enabling the final background reflection feature drift index to more accurately reflect changes in the sensor's actual operating environment.

[0096] This application's scheme quantifies the drift degree of background reflection characteristics by introducing a benchmark average echo intensity and a benchmark spectral width, and comparing them with the test average echo intensity and test spectral width in the test sensor data. This method can comprehensively evaluate the impact of changes in the background environment on sensor performance from two dimensions: echo intensity and spectral width, avoiding the one-sidedness that may be caused by a single index. By calculating the relative ratio, the drift index has better comparability and robustness, and is not affected by the absolute value. Finally, the two drift indices are fused by a weighted sum to form a comprehensive background reflection characteristic drift index, providing a more accurate and reliable input for the subsequent determination of the target signal quality index.

[0097] The above technical solution provides a specific, quantitative, and multi-dimensional method for determining the background reflection feature drift index, effectively solving the accuracy and consistency problems that may exist when determining the background reflection feature drift index of a sensor. By comprehensively considering the changes in average echo intensity and spectral width, this method makes the evaluation of the background reflection feature drift index more comprehensive and accurate, thereby improving the evaluation accuracy of the sensor's target signal quality index and enhancing the overall reliability and accuracy of sensor-based traffic flow detection methods.

[0098] In one design, the method for obtaining target sensing data by correcting initial sensing data based on the aforementioned target signal quality index may include the following steps: obtaining a first correspondence; the first correspondence includes a one-to-one correspondence between multiple signal quality index ranges and multiple signal compensation gains; using the signal compensation gain corresponding to the target signal quality index in the first correspondence as the target signal compensation gain; and correcting the initial sensing data based on the target signal compensation gain to obtain the target sensing data.

[0099] The first correspondence can be understood as a pre-established lookup table, functional relationship, or model, designed to determine an appropriate signal compensation gain based on the target signal quality index of the sensor. For example, when the target signal quality index is low, a larger signal compensation gain may be needed to enhance the signal; conversely, when the target signal quality index is high, a smaller signal compensation gain or even no compensation may be needed. The signal quality index range can be a segmentation or classification of the target signal quality index, with each range corresponding to one or a set of specific signal compensation gains. The signal compensation gain can be a multiplicative factor or an additive offset used to adjust the amplitude or intensity of the initial sensing data. The target signal compensation gain is a specific compensation value obtained by querying or calculating the first correspondence based on the currently acquired target signal quality index. Correcting the initial sensing data specifically refers to performing operations, such as multiplication or addition, on each sub-initial sensing data in the initial sensing data with the target signal compensation gain to correct for attenuation or distortion caused by signal quality issues, thereby obtaining more accurate target sensing data.

[0100] The solution in this application establishes a pre-defined correspondence, enabling the system to dynamically determine and apply an appropriate signal compensation gain based on the real-time acquired target signal quality index. When the sensor's signal quality deteriorates due to external environmental factors (such as contamination from the protective cover) or fluctuations in its own performance, the target signal quality index reflects this deterioration. Through the pre-defined correspondence, a signal compensation gain matching the current target signal quality index can be found. This gain effectively compensates for signal attenuation or distortion, thereby accurately correcting the initial sensing data. This ensures higher accuracy and reliability of the target sensing data upon which subsequent data processing relies, avoiding vehicle identification errors caused by poor original signal quality.

[0101] Through the above technical solution, this application can adaptively correct the initial sensing data according to the actual signal quality of the sensor. This dynamic compensation mechanism based on the target signal quality index effectively improves the accuracy and stability of the sensing data. Especially when the sensor's working environment is complex or its performance fluctuates, it can significantly reduce the impact of signal attenuation or distortion on traffic flow detection results, thereby improving the accuracy of vehicle identification and the overall reliability of the system.

[0102] Traditional sensor-based traffic flow detection methods, when identifying vehicles based on data features and target signal quality indices, may rely on only a single or limited set of data features for coarse judgment, resulting in insufficient identification accuracy and difficulty in effectively distinguishing different types or states of motion of vehicles. If these problems are not addressed, it may affect the accuracy of traffic flow statistics and the effectiveness of subsequent traffic management decisions. To address this, this application proposes a more refined vehicle identification method that fully utilizes the frequency characteristics, speed, and duration of sub-target sensor data to improve the accuracy and robustness of vehicle identification.

[0103] In some embodiments of this application described above, vehicle identification based on the aforementioned data features and target signal quality index specifically includes: for each sub-target sensing data in a plurality of sub-target sensing data, determining the moving speed corresponding to the sub-target sensing data based on the frequency characteristics of the sub-target sensing data; and identifying the vehicle based on the moving speed corresponding to the sub-target sensing data and the duration of the sub-target sensing data.

[0104] Specifically, the frequency characteristics of sub-target sensing data can be understood as the frequency domain representation of the echo signal received by the sensor, containing motion information of the target object (i.e., the vehicle). By analyzing these frequency characteristics, the vehicle's speed can be deduced. Speed ​​refers to the instantaneous or average rate of movement of the vehicle within the sensor's observation area. Duration refers to the length of time the vehicle is continuously detected within the sensor's observation area. In practical applications, combining data from both speed and duration allows for a more comprehensive and accurate characterization of the vehicle's motion, thus enabling effective differentiation of vehicle types. For example, the speed and duration of movement of a fast-moving sedan will differ significantly from those of a slow-moving heavy truck.

[0105] This application's solution effectively addresses the problem of insufficient vehicle recognition accuracy in existing technologies by refining the vehicle recognition process into two key steps. First, through in-depth analysis of the frequency characteristics of sub-target sensor data, the vehicle's speed can be accurately determined. Frequency characteristics, particularly Doppler shift, are physical quantities that directly reflect the relative speed of a target. Therefore, determining the speed based on frequency characteristics provides a direct quantitative indicator of the vehicle's dynamic behavior. Second, after acquiring the vehicle's speed, it is combined with the duration of the vehicle's movement within the sensor's observation area for a comprehensive judgment. Speed ​​and duration are a comprehensive reflection of the vehicle's inherent attributes and motion state. For example, large vehicles typically have longer durations and relatively lower speeds (under specific traffic conditions), while small vehicles may have shorter durations and higher speeds. Through this multi-dimensional feature fusion analysis, different types and motion patterns of vehicles can be more accurately distinguished, overcoming the limitations of relying solely on a single feature for recognition.

[0106] Through the above technical solution, this application can significantly improve the accuracy and precision of vehicle identification in traffic flow detection. By introducing the key parameter of moving speed and combining it with the duration, the system can more effectively distinguish different types of vehicles, such as cars, trucks, and motorcycles, and even identify the vehicle's driving state (such as acceleration, deceleration, or constant speed). This not only improves the reliability of traffic flow statistics but also provides richer and more accurate data support for intelligent traffic management systems, contributing to more refined traffic scheduling and congestion prediction.

[0107] In some preferred embodiments, this application is implemented as follows: Assume a sensor detects sub-target sensing data. After analyzing its frequency characteristics, the moving speed corresponding to the sub-target sensing data is determined to be 60 km / h, and the duration of the sub-target sensing data within the sensor's observation area is 0.5 seconds. The system can preset a recognition rule base, which includes the moving speed range and duration range corresponding to different vehicle types. For example, the rule base may define: the moving speed range of a small car is 50-120 km / h, and the duration range is 0.3-0.8 seconds; the moving speed range of a large truck is 30-80 km / h, and the duration range is 1.0-3.0 seconds. According to the above rules, when sub-target sensing data with a moving speed of 60 km / h and a duration of 0.5 seconds is detected, the system can identify it as a small car. In this way, combined with the dynamic characteristics of vehicles, accurate classification and identification of vehicles in traffic flow is achieved.

[0108] Specifically, in the process of determining the moving speed corresponding to the sub-target sensing data based on the frequency characteristics of the sub-target sensing data, the following methods can be used.

[0109] A fast Fourier transform is performed on the frequency characteristics of the sub-target sensing data to obtain the Doppler frequency shift of the sub-target sensing data; half of the ratio of the Doppler frequency shift to the wavelength of the sensor signal is taken as the moving speed corresponding to the sub-target sensing data.

[0110] Specifically, the Fast Fourier Transform (FFT) is an efficient algorithm for calculating the Discrete Fourier Transform, aiming to convert time-domain signals into frequency-domain signals. Here, FFT is applied to the frequency characteristics of the sub-target sensing data to extract key frequency components from complex frequency features, particularly the Doppler shift related to target movement. The Doppler shift refers to the phenomenon where the frequency of a received wave changes when there is relative motion between the wave source and receiver. In traffic flow detection, the sensor signal emitted by the sensor is reflected back after encountering a moving vehicle; the frequency of the reflected signal shifts due to the vehicle's movement—this shift is the Doppler shift. The magnitude of the Doppler shift is proportional to the vehicle's speed. In practical applications, the wavelength of the sensor signal is an inherent physical parameter of the sensor's emitted signal, usually determined during sensor design. This wavelength is one of the key parameters for calculating the moving speed. Taking half the ratio of the Doppler shift to the wavelength of the sensor signal as the moving speed corresponding to the sub-target sensing data is a classic calculation method based on the Doppler effect. Specifically, the Doppler frequency shift Δf = 2*v / λ, where v is the target's moving speed and λ is the wavelength of the sensing signal. Therefore, the moving speed v = Δf*λ / 2.

[0111] The proposed solution utilizes a Fast Fourier Transform (FFT) on the frequency characteristics of the sub-target sensing data to effectively separate the Doppler frequency shift caused by vehicle movement from the original frequency features. The Doppler frequency shift directly reflects the vehicle's radial velocity relative to the sensor. By calculating the extracted Doppler frequency shift against the known wavelength of the sensor's signal, the corresponding movement velocity of the sub-target sensing data can be accurately derived. This physics-based calculation method ensures the accuracy and reliability of the velocity measurement.

[0112] The above technical solution enables the determination of vehicle speed in a physically accurate and computationally efficient manner based on the frequency characteristics of sub-target sensing data. This method utilizes the Doppler effect, resulting in high accuracy and robustness in vehicle speed measurement, providing reliable speed information for subsequent vehicle identification.

[0113] In one design, this application further proposes a step for identifying a vehicle based on the moving speed corresponding to the sub-target sensing data and the duration of the sub-target sensing data, including: obtaining a second correspondence; the second correspondence includes a one-to-one correspondence between multiple first pieces of information and multiple second pieces of information; the first pieces of information include a range of moving speed and a range of duration, and the second pieces of information include a vehicle type; the first pieces of information into which the moving speed and duration of the sub-target sensing data corresponding to the second correspondence fall are taken as target first information; and the vehicle type corresponding to the target first information of the second correspondence is taken as the target vehicle type corresponding to the sub-target sensing data.

[0114] Specifically, obtaining the second correspondence refers to establishing a predefined mapping table or rule set. This mapping table is used to associate the combination of vehicle speed and duration with specific vehicle types. The second correspondence can be understood as a lookup table or decision rule base, aiming to standardize and accelerate the vehicle identification process. In practical applications, the second correspondence includes a one-to-one correspondence between multiple sets of first information and multiple sets of second information. The first information specifically refers to the combination of speed and duration ranges. For example, speed can be divided into low, medium, and high speed ranges, and duration into short, medium, and long duration ranges. Different combinations of these ranges define different sets of first information. The second information specifically refers to the vehicle type corresponding to these combined features, such as a small car, medium-sized truck, large bus, motorcycle, or bicycle. During the identification process, the speed and duration of the sub-target sensing data are matched with the first information defined in the second correspondence. When the speed and duration fall within a specific speed and duration range, this combination is determined as the target's first information. Subsequently, based on the second correspondence, the vehicle type corresponding to the first information of the target is found, thereby determining the target vehicle type corresponding to the sub-target sensing data.

[0115] This application's solution introduces a pre-defined second correspondence to transform dynamically changing movement speed and duration features into directly queryable vehicle type information. It is precisely this structured mapping mechanism that allows the vehicle identification process to move beyond complex real-time calculations or fuzzy matching, enabling simple lookup and matching instead of relying on complex real-time calculations or fuzzy matching. This approach effectively solves the problems of insufficient identification accuracy and low computational efficiency that may exist in traditional identification methods, providing a more reliable and efficient data foundation for subsequent traffic flow statistics.

[0116] The above technical solutions significantly improve the accuracy and efficiency of vehicle recognition. By employing a predefined second correspondence, the recognition process becomes standardized and controllable, reducing recognition errors caused by environmental changes or data fluctuations. Furthermore, this lookup table-based recognition method has lower computational complexity compared to complex pattern recognition algorithms, thereby enhancing the system's real-time processing capabilities and enabling the traffic flow detection system to analyze and respond to traffic conditions more quickly and accurately.

[0117] This application proposes a sensor-based traffic flow detection system. The sensor is protected by a cover, with the sensor positioned on the first side of the cover. A millimeter-wave radar is also mounted on the first side of the cover, and a millimeter-wave radar signal reflection device is mounted on the second side of the cover. The system includes: an acquisition device and a processing device; the acquisition device is used to acquire the target signal quality index of the sensor and the initial sensing data of the sensor based on the millimeter-wave radar and the millimeter-wave radar signal reflection device; the initial sensing data includes multiple sub-initial sensing data; the processing device is used to correct the multiple sub-initial sensing data according to the target signal quality index to obtain target sensing data; the target sensing data includes multiple sub-target sensing data; the processing device is used to extract data features from the target sensing data; the data features include the duration and frequency characteristics of each sub-target sensing data; the processing device is used to identify vehicles based on the data features and the target signal quality index.

[0118] This application modularizes the traffic flow detection process into an acquisition device and a processing device, enabling real-time evaluation of sensor operating status, adaptive correction of sensor data, and accurate vehicle identification. The acquisition device is responsible for collecting raw sensor data and evaluating signal quality, while the processing device corrects the data, extracts features, and identifies vehicles based on this data. This systematic design effectively addresses the problems of sensor contamination and difficulty in identifying vehicles with weak signals, significantly improving the accuracy and reliability of traffic flow detection and providing more precise data support for urban traffic management.

[0119] The core of the sensor-based traffic flow detection system proposed in this application lies in the collaborative work of the acquisition device and the processing device to achieve real-time evaluation of the sensor's working status, adaptive correction of the sensor data, and accurate vehicle identification.

[0120] Specifically, the sensor is equipped with a protective cover, with the sensor positioned on the first side of the cover. A millimeter-wave radar is also mounted on the first side of the cover, and a millimeter-wave radar signal reflector is mounted on the second side of the cover. The detailed configuration of this structure has already been described in the above embodiments and will not be repeated here. It is important to emphasize that this configuration provides the physical basis for the millimeter-wave radar to detect the millimeter-wave radar signal reflector through the protective cover, thereby enabling the system to assess the contamination status of the protective cover.

[0121] The system of this application includes an acquisition device and a processing device.

[0122] The acquisition device is used to acquire the target signal quality index and initial sensing data of the sensor based on the millimeter-wave radar and the millimeter-wave radar signal reflection device. The initial sensing data includes multiple sub-initial sensing data. The acquisition device can be implemented as a standalone hardware module, such as an integrated circuit board containing a millimeter-wave radar control unit, a signal acquisition unit, and a preliminary data processing unit; or it can be a software module running on a general-purpose processor, communicating with the millimeter-wave radar and sensor through an interface. As a preferred embodiment, the acquisition device can periodically trigger the millimeter-wave radar to transmit millimeter-wave signals and receive signals reflected back by the millimeter-wave radar signal reflection device. By analyzing the intensity changes of the reflected signals, the contamination level of the protective cover is roughly estimated and converted into a target signal quality index. Simultaneously, the acquisition device directly receives raw analog or digital signals from the sensor and converts them into initial sensing data. In practical applications, the acquisition device can preset a signal quality assessment model. This model determines the target signal quality index by looking up tables or simple calculations based on the attenuation of the reflected signals received by the millimeter-wave radar and environmental parameters (such as temperature and humidity). Initial sensing data can be the raw, unprocessed echo signal sequence directly output by the sensor when it detects a vehicle passing by.

[0123] Furthermore, the processing device is used to correct multiple sub-initial sensing data according to the target signal quality index to obtain target sensing data. The target sensing data includes multiple sub-target sensing data. The processing device can be a microcontroller, digital signal processor (DSP), or general-purpose computer, internally running a data correction algorithm. Specifically, the processing device can dynamically adjust a signal gain coefficient based on the target signal quality index provided by the acquisition device and apply it to each sub-initial sensing data to compensate for signal attenuation. For example, when the target signal quality index is low, the processing device can apply a larger gain coefficient to amplify the initial sensing data; when the target signal quality index is high, a smaller gain coefficient is applied. In addition, the processing device can preset multiple correction curves or functions, and select an appropriate curve or function based on the target signal quality index to perform nonlinear correction on the initial sensing data, such as through adaptive filtering or noise suppression algorithms to optimize signal quality.

[0124] In addition, the processing device is also used to extract data features from the target sensing data. These data features include the duration and frequency characteristics of each sub-target sensing data point. The processing device performs feature extraction by executing signal processing algorithms. For example, the processing device can perform time-domain analysis on each sub-target sensing data point, calculating its duration by detecting the start and end points of the signal. For frequency characteristics, the processing device can perform spectral analysis on the sub-target sensing data, for example, by using Discrete Fourier Transform (DFT) or Short-Time Fourier Transform (STFT) to obtain the frequency components of the signal and extract the dominant frequency or frequency distribution range as frequency features. As a specific implementation, the processing device can employ time-frequency analysis methods such as wavelet transform to simultaneously obtain the time-domain and frequency-domain features of the signal and extract the duration and frequency characteristics from them.

[0125] Finally, the processing device is also used to identify vehicles based on data features and the target signal quality index. The processing device internally stores a vehicle identification model or rule base. Specifically, the processing device can match the extracted duration and frequency features with a preset vehicle type feature template. For example, a threshold can be set; when the duration and frequency features fall within a certain range, the vehicle is identified as a specific type. Simultaneously, the target signal quality index can serve as a weighting factor or confidence correction parameter in the identification process. For example, when the target signal quality index is low, the processing device can appropriately reduce the matching strictness or assign a lower confidence level to the identification result to avoid misjudgment. In some preferred embodiments, the processing device can employ a machine learning-based classifier, such as a support vector machine (SVM) or neural network, to learn the mapping relationship between data features and vehicle types through training, and utilize the target signal quality index as one of the input features to improve the accuracy of identification.

[0126] The sensor-based traffic flow detection system proposed in this application forms a complete and adaptive traffic flow detection scheme by introducing the collaborative work of the acquisition device and the processing device, which effectively overcomes the data distortion problem caused by sensor contamination and difficulty in identifying weak signal vehicles in traditional methods.

[0127] Specifically, traditional systems often lack effective self-diagnostic mechanisms to detect signal attenuation when faced with sensor shield contamination, leading to continuous output of distorted data. The acquisition device in this application, through the detection of millimeter-wave radar signal reflection devices by millimeter-wave radar, can acquire the target signal quality index of the sensor in real time, providing the system with an assessment of its own "health status." This innovation enables the system to proactively perceive and quantify the impact of the external environment on sensor performance, thus providing a reliable basis for subsequent data processing.

[0128] Furthermore, addressing the challenge of identifying weak signals generated by novel micro-vehicles, the processing device of this application corrects the initial sensing data using the acquired target signal quality index. This correction mechanism effectively compensates for signal attenuation caused by pollution and weak signals, enabling the effective identification of vehicles with weak signals that might otherwise be ignored by fixed discrimination thresholds in traditional systems. For example, the processing device can dynamically adjust the signal gain based on the signal quality index, thereby increasing the strength of weak signals to reach a recognizable threshold, significantly outperforming traditional systems.

[0129] By processing the corrected target sensing data through a processing device to extract data features (including duration and frequency features) and combining them with the target signal quality index for vehicle identification, the system of this application can more accurately determine the type and number of vehicles. For example, frequency features can be used to determine the vehicle's speed, while duration can help determine the vehicle's size or type. In cases of poor signal quality, the target signal quality index can serve as a basis for adjusting the identification algorithm parameters, further improving the accuracy and robustness of the identification.

[0130] In summary, the system of this application, through the close cooperation of the acquisition and processing devices, achieves intelligent perception of sensor operating status, adaptive optimization of sensor data, and accurate vehicle identification. Compared with existing technologies, the system of this application can significantly improve the accuracy and reliability of traffic flow data, providing more precise data support for urban traffic management, thereby optimizing traffic signals, providing congestion warnings, and planning traffic resources, ultimately improving the overall operational efficiency of urban traffic and user experience.

[0131] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A sensor-based traffic flow detection method, characterized in that, The sensor is equipped with a protective cover, and the sensor is disposed on a first side of the protective cover. A millimeter-wave radar is also disposed on the first side of the protective cover, and a millimeter-wave radar signal reflecting device is also disposed on a second side of the protective cover. The method includes: The target signal quality index of the sensor and the initial sensing data of the sensor are obtained based on the millimeter-wave radar and the millimeter-wave radar signal reflection device; the initial sensing data includes multiple sub-initial sensing data. The target sensing data is obtained by correcting the plurality of sub-initial sensing data according to the target signal quality index; the target sensing data includes multiple sub-target sensing data. Extract data features from the target sensing data; the data features include the duration and frequency characteristics of each sub-target sensing data. Vehicles are identified based on the data characteristics and the target signal quality index.

2. The sensor-based traffic flow detection method according to claim 1, characterized in that, The target signal quality index of the sensor is obtained based on the millimeter-wave radar and the millimeter-wave radar signal reflection device, including: Control the transmitting antenna of the millimeter-wave radar to transmit the raw millimeter-wave radar signal; The target millimeter-wave radar signal received by the receiving antenna of the millimeter-wave radar is acquired; the target millimeter-wave radar signal is the reflected signal of the original millimeter-wave radar signal reflected by the millimeter-wave radar signal reflecting device. The contamination index of the protective shield is determined based on the target millimeter-wave radar signal; The target signal quality index of the sensor is determined based on the pollution index.

3. The sensor-based traffic flow detection method according to claim 2, characterized in that, The contamination index of the protective shield is determined based on the target millimeter-wave radar signal, including: Obtain the reference peak intensity and reference waveform characteristics of the reflected signal of the original millimeter-wave radar signal; The ratio of the difference between the reference peak intensity and the target peak intensity of the target millimeter-wave radar signal to the reference peak intensity is used as the peak intensity deviation index of the protective cover; The mean square error between the target waveform characteristics and the reference waveform characteristics of the target millimeter-wave radar signal is normalized to obtain the waveform characteristic deviation index of the protective cover. The weighted sum of the peak intensity deviation index and the waveform characteristic deviation index is used as the pollution index of the protective cover.

4. The sensor-based traffic flow detection method according to claim 2, characterized in that, Determining the target signal quality index of the sensor based on the pollution index includes: Acquire test sensing data of the sensor on the target area; the target area is a preset area in the observation area of ​​the sensor; The background reflection feature drift index of the sensor is determined based on the test sensing data; the value range of the background reflection feature drift index is 0-1. The weighted sum of the background reflection feature drift index and the contamination index of the protective cover is used as the target signal quality index of the sensor.

5. The sensor-based traffic flow detection method according to claim 4, characterized in that, Determining the background reflectance feature drift index of the sensor based on the test sensing data includes: Obtain the reference average echo intensity and reference spectral width of the sensor's sensing data of the target area; The ratio of the difference between the reference average echo intensity and the test average echo intensity of the test sensor data to the reference average echo intensity is used as the background feature drift index of the sensor. The ratio of the difference between the reference spectral width and the test spectral width of the test sensor data to the reference spectral width is used as the spectral width drift index of the sensor. The weighted sum of the background feature drift index and the spectral width drift index is used as the background reflection feature drift index of the sensor.

6. The sensor-based traffic flow detection method according to claim 1, characterized in that, The initial sensing data is corrected based on the target signal quality index to obtain target sensing data, including: Obtain the first correspondence; the first correspondence includes a one-to-one correspondence between multiple signal quality index ranges and multiple signal compensation gains; The signal compensation gain corresponding to the target signal quality index in the first correspondence is taken as the target signal compensation gain. The initial sensing data is corrected based on the target signal compensation gain to obtain the target sensing data.

7. The sensor-based traffic flow detection method according to claim 1, characterized in that, Vehicle identification based on the data features and the target signal quality index includes: For each sub-target sensing data in multiple sub-target sensing data, the moving speed corresponding to the sub-target sensing data is determined based on the frequency characteristics of the sub-target sensing data; The vehicle is identified based on the moving speed corresponding to the sub-target sensing data and the duration of the sub-target sensing data.

8. The sensor-based traffic flow detection method according to claim 7, characterized in that, Determining the moving speed corresponding to the sub-target sensing data based on the frequency characteristics of the sub-target sensing data includes: The frequency characteristics of the sub-target sensing data are subjected to a fast Fourier transform to obtain the Doppler frequency shift of the sub-target sensing data; Half of the ratio of the Doppler frequency shift to the wavelength of the sensor signal is taken as the moving speed corresponding to the sub-target sensing data.

9. A sensor-based traffic flow detection method according to claim 7, characterized in that, Vehicle identification is performed based on the moving speed corresponding to the sub-target sensing data and the duration of the sub-target sensing data, including: Obtain a second correspondence; the second correspondence includes a one-to-one correspondence between multiple first pieces of information and multiple second pieces of information; the first information includes a range of movement speed and a range of duration, and the second information includes vehicle type; The first information into which the moving speed corresponding to the sub-target sensing data and the duration of the sub-target sensing data fall in the second correspondence relationship shall be taken as the first information of the target; The vehicle type corresponding to the first information of the target in the second correspondence is taken as the target vehicle type corresponding to the sub-target sensing data.

10. A sensor-based traffic flow detection system, characterized in that, The sensor is provided with a protective cover, and the sensor is located on the first side of the protective cover. A millimeter-wave radar is also provided on the first side of the protective cover, and a millimeter-wave radar signal reflection device is also provided on the second side of the protective cover. The system includes: an acquisition device and a processing device. The acquisition device is used to acquire the target signal quality index of the sensor and the initial sensing data of the sensor based on the millimeter-wave radar and the millimeter-wave radar signal reflection device; the initial sensing data includes multiple sub-initial sensing data. The processing device is used to correct the plurality of sub-initial sensing data according to the target signal quality index to obtain target sensing data; the target sensing data includes a plurality of sub-target sensing data; The processing device is used to extract data features from the target sensing data; the data features include the duration and frequency features of each sub-target sensing data. The processing device is used to identify vehicles based on the data features and the target signal quality index.

Images