A real-time measurement method for vehicle acceleration based on distributed optical fiber sensing

Abstract

The present invention discloses a real-time vehicle acceleration measurement method based on distributed fiber optic sensing. The method sends pulses to a sensing fiber pre-laid along the road below. The DAS system collects and obtains the strain space signal at each moment when the vehicle passes through the sensing fiber. The strain space signal at each moment is analyzed for asymmetry, and the peak values ​​of the signal areas corresponding to the front and rear axles of the vehicle at each moment are found. The difference between the two peak values ​​is calculated to obtain an asymmetry index for measuring the vehicle load. The obtained asymmetry index is imported into a preset attenuation model, and the vehicle acceleration is output in real time. The present invention utilizes the strain space signal obtained at each moment when the vehicle passes through the sensing fiber and performs asymmetry analysis, establishing a correlation mechanism between the dynamic load transfer formed by the vehicle during acceleration and the asymmetry of the strain space signal, significantly improving the real-time performance and accuracy of acceleration measurement.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation systems, and in particular to a real-time vehicle acceleration measurement method based on distributed optical fiber sensing. Background Art

[0002] As intelligent transportation applications expand, roadside sensors, as core system components, must meet the demands of massive data collection and interact in real time with connected vehicles and road infrastructure to support decision-making. Current mainstream roadside sensing technologies include invasive sensors (such as induction coils) and non-invasive sensors (such as lidar, vision sensors, and millimeter-wave radar). However, these technologies all have limitations in terms of ease of installation, maintenance costs, and resistance to environmental interference. For example, the deployment and maintenance of induction coils often require disruption to the pavement structure, increasing the complexity of construction and maintenance. Vision sensors and millimeter-wave radar are sensitive to environmental conditions and prone to performance degradation in inclement weather. Although vision and radar sensors have improved their detection range to hundreds of meters compared to traditional invasive induction coils, their coverage remains limited, with a single device capable of monitoring only a portion of the road. Continuous monitoring of long highways still requires the deployment of multiple sensor nodes to meet coverage requirements, resulting in high deployment and maintenance costs.

[0003] In recent years, an emerging fiber optic sensing technology called DAS has garnered widespread attention as a key complement to traditional roadside sensing technologies, potentially addressing some of the shortcomings of traditional roadside sensing. DAS connects to existing roadside communication optical cables, converting them into vibration sensing arrays. As vehicles pass over the road, the strain caused by their vibrations is transmitted to the optical cables, which then detect the strain, enabling vehicle monitoring. This offers numerous advantages: First, DAS can achieve a detection range of tens of kilometers, meeting the needs of long-distance detection. Second, DAS can be deployed using existing communication optical cables, resulting in low installation and maintenance costs. Finally, DAS strain detection is not tied to a specific vehicle, thus avoiding privacy concerns.

[0004] Existing DAS technology for vehicle acceleration detection still has the following limitations: It fails to address the impact of dynamic parameters (such as acceleration) on the signal. Currently, only the effects of factors such as the distance between the vehicle and the optical cable, the DAS gauge length, speed, and vehicle weight on the quasi-static signal waveform have been studied. Load transfer during acceleration has been neglected, leading to theoretical biases in quasi-static strain signal analysis. This limits the ability to extract vehicle dynamic information from DAS quasi-static signals.

[0005] In addition, acceleration can cause dynamic load transfer, breaking the symmetry of the quasi-static signal. Existing methods do not analyze this, resulting in acceleration measurement relying on the velocity differential method (needing to first locate → speed measurement → time differentiation), introducing measurement delay (>1 second) and error accumulation.

[0006] The lanes far away from the optical cable have their strain strength attenuated due to the increased signal transmission distance. The existing methods have not established a unified signal compensation mechanism, and the measurement consistency across lanes is poor. Summary of the Invention

[0007] In response to the problems existing in the prior art, the present invention provides a real-time vehicle acceleration measurement method based on distributed fiber optic sensing. Through the correlation mechanism of dynamic load transfer and quasi-static strain signal asymmetry, large-scale, high-real-time, and multi-lane synchronous vehicle acceleration monitoring is achieved, solving the problems of reliance on velocity differentials and poor multi-lane consistency in the prior art.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0009] A method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing includes the following steps:

[0010] S10, sending a pulse to a sensing optical fiber pre-laid under the road along the road direction;

[0011] S20, the DAS system collects and obtains the strain spatial signal at each moment when the vehicle passes through the sensing optical fiber;

[0012] S30, performing an asymmetry analysis on the strain spatial signal at each moment, finding the peak values ​​of the signal regions corresponding to the front and rear axles of the vehicle at each moment, and calculating the difference between the two peak values ​​to obtain an asymmetry index for measuring the vehicle load;

[0013] S40: Import the obtained asymmetry index into a preset attenuation model, and output the acceleration of the vehicle in real time.

[0014] Specifically, in S30, a peak detection algorithm is first used to determine the peak with the largest absolute value from the strain space signal at a certain moment, and the peak is used as the position information of the vehicle at that moment, and the peak is defined as the center point of the vehicle. Then, the signal areas corresponding to the front and rear axles of the vehicle are determined in combination with the vehicle driving information and the vehicle wheelbase parameters. Finally, the peak detection algorithm is used to determine the local peak values ​​of the strain space signal in the signal areas corresponding to the front and rear axles of the vehicle, and the peak values ​​are respectively recorded as the peak values ​​of the signal areas corresponding to the front and rear axles of the vehicle.

[0015] Specifically, when calculating the asymmetry index in S30 , the peak value of the signal area corresponding to the front axle of the vehicle is subtracted from the peak value of the signal area corresponding to the rear axle of the vehicle, and the obtained difference is used as the asymmetry index.

[0016] Specifically, the vehicle driving information in S30 includes the vehicle's driving direction and driving track.

[0017] Specifically, the configuration process of the preset attenuation model in S40 includes:

[0018] The data of different lanes are divided into different subsets according to the distance from the lane to the sensing fiber;

[0019] Time-align the asymmetry index calculated according to the S30 process with the actual detected acceleration data to maintain a consistent data sampling rate;

[0020] Establishing the asymmetry index VLAI and acceleration The attenuation model shows a linear relationship between:

[0021]

[0022] In the above formula, the slope k and intercept b are both attenuation model parameters;

[0023] Each lane is analyzed separately, and the asymmetry index and acceleration data are fitted by the least squares method to solve the slope k and intercept b. The value of k is used to distinguish the attenuation coefficients corresponding to different lanes. Then, independent attenuation model parameters are calibrated for each lane to form a preset attenuation model.

[0024] Specifically, when the obtained asymmetry index is imported into the preset attenuation model in S40, the lane of the vehicle corresponding to the asymmetry index is first determined based on the strain space signal, and then the preset attenuation model is configured by calling the corresponding attenuation model parameters according to the determined lane.

[0025] Specifically, when determining the lane of the vehicle corresponding to the asymmetry index based on the strain spatial signal in S40, the direction of the lane of the corresponding vehicle is first determined using the direction of the vehicle's driving trajectory, and then the lane position of the corresponding vehicle is determined using the average strain intensity of the strain spatial signal among multiple lanes in the same direction.

[0026] Specifically, in S20, the backscattered signal caused by the vehicle is collected in real time by the DAS system to generate a spatiotemporal strain matrix with a dimension of M×N, where M is the number of spatial channels and N is the number of time channels; then, based on the vehicle's driving trajectory, the strain spatial signal at each moment when the vehicle passes through the sensing optical fiber is extracted from the spatiotemporal strain matrix. When only the target vehicle exists in the spatiotemporal strain matrix, the strain influence range of the target vehicle's driving process is directly obtained.

[0027] In S20, when there are multiple vehicles traveling in the spatiotemporal strain matrix, the target vehicle's driving trajectory is separated by a convolutional neural network to obtain the strain influence range of the target vehicle's driving process.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] (1) The present invention uses the vehicle passing through the sensing optical fiber to obtain the strain space signal at each moment and perform asymmetry analysis, establishing a correlation mechanism between the dynamic load transfer formed when the vehicle accelerates and the asymmetry of the strain space signal, reducing the acceleration measurement delay from seconds to milliseconds, significantly improving the real-time performance, and eliminating the accumulated errors caused by traditional differential calculations, thereby improving the measurement accuracy of vehicle acceleration.

[0030] (2) By independently calibrating the attenuation model parameters for each lane, the present invention can effectively compensate for the signal attenuation differences between different lanes, thereby improving the accuracy and consistency of multi-lane measurements.

[0031] (3) This invention supports real-time monitoring, facilitating real-time decision-making and response in intelligent transportation systems, and improving traffic safety and management efficiency. For example, in scenarios such as emergency braking and speeding monitoring, real-time and accurate acceleration data can trigger early warning or control measures more quickly, enhancing road safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the overall process of an embodiment of the present invention.

[0033] Figure 2 Schematic diagram of the strain spatial signal at a moment collected by the DAS system in the experiment in the embodiment of the present invention.

[0034] Figure 3 The original image data of the lane near the optical cable collected by the DAS system used in the experiment in the embodiment of the present invention.

[0035] Figure 4 This is image data of a vehicle trajectory spatial signal on a lane close to an optical cable extracted experimentally in an embodiment of the present invention.

[0036] Figure 5This is a comparison diagram of the real-time acceleration of a vehicle on a lane close to an optical cable and the on-board acceleration sensor obtained through experiments in an embodiment of the present invention.

[0037] Figure 6 The original image data of the lane far away from the optical cable is collected by the DAS system used in the experiment in the embodiment of the present invention.

[0038] Figure 7 This is a comparison diagram of the real-time acceleration of a vehicle on a lane far away from the optical cable and the on-board acceleration sensor obtained through experiments in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described below with reference to the accompanying drawings and examples. The embodiments of the present invention include but are not limited to the following examples.

[0040] like Figures 1 to 7 As shown, the real-time vehicle acceleration measurement method based on distributed optical fiber sensing includes the following steps:

[0041] S10. Send pulses to a sensing optical fiber pre-laid under the road along the road direction; wherein the sensing optical fiber can utilize an existing communication optical cable laid along the road as a distributed sensing optical fiber. The existing communication optical cable is usually buried in a trench or placed under a non-motorized vehicle lane or sidewalk on the side of the road; it can also be newly buried in the road according to actual application requirements, such as during road reconstruction or new road construction.

[0042] S20. The DAS system collects and obtains the spatial strain signal at each moment when a vehicle passes through the sensing optical fiber. Specifically, the DAS system collects the backscattered signal caused by the vehicle in real time, generating a spatiotemporal strain matrix of dimension M×N, where M is the number of spatial channels (the number of sampling points along the optical cable, corresponding to the spatial resolution) and N is the number of temporal channels (the number of sampling time points, corresponding to the length of the time series). The spatial strain signal at each moment when the vehicle passes through the sensing optical fiber is then extracted from the spatiotemporal strain matrix based on the vehicle's trajectory. When only the target vehicle is present in the spatiotemporal strain matrix, the strain influence range of the target vehicle's travel is directly obtained. When multiple vehicles are present in the spatiotemporal strain matrix, the target vehicle's trajectory is separated using a convolutional neural network to obtain the strain influence range of the target vehicle's travel.

[0043] S30. Perform an asymmetry analysis on the strain spatial signal at each moment, find the peak values ​​of the signal regions corresponding to the front and rear axles of the vehicle at each moment, and calculate the difference between the two peak values ​​to obtain a Vehicle Load Asymmetry Index (VLAI) to measure the vehicle load asymmetry.

[0044] Specifically, a peak detection algorithm is first used to determine the peak with the largest absolute value from the strain space signal at a certain moment, which is used as the vehicle's position information at that moment and is defined as the vehicle's center point. The vehicle's driving information and wheelbase parameters are then combined to determine the signal areas corresponding to the front and rear axles of the vehicle. Finally, a peak detection algorithm is used to determine the local peaks of the strain space signal in the signal areas corresponding to the front and rear axles of the vehicle, and these are recorded as the peaks of the signal areas corresponding to the front and rear axles of the vehicle. When calculating the asymmetry index, the peak value of the signal area corresponding to the front axle of the vehicle is subtracted from the peak value of the signal area corresponding to the rear axle of the vehicle, and the difference is used as the asymmetry index. Vehicle driving information includes the vehicle's driving direction and trajectory.

[0045] S40: Import the obtained asymmetry index into a preset attenuation model, and output the acceleration of the vehicle in real time.

[0046] Specifically, the configuration process of the preset attenuation model in S40 includes:

[0047] The data of different lanes are divided into different subsets according to the distance from the lane to the sensing fiber;

[0048] The asymmetry index calculated according to the S30 process is time-aligned with the actual measured acceleration data. A sliding average filter with a time window (e.g., 0.5 seconds) can be applied to the VLAI and acceleration data to eliminate high-frequency noise. The sensor data is upsampled to maintain the same data sampling rate as the VLAI. For example, if the DAS system sampling frequency is 250 Hz, the acceleration sensor is upsampled to the same frequency.

[0049] Establishing the asymmetry index VLAI and acceleration The attenuation model shows a linear relationship between:

[0050]

[0051] In the above formula, the slope k and intercept b are both attenuation model parameters;

[0052] Each lane is analyzed separately, using the least squares method to fit the asymmetry index and acceleration data to determine the slope k and intercept b. The value of k is used to distinguish the attenuation coefficients corresponding to different lanes. Independent attenuation model parameters are then calibrated for each lane to form a pre-defined attenuation model. For individual analysis, data segments under typical vehicle operating conditions, such as constant speed, acceleration, and deceleration, can be selected to exclude anomalous segments that exhibit signal distortion or noise.

[0053] Specifically, when the obtained asymmetry index is imported into the preset attenuation model in S40, the lane corresponding to the vehicle in question is first determined based on the strain spatial signal. Then, the preset attenuation model is configured based on the determined lane by calling the corresponding attenuation model parameters. When determining the lane corresponding to the vehicle in question based on the strain spatial signal, the direction of the vehicle's lane is first determined using the direction of the vehicle's travel trajectory. Then, the average strain intensity of the strain spatial signal is used to determine the vehicle's lane position among multiple lanes in the same direction. The average strain intensity of the strain spatial signal refers to the average strain amplitude over the entire range of the strain spatial signal.

[0054] The effectiveness of the method of the present invention was verified by the following experiments.

[0055] The experimental environment was a road on the Sichuan University campus. A communication fiber optic cable laid beneath the non-motorized vehicle lane served as the sensing fiber. The cable ran parallel to the road, and a distributed fiber acoustic sensor (DAS) device was deployed to connect to the 1.3 km long communication fiber. During data collection, a 250 Hz temporal sampling rate, a 3.2 m spatial sampling interval, and a 3.2 m gauge length were used.

[0056] The experiment used a privately owned sedan equipped with a capacitive triaxial accelerometer to record vehicle speed in real time. This acceleration data served as a reference for actual vehicle acceleration and was compared with the experimental results obtained using the acceleration measurement scheme presented in this paper. The accelerometer's measurement frequency was 200 Hz, enabling more detailed tracking of vehicle acceleration changes. The overall experimental speed was kept within a relatively low safety range, and the acceleration was controlled within a range of [-0.2g to 4g] to ensure the safety of the experimental process.

[0057] The DAS system uses backward Rayleigh scattered light to collect the original vibration signal of the vehicle, and then phase demodulates the collected raw data to obtain the strain signal. A low-pass filter is used to denoise the data to remove environmental noise and high-frequency noise. The strain signal is an M×N matrix. The detection window size win_size is set according to the length of the target vehicle to determine the analysis range centered on the center point of the vehicle. Starting from the first time channel where the DAS records the vehicle's travel, it is processed sequentially to the last time channel. Through the peak detection algorithm, the center point of the vehicle strain signal collected by the DAS is found in the spatial channel direction (along the length of the optical cable), and its coordinates are marked as (x_pos, y_pos), where: x_pos represents the current position of the vehicle on the optical cable (spatial channel index), and y_pos represents the current time channel index. The waveform of the vehicle strain signal collected by the DAS is as follows Figure 2 As shown, it can be obtained from Figure 2Three peaks of the vehicle strain signal are marked with dots. The peak detection algorithm locates these three peaks within the spatial signal range. The peaks on either side have opposite signs to the peak in the center, and the center peak has the largest absolute amplitude. The center peak is then used as the center point of the vehicle strain signal collected by the DAS. The vehicle's position is updated in real time based on the DAS's sampling rate, ensuring real-time and accurate positioning. In this experiment, the vehicle's position information is updated at a frequency of 250 Hz, the same as the DAS's data sampling frequency.

[0058] Determine the real-time position of the vehicle on the optical cable to provide a benchmark for subsequent window division. Search for the local maximum of the strain signal in the left and right windows and record the peak value. The spatial range of the left window is defined as from x_pos -win_size to x_pos (the area to the left of the vehicle center point). The spatial range of the right window is defined as from x_pos to x_pos +win_size (the area to the right of the vehicle center point). The division range of the left and right windows needs to be dynamically adjusted according to the vehicle wheelbase (such as 2.8-3.5m) and the optical cable gauge length (such as 3.2m) to match the strain areas corresponding to the front and rear axles. The peak values ​​found in the left and right windows are Figure 2 The peak values ​​on both sides are marked with dots. The subtraction of the left and right window peaks is used to measure the VLAI collected by the DAS. The acceleration is calculated based on the parameters of the corresponding lane according to the vehicle's real-time position.

[0059] like Figure 3 The image shows the vehicle driving data collected by the DAS system. The thick gray lines mark the vehicle driving trajectory. It can be seen that the vehicle driving trajectory is affected by environmental noise interference and other road events. Figure 3 The color brightness changes of the vehicle trajectory data image (the brighter the image, the higher the strain intensity) are directly related to the vehicle acceleration and deceleration events. Figure 4 This is the result image after using the convolutional neural network to extract the vehicle driving signal. It can be found that the convolutional neural network removes the driving tracks of other vehicles and environmental noise and only retains the driving track of the target vehicle. Figure 5 The real-time acceleration measurement method proposed by the present invention is Figure 4 The data shows a high degree of agreement between the acceleration recognition results and the on-board acceleration results, confirming that the acceleration measurement accuracy of the present invention meets the requirements for real-time acceleration measurement accuracy. The DAS system has a sampling frequency of 250 Hz, which means that millisecond-level acceleration monitoring can be achieved. Figure 6 is the original strain signal image far away from the cable lane, and Figure 3 In contrast, the overall intensity of the strain signal is attenuated (darker color). This is because the far lane is far away from the optical fiber and the strain is attenuated during the propagation process. Figure 7This figure compares real-time acceleration in lanes away from the fiber optic cable using the attenuation compensation model. The cross-lane acceleration measurements using the attenuation model are similar to those from the onboard accelerometer, demonstrating the effectiveness of the attenuation model. The model supports stable acceleration measurement in multi-lane scenarios.

[0060] The above embodiments are only preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any changes that adopt the design principles of the present invention and any changes made through non-creative work on this basis should fall within the scope of protection of the present invention.

Claims

1. A method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing, characterized in that: The following steps are involved: S10, sending a pulse to a sensing optical fiber pre-laid under the road along the road direction; S20, the DAS system collects and obtains the strain spatial signal at each moment when the vehicle passes through the sensing optical fiber; S30, performing an asymmetry analysis on the strain spatial signal at each moment, finding the peak values ​​of the signal regions corresponding to the front and rear axles of the vehicle at each moment, and calculating the difference between the two peak values ​​to obtain an asymmetry index for measuring the vehicle load; S40, importing the obtained asymmetry index into a preset attenuation model, and outputting the acceleration of the vehicle in real time; The configuration process of the preset attenuation model includes: The data of different lanes are divided into different subsets according to the distance from the lane to the sensing fiber; Time-align the asymmetry index calculated according to the S30 process with the actual detected acceleration data to maintain a consistent data sampling rate; Establishing the asymmetry index VLAI and acceleration The attenuation model shows a linear relationship between: In the above formula, the slope k and intercept b are both attenuation model parameters; Each lane is analyzed separately, and the asymmetry index and acceleration data are fitted by the least squares method to solve the slope k and intercept b. The value of k is used to distinguish the attenuation coefficients corresponding to different lanes. Then, independent attenuation model parameters are calibrated for each lane to form a preset attenuation model.

2. The method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing according to claim 1, characterized in that: In S30, a peak detection algorithm is first used to determine the peak with the largest absolute value from the strain space signal at a certain moment, and the peak value is used as the position information of the vehicle at that moment, and the peak value is defined as the center point of the vehicle. Then, the signal areas corresponding to the front and rear axles of the vehicle are determined in combination with the vehicle driving information and the vehicle wheelbase parameter. Finally, a peak detection algorithm is used to determine the local peak values ​​of the strain space signal in the signal areas corresponding to the front and rear axles of the vehicle, respectively, and the peak values ​​are recorded as the peak values ​​of the signal areas corresponding to the front and rear axles of the vehicle, respectively.

3. The method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing according to claim 2, characterized in that: When calculating the asymmetry index in S30 , the peak value of the signal region corresponding to the front axle of the vehicle is subtracted from the peak value of the signal region corresponding to the rear axle of the vehicle, and the obtained difference is used as the asymmetry index.

4. The method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing according to claim 2, characterized in that: The vehicle driving information includes the vehicle's driving direction and driving track.

5. The method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing according to claim 1, characterized in that: When the obtained asymmetry index is introduced into the preset attenuation model in S40, the lane of the vehicle corresponding to the asymmetry index is first determined based on the strain space signal, and then the preset attenuation model is configured by calling the corresponding attenuation model parameters according to the determined lane.

6. The method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing according to claim 5, characterized in that: When determining the lane in which the vehicle corresponding to the asymmetry index is located based on the strain spatial signal in S40, the direction of the lane in which the corresponding vehicle is located is first determined using the direction of the vehicle's driving trajectory. Then, among multiple lanes in the same direction, the average strain intensity of the strain spatial signal is used to determine the lane position in which the corresponding vehicle is located.

7. The method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing according to any one of claims 1 to 6, characterized in that: In S20, the backscattered signal caused by the vehicle is collected in real time by the DAS system to generate a spatiotemporal strain matrix with a dimension of M×N, where M is the number of spatial channels and N is the number of temporal channels. Then, based on the vehicle's driving trajectory, the strain spatial signal at each moment when the vehicle passes through the sensing optical fiber is extracted from the spatiotemporal strain matrix. When only the target vehicle exists in the spatiotemporal strain matrix, the strain influence range of the target vehicle during its driving process is directly obtained.

8. The method for real-time measurement of vehicle acceleration based on distributed optical fiber sensing according to claim 7, characterized in that: When there are multiple vehicles traveling in the spatiotemporal strain matrix, the target vehicle's driving trajectory is separated by a convolutional neural network to obtain the strain influence range of the target vehicle's driving process.

Images