A vehicle prediction method based on distributed optical fiber sensing and adaptive correction

Abstract

A vehicle prediction method based on distributed optical fiber sensing and adaptive correction. This method locates vehicles through the full spatio-temporal information collected by an optical fiber sensor array, corrects the positioning results by using the continuity of vehicle movement and driving history, and establishes a dynamic model based on the short-term position change of the vehicle to achieve accurate prediction of vehicle information. This method does not require the use of specially designed optical fiber structures or complex optical fiber laying methods and is applicable to the communication optical cables buried by the roadside. The present invention has the advantages of simple implementation, rapid response, low cost, wide coverage, etc., can effectively promote the construction and development of the Internet of Things network, is applicable to aspects such as vehicle monitoring, intelligent transportation, and smart city, and has great significance.

Description

Technical Field

[0001] The present invention belongs to the technical field of distributed optical fiber sensing and application, and in particular relates to a vehicle prediction method based on distributed optical fiber sensing and adaptive correction. Background Art

[0002] Distributed fiber optic sensing systems use optical fiber as a sensing medium and use Rayleigh scattering inside the optical fiber for sensing. When the outside world applies strain to the optical fiber, the change in the length and refractive index of the optical fiber causes the phase of the Rayleigh scattered light to change. By demodulating the phase change, the external information can be quantitatively sensed. Due to the low cost, small size, high temperature resistance, and corrosion resistance of the optical fiber itself, it has been widely used in perimeter security, geophysics and other fields in recent years. With the help of communication optical cables in cities, distributed fiber optic sensing systems are used to sense vehicle information on the road, so as to count the traffic volume and speed, which is of great significance to urban traffic management.

[0003] Predicting the vehicle status can obtain the vehicle's driving trajectory in the future, including the vehicle's route, arrival time, stay time, etc., which is of great significance to the navigation system and vehicle scheduling. In addition, predicting the short-term distance, speed, lane change and other factors of the vehicle can effectively reduce the risk of traffic accidents and protect people's lives and property. Vehicle prediction is usually based on the current state to predict the state at the next moment, and the acquisition of the current vehicle state usually requires the use of fixed vehicle detectors, such as infrared sensors, geomagnetic sensors, video detectors, etc. However, these instruments are susceptible to environmental influences, complex installation methods, and limited detection accuracy. Due to their high prices, they are usually only fixed at certain locations on the road, and the vehicle status cannot be obtained at all times, making it difficult to make accurate predictions of the vehicle status continuously. Summary of the invention

[0004] Based on the above problems, the present invention provides a vehicle prediction method based on distributed optical fiber sensing and adaptive correction to solve the technical problems of existing vehicle prediction methods such as high cost, limited scenarios and low prediction accuracy.

[0005] The technical solution adopted by the present invention is as follows:

[0006] A vehicle prediction method based on distributed optical fiber sensing and adaptive correction comprises the following steps:

[0007] Step 1: The optical fiber sensing system transmits detection light pulses to the optical fibers laid on both sides of the road, and quantitatively receives the returned sensing signals R(l,t), where l represents the axial distance of the sensing optical fiber and t represents the receiving time;

[0008] Step 2: Perform phase demodulation on the sensing signal R(l,t), use parallel operations to accelerate the demodulation process, and obtain the phase information Φ(l,t) in real time;

[0009] Step 3: Calculate the standard deviation of the phase, preliminarily determine the vehicle position range, select N sensing channels within this range, and use the channel with the highest signal-to-noise ratio as the reference channel. Calculate the time delay Δτ = [Δτ1, Δτ2,..., Δτ N , where 1, 2,... N represent the number of sensing channels; this process ensures the accuracy of the time difference of arrival calculation, and N is determined according to actual needs;

[0010] Step 4: To compensate for the actual uncertainties, add a random time offset to the time delay Δτ, then use the linear TDOA positioning algorithm to locate the vehicle and repeat this process a preset number of times to average the located vehicle position where is the polar coordinate angle;

[0011] Step 5: According to obtain the two-dimensional position of the vehicle, match the best landing point according to the vehicle's driving trajectory, and adaptively correct the positioning result to obtain the vehicle position (x, y);

[0012] Step 6: Establish a position-time model within W time windows following the recent trajectory changes of the vehicle, and accurately predict the vehicle position (x’, y’) at subsequent times based on this position-time model;

[0013] Step 7: Update the sensing channels and time windows, repeat the above process, and achieve continuous prediction of the vehicle.

[0014] Furthermore, the parallel operations use GPU multi-threading, FPGA multi-logic units, or other parallel operation methods. The time for this process is usually on the order of microseconds, during which the vehicle state hardly changes.

[0015] Furthermore, the time offset follows a normal distribution with a mean of 0 and a standard deviation of σ. This process compensates for errors caused by non-constant wave speed, dispersion effects, or other reasons.

[0016] Furthermore, the adaptive correction specifically includes: According to Based on the TDOA positioning algorithm, the two-dimensional position (x^, y^) of the vehicle is obtained. Based on the previously traveled path, the vehicle trajectory curve is fitted, and the optimal position (x^’, y^’) is matched. The deviation between the two is calculated. If the deviation between the two is less than the spatial resolution of the system, the position of the vehicle is (x, y) = [(x^, y^) + (x^’, y^’)] / 2; otherwise, the position of the vehicle (x, y) = (x^’, y^’); the spatial resolution of the sensing system is determined by the detection pulse, usually on the order of meters, and can accurately locate the vehicle.

[0017] Further, the position-time model is an algebraic polynomial relationship. The specific establishment of the position-time model is as follows:

[0018] When driving at a constant speed, P(t) = P0 + v·t;

[0019] When driving with uniform acceleration, P(t) = P0 + v0·t + a·t 2 ;

[0020] For other normal driving states, P(t) = P0 + a·t + b·t 2 + c·t 3 。

[0021] Therefore, a corresponding model can be established according to multiple recent position-time relationships. The size of the time window W is related to the sampling frequency of the system and should be determined according to the actual scenario.

[0022] Further, based on the vehicle position (x’, y’) predicted in step 6, the vehicle speed is predicted by taking the derivative with respect to time, and the vehicle acceleration is further predicted by taking the derivative with respect to time.

[0023] Further, the fiber optic sensing system includes a laser, an optical coupler, a modulator, a circulator, a photodetector, an ADC acquisition device, and a computing device; the laser is connected to two optical couplers. The first optical coupler is connected to the modulator to generate a detection optical pulse. The detection optical pulse enters the optical fiber through the circulator, and the generated Rayleigh scattered light returns to the second optical coupler through the circulator. The second optical coupler is connected to the detector, and after passing through the ADC acquisition device, further operations are performed using the computing device.

[0024] Further, the laser uses a narrow linewidth laser, and the detection optical pulse uses a single-frequency pulse with a pulse width in the nanosecond range.

[0025] Further, the light output from the laser is connected to the second optical coupler as the local oscillator light. The second optical coupler receives the Rayleigh scattered light, and the two parts of light are beat to obtain a sensing signal.

[0026] The advantages and technical effects of the present invention are as follows:

[0027] 1. The present invention is applicable to existing distributed optical fiber systems, which have low cost, simple implementation, and wide coverage. Based on optical fiber as the sensing medium, it can work normally in harsh environments such as high-altitude snow-capped mountains and extreme weather such as heavy fog, and a high sampling frequency enables continuous prediction of vehicles.

[0028] 2. The present invention does not require the use of special-structured optical fibers or complex optical fiber laying methods, and is applicable to communication optical cables buried beside roads, with strong practicability. Relying on the meter-level spatial resolution and multi-channel characteristics of distributed optical fiber sensing technology, it improves the TDOA algorithm to locate the position of vehicles, adaptively corrects the vehicle position by matching the driving trajectory, and establishes a simple position-time dynamic model. These two, the model and the TDOA algorithm, complement each other to achieve fast and accurate prediction of vehicle information. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a structural device diagram of a distributed optical fiber sensing system.

[0030] Figure 2 It is a flowchart of the vehicle prediction method based on distributed optical fiber sensing and adaptive correction of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the following solutions are only explanatory descriptions of the inventive concept, and the specific solutions are not limited thereto. In addition, for the sake of description, only parts related to the present invention are shown in the drawings, rather than the entire process.

[0032] Please refer to Figure 1 , Figure 1 which is a structural device diagram of a distributed optical fiber sensing system, responsible for signal acquisition and demodulation. As shown in the figure, it mainly includes a laser, an optical coupler, a modulator, a circulator, a photodetector, and an ADC acquisition device, and finally a computing device is used for operation. To reduce the influence of phase noise, a narrow linewidth laser is used for the laser, which is responsible for outputting continuous light. After being divided into two parts by the coupler, one part is used as the local oscillator light, and the other part passes through the modulator to generate a detection optical pulse. The pulse uses a single-frequency pulse with a pulse width in the nanosecond range to ensure that the spatial resolution of the system is in the meter range. Then the detection optical pulse enters the optical fiber through port 2 of the circulator, and the Rayleigh scattered light returns through port 3 of the circulator and is beat with the local oscillator light by a 2×2 coupler. Finally, it is converted into an electrical signal by a balanced detector, acquired by the ADC, and then a series of operations are performed using a computing device to achieve vehicle prediction.

[0033] Please refer to Figure 2 , Figure 2It is the flowchart of the vehicle prediction method based on distributed optical fiber sensing and adaptive correction. A vehicle prediction method based on distributed optical fiber sensing and adaptive correction includes the following steps:

[0034] Step 1: The optical fiber sensing system emits a detection optical pulse to the optical fiber and quantitatively receives the returned sensing signal R(l,t), where l represents the axial distance of the sensing optical fiber and t represents the reception time;

[0035] Step 2: Perform phase demodulation on the sensing signal R(l,t), use GPU multi-threaded parallel computing to accelerate the demodulation process, and obtain the phase information Φ(l,t) in real time;

[0036] Step 3: Calculate the standard deviation of the phase, preliminarily determine the vehicle position range, select 11 sensing channels within this range and use the channel with the highest signal-to-noise ratio as the reference channel, and calculate the time delay amount △τ = [△τ1, △τ2,..., △τ 11 , where 1, 2,... 11 represent the number of sensing channels;

[0037] Step 4: Add a time offset that follows a normal distribution with a mean of 0 and a standard deviation of 0.5 ms to the time delay amount △τ, then use the linear TDOA positioning algorithm to locate the vehicle and repeat this process 10 times to average the located vehicle position where is the polar coordinate angle, which together with R represents the two-dimensional position of the vehicle;

[0038] Step 5: According to locate the two-dimensional position of the vehicle, obtain the vehicle position according to the vehicle driving trajectory matching, and adaptively correct the positioning result to obtain the vehicle position (x,y); The adaptive correction positioning specifically includes: According to Based on the TDOA positioning algorithm, obtain the two-dimensional position (x^,y^) of the vehicle, fit the vehicle trajectory curve based on the previous driving path, match to obtain the best position (x^’,y^’), calculate the deviation between the two. If the deviation between the two is less than 5 m, the position of the vehicle is (x,y) = [(x^,y^)+(x^’,y^’)] / 2, otherwise correct the position of the vehicle (x,y) = (x^’,y^’).

[0039] Step 6: Establish a position-time model based on the trajectory changes of the vehicle within the recent 50 time windows, and accurately predict the position (x’, y’) of the vehicle at subsequent moments based on this model; since the motion law of the vehicle is relatively simple, the established position-time model is usually an algebraic polynomial relationship, and establishing the position-time model is a very fast process, so real-time prediction can be achieved. If the vehicle encounters some kind of accident, the established position-time model will be too different from the actual position of the vehicle at this time, so it will be compared with the positioning result and corrected in a timely manner to continue the subsequent tracking and prediction.

[0040] The driving process of the vehicle is usually stable, and the position-time model can be established as follows:

[0041] When driving at a constant speed, P(t) = P0 + v0·t;

[0042] When driving with uniform acceleration, P(t) = P0 + v0·t + a·t 2 ;

[0043] For other normal driving states, P(t) = P0 + A·t + B·t 2 + C·t 3 ;

[0044] where P0 is the initial position of the vehicle, P(t) is the position of the vehicle at the current moment, t is the time, v0 is the initial speed of the vehicle, a is the acceleration of the vehicle, and A, B, C are the polynomial coefficients obtained by fitting;

[0045] Corresponding models can be established according to the recent multiple position-time relationships. The size of the time window is related to the sampling frequency of the system and should be determined according to the actual scenario.

[0046] Based on the vehicle position predicted in Step 6, the vehicle speed can be predicted by taking the derivative with respect to time, and the vehicle acceleration can be predicted by further taking the derivative with respect to time.

[0047] Step 7: Update the sensing channels and time windows, and repeat the above process to achieve continuous prediction of the vehicle.

[0048] The above are only the preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, and improvements made by those skilled in the art within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A vehicle prediction method based on distributed optical fiber sensing and adaptive correction, characterized in that: The following steps are involved: Step 1: The optical fiber sensing system transmits detection light pulses to the optical fibers laid on both sides of the road, and quantitatively receives the returned sensing signals R(l,t), where l represents the axial distance of the sensing optical fiber and t represents the receiving time; Step 2: Perform phase demodulation on the sensing signal R(l,t), use parallel computing to accelerate the demodulation process, and obtain phase information Φ(l,t) in real time; Step 3: Calculate the phase difference, preliminarily determine the vehicle position range, select N sensor channels within the range and use the channel with the highest signal-to-noise ratio as the reference channel, and calculate the time delay of each sensor channel relative to the reference channel △τ=[△τ1,△τ2,...,△τ N ], where 1, 2, ... N represents the number of sensing channels; Step 4: Add a random time offset to the time delay △τ, then use the linear TDOA positioning algorithm to locate the vehicle and repeat this process for a preset number of times to average the vehicle position (R, φ), where φ is the polar coordinate angle; Step 5: According to (R, φ), the two-dimensional position of the vehicle is obtained, the optimal landing point is obtained according to the vehicle's driving trajectory, and the positioning result is adaptively corrected to obtain the vehicle position (x, y); Step 6: Establish a position-time model within W time windows based on the recent trajectory changes of the vehicle, and accurately predict the vehicle position (x', y') at the next moment based on this position-time model; Step 7: Update the sensing channel and time window and repeat the above process to achieve continuous prediction of the vehicle.

2. A vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 1, characterized in that: The parallel computing uses GPU multi-threading, FPGA multi-logic units or other parallel computing methods.

3. The vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 1, characterized in that: The time offset follows a normal distribution with a mean of 0 and a standard deviation of σ.

4. The vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 1, characterized in that: The adaptive correction specifically includes: according to (R, φ), based on the TDOA positioning algorithm, the two-dimensional position (x^, y^) of the vehicle is obtained, the vehicle trajectory curve is fitted based on the previous driving path, and the optimal position (x^', y^') is obtained by matching, and the deviation between the two is calculated. The deviation between the two is less than the spatial resolution of the system. The position of the vehicle is (x, y) = [(x^, y^) + (x^', y^')] / 2, otherwise the position of the vehicle is corrected to (x, y) = (x^', y^'); the spatial resolution of the sensing system is determined by the detection pulse.

5. The vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 1, characterized in that: The position-time model is an algebraic polynomial relationship, and the position-time model is specifically established as follows: When driving at a constant speed, P(t)=P0+v0·t; When driving with uniform acceleration, P(t)=P0+v0·t+a·t 2 ; Other states of normal driving, P(t)=P0+A·t+B·t 2 +C·t 3 ; Among them, P0 is the initial position of the vehicle, P(t) is the current position of the vehicle, t is time, v0 is the initial speed of the vehicle, a is the acceleration of the vehicle, and A, B, and C are the polynomial coefficients obtained by fitting.

6. The vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 1, characterized in that: Based on the vehicle position (x', y') predicted in step 6, the vehicle speed is predicted by taking the time derivative, and the vehicle acceleration is further predicted by taking the time derivative.

7. The vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 1 is characterized in that: The optical fiber sensing system includes a laser, an optical coupler, a modulator, a circulator, a photodetector, an ADC acquisition device and a computing device; the laser is connected to two optical couplers, optical coupler No. 1 is connected to the modulator to generate a detection light pulse, the detection light pulse passes through the circulator and enters the optical fiber, the generated Rayleigh scattered light passes through the circulator and returns to optical coupler No. 2, the optical coupler No. 2 is connected to the detector, passes through the ADC acquisition device, and uses the computing device for further calculation.

8. The vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 7, characterized in that: The laser is a narrow line width laser, and the detection light pulse is a single frequency pulse with a pulse width in the nanosecond order.

9. The vehicle prediction method based on distributed optical fiber sensing and adaptive correction according to claim 7, characterized in that: The light from the laser is connected to the second optical coupler as the local oscillator light. The second optical coupler receives the Rayleigh scattered light, and the sensing signal is obtained after the two parts of light beat each other.

Images