FMCW laser radar nonlinear pre-correction method based on depth adaptive iteration

Abstract

The invention discloses an FMCW laser radar nonlinear pre-correction method based on depth adaptive iteration, and the method comprises the steps: measuring a time-frequency curve of a laser under a current driving voltage, generating a difference signal of an ideal sweep frequency function, and carrying out the iterative optimization of the driving voltage through an adaptive algorithm until a nonlinear error is minimized. According to the method, the effective linearization frequency modulation bandwidth proportion can be remarkably improved, the convergence speed is higher, and the method is suitable for correction of FMCW lasers with different non-linear degrees and can be used for an FMCW laser radar system with high precision and a large tuning range.

Description

Technical Field

[0001] The present invention relates to the technical field of frequency modulated continuous wave lidar, and particularly relates to a nonlinear pre-correction method for FMCW lidar based on depth adaptive iteration. Background Art

[0002] Frequency modulated continuous wave (FWCM) lidar has the advantages of anti-environmental light interference, high measurement resolution, high measurement accuracy, etc. Its ranging performance depends to a great extent on the sweep linearity of the laser source. Due to the inherent characteristics of the frequency modulated laser, the relationship between the uncorrected output optical frequency and time is not strictly linear. In practical applications, it is necessary to perform nonlinear pre-correction on the frequency modulated laser.

[0003] Currently, the nonlinear pre-correction of frequency modulated lasers is mainly achieved through the current iteration algorithm, that is, by measuring the time-frequency curve of the relationship between the frequency and time of the laser, and combining the iteration algorithm to update the drive voltage signal to make the output optical frequency change linearly (1. X. Zhang, J. Pouls, and M. Wu. “Laser frequency sweep linearizationby iterative learning pre-distortion for FMCW LiDAR.” Opt. Express 27(7), 9965-9974 (2019). 2. Li P, Zhang Y T, Yao J Q. Rapid linear frequency swept frequency-modulated continuous wave laser source using iterative pre-distortion algorithm[J]. RemoteSensing, 2022,14(14): 3455-3465.). However, since the fixed proportional coefficient cannot be dynamically adjusted according to the error, this method requires multiple iterations or a large number of complex operation calculations during operation, and cannot quickly achieve nonlinear correction. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention proposes a nonlinear pre-correction method for FMCW lidar based on depth adaptive iteration, for the rapid correction of FMCW lasers with different nonlinear degrees.

[0005] To solve the above technical problems, the technical solution of the present invention is: a non-linear pre-correction method for FMCW lidar based on depth adaptive iteration, comprising the following steps:

[0006] Step 1: Measure the optical frequency signal output by the FMCW laser under the initial drive voltage signal to obtain the initial time-frequency curve characterizing the relationship between optical frequency and time, and calculate the initial residual non-linearity at the same time;

[0007] Step 2: Use the least squares method to fit the current time-frequency curve to a linear function, remove the intercept term of the linear function, and generate an ideal sweep function;

[0008] Step 3: Obtain the non-linear frequency error signal by calculating the difference between the measured time-frequency curve and the ideal sweep function point by point , and record its maximum absolute deviation value ;

[0009] Step 4: Convert the error signal into a compensation voltage signal , and adaptively iterate the proportional coefficient according to the following rules :

[0010] Updated proportional coefficient: , is the number of iterations, when is the initial proportional coefficient;

[0011] When is greater than or equal to 1 GHz, [0.1, 1];

[0012] When is less than 1 GHz, [-0.1, -0.9];

[0013] Step 5: Superimpose the compensation voltage signal onto the current drive voltage signal to generate an updated drive voltage signal;

[0014] Step 6: Measure the time-frequency curve output by the laser under the new drive voltage signal and recalculate the residual non-linearity;

[0015] Step 7: Repeat Steps 1 to 6. When it satisfies that the current residual non-linearity value is greater than the historical minimum value continuously for three times, terminate the iteration and output the corresponding historical optimal drive voltage signal.

[0016] Preferably, the time-frequency curve is measured by an unbalanced Mach-Zehnder interferometer.

[0017] Preferably, the drive voltage signal can also be converted into a drive current signal.

[0018] Preferably, the amplitude range of the laser drive voltage signal is 0 - 5V.

[0019] Preferably, the drive voltage signal is a triangular wave signal or a sawtooth wave signal.

[0020] Preferably, the initial proportionality coefficient in step 4 can be calibrated manually or predicted by computer deep learning.

[0021] The present invention has the following characteristics and beneficial effects:

[0022] In the present invention, by dynamically updating the ideal sweep function during the iteration process and optimizing the sweep slope in real time, the ratio of the effective linear frequency modulation bandwidth within the maximum tunable bandwidth of the laser is improved. In the embodiments of the present invention, this ratio can reach up to 98% at most.

[0023] The present invention uses an adaptive proportionality coefficient and step size to achieve efficient compensation for the non - linear error signal, making the convergence speed faster, significantly reducing the number of iterations, being compatible with FMCW lasers with different non - linear degrees, and tolerating initial frequency deviations at the GHz level. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a flowchart of an embodiment of a non - linear pre - correction method for an FMCW lidar based on deep adaptive iteration according to the present invention.

[0025] Figure 2 is a comparison diagram of the drive voltage signal waveforms before and after iteration in the embodiment of the present invention.

[0026] Figure 3 is a result diagram of the time - frequency curve before non - linear pre - correction in the embodiment of the present invention.

[0027] Figure 4 is a result diagram of the time - frequency curve after non - linear pre - correction in the embodiment of the present invention.

[0028] Figure 5 is a comparison diagram of the time - frequency curve results before and after non - linear pre - correction of the comparative example. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] The following further describes the present invention in detail with reference to the drawings and embodiments, but the protection scope of the present invention should not be limited thereby.

[0030] A non - linear pre - correction method for an FMCW lidar based on deep adaptive iteration provided by the present invention, asFigure 1 As shown in the figure, it includes the following steps:

[0031] Step 1: Measure the time-frequency curve obtained by the FMCW laser under the drive of the drive voltage signal, and simultaneously calculate the residual nonlinearity at the current moment. The time-frequency curve is used to characterize the relationship between the optical frequency and time.

[0032] Specifically, in this embodiment, a triangular wave signal with a modulation frequency of 10 kHz and a voltage amplitude range of 0 - 1750 mV is selected and input into the signal generator as the initial drive voltage signal. The initial drive voltage signal is input into the DFB laser, and at this time, the laser temperature is controlled at 25 degrees. The laser output signal is input into an unbalanced Mach-Zehnder interferometer with a 5 m delay fiber, and the initial time-frequency curve is obtained through the phase demodulation algorithm, and the initial residual nonlinearity is calculated.

[0033] The calculation method of the residual nonlinearity is as follows: First, calculate the sum of squared residuals of the time-frequency curve, which is used to represent the cumulative deviation between the time-frequency curve and the ideal linear model; calculate the total sum of squares of the time-frequency curve, which is used to represent the total variation degree of the time-frequency curve relative to its mean value; and the ratio of the sum of squared residuals to the total sum of squares is the residual nonlinearity (using these two items to calculate the coefficient of determination in linear regression to measure the linearity of the time-frequency curve. And the residual nonlinearity is 1 - the coefficient of determination).

[0034] Specifically, as Figure 3 shown; the residual nonlinearity is , mainly determined by the linear regression coefficient , that is . Where is the sum of squared residuals, is the total sum of squares. For a time-frequency curve with N sample points, its , . Where is the time-frequency curve, is the linear function after fitting the time-frequency curve. The subscript represents the number of iterative processes in the nonlinear correction process, is the mean value of the time-frequency curve . In this embodiment, the residual nonlinearity is .

[0035] Step 2: Use the least squares method to fit the current time-frequency curve into a linear function, remove the intercept term of the linear function, and generate an ideal sweep function. The initial linear function obtained by fitting is , and after removing the intercept term, the ideal sweep function = .

[0036] Step 3: By calculating the difference between the measured time-frequency curve and the ideal sweep function point by point, the non-linear frequency error signal is obtained. , and record its maximum absolute deviation value. . In this embodiment, the maximum absolute deviation value is recorded. is 2.368 GHz.

[0037] It should be noted that is obtained by calculating the difference point by point and cannot be expressed by a specific function.

[0038] Step 4: Convert the error signal into a compensation voltage signal , and adaptively iterate the proportionality coefficient according to the following rules:

[0039] Updated proportionality coefficient: , where k is the number of iterations. When k = 0, is the initial proportionality coefficient, which is selected according to historical experimental data. ;

[0040] When is greater than or equal to 1 GHz, take 0.5;

[0041] When is less than 1 GHz, take -0.5.

[0042] Step 5: Superimpose the compensation voltage signal onto the current drive voltage signal to generate an updated drive voltage signal;

[0043] Step 6: Measure the time-frequency curve of the laser output under the new drive voltage signal and recalculate the residual non-linearity;

[0044] Step 7: Repeat Steps 1 to 6. When it satisfies that the current residual non-linearity value is greater than the historical minimum value for three consecutive times, terminate the iteration and output the corresponding historical optimal drive voltage signal.

[0045] In this embodiment, the initial drive voltage signal and the optimal drive voltage signal after 14 iterations are as Figure 2 shown. From this figure, the iterative change trend of the drive voltage signal can be seen.

[0046] In this embodiment, after 14 iterations, the up-sweep time-frequency curve and the residual non-linearity at this time are as Figure 4 shown. At this time, the effective linearized FM bandwidth accounts for 98% of the maximum tunable bandwidth. Comparing Figure 3 and Figure 4 , the residual non-linearity after 14 iterations is from Reduce to , where the residual nonlinearity is reduced to 0.0016% of the initial residual nonlinearity after processing.

[0047] Comparative example

[0048] An iterative learning pre-distortion lidar for linearizing FMCW of laser frequency scanning establishes a fixed ideal frequency sweep function by tuning the relationship between bandwidth and time (2B / T), and uses the difference between the time-frequency curve and the ideal frequency sweep function (the difference is used to obtain the error signal) for iterative correction, that is, by changing the drive voltage signal to change the time-frequency relationship of the output signal, so as to continuously reduce the nonlinearity of the time-frequency curve. Its effective linearized FM bandwidth accounts for 80% of the maximum tunable bandwidth.

[0049] Through Figure 5 It can be seen that the residual nonlinearity is the value marked in the figure . After 256 iterations, the residual nonlinearity is reduced from Reduce to .

[0050] By comparing this embodiment with the comparative example, it can be seen that the effective linearized FM bandwidth of this embodiment accounts for the maximum tunable bandwidth has been significantly improved.

[0051] In addition, after 14 iterations in this embodiment, the residual nonlinearity is reduced to 0.0016% of the initial residual nonlinearity after correction, while the comparative example is reduced to 0.0043% of the initial residual nonlinearity after 256 iterations of correction. It can be seen that the correction performance of the technical solution of this embodiment is significantly improved, and only a small number of iterations are required, and the correction efficiency is also greatly improved. It shows that the real-time nonlinear correction device and method of the FMCW lidar of the present invention can significantly suppress the FM nonlinearity and improve the ranging accuracy of the FMCW lidar.

[0052] The above has described the embodiments of the present invention in detail with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, without departing from the principle and spirit of the present invention, various changes, modifications, substitutions and variations to these embodiments including components still fall within the protection scope of the present invention.

Claims

1. A non - linear pre - correction method for FMCW lidar based on depth - adaptive iteration, characterized in that, It includes the following steps: Step 1: Measure the time-frequency curve obtained by the FMCW laser under the drive of the drive voltage signal, and simultaneously calculate the residual nonlinearity at the current moment. The time-frequency curve is used to characterize the relationship between the optical frequency and time; Step 2: Obtain the ideal sweep function using the time-frequency curve, and then calculate the difference between the measured time-frequency curve and the ideal sweep function point by point to obtain the error signal of the nonlinear frequency; Step 3: Convert the error signal into a compensation voltage signal, and superimpose the compensation voltage signal on the drive voltage signal in Step 1 to generate an updated drive voltage signal; Step 4: Use the updated drive voltage signal as the input to drive the FMCW laser to obtain a new time-frequency curve, and recalculate the residual nonlinearity; Step 5: Repeat Steps 1-4. When the current residual nonlinearity value is greater than the historical minimum three times in a row, terminate the iteration and output the corresponding historical optimal drive voltage signal.

2. The method according to claim 1, wherein The time-frequency curve is measured by an unbalanced Mach-Zehnder interferometer.

3. The method according to claim 1, wherein The calculation method of the residual nonlinearity is: calculate the sum of squared residuals and the total sum of squares of the time-frequency curve respectively. The ratio of the sum of squared residuals to the total sum of squares is the residual nonlinearity.

4. The method according to claim 1, wherein The method for obtaining the ideal sweep function is: obtain a linear function by fitting the time-frequency curve, and remove the intercept term of the linear function to generate the ideal sweep function.

5. The method according to claim 4, characterized in that, The fitting method is the least squares method.

6. The method according to claim 1, wherein The method for converting the error signal into a compensation voltage signal is: ； Among them, is the compensation voltage signal at time t, is the error signal at time t, is the adaptive iteration proportionality coefficient.

7. The method according to claim 6, wherein The iterative method of the adaptive iteration proportionality coefficient is: , is the number of iterations, when is the initial proportionality coefficient; When is greater than or equal to 1 GHz, [0.1, 1]; when is less than 1 GHz, [-0.1, -0.9], where is the maximum absolute deviation value.

8. The method according to claim 7, wherein The initial proportionality coefficient can be calibrated manually or predicted by a deep learning model based on historical data.

9. The method according to any one of claims 1-8, characterized in that, The amplitude range of the drive voltage signal is 0-5V.

10. The method according to any one of claims 1-8, characterized in that, The drive voltage signal is a triangular wave signal or a sawtooth wave signal.

11. The method according to any one of claims 1-8, characterized in that, The drive voltage signal is replaced with a drive current signal.

Images