Laser full-waveform signal high-speed real-time processing system and method

Abstract

The invention discloses a high-speed real-time processing system and method for a laser full-waveform signal, and provides a waveform fitting method based on random sampling consistency, so that outliers in a cut waveform can be quickly removed, a Gaussian waveform with an optimal signal-to-noise ratio can be extracted from an original sampling waveform, and the recognition speed and precision of a waveform peak value can be effectively improved. High-speed real-time processing of full-waveform signals is achieved, an assembly line judgment mode is adopted, the processing target of completing waveform cutting in a clock period is achieved, and high-precision distance measurement can be achieved by obtaining position information of echoes and reference waves.

Description

Technical Field

[0001] This invention relates to a high-speed real-time processing system and method for laser full-waveform signals, belonging to the field of full-waveform laser ranging technology. Background Technology

[0002] Full-waveform laser ranging is a typical distance measurement method that plays an important role in deep space exploration, inter-satellite ranging, land surveying, and industrial measurement. Full-waveform laser ranging systems use high-speed ADCs to achieve full-waveform sampling of the laser echo signal, obtaining digitized reference and echo waveforms. This allows for the determination of the time difference between the reference and echo signals through digital signal processing. Common full-waveform signal processing methods involve filtering, waveform trimming, and waveform fitting. Common waveform fitting methods include linear Gaussian fitting, weighted linear Gaussian fitting, iterative weighted linear Gaussian fitting, and the Levenberg-Marquardt (LM) method. These methods involve exponential operations and iterative solutions, which limit the speed of real-time full-waveform processing. Summary of the Invention

[0003] The technical problem solved by this invention is to address the issues of slow real-time processing speed and insufficient accuracy of traditional ranging technologies in the current technology, and to propose a high-speed real-time processing system and method for laser full waveform signals.

[0004] The present invention solves the above-mentioned technical problem through the following technical solution:

[0005] A high-speed real-time processing system for laser full-waveform signals includes a full-waveform sampling circuit, a laser control circuit, an FPGA processing center, a pulsed laser, an optical transmitting system, an optical receiving system, an echo detection circuit, and a reference wave detection circuit, wherein:

[0006] The FPGA processing center performs data processing based on the echo analog signal transmitted by the full waveform sampling circuit and the reference wave analog signal transmitted by the reference wave detection circuit, and generates corresponding control signals to send to the laser control circuit.

[0007] The laser control circuit receives control signals sent by the FPGA processing center, converts them into drive signals usable by the pulsed laser, and sends them to the pulsed laser.

[0008] The pulsed laser receives the drive signal sent by the laser control circuit and sends corresponding laser pulses to the optical emission system and the reference wave detection circuit respectively.

[0009] An optical emission system that adjusts the direction of the laser pulses emitted by a pulsed laser and radiates them outwards;

[0010] The reference wave detection circuit converts the laser pulse into a reference wave analog signal through photoelectric conversion and sends it to the full waveform sampling circuit.

[0011] The optical receiving system receives the backlight signal reflected from the target, performs stray light suppression processing, and then sends it to the echo detection circuit.

[0012] The echo detection circuit receives the echo signal after stray light suppression processing, performs photoelectric conversion to obtain the echo analog signal, and sends it to the full waveform sampling circuit.

[0013] The FPGA processing center includes a laser repetition rate control module, a waveform acquisition and buffering module, a waveform trimming and fitting module, and a distance calculation module, wherein:

[0014] The laser repetition rate control module controls the pulsed laser to emit a laser pulse once, receives the reference wave signal and the return light signal, and sends them to the waveform acquisition and buffering module respectively.

[0015] The waveform acquisition and caching module caches the reference wave signal and the return light signal respectively and sends them to the waveform trimming and fitting module;

[0016] The waveform trimming and fitting module performs pipeline decision-making, outputs reference wave signals and backlight signals that pass the pipeline decision-making, and calculates the waveform peak positions of the waveform data after outlier trimming using a fitting algorithm.

[0017] The distance calculation module calculates the distance to the peak position of the waveform based on the peak positions of the reference wave signal and the return light signal.

[0018] The method for pipeline decision-making in the waveform trimming and fitting module is as follows:

[0019] After reading the reference wave signal and the return signal, collect waveform data at 3 points. If the waveform data at 3 consecutive points are all greater than the preset threshold, the current waveform is considered to be a valid waveform. Then, extract the first N / 2-3 waveform sampling data and the last N / 2 waveform sampling data of the selected 3 points, as well as the count value of the first point of the selected 3 points, as waveform data without outlier clipping.

[0020] The waveform acquisition and buffering module converts the reference wave signal into an electrical analog signal after photoelectric conversion, and then performs data acquisition through a dual-channel ADC. The pulsed laser generated by the pulsed laser is reflected by the detection target to obtain the return light signal, which is then converted into an electrical analog signal through photoelectric conversion, and then performed data acquisition again by the dual-channel ADC.

[0021] After receiving the waveform data after pipeline decision and waveform clipping, the waveform trimming and fitting module uses a fitting algorithm to calculate the peak positions of the waveform data of the reference wave signal and the return light signal, and solves the distance between the peak positions. The method is: d=c(μ2-μ1) / (2f), where μ1 and μ2 are the peak positions of the waveform data of the reference wave signal and the return light signal, respectively, and d is the distance between the peak positions.

[0022] The method for calculation using the fitting algorithm is as follows:

[0023] The obtained waveform data is logarithmically transformed into a quadratic curve after logarithmic transformation of the Gaussian signal. The full waveform signal is then described using a signal processing model. The waveform data is as follows:

[0024]

[0025] In the formula, t is the sampling time, and A, τ and w are the pulse amplitude, pulse time scale and pulse width, respectively;

[0026] The Gaussian signal after taking the logarithm is:

[0027]

[0028] The quadratic curve of the Gaussian signal transformation after taking the logarithm is as follows:

[0029] z = a·t 2 +b·t+c

[0030]

[0031] In the formula, z, a, b, and c are all Gaussian correlation coefficients.

[0032] The method for outlier clipping of waveform data is as follows:

[0033] Outliers in the trimmed waveform data are identified and removed using a random sampling consensus method. Random data is randomly selected for quadratic curve fitting. Data points conforming to the quadratic curve are designated as inliers, and their counts are reset to zero. A quadratic curve model is fitted using these inliers, and each inlier in the current waveform data is evaluated. The counted inliers are compared with the saved maximum inlier count. If the current inlier count is larger, the parameters of the fitted quadratic curve model are updated to reflect the current inlier count, and the maximum inlier count is also updated accordingly. The current optimal quadratic curve model is then used again to evaluate each inlier in the current waveform data until the proportion of the maximum inlier count in the dataset exceeds a preset threshold. The optimal quadratic curve is then output, and the waveform data corresponding to this optimal quadratic curve is obtained as the waveform data after pipelined decision-making and waveform trimming.

[0034] The method for determining interior points is as follows:

[0035] Calculate the distance from each data point in the current waveform data to the fitted quadratic curve model. If the distance is greater than the distance threshold, it is determined to be an outside point; otherwise, it is recorded as an inside point. The inside point count is incremented by 1 and the count is compared with the maximum saved inside point count.

[0036] The optimal quadratic curve parameters (a, b, c) corresponding to the waveform data after pipeline decision and waveform trimming are the final fitted values ​​of the waveform data output by the waveform trimming and fitting module. The peak positions of the reference wave signal and the return light signal are calculated based on the final fitted values.

[0037] t ref =-b ref / (2a ref ),t echo =-b echo / (2a echo )

[0038] In the formula, t ref b is the peak position of the reference wave signal waveform. ref For the optimal quadratic curve parameters b and a of the reference wave signal waveform ref The optimal quadratic curve parameters a and t are used to define the reference wave signal waveform. echo b is the peak position of the reference wave signal waveform. echo For the optimal quadratic curve parameters b and a of the reference wave signal waveform echo The optimal quadratic curve parameter 'a' is used to define the reference wave signal waveform.

[0039] The method for calculating the distance d at the peak position of the waveform is as follows:

[0040] d=c(T echo +t echo -T ref -t ref ) / (2f)

[0041] In the formula, c is the speed of light, and T is the speed of light. ref For the reference wave signal to satisfy the condition that three consecutive points are greater than the threshold, T is the count value of the first of the three points. echo f is the count value of the first of the three points when the echo signal satisfies the condition that three consecutive points are greater than the threshold, and f is the full waveform sampling frequency.

[0042] A processing method based on a high-speed real-time processing system for full-waveform laser signals, comprising:

[0043] The pulsed laser is controlled by the FPGA processing center to emit a single laser pulse.

[0044] Simultaneously with the emission of laser pulses, the dual-channel ADC of the full waveform sampling circuit begins to synchronously acquire the reference wave signal and the return light signal, and buffers them in the internal FIFO of the FPGA processing center.

[0045] Simultaneously, the FPGA processing center reads and counts the AD-quantized waveform data from the internal FIFO.

[0046] The read waveform data is subjected to pipeline decision, and the reference wave signal and the return signal that pass the pipeline decision are output. The waveform peak position of the waveform data after outlier pruning is calculated using a fitting algorithm.

[0047] The distance between the peak positions of the waveforms is calculated based on the peak positions of the reference wave signal and the return signal.

[0048] The advantages of this invention compared to the prior art are:

[0049] (1) The present invention provides a high-speed real-time processing system and method for laser full waveform signals. It adopts a waveform fitting method based on random sampling consistency, which can quickly remove outliers in the clipped waveform and extract the Gaussian waveform with the best signal-to-noise ratio from the original sampled waveform. It can effectively improve the recognition speed and accuracy of waveform peaks and realize high-speed real-time processing of full waveform signals. It can solve the limitations on the speed of real-time processing of full waveforms in the exponential operation and iterative solution operations involved in linear Gaussian fitting, weighted linear Gaussian fitting, iterative weighted linear Gaussian fitting, LM method and other methods.

[0050] (2) The present invention adopts a pipeline decision method of 3-point decision method. If 3 consecutive points are greater than the threshold, it is considered as a valid waveform. This realizes the completion of waveform trimming within one clock cycle. For existing valid waveform identification methods of full waveform laser ranging, including first-order local linear analysis, it is necessary to cache the sampled data and judge the incremental relationship between the sampled data to determine whether the waveform is a valid waveform. This invention can reduce noise interference and achieve higher accuracy. Attached Figure Description

[0051] Figure 1 Block diagram of the high-speed real-time processing system for full-waveform laser signals provided by the present invention;

[0052] Figure 2 A schematic diagram of the rapid echo waveform identification and cropping method provided by the present invention;

[0053] Figure 3 This is a schematic diagram of the echo waveform fitting method based on random sampling consistency provided by the present invention. Detailed Implementation

[0054] A high-speed real-time processing system and method for laser full-waveform signals is proposed. A waveform fitting method based on random sampling consistency is proposed, which can quickly remove outliers in the clipped waveform and extract the Gaussian waveform with the optimal signal-to-noise ratio from the original sampled waveform. This can effectively improve the recognition speed and accuracy of waveform peaks and realize high-speed real-time processing of full-waveform signals. The pipeline decision method is also adopted to achieve the processing goal of completing waveform clipping within one clock cycle. High-precision distance measurement can be achieved by obtaining the position information of the echo and reference wave.

[0055] A high-speed real-time laser full-waveform signal processing system, including:

[0056] The system includes a full-waveform sampling circuit, a laser control circuit, an FPGA processing center, a pulsed laser, an optical transmitting system, an optical receiving system, an echo detection circuit, and a reference wave detection circuit, among which:

[0057] The FPGA processing center performs data processing based on the echo analog signal transmitted by the full waveform sampling circuit and the reference wave analog signal transmitted by the reference wave detection circuit, and generates corresponding control signals to send to the laser control circuit.

[0058] The laser control circuit receives control signals sent by the FPGA processing center, converts them into drive signals usable by the pulsed laser, and sends them to the pulsed laser.

[0059] The pulsed laser receives the drive signal sent by the laser control circuit and sends corresponding laser pulses to the optical emission system and the reference wave detection circuit respectively.

[0060] An optical emission system that adjusts the direction of the laser pulses emitted by a pulsed laser and radiates them outwards;

[0061] The reference wave detection circuit converts the laser pulse into a reference wave analog signal through photoelectric conversion and sends it to the full waveform sampling circuit.

[0062] The optical receiving system receives the backlight signal reflected from the target, performs stray light suppression processing, and then sends it to the echo detection circuit.

[0063] The echo detection circuit receives the echo signal after stray light suppression processing, performs photoelectric conversion to obtain the echo analog signal, and sends it to the full waveform sampling circuit.

[0064] The FPGA processing center includes a laser repetition rate control module, a waveform acquisition and buffering module, a waveform trimming and fitting module, and a distance calculation module, among which:

[0065] The laser repetition rate control module controls the fiber laser to emit one laser pulse.

[0066] The waveform acquisition and buffering module works as follows: After the laser emits light, the reference light signal is converted into an analog electrical signal by a photoelectric conversion circuit, and then sampled by a high-speed ADC. Simultaneously, the laser light emitted by the laser is reflected by the detection target (return signal), converted into an analog electrical signal by a photoelectric conversion circuit, and then sampled by the high-speed ADC. The FPGA receives the reference wave signal and the return signal acquired by the high-speed ADC, buffers them in an internal FIFO, and sends them to the waveform trimming and fitting module.

[0067] The waveform trimming and fitting module performs pipelined decision-making, outputs reference wave signals and backlight signals that pass the pipelined decision-making, and calculates the waveform peak positions of the waveform data after outlier trimming using a fitting algorithm.

[0068] The distance calculation module calculates the distance to the peak position of the waveform based on the peak positions of the reference wave signal and the return light signal.

[0069] The method for pipeline decision-making in the waveform trimming and fitting module is as follows:

[0070] After reading the reference wave signal and the return signal, three waveform data points are collected. If all three consecutive waveform data points are greater than the preset threshold, the current waveform is considered a valid waveform. Then, the first N / 2-3 waveform sampling data points and the last N / 2 waveform sampling data points of the selected three waveform data points, as well as the count value of the first point of the selected three waveform data points, are taken as waveform data without outlier clipping.

[0071] After receiving the waveform data after pipeline decision and waveform clipping, the waveform trimming and fitting module uses a fitting algorithm to calculate the peak positions of the waveform data of the reference wave signal and the return light signal, and solves the distance between the peak positions. The method is: d=c(μ2-μ1) / (2f), where μ1 and μ2 are the peak positions of the waveform data of the reference wave signal and the return light signal, respectively, and d is the distance between the peak positions.

[0072] The method for calculation using the fitting algorithm is as follows:

[0073] The obtained waveform data is logarithmically transformed into a quadratic curve after logarithmic transformation of the Gaussian signal. The full waveform signal is then described using a signal processing model. The waveform data is as follows:

[0074]

[0075] In the formula, t is the sampling time, and A, τ and w are the pulse amplitude, pulse time scale and pulse width, respectively;

[0076] The Gaussian signal after taking the logarithm is:

[0077]

[0078] The quadratic curve of the Gaussian signal transformation after taking the logarithm is as follows:

[0079] z = a·t 2 +b·t+c

[0080] In the formula, the expressions for z, a, b, and c are as follows:

[0081]

[0082] The method for outlier clipping of waveform data is as follows:

[0083] Outliers in the trimmed waveform data are identified and removed using a random sampling consensus method. Random data is randomly selected for quadratic curve fitting. Data points conforming to the quadratic curve are designated as inliers, and their counts are reset to zero. A quadratic curve model is fitted using these inliers, and each inlier in the current waveform data is evaluated. The counted inliers are compared with the saved maximum inlier count. If the current inlier count is larger, the parameters of the fitted quadratic curve model are updated to reflect the current inlier count, and the maximum inlier count is also updated accordingly. The current optimal quadratic curve model is then used again to evaluate each inlier in the current waveform data until the proportion of the maximum inlier count in the dataset exceeds a preset threshold. The optimal quadratic curve is then output, and the waveform data corresponding to this optimal quadratic curve is obtained as the waveform data after pipelined decision-making and waveform trimming.

[0084] The method for determining interior points is as follows:

[0085] Calculate the distance from each data point in the current waveform data to the fitted quadratic curve model. If the distance is greater than the distance threshold, it is determined to be an outside point; otherwise, it is recorded as an inside point. The inside point count is incremented by 1 and the count is compared with the maximum saved inside point count.

[0086] The processing method implemented by the high-speed real-time processing system for laser full-waveform signals includes:

[0087] The pulsed laser is controlled by the FPGA processing center to emit a single laser pulse.

[0088] Simultaneously with the emission of laser pulses, the dual-channel ADC of the full waveform sampling circuit begins to synchronously acquire the reference wave signal and the return light signal, and buffers them in the internal FIFO of the FPGA processing center.

[0089] Simultaneously, the FPGA processing center reads and counts the AD-quantized waveform data from the internal FIFO.

[0090] The read waveform data is subjected to pipeline decision, and the reference wave signal and the return signal that pass the pipeline decision are output. The waveform peak position of the waveform data after outlier pruning is calculated using a fitting algorithm.

[0091] The distance between the peak positions of the waveforms is calculated based on the peak positions of the reference wave signal and the return signal.

[0092] The following description, in conjunction with the accompanying drawings and preferred embodiments, provides further details:

[0093] In the current embodiment, the high-speed real-time processing system for laser full-waveform signals is as follows: Figure 1 As shown, the system comprises: a full-waveform sampling circuit for acquiring full-waveform signals; a laser control circuit for generating drive signals for the pulsed laser; an FPGA for laser control, data buffering, waveform trimming and fitting, and distance calculation; a pulsed laser for emitting laser pulses; an optical emission system for beam pointing; an optical receiving system for stray light suppression and backlight reception; an echo detection circuit for photoelectric conversion to generate analog echo signals; and a reference wave detection circuit for photoelectric conversion to generate analog reference wave signals.

[0094] The specific steps of the high-speed real-time processing method for laser full-waveform signals are as follows:

[0095] 1) The FPGA controls the fiber laser to emit a single laser pulse;

[0096] 2) At the same time as the laser pulse is emitted, the dual-channel high-speed ADC starts to synchronously acquire the reference optical signal and the echo signal, and buffers them into the FPGA's internal FIFO;

[0097] 3) Simultaneously with the emission of the laser pulse, the FPGA synchronously reads the AD quantization data from the internal FIFO and counts it;

[0098] 4) Data synchronously read from the internal FIFO is pipelined, and a pipelined decision method is used. A valid waveform is considered if three consecutive points exceed a threshold. The data is then extracted as follows: N / 2 - 3 samples before and N / 2 samples after these three points, along with the count value of the first point (echo value T). echo And the reference wave is T ref ), as input to the subsequent real-time fitting module, such as Figure 2 As shown;

[0099] 5) After receiving N waveform data, the real-time fitting module uses a random sampling consensus method to quickly remove outliers in the cropped waveform and obtain a Gaussian waveform with a high signal-to-noise ratio. Then, it uses a fitting algorithm to solve for the peak positions μ1 and μ2 of the reference wave and echo.

[0100] 6) Calculate the distance based on the peak positions of the reference wave and the echo, d = c(μ2-μ1) / (2f).

[0101] In the specific steps of the high-speed real-time processing method for laser full-waveform signals, the data synchronously read from the internal FIFO by the FPGA is pipelined. A pipelined decision method is used, where three consecutive points exceeding a threshold are considered a valid waveform. This achieves waveform trimming within one clock cycle, extracting only the N / 2-3 samples before and the N / 2 samples after those three points as input to the subsequent real-time fitting module. Figure 2 As shown;

[0102] In the specific steps of the high-speed real-time processing method for laser full-waveform signals, in the waveform fitting section, the logarithm of the acquired data is first taken to convert the Gaussian signal into a quadratic curve, and the standard Gaussian model is used to describe the full-waveform signal.

[0103]

[0104] In the formula, t is the sampling time, and A, τ, and w are the pulse amplitude, pulse time scale, and pulse width, respectively. Taking the logarithm of both sides yields:

[0105]

[0106] Therefore, the complete waveform signals of the reference wave and the echo can be represented as a quadratic curve: z = a·t 2 +b·t+c.

[0107] In the specific steps of the high-speed real-time processing method for laser full-waveform signals, to address the problem of inaccurate fitting algorithms caused by outliers, a random sampling consensus method is used to quickly identify and remove outliers in the cropped waveform, such as... Figure 3 As shown. Six data points are randomly selected for quadratic curve fitting. Simultaneously, the count of inliers (data points conforming to the fitted quadratic curve) is reset to zero. Using the quadratic curve model obtained from the curve fitting, inlier determination is performed on all data points of the current waveform. The determination method is to calculate the distance ζ from each point to the curve. If ζ is greater than a threshold, it is determined as an outside point; otherwise, it is an inside point, and the inlier count is incremented by 1. The counted inlier count is compared with the saved maximum inlier count. If the current inlier count is larger, the optimal quadratic curve parameters are updated to the current parameters, and the maximum inlier count is also updated to the current value. Then, using the current optimal quadratic curve parameters, inlier determination is performed again on all data points of the current waveform until the proportion of the maximum inlier count in the dataset exceeds a set value. The resulting optimal quadratic curve parameters (a, b, c) are then used as the final fitted values.

[0108] In the specific steps of the high-speed real-time processing method for laser full-waveform signals, the peak positions of the reference wave and echo are calculated by determining the distance between them. The peak positions t of the reference wave and echo are calculated using the (a,b,c) values ​​of the reference wave and echo obtained in step 5). ref =-b ref / (2a ref ),t echo =-b echo / (2a echo The distance measured in this study can be obtained from the following formula:

[0109] d=c(T echo +t echo -T ref -t ref ) / (2f)

[0110] In the formula, c is the speed of light, and T is the speed of light. ref For the reference wave signal to satisfy the condition that three consecutive points are greater than the threshold, T is the count value of the first of the three points. echo f is the count value of the first of the three points when the echo signal satisfies the condition that three consecutive points are greater than the threshold, and f is the full waveform sampling frequency.

[0111] The waveform fitting method based on random sampling consensus proposed in this patent can quickly remove outliers in the trimmed waveform, effectively improve the identification speed and accuracy of waveform peaks, and realize high-speed real-time processing of the entire waveform signal. At the same time, the pipeline decision method is adopted to achieve the processing goal of completing waveform trimming within one clock cycle, and can realize high-precision distance measurement by obtaining the position information of the echo and reference wave.

[0112] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

[0113] The contents not described in detail in this specification are common knowledge to those skilled in the art.

Claims

1. A high-speed real-time processing system for laser full-waveform signals, characterized in that: It includes a full waveform sampling circuit, a laser control circuit, an FPGA processing center, a pulsed laser, an optical transmitting system, an optical receiving system, an echo detection circuit, and a reference wave detection circuit, among which: The FPGA processing center performs data processing based on the echo analog signal transmitted by the full waveform sampling circuit and the reference wave analog signal transmitted by the reference wave detection circuit, and generates corresponding control signals to send to the laser control circuit. The laser control circuit receives control signals sent by the FPGA processing center, converts them into drive signals usable by the pulsed laser, and sends them to the pulsed laser. The pulsed laser receives the drive signal sent by the laser control circuit and sends corresponding laser pulses to the optical emission system and the reference wave detection circuit respectively. An optical emission system that adjusts the direction of the laser pulses emitted by a pulsed laser and radiates them outwards; The reference wave detection circuit converts the laser pulse into a reference wave analog signal through photoelectric conversion and sends it to the full waveform sampling circuit. The optical receiving system receives the backlight signal reflected from the target, performs stray light suppression processing, and then sends it to the echo detection circuit. The echo detection circuit receives the echo signal after stray light suppression processing, performs photoelectric conversion to obtain the echo analog signal, and sends it to the full waveform sampling circuit.

2. The high-speed real-time processing system for laser full-waveform signals according to claim 1, characterized in that: The FPGA processing center includes a laser repetition rate control module, a waveform acquisition and buffering module, a waveform trimming and fitting module, and a distance calculation module, wherein: The laser repetition rate control module controls the pulsed laser to emit a laser pulse once, receives the reference wave signal and the return light signal, and sends them to the waveform acquisition and buffering module respectively. The waveform acquisition and caching module caches the reference wave signal and the return light signal respectively and sends them to the waveform trimming and fitting module; The waveform trimming and fitting module performs pipeline decision-making, outputs reference wave signals and backlight signals that pass the pipeline decision-making, and calculates the waveform peak positions of the waveform data after outlier trimming using a fitting algorithm. The distance calculation module calculates the distance to the peak position of the waveform based on the peak positions of the reference wave signal and the return light signal.

3. The high-speed real-time laser full-waveform signal processing system according to claim 2, characterized in that: The method for pipeline decision-making in the waveform trimming and fitting module is as follows: After reading the reference wave signal and the return signal, collect waveform data at 3 points. If the waveform data at 3 consecutive points are all greater than the preset threshold, the current waveform is considered to be a valid waveform. Then, extract the first N / 2-3 waveform sampling data and the last N / 2 waveform sampling data of the selected 3 points, as well as the count value of the first point of the selected 3 points, as waveform data without outlier clipping. The waveform acquisition and buffering module converts the reference wave signal into an electrical analog signal after photoelectric conversion, and then performs data acquisition through a dual-channel ADC. The pulsed laser generated by the pulsed laser is reflected by the detection target to obtain the return light signal, which is then converted into an electrical analog signal through photoelectric conversion, and then performed data acquisition again by the dual-channel ADC.

4. The high-speed real-time processing system for laser full-waveform signals according to claim 3, characterized in that: After receiving the waveform data after pipeline decision and waveform clipping, the waveform trimming and fitting module uses a fitting algorithm to calculate the peak positions of the waveform data of the reference wave signal and the return light signal, and solves the distance between the peak positions. The method is: d=c(μ2-μ1) / (2f), where μ1 and μ2 are the peak positions of the waveform data of the reference wave signal and the return light signal, respectively, and d is the distance between the peak positions.

5. The high-speed real-time processing system for laser full-waveform signals according to claim 4, characterized in that: The method for calculation using the fitting algorithm is as follows: The obtained waveform data is logarithmically transformed into a quadratic curve after logarithmic transformation of the Gaussian signal. The full waveform signal is then described using a signal processing model. The waveform data is as follows: In the formula, t is the sampling time, and A, τ and w are the pulse amplitude, pulse time scale and pulse width, respectively; The Gaussian signal after taking the logarithm is: The quadratic curve of the Gaussian signal transformation after taking the logarithm is as follows: z=a·t 2 +b·t+c In the formula, z, a, b, and c are all Gaussian correlation coefficients.

6. The high-speed real-time processing system for laser full-waveform signals according to claim 4, characterized in that: The method for outlier clipping of waveform data is as follows: Outliers in the trimmed waveform data are identified and removed using a random sampling consensus method. Random data is randomly selected for quadratic curve fitting. Data points conforming to the quadratic curve are designated as inliers, and their counts are reset to zero. A quadratic curve model is fitted using these inliers, and each inlier in the current waveform data is evaluated. The counted inliers are compared with the saved maximum inlier count. If the current inlier count is larger, the parameters of the fitted quadratic curve model are updated to reflect the current inlier count, and the maximum inlier count is also updated accordingly. The current optimal quadratic curve model is then used again to evaluate each inlier in the current waveform data until the proportion of the maximum inlier count in the dataset exceeds a preset threshold. The optimal quadratic curve is then output, and the waveform data corresponding to this optimal quadratic curve is obtained as the waveform data after pipelined decision-making and waveform trimming.

7. The high-speed real-time processing system for laser full-waveform signals according to claim 6, characterized in that: The method for determining interior points is as follows: Calculate the distance from each data point in the current waveform data to the fitted quadratic curve model. If the distance is greater than the distance threshold, it is determined to be an outside point; otherwise, it is recorded as an inside point. The inside point count is incremented by 1 and the count is compared with the maximum saved inside point count.

8. The high-speed real-time processing system for laser full-waveform signals according to claim 6, characterized in that: The optimal quadratic curve parameters (a, b, c) corresponding to the waveform data after pipeline decision and waveform trimming are the final fitted values ​​of the waveform data output by the waveform trimming and fitting module. The peak positions of the reference wave signal and the return light signal are calculated based on the final fitted values. t ref =-b ref / (2a ref )、t echo =-b echo / (2a echo ) In the formula, t ref b is the peak position of the reference wave signal waveform. ref For the optimal quadratic curve parameters b and a of the reference wave signal waveform ref The optimal quadratic curve parameter 'a' is used to define the reference wave signal waveform. t echo b is the peak position of the reference wave signal waveform. echo For the optimal quadratic curve parameters b and a of the reference wave signal waveform echo The optimal quadratic curve parameter 'a' is used to define the reference wave signal waveform.

9. The high-speed real-time processing system for laser full-waveform signals according to claim 8, characterized in that: The method for calculating the distance d at the peak position of the waveform is as follows: d=c(T echo +t echo -T ref -t ref ) / (2f) In the formula, c is the speed of light, and T is the speed of light. ref For the reference wave signal to satisfy the condition that three consecutive points are greater than the threshold, T is the count value of the first of the three points. echo f is the count value of the first of the three points when the echo signal satisfies the condition that three consecutive points are greater than the threshold, and f is the full waveform sampling frequency.

10. A processing method implemented by the high-speed real-time laser full-waveform signal processing system according to claim 6, characterized in that... include: The pulsed laser is controlled by the FPGA processing center to emit a single laser pulse. Simultaneously with the emission of laser pulses, the dual-channel ADC of the full waveform sampling circuit begins to synchronously acquire the reference wave signal and the return light signal, and buffers them in the internal FIFO of the FPGA processing center. Simultaneously, the FPGA processing center reads and counts the AD-quantized waveform data from the internal FIFO. The read waveform data is subjected to pipeline decision, and the reference wave signal and the return signal that pass the pipeline decision are output. The waveform peak position of the waveform data after outlier pruning is calculated using a fitting algorithm. The distance between the peak positions of the waveforms is calculated based on the peak positions of the reference wave signal and the return signal.

Images