Scanning control method and system for multi-target cruise and storage medium

Abstract

The invention discloses a scanning control method and system for multi-target cruise and a storage medium, and the method comprises the steps: taking a priori target angle matrix of a corner reflector as a target track, and controlling a holder to carry out open-loop coarse positioning scanning so as to point to a priori target angle, laser emitted by the semiconductor laser is reflected by the corresponding corner reflectors and then is emitted to the position sensitive detector; the light spot center offset of the laser is calculated through a position sensitive detector, and the light spot center offset is mapped into an angle compensation instruction of the holder so as to carry out closed-loop micro-positioning regulation and control on the holder; and calculating a signal evaluation function value of the holder at the current angle after closed-loop fine tuning according to the intensity signal of the light spot, and triggering the spectrometer to perform interference spectrum acquisition based on the signal evaluation function value.

Description

Technical Field

[0001] This invention relates to the field of gas monitoring technology, and specifically to a scanning control method, system, and storage medium for multi-target cruise. Background Technology

[0002] Open-Path Fourier Transform Infrared Spectroscopy (OP-FTIR) is a non-contact, multi-component, real-time, continuous gas monitoring method widely used for high-precision monitoring of air pollution sources such as fugitive emissions from industrial parks and agricultural non-point source emissions. Currently, commonly used system structures include two deployment methods: through-beam and reflective. The reflective structure integrates the infrared light source and detector on one side of the main unit. The infrared beam is collimated by a Cassegrain telescope and then directed to a corner mirror on the opposite side. After returning to the main unit, it is received by the spectrometer, achieving optical path reversal. In polluted gas clouds, the infrared beam undergoes two absorptions, effectively doubling the optical path length compared to through-beam systems. This enhances the detection capability for low-concentration gases and offers advantages such as flexible deployment and convenient maintenance.

[0003] The U.S. Environmental Protection Agency (EPA) has released a technical method, OTM-10, which combines an open-path gas monitoring system with Vertical Radial Plume Mapping (VRPM) technology. By deploying multiple corner mirrors at different heights and horizontal positions and controlling the main unit to sequentially align with each mirror, infrared absorption spectra along different paths are acquired. This allows for the inversion of the two-dimensional concentration field of the pollution plume, which, combined with meteorological parameters such as wind speed, enables quantitative estimation of surface source emission fluxes. This method has significant application value in emission source monitoring, environmental enforcement, and source intensity identification.

[0004] However, to achieve the high-quality path integral concentration required for VRPM inversion accuracy, it is essential to ensure the stability of the signal-to-noise ratio (SNR) of each path during multi-path spectral acquisition. In practical applications, the stability of the SNR is highly dependent on the alignment accuracy between the host and the corner mirror. Especially during multi-target cyclic scanning, alignment errors significantly affect spectral quality, thereby impacting concentration calculation and flux inversion results. Furthermore, due to the temporal sequence of scanning paths, data from different paths exhibit time shifts. If the scanning cycle is too long, it will introduce significant time variability errors, leading to distortion in the plume concentration field construction and further affecting the accuracy of emission flux calculation.

[0005] Among the relevant technologies, the literature "Development of Ground-based FTIR Remote Sensing System for Water Vapor and Its Stable Isotopes, Wu Peng, Doctoral Dissertation in Engineering, University of Science and Technology of China" focuses on the overall design and deployment of the FTIR spectral system, emphasizing spectral detection accuracy and data inversion algorithms, but does not solve the problems of multi-path automatic scanning and rapid target alignment, especially lacking technical implementation at the cruise control algorithm level; it mainly adopts manual calibration + fixed scanning process, without introducing real-time feedback and feedforward control logic, and the scanning accuracy is limited by manual adjustment and single angle step control. The patent application document with publication number CN116414156A addresses the issues of angular velocity estimation error and hysteresis in real-time response of gimbals. It proposes a closed-loop control strategy based on Luenberger observer and LQR optimal control, aiming to improve the stability and real-time performance of gimbal response. However, it does not consider path-level scheduling and spatial target scanning control. Its control end achieves optimal angle / angular velocity control through LQR, relying on the system modeling accuracy and state-space observer to estimate angular velocity. It has a strong response to large-angle trajectories, and its accuracy depends on LQR parameter tuning. It does not involve signal feedback or dynamic scheduling mechanisms and is mainly aimed at low-speed control scenarios with fixed targets. It cannot handle small error compensation at the PSD level and multi-path alignment scheduling. The patent application document with publication number CN118365707A is mainly used for visual target pose recognition and path compensation in image recognition environment. It adopts image enhancement, feature matching, IMU fusion and reinforcement learning to solve target tracking in complex image environment. It relies on image feature matching and is not suitable for textureless infrared scenes. It realizes visual tracking target localization through image processing chain such as feature point extraction, image matching, Bayesian-reinforcement learning pose optimization, etc. It emphasizes image / IMU data fusion and path repair mechanism. The pose accuracy depends on the image and IMU quality, is easily affected by external interference, and lacks mechanical control accuracy and spectral sampling guarantee mechanism. Summary of the Invention

[0006] The technical problem to be solved by this invention is how to improve the efficiency and alignment accuracy of automatic multi-target scanning.

[0007] The present invention solves the above-mentioned technical problems through the following technical means:

[0008] A scanning control method for multi-target cruise is proposed, which controls a gimbal. A spectrometer is connected to the gimbal, and gas monitoring is achieved along different paths as the gimbal angle changes. A telescope is placed along the infrared light output path of the spectrometer, and a semiconductor laser is arranged in the telescope housing. A position-sensitive detector is arranged in the spectrometer housing. The method includes:

[0009] Using the prior target angle matrix of the corner reflector as the target trajectory, the gimbal is controlled to perform an open-loop coarse positioning scan to point to the prior target angle, so that the laser emitted by the semiconductor laser is reflected by the corresponding corner reflector and hits the position-sensitive detector.

[0010] The laser spot center offset is calculated by a position-sensitive detector, and the spot center offset is mapped to the angle compensation command of the gimbal to perform closed-loop micro-positioning control of the gimbal.

[0011] The signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning is calculated based on the intensity signal of the light spot, and the spectrometer is triggered to acquire the interferometric spectrum based on the signal evaluation function value.

[0012] Furthermore, before controlling the gimbal to perform an open-loop coarse positioning scan to point to the target angle using the prior target angle matrix of the corner reflector as the target trajectory, the method further includes:

[0013] The gimbal is adjusted so that the infrared light emitted by the spectrometer returns to the interferometer via the corresponding corner reflector and maximizes the peak value of the ZPD signal in the interferogram. The angle parameter of the gimbal at this time is read as the prior target angle of the corresponding corner reflector.

[0014] The prior target angle matrix is ​​constructed based on the prior target angles of the corner reflectors at different positions.

[0015] Furthermore, the step of using the prior target angle matrix of the corner reflector as the target trajectory and controlling the gimbal to perform an open-loop coarse positioning scan to point to the target angle includes:

[0016] The prior target angles of the corresponding angle reflectors in the prior angle matrix are converted into motor pulses;

[0017] Open-loop coarse positioning scanning is performed using a motor pulse-controlled gimbal to point to the prior target angle.

[0018] Furthermore, before calculating the laser spot center offset using a position-sensitive detector and mapping the spot center offset to a gimbal angle compensation command for closed-loop micro-positioning control of the gimbal, the method further includes:

[0019] During the process of the gimbal turning from one prior target angle to the next prior target angle, the angle deviation component of the gimbal is predicted based on unscented Kalman filtering, and the position drift of the gimbal is compensated for based on the angle deviation.

[0020] Furthermore, the step of predicting the angular deviation component of the gimbal based on unscented Kalman filtering during gimbal rotation, and performing feedforward compensation for the gimbal's position drift based on the angular deviation, includes:

[0021] A state prediction model for the gimbal is constructed based on a nonlinear state transition function, and an observation model is constructed based on an observation function.

[0022] Generate the first sigma point set based on the current state vector;

[0023] The first sigma point set is propagated through a nonlinear state transition function to obtain the second sigma point set, and the mean and state covariance of the state variable prediction results are calculated based on the second sigma point set.

[0024] The first sigma point set is propagated through the observation function to obtain the third sigma point set, and the mean, observation covariance, and state-observation covariance of the observation vector prediction results are calculated based on the third sigma point set.

[0025] Kalman gain is calculated based on observation covariance and state-observation covariance, and the state vector and state covariance are updated based on the calculated Kalman gain, combined with the prediction mean and observation data.

[0026] The updated state vector is used as the current state vector to generate a new first sigma point set. This process is iterated until the termination condition is met, and the average angle prediction of the gimbal is obtained.

[0027] The angle deviation component is obtained by subtracting the mean of the predicted angle from the corresponding prior target angle in the prior target angle matrix.

[0028] Furthermore, the step of calculating the laser spot center offset using a position-sensitive detector and mapping the spot center offset to a gimbal angle compensation command for closed-loop micro-positioning control of the gimbal includes:

[0029] The position of the reflected laser spot is calculated based on the four-electrode current values ​​of the position-sensitive detector;

[0030] The position of the light spot is compared with the expected position of the light spot to obtain the horizontal and vertical displacement errors, and the horizontal and vertical displacement errors are mapped to the angular displacement compensation amount of the corner reflector.

[0031] Based on the angular displacement compensation amount, a compensation angle command is generated to control the motor connected to the gimbal, thereby performing closed-loop micro-positioning control of the gimbal.

[0032] Furthermore, the step of generating a compensation angle command based on the angular displacement compensation amount to control the motor connected to the gimbal, and performing closed-loop micro-positioning control of the gimbal, includes:

[0033] The angular displacement compensation amount is converted into a compensation angle command using a PID controller. The control parameters of the PID controller are adjusted by an artificial neural network based on the dynamic change trend of the target spot offset error. The input of the artificial neural network is the current error state vector, and the output is the optimal combination of gain parameters of the PID controller.

[0034] The motor connected to the gimbal is controlled based on the compensation angle command to perform closed-loop micro-positioning control of the gimbal.

[0035] Furthermore, the artificial neural network is a feedforward neural network pre-trained using the backpropagation algorithm. During training, the network weight parameters are corrected through online error feedback. The objective function used is:

[0036]

[0037] In the formula, θ meas (t) represents the angular displacement measured by the system, θ target (t) represents the angular displacement compensation amount obtained from the mapping.

[0038] Furthermore, the step of calculating the signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning based on the intensity signal of the light spot, and triggering the spectrometer interferometric spectrum acquisition based on the signal evaluation function value, includes:

[0039] The normalized signal-to-noise ratio is calculated based on the intensity signal of the light spot and used as the signal evaluation function value at the current angle of the gimbal.

[0040] When the signal evaluation function value is greater than or equal to the upper limit threshold of intensity determination, the spectrometer is triggered to acquire interferometric spectrum.

[0041] When the signal evaluation function value is greater than the lower threshold of intensity judgment but less than the upper threshold of intensity judgment, the gimbal pointing angle is optimized based on the Bayesian optimization fine angle locking method, and then the spectrometer is triggered to acquire the interferometric spectrum.

[0042] When the signal evaluation function value is less than or equal to the lower limit threshold for intensity determination, the current scanning path is determined to be abnormal.

[0043] Furthermore, the refined angle locking method based on Bayesian optimization optimizes the gimbal pointing angle, including:

[0044] A black-box objective function is constructed using the current angle of the gimbal as input and the intensity of the light spot signal received by the position-sensitive detector as output.

[0045] Modeling the black-box objective function using a Gaussian process model on the sampling point set;

[0046] The improvement potential function of the current angle is calculated using the expected improvement criterion to select the next actual sampling angle;

[0047] Based on the determined actual sampling angle, the gimbal is rotated and the intensity of the light spot signal received by the position-sensitive detector is calculated. The actual sampling angle is then added to the sampling point set and the Gaussian process model is updated.

[0048] Repeatedly update the high-speed process model until the iteration termination condition is met to obtain the optimal pointing angle of the gimbal.

[0049] Furthermore, the modeling of the Gaussian process model of the black-box objective function on the sampling point set is expressed as:

[0050]

[0051] In the formula, μ(θ,φ) represents the expected estimate of the function value at (θ,φ) by the Gaussian process model, and k((θ,φ),(θ) = φ(θ,φ) = φ(θ,φ). ′ ,φ ′ J(θ,φ) is the covariance Gaussian kernel function, J(θ,φ) represents the value of the objective function, and the inputs are the parameter vectors θ and φ. Let be the notation for a Gaussian process, representing the value of the function at any finite sampling point as a V-variable Gaussian distribution.

[0052] Furthermore, the formula for calculating the improvement potential function of the current angle using the expected improvement criterion is expressed as follows:

[0053] EI(θ,φ)=(μ-J + )Φ(Z)+σφ(Z),

[0054] In the formula, μ and σ represent the mean and standard deviation of the Gaussian process prediction, respectively, and Φ(·) and φ(·) are the cumulative and density functions of the standard normal distribution, respectively. + The current optimal objective function value can be represented as J. + =max i J(θ i ,φ i ), where Φ(Z) represents the cumulative distribution function (CDF) of the standard normal distribution, used to give the probability that the random variable is less than a certain value, i.e.:

[0055]

[0056] φ(Z) represents the probability density function (PDF) of the standard normal distribution, used to describe the density of the standard normal distribution at a point, and is given by:

[0057]

[0058] EI(θ,φ) represents the expected improvement value, indicating the expected magnitude by which the target value exceeds the current optimal value after sampling at (θ,φ). It is often used in Bayesian optimization calculations.

[0059] Furthermore, when it is determined that the current scan path is abnormal, the method further includes:

[0060] Calculate the time difference between the collection timestamp of the abnormal path and the initial path time, and add the abnormal path to the abnormal path set when the absolute value of the difference between the time difference and the preset time offset exceeds the set threshold.

[0061] The abnormal paths in the abnormal path set are sorted according to their comprehensive priority score and then back-scanned.

[0062] If the backscan is successful, the original data of the abnormal path is replaced. If the backscan fails, the missing data points are filled by interpolation or by using valid data from the previous round.

[0063] Furthermore, the position-sensitive detector employs a two-dimensional PSD sensor.

[0064] Furthermore, the gimbal is a two-degree-of-freedom two-dimensional rotating gimbal, and the gimbal's adjustable angles include horizontal angle and pitch angle.

[0065] Furthermore, this invention proposes a scanning control system for multi-target cruise, used to control a gimbal. A spectrometer is connected to the gimbal, enabling gas monitoring along different paths as the gimbal angle changes. A telescope is positioned along the infrared light output path of the spectrometer, with a semiconductor laser housed within the telescope housing and a position-sensitive detector housed within the spectrometer housing. The control system includes:

[0066] The coarse positioning module is used to control the gimbal to perform open-loop coarse positioning scanning with the prior target angle matrix of the corner reflector as the target trajectory, so that the laser emitted by the semiconductor laser is reflected by the corresponding corner reflector and hits the position-sensitive detector.

[0067] The closed-loop feedback module is used to calculate the laser spot center offset through the position-sensitive detector and map the spot center offset into the gimbal angle compensation command to perform closed-loop micro-positioning control of the gimbal.

[0068] The path optimization module is used to calculate the signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning based on the intensity signal of the light spot, and to trigger the spectrometer to perform interferometric spectrum acquisition based on the signal evaluation function value.

[0069] Furthermore, the present invention also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the scanning control method for multi-target cruise as described above.

[0070] Furthermore, the present invention also proposes a computer program product, comprising a computer program that is executed by a processor to implement the steps of the scanning control method for multi-target cruise as described above.

[0071] The advantages of this invention are:

[0072] (1) This invention first uses the prior target angle matrix of the corner reflector as the target trajectory, and controls the gimbal to perform open-loop coarse positioning scan to point to the prior target angle, so as to achieve rapid coarse positioning. Since calling the prior target angle instead of real-time search can effectively shorten the scanning adjustment time of different corner reflectors, the impact of spatiotemporal fluctuations on the final measurement results can be reduced. Then, combined with the semiconductor laser and position sensitive detector introduced in this invention, a closed-loop feedback is performed. The laser emitted by the semiconductor laser is reflected by the corresponding corner reflector and hits the position sensitive detector. The position sensitive detector calculates the laser spot center offset in real time. By acquiring the reflected spot offset in real time and calculating the optical axis deviation, the small angle correction is guided, thereby enhancing the alignment robustness and spectral signal-to-noise ratio of the system, improving the acquired spectral quality, and overcoming the problem that if the scanning cycle is too long, the time variability error will be introduced due to the time offset of different path data, resulting in the distortion of the plume concentration field construction.

[0073] (2) This invention predicts the angle deviation component of the gimbal based on unscented Kalman filtering during the gimbal rotation process, and performs feedforward compensation for the position drift of the gimbal based on the angle deviation, thereby further improving the positioning accuracy and ensuring that the position sensitive detector can acquire the laser spot of the semiconductor laser. Then, the semiconductor laser and the position sensitive detector are used for closed-loop feedback to effectively eliminate residual error and ensure that the gimbal achieves sub-angle level alignment accuracy. Therefore, this invention proposes a control mode of "dual feedback + dual closed loop (motion prediction + position sensitive detector feedback)". This control strategy takes into account both response speed and positioning accuracy, improves the alignment stability of the system in complex dynamic environments, significantly compresses the single target alignment time, and effectively reduces repeated positioning error.

[0074] (3) In practical applications, although the position-sensitive detector can capture the light spot, there is a certain gap between the position of the light spot and the ideal position. This indicates that the light intensity of the captured optical disc has not reached the maximum, the collected infrared spectrum noise is large, and the quantitative result error is large. Therefore, the angle of the gimbal is further refined by the Bayesian optimization fine angle locking method, which greatly improves the accuracy of target alignment, effectively improves the signal recovery rate of high dynamic scene, realizes high confidence intelligent locking, and then obtains high signal-to-noise ratio spectrum.

[0075] (4) When a signal is abnormal or reflected light is lost in a certain path, the system can actively perform backscan or jump to the next path. The task scheduling and backscan mechanism ensures the data completion of each round and ensures the data integrity required by throughput inversion technologies such as VRPM.

[0076] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0077] Figure 1 This is a flowchart illustrating a scanning control method for multi-target cruise proposed in one embodiment of the present invention;

[0078] Figure 2 This is a schematic flowchart of a scanning control method for multi-target cruise in one embodiment of the present invention;

[0079] Figure 3 This is a schematic diagram of the PSD spot position in one embodiment of the present invention;

[0080] Figure 4 This is a schematic diagram of the system structure of an open optical path Fourier transform infrared spectrum according to an embodiment of the present invention;

[0081] Figure 5 This is a schematic diagram of a scanning control system for multi-target cruise proposed in one embodiment of the present invention. Detailed Implementation

[0082] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0083] like Figure 1 As shown, the first embodiment of the present invention proposes a scanning control method for multi-target cruise, which is used to control a gimbal. A spectrometer is connected to the gimbal, and gas monitoring is achieved on different paths as the angle of the gimbal changes. A telescope is set on the infrared light output path of the spectrometer, and a semiconductor laser is arranged on the telescope housing. A position-sensitive detector is arranged on the spectrometer housing. The method includes the following steps:

[0084] S10. Using the prior target angle matrix of the corner reflector as the target trajectory, control the gimbal to perform open-loop coarse positioning scan to point to the prior target angle, so that the semiconductor laser emits laser light that is reflected by the corresponding corner reflector and hits the position-sensitive detector.

[0085] S20. Calculate the laser spot center offset using a position-sensitive detector, and map the spot center offset into an angle compensation command for the gimbal to perform closed-loop micro-positioning control of the gimbal.

[0086] S30. Calculate the signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning based on the intensity signal of the light spot, and trigger the spectrometer to acquire the interferometric spectrum based on the signal evaluation function value.

[0087] It should be noted that in actual operation, each OP-FTIR spectrometer is used in combination with multiple corner reflectors placed at different locations to achieve quantitative analysis of target gases on multiple measurement paths. In order to improve the signal-to-noise ratio of the measured spectrum and the measurement efficiency between the corner reflectors, this embodiment proposes a precise and fast control method. First, the prior target angle matrix constructed by multiple corner reflectors is used as the prior positioning reference of the high-precision gimbal to achieve rapid coarse positioning of the gimbal. To further improve the alignment accuracy of the system in dynamic environments and prevent the accumulation of angle drift caused by gimbal mechanical gaps, temperature drift or external disturbances, this invention introduces a closed-loop feedback control module composed of a semiconductor laser and a position-sensitive detector, namely a two-dimensional PSD (Position-Sensitive Detector). Then, a closed-loop feedback is achieved by combining the semiconductor laser and position-sensitive detector introduced this time. The laser emitted by the semiconductor laser is reflected by the corresponding corner mirror and hits the position-sensitive detector. The position-sensitive detector calculates the laser spot center offset in real time. By acquiring the offset of the reflected spot in real time and calculating the optical axis deviation, the system can guide the correction of small angles, thereby enhancing the alignment robustness and spectral signal-to-noise ratio of the system and improving the quality of the acquired spectrum.

[0088] As a further preferred technical solution, before step S10: using the prior target angle matrix of the corner reflector as the target trajectory, controlling the gimbal to perform open-loop coarse positioning scan to point to the target angle, the method further includes the following steps:

[0089] S01. Adjust the gimbal so that the infrared light emitted by the spectrometer returns to the interferometer through the corresponding corner reflector and maximizes the peak value of the ZPD signal in the interferogram. Read the angle parameter of the gimbal at this time as the prior target angle of the corresponding corner reflector.

[0090] S02. Construct the prior target angle matrix based on the prior target angles of the corner reflectors at different positions.

[0091] Specifically, in this embodiment, the controlled object gimbal is a two-degree-of-freedom two-dimensional rotating platform, driven by two two-phase stepper motors, which control the horizontal and pitch angle movements of the platform respectively. During multi-target cruise scanning, this embodiment first adopts an open-loop control strategy based on the prior target angle matrix of corner reflectors. In the initial deployment stage of the optical path, the high-precision gimbal at the host end is manually adjusted to ensure that the infrared light returns to the interferometer via corner reflector #1 and maximizes the peak value of the ZPD (Zero Path Difference) signal in the interferogram. The pitch and azimuth angles of the gimbal at this time are read as the spatial coordinates of corner reflector #1. Similarly, the angle parameters of the remaining corner reflectors are obtained sequentially to construct a complete prior target angle matrix, which serves as the prior input for the subsequent automatic scanning path.

[0092] Specifically, the prior target angle matrix constructed in this embodiment, which includes the standard angle configuration of all corner mirrors, can be represented as:

[0093]

[0094] In the formula, θ1~θ N These are the prior target horizontal angles corresponding to the N corner mirrors, φ1~φ N These are the prior target elevation angles corresponding to the N corner mirrors.

[0095] As a further preferred technical solution, step S10: using the prior target angle matrix of the corner reflector as the target trajectory, controlling the gimbal to perform an open-loop coarse positioning scan to point to the target angle, specifically includes:

[0096] S11. Convert the prior target angles of the corresponding angle reflectors in the prior target angle matrix into motor pulses;

[0097] S12. Based on the motor pulse control gimbal, perform open-loop coarse positioning scan to point to the prior target angle.

[0098] Specifically, in this embodiment, the prior target angle of the corresponding corner reflector is first obtained from the prior target angle matrix:

[0099] (θ cmd ,φ cmd )=M(j),j=1,…,N

[0100] Where M(j) represents the prior position coordinates of the j-th target angle reflector, i.e., the preset angle, and θ cmd ,φ cmd These are the azimuth and elevation angles, respectively.

[0101] The prior target angle is then converted into motor pulses as follows:

[0102]

[0103] Where, n θ ,n φ This indicates the number of pulses the controller needs to send to the actuator motor; θ0, φ0 represent the initial angle of the pan-tilt unit read by the encoder built into the pan-tilt unit, and α... θ ,α φ The "unit pulse angle resolution" indicates the angle of rotation of the gimbal for each pulse emitted. This indicates rounding down to the nearest integer to ensure that the control does not exceed the limit.

[0104] This embodiment employs a two-degree-of-freedom two-dimensional rotary platform, driven by two two-phase stepper motors, which control the platform's horizontal and pitch angle movements respectively. Angular displacement control is achieved by precisely controlling the number of motor pulses, thereby enabling accurate positioning. Simultaneously, the pulse frequency is used to regulate the rotational speed and acceleration, ensuring smooth transitions and rapid switching between different targets and guaranteeing stable platform operation.

[0105] This embodiment uses a pre-constructed prior target angle matrix of the corner reflector as a prior positioning reference for coarse positioning of the high-precision gimbal. By calling the prior angle instead of real-time search, the scanning and adjustment time of different reflectors can be effectively shortened, thereby reducing the impact of spatiotemporal fluctuations on the final measurement results.

[0106] As a further preferred technical solution, before step S20: calculating the laser spot center offset using a position-sensitive detector and mapping the spot center offset to an angle compensation command for the gimbal to perform closed-loop micro-positioning control of the gimbal, the method further includes the following steps:

[0107] S20' During the process of the gimbal turning from one prior target angle to the next prior target angle, the angle deviation component of the gimbal is predicted based on the unscented Kalman filter, and the position drift of the gimbal is compensated by feedforward based on the angle deviation.

[0108] It should be noted that, theoretically, after coarse positioning, the focus should shift to reading the position of the PSD receiving spot for finer control. However, in practical applications, in order to improve the detection sensitivity of the PSD, the reflection angle of the semiconductor laser is very small, and the response area of ​​the PSD is effective. Simply relying on coarse positioning for qualitative control may not be able to acquire the spot on the PSD. Therefore, in this embodiment, after coarse positioning, dynamic prediction and error compensation control using the unscented Kalman filter algorithm (UKF) are performed to fine-tune the coarse positioning, ensuring that the laser spot of the semiconductor laser can be acquired on the PSD.

[0109] On the other hand, during the process of the gimbal turning from one target angle to the next, non-ideal dynamic behaviors such as position overshoot caused by structural inertia, hysteresis drift caused by changes in drive response and load, and dynamic deviation between control commands and mechanical response will introduce an angle offset error ε. θ ,ε φ However, due to factors such as motor bearing friction, the gimbal movement process exhibits nonlinear dynamic drift or non-Gaussian distribution characteristics of feedback signals. This embodiment employs an unscented Kalman filter to implement a control algorithm for dynamic prediction and error compensation of the gimbal angle, thereby achieving nonlinear state estimation and feedforward compensation of the gimbal and effectively improving the robustness and adaptability of system state prediction.

[0110] As a further preferred technical solution, step S20': during the process of the gimbal turning from one prior target angle to the next prior target angle, the angle deviation component of the gimbal is predicted based on unscented Kalman filtering, and the position drift of the gimbal is compensated for based on the angle deviation, specifically including the following steps:

[0111] S21' Construct a state prediction model for the gimbal based on a nonlinear state transition function, and construct an observation model based on an observation function;

[0112] Specifically, (1) constructing a state-space model:

[0113] Let the gimbal state vector be:

[0114]

[0115] Where, θ k , φ k The azimuth and elevation angles at the current time k are respectively. Let k be the angular velocities of the azimuth and pitch angles at time k. Then it represents the angular acceleration of the azimuth and elevation angles at time k.

[0116] The state prediction model is established using a nonlinear state transition function as follows:

[0117] x k =f(x) k-1 )+w k-1

[0118] Where, x k For the state vector, the nonlinear state transition function To set the time step;

[0119] Process noise Reflecting structural dynamic modeling and external disturbances, It follows a Gaussian distribution, where Q is the process noise covariance matrix, and can be expressed as:

[0120]

[0121] in, Let be the process noise variances for azimuth angle θ and elevation angle φ, respectively. Indicates azimuth angular velocity and pitch angular velocity Process noise variance, azimuth acceleration and pitch acceleration The process noise variance. The initial setting parameters can be estimated based on the actual measured variance of the gimbal's built-in motor response error.

[0122] (2) Constructing an observation space model:

[0123] The observation vector comes from the encoder and gyroscope, and is represented as follows:

[0124]

[0125] in, The accuracy is determined by the motor encoder. It was estimated using the difference angle.

[0126] The observation space model can be represented as:

[0127] z k =h(x k )+v k

[0128] Among them, z k For the observation vector, v k ~(0,R), where R is the observation noise covariance matrix, can be expressed as:

[0129]

[0130] S22' Generate the first sigma point set based on the current state vector;

[0131] Specifically, this embodiment generates the first sigma point set based on the state vector through unscented transformation. The specific process is as follows: the state dimension is n=6, generating 2n+1=13 sigma points:

[0132]

[0133] Where λ=α 2 (n+k)-n, default α=10 -3 k = 0, β = 2; χ0 is the first Sigma point, which is equal to the mean of the predicted state vector at the current time. Let χ be the mean of the state estimate at time k.i For the i-th Sigma point (i = 1, 22, ..., n) generated, we have in[·] i Represents the i-th column of the matrix; P is the state estimate at the previous time step (k-1), which is the mean of the state vector after the last iteration update in UKF. k-1 P k Let χ be the state covariance matrix of the previous time step (k-1) and the current time step k, respectively. i+n This is the (i+n)th Sigma point generated.

[0134] S23' Propagate the first sigma point set through a nonlinear state transition function to obtain the second sigma point set, and calculate the mean and state covariance of the state variable prediction results based on the second sigma point set;

[0135] Specifically, the state propagation and statistical estimation process implemented in this embodiment is as follows:

[0136] Propagating the first sigma point set through a nonlinear state transition function, we obtain the second sigma point set:

[0137]

[0138] Predicted state mean:

[0139]

[0140] The predicted state covariance is:

[0141]

[0142] Weight settings:

[0143]

[0144] in, For the i-th propagated predicted Sigma point, it is derived from the i-th Sigma point at the previous time step. It is obtained through the nonlinear state transition function f(·), where k represents the time index (time step), and k|k-1 represents the prediction of time k|k-1 based on the information at time k-1; The weights used to calculate the mean are used to control the contribution of each point when performing a weighted summation of sigma points. The predicted state mean vector is the predicted value obtained by weighted averaging of the propagated sigma points, P. k|k-1 W is the covariance matrix of the predicted state, describing the uncertainty of the predicted state. i (c)To calculate the weights of the covariance, The predicted state mean vector is the predicted value obtained by weighted averaging of the propagated sigma points, (·). c This refers to the transpose operation of a vector or matrix, specifically the transpose of the outer product, W. i The weights are assigned to other sigma points (i = 1, 2, ..., n), with the mean weight and covariance weight being the same.

[0145] S24. Propagate the first sigma point set through the observation function to obtain the third sigma point set, and calculate the mean, observation covariance and state-observation covariance of the observation vector prediction results based on the third sigma point set.

[0146] Specifically, the process of constructing observational predictions and related covariance is as follows:

[0147] Calculate the observation prediction for each sigma point, propagate the first sigma point set through the observation function to obtain the third sigma point set, then:

[0148]

[0149] Predicted mean of observation:

[0150]

[0151] Observation covariance:

[0152]

[0153] State-observation covariance:

[0154]

[0155] In the formula, Let be the sigma point of the i-th observation space, and let represent the i-th sigma point in the prediction space. The value mapped to the measurement space by the observation function h(·) The predicted observation mean (measured prediction value) is obtained by weighted averaging of all observation spatial sigma points, S. k P is the predicted observation covariance matrix, describing the uncertainty of the predicted measurements. xz The cross-covariance matrix between the state and the observation describes the linear correlation between the predicted state and the predicted observation, and is used to calculate the Kalman gain. Let i be the i-th sigma point in the predicted state space.

[0156] S25' Calculate the Kalman gain based on the observation covariance and the state-observation covariance, and update the state vector and state covariance based on the calculated Kalman gain combined with the prediction mean and observation data;

[0157] Specifically, the process of implementing Kalman gain and state update is as follows:

[0158] Kalman gains include:

[0159]

[0160] Status Update:

[0161]

[0162] Covariance update:

[0163]

[0164] Among them, K k For Kalman gain, The updated state estimate means (the optimal estimate combining prediction and observation information). To predict the state mean (the result obtained from time updates), The predicted mean is obtained by mapping the predicted sigma points to the measurement space and then weighting them.

[0165] S26', and then repeat steps S22' to S25' using the updated state vector as the current state vector, iterating until the termination condition is met, and predicting the average angle of the gimbal, including the average horizontal angle prediction. and pitch angle prediction mean

[0166] S27'. Subtract the predicted angle mean from the corresponding prior target angle in the prior target angle matrix to obtain the angle deviation component.

[0167] Specifically, the formula for the angular deviation component is expressed as follows:

[0168]

[0169] The angular deviation component includes the horizontal angular deviation component Δθ. pre and pitch angle deviation component Δφ pre , and Let represent the prior target angle corresponding to the j-th reflector.

[0170] In this embodiment, the host computer can record the historical step response data of the gimbal rotation, establish an acceleration-angle error prediction model through a Kalman filter, and insert a feedforward correction value before the gimbal rotation command is executed to improve the accuracy of angle arrival.

[0171] As a further preferred technical solution, step S20: calculating the laser spot center offset using a position-sensitive detector, and mapping the spot center offset to an angle compensation command for the gimbal to perform closed-loop micro-positioning control of the gimbal, specifically includes the following steps:

[0172] S21. Calculate the position of the reflected laser spot based on the four-electrode current values ​​of the position-sensitive detector;

[0173] Specifically, such as Figure 3 As shown, this embodiment acquires the current of the four electrodes in real time and calculates the position of the light spot in the two-dimensional coordinate system as follows:

[0174]

[0175] Where, x psd ,y psd Here, L represents the horizontal and vertical coordinates of the light spot on the PSD, L is the side length of the PSD's photosensitive surface, and I1 to I4 are the current values ​​output by the four motors of the PSD. This represents the total current across the four electrodes.

[0176] S22. Compare the position of the light spot with the desired position of the light spot to obtain the horizontal and vertical displacement errors and map the horizontal and vertical displacement errors to the angular displacement compensation amount of the corner reflector.

[0177] Specifically, in this embodiment, the obtained spot offset is compared with the preset center position to obtain the horizontal and vertical displacement errors. Furthermore, the horizontal and vertical displacement errors are linearly transformed and mapped to the compensation angle command of the target corner reflector using a PI controller.

[0178]

[0179] Where, x c ,y c Let Δθ be the desired position of the light spot. psd ,Δφ psd These are the angular displacement compensation amounts in the azimuth and pitch directions, respectively. These are the mapping coefficients obtained through system calibration.

[0180] S23. Generate compensation angle commands based on angular displacement compensation to control the motor connected to the gimbal, and perform closed-loop micro-positioning control of the gimbal.

[0181] Furthermore, in this embodiment, the angular displacement compensation amount can be input to a PID controller to generate precise control commands to drive a high-precision stepper motor, thereby achieving real-time fine-tuning of the corner reflector. The PID controller is used to accurately map the spot offset to the angular displacement compensation amount required for motor drive, and to achieve high-precision attitude adjustment of the system. Taking into account the system's dynamic response characteristics, nonlinear disturbances, and measurement noise, it is ensured that the corner reflector can still achieve rapid and stable alignment under minor disturbances.

[0182] Therefore, this invention proposes a "dual feedback + dual closed loop (motion prediction + position-sensitive detector feedback)" control mode. "Dual closed loop control" refers to a dual feedback adjustment mechanism in which the position-sensitive detector (PSD) and the predictive control model participate simultaneously during the gimbal rotation process and the target signal acquisition process. It includes two closed loops: (1) Position prediction-feedback closed loop: using the prior angle matrix of the target reflector and the spot position prediction model, the angle position of the target in the next step is predicted in advance. Feedback loop: the spot position information collected by the position-sensitive detector in real time is compared with the predicted value, the angle deviation is calculated and corrected immediately. This significantly reduces the time offset caused by excessively long scanning cycle or asynchronous path data, enabling the gimbal to lock the target angle quickly and stably. (2) Light intensity detection-feedback closed loop: Prediction loop: predict the trend of spot intensity change based on historical scanning data and laser emission characteristics. Feedback loop: compare the actual collected spot intensity signal with the predicted value, and adjust the micro angle of the gimbal in real time to ensure that the signal is collected at the peak point of light intensity, which is used to enhance the spectral signal-to-noise ratio and reduce the quantitative error caused by light intensity fluctuations during the acquisition process. It can maintain continuous tracking and stable acquisition of targets in dynamic or interference environments, reduce deviations in both position and time, and avoid inversion errors caused by data asynchrony or light intensity attenuation.

[0183] Preferably, to further improve the response speed and tracking accuracy of the control system to target offset, this embodiment constructs an adaptive gain adjustment PID control based on an artificial neural network (ANN) on the basis of a traditional PID controller. This algorithm can adjust the various control parameters of the PID in real time according to the dynamic change trend of the target spot offset error, ensuring that the system always maintains fast response and stable convergence control characteristics under different operating conditions.

[0184] Specifically, the artificial neural network is a feedforward neural network pre-trained using the backpropagation algorithm. During training, the network weight parameters are corrected through online error feedback. The objective function used is:

[0185]

[0186] In the formula, θ meas (t) represents the angular displacement measured by the system, θ target(t) represents the angular displacement compensation amount obtained from the mapping.

[0187] The input to the trained artificial neural network is the current error state vector of the system:

[0188]

[0189] The output is the optimal combination of gain parameters Y at time point t. t :

[0190] Y t =[K P (t),K I (t),K D (t)]

[0191] Where e(t) is the current position error of the light spot, and X t Let K be the error state vector. P (t), K I (t), K D (t) represents the adaptive adjustment ratio.

[0192] Furthermore, based on the optimal combination of gain parameters, the PID controller output is:

[0193]

[0194] Among them, u(t) is the output angular displacement compensation control quantity.

[0195] It should be noted that this embodiment acquires the current of the four electrodes in real time, calculates the spot offset, and maps it into a micro-angle compensation command through a proportional-integral-derivative control law based on an online learning gain adjustment method using an artificial neural network (ANN). This command is then sent to the servo driver to achieve secondary angle fine-tuning. This ultimately forms a closed-loop control system consisting of a semiconductor laser, a PSD detector, a control algorithm, a motor drive module, and a corner mirror, achieving high responsiveness, high precision, and low latency automatic alignment capabilities. This closed-loop structure effectively suppresses spot offset caused by external disturbances, environmental noise, and other factors, improving system stability and robustness.

[0196] As a further preferred technical solution, step S30: calculating the signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning based on the intensity signal of the light spot, and triggering the spectrometer interferometric spectrum acquisition based on the signal evaluation function value, specifically includes the following steps:

[0197] S31. Calculate the normalized signal-to-noise ratio based on the intensity signal of the light spot as the signal evaluation function value of the gimbal at the current angle;

[0198] Specifically, using the spot intensity signal received by the PSD detector as the main input, a model quality evaluation function is constructed, with the normalized signal-to-noise ratio (SNR) as the criterion:

[0199]

[0200] Where J(θ,φ) is the signal evaluation function value at the current angle, used to determine whether the alignment accuracy requirements are met, and S peak σ represents the peak current (μA or V) of the current frame spot in the PSD. bg This represents the standard deviation of the Beijing signal or the zero optical path difference condition during the sampling period.

[0201] S32. When the signal evaluation function value is greater than or equal to the upper limit threshold of intensity determination, the spectrometer is triggered to acquire the interferometric spectrum.

[0202] S33. When the signal evaluation function value is greater than the lower limit threshold of intensity judgment and less than the upper limit threshold of intensity judgment, the gimbal pointing angle is optimized based on the Bayesian optimization fine angle locking method, and then the spectrometer is triggered to acquire the interferometric spectrum.

[0203] S34. When the signal evaluation function value is less than or equal to the lower limit threshold of intensity determination, the current scanning path is determined to be abnormal.

[0204] Specifically, a signal strength determination upper limit threshold J is set. th0 And the lower limit threshold J for intensity determination th1 The following guidelines shall be followed:

[0205]

[0206] like Figure 2 As shown, the invalid region is the response range determined by system structural errors and the influence of PSD itself, while the ideal region is the spot position region corresponding to the spectral signal-to-noise ratio obtained from previous experience when it is in the reliable region. i This is the model status flag for the current path i. If flag i =1 will directly trigger the spectrometer to acquire data; if flag i If the value is 0, the fine-grained angle locking method based on Bayesian optimization is entered, flag. i =-1 indicates an abnormal signal, and abnormal path repair is performed.

[0207] It should be noted that after the light spot response is obtained on the PSD, whether it is the optimal angle to trigger spectral acquisition, whether further fine-tuning the angle search is performed, or whether the current path is marked as "signal abnormality" depends on the signal evaluation results and will be handled by the subsequent task scheduling module. Due to a series of reasons such as system structure, mechanical deformation, and external disturbances, after the PSD closed-loop feedback adjustment, the PSD may be able to capture a light spot, but the position of the light spot may differ from the ideal target. At this time, the light intensity is not at its maximum, so the acquired infrared spectrum has large noise and the quantitative result has large error. Therefore, further fine-tuning is required.

[0208] As a further preferred technical solution, in step S33, the refined angle locking method based on Bayesian optimization optimizes the gimbal pointing angle, specifically including the following steps:

[0209] S331. Using the current angle of the gimbal as input and the intensity of the light spot signal received by the position-sensitive detector as output, construct a black-box objective function;

[0210] Specifically, this embodiment constructs a black-box objective function:

[0211] J(θ,φ)=S PSD (θ,φ)

[0212] The black-box objective function takes the current azimuth and pitch angles of the gimbal as input and outputs the intensity of the light spot signal received by the PSD. The function is non-differentiable and is affected by factors such as optical path interference and specular reflection; therefore, a sampling-based model is required. PSD (θ,φ) represents the light signal intensity value of the reflected laser spot detected by the PSD receiver under a given azimuth angle θ and elevation angle φ. It is the output value of the black box objective function.

[0213] S332. Model the black-box objective function using a Gaussian process model on the sampling point set;

[0214] Specifically, Gaussian process modeling is based on the set of sampled points. The process of training a Gaussian is as follows:

[0215]

[0216] Where μ(θ,φ) represents the expected estimate of the function value at (θ,φ) by the Gaussian process model, which is usually 0 or a constant. Through sampling and dynamic learning, J(θ,φ) represents the function value of the black-box objective function at the input (θ,φ). Let μ(θ,φ) represent a Gaussian process, where μ(θ,φ) is the expected value at (θ,φ).

[0217] k((θ,φ),(θ ′ ,φ′ )) is the covariance Gaussian kernel function, which can be expressed as:

[0218]

[0219] in, Let l be the variance of the function value, and l be the length scale, which determines the rate of change. 2 This represents the square of the Euclidean distance between the two.

[0220] It should be noted that this embodiment uses the covariance Gaussian kernel function to describe the similarity of function values ​​between any two angle points. The larger the kernel function value, the closer (θ,φ) is to (θ) and (θ) is to (θ). ′ ,φ ′ The more similar the signals are, the smaller the change in function values. The established model allows for probabilistic modeling and uncertainty assessment of the signal response function across the entire angular space.

[0221] S333. Calculate the improvement potential function of the current angle using the expected improvement criterion and select the next actual sampling angle;

[0222] Specifically, this embodiment uses the expected improvement criterion to calculate the improvement potential function of the current angle as the sampling point selection strategy, and the formula is expressed as:

[0223] EI(θ,φ)=(μ-J + )Φ(Z)+σφ(Z),

[0224] In the formula, μ and σ represent the mean and standard deviation of the Gaussian process prediction, respectively, and Φ(·) and φ(·) are the cumulative and density functions of the standard normal distribution, respectively. + Let J represent the optimal value of the objective function known at present, serving as a benchmark to calculate the potential improvement of the current prediction compared to the best result. Φ(Z) represents the cumulative distribution function (CDF) of the standard normal distribution, used to measure the improvement of the current point compared to the optimal value J. + The probability φ(Z) represents the probability density function (PDF) of the standard normal distribution, which is used to measure the contribution of prediction uncertainty to the improvement potential, and EI(θ,φ) represents the expected improvement.

[0225] This embodiment achieves precise and rapid locking of the left and right points of the signal by maximizing the EI function to select the next actual sampling angle.

[0226] S334. Based on the determined actual sampling angle, control the gimbal rotation and calculate the intensity of the light spot signal received by the position-sensitive detector, and add the actual sampling angle to the sampling point set and update the Gaussian process model.

[0227] Specifically, in this embodiment, the maximum point of the EI function is used as the control command:

[0228] (θ next ,φ next =argmaxEI(θ,φ)

[0229] Control the pan-tilt unit to move to (θ) next ,φ next ), θ next ,φ next The azimuth (θ) and pitch (φ) of the gimbal are represented by the angles of the gimbal for the next sampling, and argmaxEI(θ,φ) represents all possible angles. In the process, find the point with the largest EI(θ,φ) (desired improvement value); and read the corresponding PSD spot intensity as J. k+1 Add to dataset D k+1 And update Model.

[0230] S335. Repeat the high-speed process model update until the iteration termination condition is met to obtain the optimal pointing angle of the gimbal.

[0231] Specifically, the maximum value of EI is less than ∈, and the maximum iteration K is reached when one of the following conditions is met: (1) EI is less than ∈; (2) the maximum iteration K is reached. max (3) When there is no significant improvement after n consecutive iterations, the iteration terminates. At this point, the optimal point is output, and the gimbal is precisely pointed to this angle to complete the final lock, which is represented as:

[0232]

[0233] Where, θ * ,φ * J represents the optimal azimuth and elevation angles for the final lock-in. i This represents the intensity of the light spot signal obtained at the i-th sampling angle position. This indicates that J was found among all sampling points. i The largest point, that is, the best observation angle in history.

[0234] It should be noted that in this embodiment of the multi-target gimbal scanning system, due to system structural errors, disturbances, and nonlinear dynamic behavior, there is a slight deviation between the actual optimal alignment angle of the target reflector and the preset angle. Traditional gradient- and grid-based optimization methods often suffer from slow convergence, local optima, and long processing times when facing non-smooth or noisy feedback functions. Therefore, this invention introduces a refined angle locking method based on Bayesian optimization. On the basis of coarse alignment, a global posterior estimate of the PSD signal intensity function is established through a probabilistic model, gradually approximating the maximum signal intensity value to achieve high-confidence intelligent locking and thus obtain a high signal-to-noise ratio spectrum.

[0235] As a further preferred technical solution, in step S34, when it is determined that the current scan path is abnormal, the method further includes the following steps:

[0236] S341. Calculate the time difference between the collection timestamp of the abnormal path and the initial path time, and add the abnormal path to the abnormal path set when the absolute value of the difference between the time difference and the preset time offset exceeds the set threshold.

[0237] It should be noted that this initial path is the timestamp recorded when the first valid sample is completed in the current scanning task. It serves as a reference time for comparison with subsequent acquisition times.

[0238] S342. Perform a backscan on the abnormal paths in the abnormal path set, sorted by comprehensive priority score.

[0239] S343. If the backscan is successful, replace the original data of the abnormal path. If the backscan fails, fill the missing data points by interpolation or by using the valid data from the previous round.

[0240] Specifically, when flag i = -1 indicates that the signal of this path is abnormal, and it is added to the abnormal path set, defined as:

[0241] P abnormal ={i∈P|flag i}=-1

[0242] Perform time difference threshold judgment: the timestamp for each path is t. i If the initial path time is t1, then:

[0243] Δt i =t i -t1

[0244] like If the value exceeds the set threshold, it is considered a synchronization mismatch path, and P is added. abnormal .

[0245] It should be understood that if the time difference does not exceed the set threshold, it means that the time difference of the path is within an acceptable range and is considered a normally synchronized path. It will not be added to the abnormal path set and will continue to be executed according to the normal scanning or data processing process.

[0246] It should be noted that in this embodiment, the condition for the abnormal path set pabmormal is flag_i = -1. This is a preliminary screening condition, which can be used to determine whether a path is abnormal based on the path signal status (such as signal interruption, packet loss, PSD signal abnormality, etc.). Furthermore, a time synchronization constraint is introduced into the path anomaly detection to filter out paths that, although their signals are abnormal, do not affect the measurement accuracy, while identifying those abnormal paths that will truly cause result deviations. This can also be understood as: synchronization mismatch paths are a subset of abnormal paths, hence the inclusion in pabmormal.

[0247] Specifically, to ensure data integrity and sampling robustness during multi-path scanning, this embodiment designs an intelligent task scheduling and anomaly self-repair mechanism. Based on the path signal strength S... i Acquisition time offset Δt i Historical path anomaly rate R i A priority scoring function is constructed using multiple parameters, including (long-term statistical values).

[0248]

[0249] Where w1, w2, and w3 are adjustable weighting factors, and Score i The higher the value, the higher the scheduling priority.

[0250] This embodiment applies to all flags. i Paths with a value of -1 are ranked according to Score. i After sorting the values, the scan proceeds sequentially. For abnormal paths, a backscan mechanism is implemented, i.e.:

[0251] If |P abnormal If |>0, it indicates that at least one scan path is currently identified as abnormal. A backscan will then be performed based on the degree of abnormality (the degree of abnormality mainly refers to the deviation or degradation of the scan path in terms of data acquisition quality; paths with a high degree of abnormality will be prioritized for backscanning and repair).

[0252]

[0253] Among them, P rescan This indicates that in the abnormal path set P abnormal The subset of paths selected from the data that require priority execution of the backscan task. The retrace priority score for path i is calculated by comprehensively considering multiple factors such as path signal strength, historical anomaly rate, time offset, and task urgency. θ represents the retrace priority threshold, which is only applied when the path's priority score is within a certain range. The path will only be added to P if the value is greater than θ. rescan It also performs a backscan to avoid invalid or low-return scans.

[0254] The success of the retracement is determined by verifying the received signal or data quality. Specifically, after the retracement is executed, the device should acquire the latest spot intensity signal for the corresponding path if it exceeds the set threshold, or if the signal strength after the retracement is significantly stronger than the signal strength at the time of the anomaly. If the retracement fails, the missing data points are filled by interpolation or using valid data from the previous round to ensure that the final data used for flux inversion meets the requirements.

[0255]

[0256] in, This represents the final status flag of the m-th sampling path after the backscan and data repair process is completed. P represents the set of all sampling paths to be verified or repaired, and m represents the index or number of the path in set P, used to uniquely identify a sampling path. In other words, a backscan can only be considered successful if all paths have completed data backscanning, interpolation, or data replacement processes, and all data integrity and validity requirements are met.

[0257] This embodiment constructs a priority scoring function by setting a scheduling algorithm and combining multiple indicators such as the current target pointing status, signal strength evaluation, and path timestamp offset. Upon detecting an abnormal path (weak signal, high latency, target loss), the algorithm automatically adjusts the scanning priority of the abnormal path, skips the abnormal path, and compensates for the loss based on the priority scoring function. An automatic abnormal path backscanning mechanism replaces the original abnormal path data if the backscan is successful; if the backscan fails, the loss is filled through interpolation or compensation. This ensures the integrity and consistency of path data in each round of VRPM or MRPM measurement, improving the reliability of pollution flux inversion.

[0258] Specifically, this embodiment automatically creates a working directory named after the scan start time (format: yyyymmddHHMMSS) and loads the prior angle matrix as the target trajectory. The system first performs coarse positioning based on open-loop commands, guiding the gimbal to quickly point to the target angle; then, it performs dynamic error prediction compensation based on unscented Kalman filtering (UKF) to feedforward correction of environmental disturbances and control residuals; next, it acquires spot offset information through the PSD detection module, calculates the error, and completes micro-closed-loop adjustment. Then, it calculates the system's current signal-to-noise ratio (SNR) J = S / N in real time and determines the scan path status based on two set thresholds (high and low). When the SNR is better than the upper threshold, it directly triggers FTIR interferometric spectrum acquisition and saves it, marking the timestamp t. kIf the signal-to-noise ratio falls between these two values, a fine-tuning phase based on a probabilistic model is initiated to further improve positioning accuracy. If the signal-to-noise ratio is below the lower threshold, the current path is deemed abnormal, and the task scheduling and error correction process begins. If it is determined that all scanned targets have been traversed, the process terminates; otherwise, it returns and continues executing the control logic for the next target.

[0259] It should be noted that this embodiment performs coarse positioning based on the prior matrix of the mirror angle; it uses unscented Kalman filtering to predict and compensate for mechanical errors and external disturbances, and combines closed-loop feedback from a semiconductor laser-position-sensitive detector to achieve sub-arcsecond fine-tuning; it completes fine locking through signal-to-noise ratio dual-threshold decision and probability model optimization, automatically triggering Fourier infrared interferometry spectrum acquisition. Through a hierarchical architecture of "coarse positioning → prediction compensation → closed-loop fine-tuning → intelligent evaluation," the difficulty of precise locking in rapid cruise scanning of mirror arrays is overcome. This control algorithm has significant advantages such as high precision, high automation, and strong adaptability, significantly improving optical path stability and scanning efficiency, and realizing high-precision, fully automatic, and intelligent adaptive control of the gimbal in complex scenarios.

[0260] Furthermore, the system structure of open optical path Fourier transform infrared spectroscopy is as follows: Figure 4 As shown, the system mainly includes a spectrometer, a corner mirror, a semiconductor laser, a position-sensitive detector (PSD), and a supporting data processing module, as well as a host computer and its integrated software for gimbal control and spectral analysis. The spectrometer is connected to a high-precision gimbal via a metal hardware connection, allowing for gas monitoring along different paths as the gimbal angle changes. To ensure consistency between the semiconductor laser and infrared light paths, the semiconductor laser is placed on the transceiver telescope housing, while the PSD is placed on the spectrometer housing, which is adjacent to the telescope. The infrared spectrum, PSD data processing unit, and high-precision 2D gimbal are connected to the host computer via USB serial port and RS232 / 485 serial port for data and command transmission. The spectrometer and corner mirror combine to form an infrared spectrum after two absorptions by the polluted gas cloud. Combined with the spectral quantitative analysis software on the host computer, accurate measurement of the target gas is achieved. The high-precision gimbal is the controlled object of the entire control system, adjusting its horizontal and vertical angles by receiving commands from the host computer. The semiconductor laser, corner mirror, PSD, and PSD data processing unit together form a feedback loop for closed-loop control, enabling refined gimbal control.

[0261] Specifically, another embodiment of the present invention proposes a scanning control system for multi-target cruise, deployed on a host computer for controlling a gimbal. A spectrometer is connected to the gimbal, and gas monitoring is achieved along different paths as the gimbal angle changes. A telescope is installed along the infrared light output path of the spectrometer, a semiconductor laser is arranged on the telescope housing, and a position-sensitive detector is arranged on the spectrometer housing. The control system includes:

[0262] The coarse positioning module 10 is used to control the gimbal to perform open-loop coarse positioning scanning with the prior target angle matrix of the corner reflector as the target trajectory, so that the laser emitted by the semiconductor laser is reflected by the corresponding corner reflector and hits the position-sensitive detector.

[0263] The closed-loop feedback module 20 is used to calculate the laser spot center offset through the position-sensitive detector and map the spot center offset into the angle compensation command of the gimbal to perform closed-loop micro-positioning control of the gimbal.

[0264] The path optimization module 30 is used to calculate the signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning based on the intensity signal of the light spot, and to trigger the spectrometer to perform interferometric spectrum acquisition based on the signal evaluation function value.

[0265] As a further preferred technical solution, the control system also includes a priori matrix initialization module, used for:

[0266] The gimbal is adjusted so that the infrared light emitted by the spectrometer returns to the interferometer via the corresponding corner reflector and maximizes the peak value of the ZPD signal in the interferogram. The angle parameter of the gimbal at this time is read as the prior target angle of the corresponding corner reflector.

[0267] The prior target angle matrix is ​​constructed based on the prior target angles of the corner reflectors at different positions.

[0268] As a further preferred technical solution, the coarse positioning module 10 includes:

[0269] The conversion unit is used to convert the prior target angles of the corresponding angle reflectors in the prior target angle matrix into motor pulses;

[0270] The coarse positioning unit is used to perform open-loop coarse positioning scanning based on motor pulse control of the gimbal to point to the prior target angle.

[0271] As a further preferred technical solution, the system also includes a motion compensation module, used for:

[0272] During the process of the gimbal turning from one prior target angle to the next prior target angle, the angle deviation component of the gimbal is predicted based on unscented Kalman filtering, and the position drift of the gimbal is compensated for based on the angle deviation.

[0273] As a further preferred technical solution, the motion compensation module includes:

[0274] The model building unit is used to build a state prediction model of the gimbal based on a nonlinear state transition function, and to build an observation model based on an observation function.

[0275] The unscented transformation unit is used to generate the first sigma point set based on the current state vector;

[0276] The first nonlinear propagation unit is used to propagate the first sigma point set through a nonlinear state transition function to obtain the second sigma point set, and calculate the mean and state covariance of the state variable prediction results based on the second sigma point set.

[0277] The second nonlinear propagation unit is used to propagate the first sigma point set through the observation function to obtain the third sigma point set, and calculate the mean, observation covariance and state-observation covariance of the prediction results of the observation vector based on the third sigma point set.

[0278] The state update unit is used to calculate the Kalman gain based on the observation covariance and the state-observation covariance, and update the state vector and state covariance based on the calculated Kalman gain combined with the prediction mean and observation data.

[0279] The predictive mean calculation unit is used to generate a new first sigma point set by using the updated state vector as the current state vector. This process is iterated until the termination condition is met, and the predicted mean angle of the gimbal is obtained.

[0280] An angle deviation component calculation unit is used to subtract the predicted mean angle from the corresponding prior target angle in the prior target angle matrix to obtain the angle deviation component.

[0281] As a further preferred technical solution, the closed-loop feedback module 20 specifically includes:

[0282] The spot position calculation unit is used to calculate the position of the reflected laser spot based on the four electrode current values ​​of the position-sensitive detector.

[0283] The angular displacement compensation calculation unit is used to compare the spot position with the desired spot position, obtain the horizontal and vertical displacement errors, and map the horizontal and vertical displacement errors into the angular displacement compensation amount of the corner reflector.

[0284] The micro-positioning control unit is used to generate compensation angle commands based on the angular displacement compensation amount to control the motor connected to the gimbal, and to perform closed-loop micro-positioning control of the gimbal.

[0285] As a further preferred technical solution, the micro-positioning control unit is specifically used for:

[0286] The angular displacement compensation amount is converted into a compensation angle command using a PID controller. The control parameters of the PID controller are adjusted by an artificial neural network based on the dynamic change trend of the target spot offset error. The input of the artificial neural network is the current error state vector, and the output is the optimal combination of gain parameters of the PID controller.

[0287] Based on the compensation angle command, the motor connected to the gimbal is controlled to perform closed-loop micro-positioning adjustment of the gimbal.

[0288] As a further preferred technical solution, the path optimization module 30 specifically includes:

[0289] The signal evaluation function value calculation unit is used to calculate the normalized signal-to-noise ratio as the signal evaluation function value of the gimbal at the current angle based on the intensity signal of the light spot;

[0290] The judgment unit is used to trigger the spectrometer to acquire interferometric spectrum when the signal evaluation function value is greater than or equal to the upper limit threshold of intensity judgment; when the signal evaluation function value is greater than the lower limit threshold of intensity judgment but less than the upper limit threshold of intensity judgment, it optimizes the gimbal pointing angle based on the Bayesian optimization fine angle locking method of the fine optimization module, and then triggers the spectrometer to acquire interferometric spectrum; when the signal evaluation function value is less than or equal to the lower limit threshold of intensity judgment, it determines that the current scanning path is abnormal.

[0291] It should be noted that other embodiments or specific implementation methods of the scanning control system for multi-target cruise described in this invention can refer to the above-described method embodiments, and will not be repeated here.

[0292] Furthermore, another embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the scanning control method for multi-target cruise as described in the first embodiment above.

[0293] Furthermore, another embodiment of the present invention provides a computer program product, including a computer program that is executed by a processor to implement the steps of the scanning control method for multi-target cruise as described in the first embodiment above.

[0294] It should be noted that the computer-readable medium disclosed in this embodiment may be a computer-readable signal medium, a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0295] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform a zero-sample image anomaly detection method according to the above embodiments.

[0296] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server.

[0297] In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0298] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0299] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0300] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" or "several" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0301] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A scanning control method for multi-target cruise, characterized in that, The method is used to control a gimbal. A spectrometer is connected to the gimbal, and gas monitoring is achieved along different paths as the gimbal angle changes. A telescope is positioned along the infrared light output path of the spectrometer, and a semiconductor laser is housed within the telescope's housing. A position-sensitive detector is housed within the spectrometer's housing. The method includes: Using the prior target angle matrix of the corner reflector as the target trajectory, the gimbal is controlled to perform an open-loop coarse positioning scan to point to the prior target angle, so that the laser emitted by the semiconductor laser is reflected by the corresponding corner reflector and hits the position sensitive detector. The laser spot center offset is calculated by a position-sensitive detector, and the spot center offset is mapped to the angle compensation command of the gimbal to perform closed-loop micro-positioning control of the gimbal. The signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning is calculated based on the intensity signal of the light spot, and the spectrometer is triggered to acquire the interferometric spectrum based on the signal evaluation function value.

2. The scanning control method for multi-target cruise as described in claim 1, characterized in that, Before controlling the gimbal to perform an open-loop coarse positioning scan to point to the target angle using the prior target angle matrix of the corner reflector as the target trajectory, the method further includes: The gimbal is adjusted so that the infrared light emitted by the spectrometer returns to the interferometer via the corresponding corner reflector and maximizes the peak value of the ZPD signal in the interferogram. The angle parameter of the gimbal at this time is read as the prior target angle of the corresponding corner reflector. The prior target angle matrix is ​​constructed based on the prior target angles of the corner reflectors at different positions.

3. The scanning control method for multi-target cruise as described in claim 1, characterized in that, The step of using the prior target angle matrix of the corner reflector as the target trajectory and controlling the gimbal to perform open-loop coarse positioning scan to point to the target angle includes: The prior target angles of the corresponding angle reflectors in the prior target angle matrix are converted into motor pulses; Open-loop coarse positioning scanning is performed using a motor pulse-controlled gimbal to point to the prior target angle.

4. The scanning control method for multi-target cruise as described in claim 1, characterized in that, Before calculating the laser spot center offset using a position-sensitive detector and mapping the spot center offset to a gimbal angle compensation command for closed-loop micro-positioning control of the gimbal, the method further includes: During the process of the gimbal turning from one prior target angle to the next prior target angle, the angle deviation component of the gimbal is predicted based on unscented Kalman filtering, and the position drift of the gimbal is compensated for based on the angle deviation.

5. The scanning control method for multi-target cruise as described in claim 4, characterized in that, The process of predicting the angle deviation component of the gimbal during rotation based on unscented Kalman filtering, and performing feedforward compensation for the position drift of the gimbal based on the angle deviation, includes: A state prediction model for the gimbal is constructed based on a nonlinear state transition function, and an observation model is constructed based on an observation function. Generate the first sigma point set based on the current state vector; The first sigma point set is propagated through a nonlinear state transition function to obtain the second sigma point set, and the mean and state covariance of the state variable prediction results are calculated based on the second sigma point set. The first sigma point set is propagated through the observation function to obtain the third sigma point set, and the mean, observation covariance, and state-observation covariance of the observation vector prediction results are calculated based on the third sigma point set. Kalman gain is calculated based on observation covariance and state-observation covariance, and the state vector and state covariance are updated based on the calculated Kalman gain, combined with the prediction mean and observation data. The updated state vector is used as the current state vector to generate a new first sigma point set. This process is iterated until the termination condition is met, and the average angle prediction of the gimbal is obtained. The angle deviation component is obtained by subtracting the mean of the angle prediction from the corresponding prior target angle in the prior target angle matrix.

6. The scanning control method for multi-target cruise as described in claim 1, characterized in that, The step of calculating the laser spot center offset using a position-sensitive detector and mapping the spot center offset to an angle compensation command for the gimbal to perform closed-loop micro-positioning control of the gimbal includes: The position of the reflected laser spot is calculated based on the four-electrode current values ​​of the position-sensitive detector; The position of the light spot is compared with the expected position of the light spot to obtain the horizontal and vertical displacement errors, and the horizontal and vertical displacement errors are mapped to the angular displacement compensation amount of the corner reflector. Based on the angular displacement compensation amount, a compensation angle command is generated to control the motor connected to the gimbal, thereby performing closed-loop micro-positioning control of the gimbal.

7. The scanning control method for multi-target cruise as described in claim 6, characterized in that, The process of generating compensation angle commands based on angular displacement compensation to control the motors connected to the gimbal, and performing closed-loop micro-positioning control of the gimbal, includes: The angular displacement compensation amount is converted into a compensation angle command using a PID controller. The control parameters of the PID controller are adjusted by an artificial neural network based on the dynamic change trend of the target spot offset error. The input of the artificial neural network is the current error state vector, and the output is the optimal combination of gain parameters of the PID controller. The motor connected to the gimbal is controlled based on the compensation angle command to perform closed-loop micro-positioning control of the gimbal.

8. The scanning control method for multi-target cruise as described in claim 7, characterized in that, The artificial neural network is a feedforward neural network pre-trained using the backpropagation algorithm. During training, the network weight parameters are corrected through online error feedback. The objective function used is: In the formula, θ meas (t) represents the angular displacement measured by the system, θ target (t) represents the angular displacement compensation amount obtained from the mapping.

9. The scanning control method for multi-target cruise as described in claim 1, characterized in that, The step of calculating the signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning based on the intensity signal of the light spot, and triggering the spectrometer interferometric spectrum acquisition based on the signal evaluation function value, includes: The normalized signal-to-noise ratio is calculated based on the intensity signal of the light spot and used as the signal evaluation function value at the current angle of the gimbal. When the signal evaluation function value is greater than or equal to the upper limit threshold of intensity determination, the spectrometer is triggered to acquire interferometric spectrum. When the signal evaluation function value is greater than the lower threshold of intensity judgment but less than the upper threshold of intensity judgment, the gimbal pointing angle is optimized based on the Bayesian optimization fine angle locking method, and then the spectrometer is triggered to acquire the interferometric spectrum. When the signal evaluation function value is less than or equal to the lower limit threshold for intensity determination, the current scanning path is determined to be abnormal.

10. The scanning control method for multi-target cruise as described in claim 9, characterized in that, The refined angle locking method based on Bayesian optimization optimizes the gimbal pointing angle, including: A black-box objective function is constructed using the current angle of the gimbal as input and the intensity of the light spot signal received by the position-sensitive detector as output. Modeling the black-box objective function using a Gaussian process model on the sampling point set; The improvement potential function of the current angle is calculated using the expected improvement criterion to select the next actual sampling angle; Based on the determined actual sampling angle, the gimbal is rotated and the intensity of the light spot signal received by the position-sensitive detector is calculated. The actual sampling angle is then added to the sampling point set and the Gaussian process model is updated. Repeatedly update the high-speed process model until the iteration termination condition is met to obtain the optimal pointing angle of the gimbal.

11. The scanning control method for multi-target cruise as described in claim 9, characterized in that, The modeling of the black-box objective function using a Gaussian process model on the sampling point set is expressed as follows: In the formula, μ(θ,φ) represents the expected estimate of the function value at (θ,φ) by the Gaussian process model, and k((θ,φ),(θ) = φ(θ,φ) = φ(θ,φ). ′ ,φ ′ J(θ,φ) is the covariance Gaussian kernel function, J(θ,φ) represents the value of the objective function, and the inputs are the parameter vectors θ and φ. Let be the notation for a Gaussian process, representing the value of the function at any finite sampling point as a V-variable Gaussian distribution.

12. The scanning control method for multi-target cruise as described in claim 9, characterized in that, The formula for calculating the improvement potential function of the current angle using the expected improvement criterion is expressed as follows: In the formula, μ and σ represent the mean and standard deviation of the Gaussian process prediction, respectively, and Φ(·) and φ(·) are the cumulative and density functions of the standard normal distribution, respectively. + J represents the optimal value of the objective function known at present, serving as a benchmark to calculate the potential improvement of the current prediction compared to the best result. Φ(Z) represents the cumulative distribution function of the standard normal distribution, used to measure the improvement of the current point compared to the optimal value. + The probability φ(Z) represents the probability density function of the standard normal distribution, which is used to measure the contribution of prediction uncertainty to the improvement potential, and EI(θ,φ) represents the expected improvement.

13. The scanning control method for multi-target cruise as described in claim 9, characterized in that, When the current scan path is determined to be abnormal, the method further includes: Calculate the time difference between the collection timestamp of the abnormal path and the initial path time, and add the abnormal path to the abnormal path set when the absolute value of the difference between the time difference and the preset time offset exceeds the set threshold. The abnormal paths in the abnormal path set are sorted according to their comprehensive priority score and then back-scanned. If the backscan is successful, the original data of the abnormal path is replaced. If the backscan fails, the missing data points are filled by interpolation or by using valid data from the previous round.

14. A scanning control system for multi-target cruise, characterized in that, The system is used to control the pan-tilt unit. A spectrometer is connected to the pan-tilt unit, and gas monitoring is achieved along different paths as the pan-tilt unit angle changes. A telescope is installed along the infrared light output path of the spectrometer, and a semiconductor laser is housed within the telescope's housing. A position-sensitive detector is housed within the spectrometer's housing. The control system includes: The coarse positioning module is used to control the gimbal to perform open-loop coarse positioning scanning with the prior target angle matrix of the corner reflector as the target trajectory, so that the laser emitted by the semiconductor laser is reflected by the corresponding corner reflector and hits the position-sensitive detector. The closed-loop feedback module is used to calculate the laser spot center offset through the position-sensitive detector and map the spot center offset into the gimbal angle compensation command to perform closed-loop micro-positioning control of the gimbal. The path optimization module is used to calculate the signal evaluation function value of the gimbal at the current angle after closed-loop fine-tuning based on the intensity signal of the light spot, and to trigger the spectrometer to perform interferometric spectrum acquisition based on the signal evaluation function value.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the scanning control method for multi-target cruise as described in any one of claims 1-13.

Images