A vibration compensation optimization method and device suitable for an atomic gravimeter

Abstract

The application provides a vibration compensation optimization method and device suitable for an atomic gravimeter, sets to-be-optimized parameters of the atomic gravimeter and generates a plurality of initial parameter samples, applies the initial parameter samples to the atomic gravimeter to perform an atomic interference experiment, obtains corresponding target function values, collectively takes all the initial parameter samples and the corresponding target function values as an observation data set, performs a plurality of iterations, evaluates newly generated un-evaluated samples according to the observation data set in each iteration, obtains a local optimal sample of the iteration, and adds the local optimal sample and the corresponding target function value to the observation data set, so that a global optimal sample can be obtained through the plurality of iterations, and the global optimal sample is applied to the atomic gravimeter to achieve a relatively optimal compensation effect and improve the accuracy of the atomic gravimeter. Since only a small number of atomic interference experiments are required to obtain the target function values among all the generated samples, the efficiency of achieving ideal compensation accuracy is greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of quantum precision measurement and parameter optimization technology, and more specifically, relates to a vibration compensation optimization method and device suitable for atomic gravimeters. Background Technology

[0002] In existing technologies, atomic gravimeters, with their ultra-high measurement accuracy at the microgal level, have significant application value in fields such as geophysical exploration, inertial navigation, and fundamental physics research. However, their performance is highly susceptible to interference from environmental vibrations. Vibrations drive the Raman mirror to move, which in turn modulates the Raman laser phase through the Doppler effect, introducing severe additional phase noise. This can lead to blurred or even completely obliterated interference fringes, degrading the uncertainty of a single measurement to hundreds of microgals.

[0003] To mitigate this effect, software vibration compensation technology is widely used. This technology reconstructs and subtracts the vibration phase by synchronously acquiring high-sensitivity seismograph signals and combining them with the velocity sensitivity function of an atomic interferometer. However, due to limitations in sensor response characteristics and system transmission paths, existing technologies have the following drawbacks: Sensor response mismatch: Although both the seismometer and the Raman mirror are excited by ground vibrations, their response paths to the same vibration differ due to their different installation locations and asymmetrical support structures. This difference manifests in two aspects: First, there is a three-axis orientation-dependent error, meaning that the three-axis sensitive axis system of the seismometer and the measurement axis (i.e., the sensitive axis) of the gravimeter have attitude deviations (i.e., misalignment), and the three-axis channels may have inconsistent sensitivity and non-ideal orthogonality, causing their raw output to be unable to be directly used to accurately characterize motion along the sensitive axis. Second, there is a dynamic mismatch in the transmission path, meaning that the mechanical transmission path from the seismometer installation point to the Raman mirror introduces frequency-related amplitude attenuation and phase lag, causing the seismometer signal to deviate from the actual motion of the mirror in both amplitude and timing.

[0004] To address the aforementioned mismatch issue, existing technologies have introduced three-dimensional compensation coefficients and gain-delay models. However, efficiently and accurately determining these key parameters remains another major challenge. Traditional methods, such as grid search, require traversing all parameter combinations, and their computational complexity increases exponentially with the parameter dimension. For example, for three-dimensional compensation coefficients, sampling 100 points per dimension requires up to 10... 6 (100×100×100) evaluations, because each evaluation requires a complete atomic interference experiment, a single experiment involves physical processes such as cold atom preparation, Raman pulse interaction, and fluorescence detection, which typically take several to tens of seconds, resulting in 10 6This evaluation is unacceptable in terms of both time and economic costs. Furthermore, when the parameter dimension expands to five dimensions (such as simultaneously optimizing three-dimensional compensation coefficients and gain delay parameters), the number of evaluations required for grid search explodes exponentially, making it practically impossible to obtain a sufficiently accurate parameter solution within a finite timeframe. While manual trial-and-error methods do not require traversal, their parameter adjustment direction lacks theoretical guidance, convergence speed is extremely slow, and results are highly dependent on the operator's experience, making it difficult to guarantee compensation accuracy.

[0005] Therefore, there is an urgent need for a technical solution that can achieve efficient, automatic, and high-precision calibration of vibration compensation parameters with fewer atomic interferometry experiments, so as to support the engineering application of atomic gravimeters in dynamic environments such as mobile platforms in the field. Summary of the Invention

[0006] The problem this invention aims to solve is how to improve the efficiency of achieving ideal compensation accuracy while simultaneously enhancing the compensation accuracy of atomic gravimeters.

[0007] Firstly, a vibration compensation optimization method suitable for atomic gravimeters is provided, including: Set at least one parameter to be optimized, which is related to the vibration compensation optimization of the atomic gravimeter; In this process, multiple parameter values ​​are generated within a preset range for each parameter to be optimized, and the parameter values ​​corresponding to all parameters to be optimized are randomly combined to obtain multiple initial parameter samples. Obtain the objective function value corresponding to each initial parameter sample, and use all initial parameter samples and their corresponding objective function values ​​as the observation dataset. The optimization of each parameter to be optimized is carried out through multiple rounds of iteration. In each round of iteration, multiple parameter values ​​are generated within a preset range of each parameter to be optimized. The parameter values ​​corresponding to all parameters to be optimized are randomly combined to obtain multiple new parameter samples as unevaluated samples. The local optimal sample among all unevaluated samples is obtained based on the observation dataset. The local optimal sample and its corresponding objective function value are added to the observation dataset, and the updated observation dataset is applied to the next round of iteration. After the iteration is completed, the parameter sample with the smallest objective function value in the observation dataset is selected as the global optimal sample, and the global optimal sample is applied to the vibration compensation optimization.

[0008] Preferably, at least one parameter to be optimized includes: Construct a three-dimensional compensation model for compensating for reference velocities in three-dimensional directions, and construct a delay compensation model for compensating for time delays; The three-dimensional compensation parameters are obtained from the three-dimensional compensation model, and the delay compensation parameters are obtained from the delay compensation model. Both the 3D compensation parameter and the delay compensation parameter are used as parameters to be optimized.

[0009] Preferably, the expression for the three-dimensional compensation model is: ; in, The reference velocity after three-dimensional compensation. For three-dimensional compensation parameters, This is the original reference speed.

[0010] The preferred expression for the delay compensation model is: ; in, The reference speed after delay compensation. The reference velocity after three-dimensional compensation. For equivalent amplitude gain, This is a time delay.

[0011] Preferably, the objective function value corresponding to each initial parameter sample is obtained, specifically including: The corresponding initial parameter samples are applied to the vibration compensation optimization, and the atomic gravimeter outputs the optimized interference fringe results. The theoretical fitting values ​​corresponding to the initial parameter samples are obtained based on the interference fringe results; Based on the theoretical fitting values ​​corresponding to the initial parameter samples and the atomic population probability of each scanning point on each interference fringe result, the objective function value corresponding to the initial parameter samples is obtained.

[0012] Preferably, the theoretical fitting values ​​corresponding to the initial parameter samples are obtained based on the interference fringe results, specifically including: The interference phase corresponding to the initial parameter sample is obtained from the interference fringe results; The theoretical fitting values ​​corresponding to the initial parameter samples are obtained based on the interference phase.

[0013] Preferably, the expression for the theoretical fitted value is: ; Where P is the theoretical fitted value, and a, b, and c are all fitted parameters. This is the interference phase.

[0014] Preferably, the objective function value corresponding to the initial parameter sample is obtained based on the theoretical fitting value corresponding to the initial parameter sample and the atomic population probability of each scanning point on each interference fringe result, specifically including: The expression for the objective function value is: ; in, Initial parameter samples The objective function value, N represents the total number of interference fringes in the interference fringe results. k The number of scan points for the k-th interference fringe. Let be the atomic population probability of the k-th interference fringe at the i-th scan point. These are the theoretical fitted values ​​corresponding to the initial parameter samples.

[0015] Preferably, the locally optimal sample among all unevaluated samples is obtained from the observed dataset, specifically including: The kernel matrix is ​​obtained based on the updated observation dataset corresponding to the previous iteration. Based on the kernel matrix, the updated observation dataset corresponding to the previous iteration, and all unevaluated samples generated in the current iteration, a joint probability distribution is performed to obtain the prediction mean and standard deviation for the corresponding iteration round. The excellence values ​​of all unevaluated samples in the current iteration are obtained based on the objective function value corresponding to the best parameter sample up to the current iteration, the predicted mean and standard deviation corresponding to the current iteration. The local optimum sample among all unevaluated samples in the current iteration is obtained based on the excellence value of all unevaluated samples.

[0016] Preferably, the excellence value of all unevaluated samples in the current iteration is obtained based on the objective function value corresponding to the best parameter sample up to the current iteration, the predicted mean and standard deviation corresponding to the current iteration, specifically including: The expression for the excellence value is: ; in, Unevaluated sample The excellence value, This is the predicted mean for this iteration. This represents the standard deviation corresponding to this iteration. This represents the objective function value corresponding to the optimal parameter sample up to the current iteration round. For balancing parameters, Let be the cumulative distribution function of the standard normal distribution. Let be the probability density function of the standard normal distribution. When the standard deviation is 0, the excellence value is set to 0.

[0017] In a second aspect, a vibration compensation optimization device suitable for an atomic gravimeter is provided, comprising at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the processor for performing a vibration compensation optimization method suitable for an atomic gravimeter.

[0018] Thirdly, the present invention also provides a non-volatile computer storage medium storing computer-executable instructions that, when executed by one or more processors, cause the processors to perform the method of the first aspect.

[0019] Fourthly, a chip is provided, comprising: a processor and an interface for calling and running a computer program stored in memory, performing the method as described in the first aspect.

[0020] Fifthly, a computer program product containing instructions is provided that, when executed on a computer or processor, causes the computer or processor to perform the method as described in the first aspect.

[0021] In a sixth aspect, a vibration compensation optimization system suitable for an atomic gravimeter is provided, comprising a vibration compensation optimization device suitable for an atomic gravimeter as described in the second aspect, and using a vibration compensation optimization method suitable for an atomic gravimeter as described in the first aspect.

[0022] Unlike existing technologies, the present invention has at least the following beneficial effects: The parameters to be optimized for the atomic gravimeter are set, and multiple initial parameter samples are generated. These samples are then applied to the atomic gravimeter for atomic interferometry experiments to obtain the corresponding objective function values. All initial parameter samples and their corresponding objective function values ​​are used together as the observation dataset for multiple iterations. In each iteration, the newly generated unevaluated samples are evaluated based on the observation dataset to obtain the locally optimal samples for that iteration. The locally optimal samples and their corresponding objective function values ​​are then added to the observation dataset. Through multiple iterations, the globally optimal samples are obtained and applied to the atomic gravimeter to achieve a relatively optimal compensation effect and improve the accuracy of the atomic gravimeter. Since only a small number of atomic interferometry experiments are needed to obtain the objective function values ​​from all generated samples, the efficiency of achieving ideal compensation accuracy is greatly improved. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0024] Figure 1 This is a flowchart of a vibration compensation optimization method for an atomic gravimeter provided in an embodiment of the present invention; Figure 2 This is a flowchart of a vibration compensation optimization method for an atomic gravimeter provided in an embodiment of the present invention for obtaining the parameters to be optimized; Figure 3 This is a flowchart of a method for obtaining the objective function value in a vibration compensation optimization method applicable to an atomic gravimeter, provided by an embodiment of the present invention. Figure 4 This is a flowchart of a vibration compensation optimization method for atomic gravimeters provided in an embodiment of the present invention for obtaining locally optimal samples; Figure 5 This is a schematic diagram of a vibration compensation optimization device for an atomic gravimeter provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0026] Unless the context otherwise requires, throughout the specification and claims, the term "comprising" is interpreted as openly inclusive, meaning "including, but not limited to." In the description of the specification, terms such as "one embodiment," "some embodiments," "exemplary embodiment," "example," "specific example," or "some examples" are intended to indicate that a particular feature, structure, material, or characteristic associated with that embodiment or example is included in at least one embodiment or example of this disclosure. The illustrative representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, a particular feature, structure, material, or characteristic may be included in any suitable manner in any one or more embodiments or examples; that is, although they may be incorporated in embodiments or examples using the above terms for reasons such as order and position, it does not limit them to be incorporated in combination by a single embodiment or example.

[0027] In the description of this invention, 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 with "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more. Furthermore, for example, the description may use the prefix "A" or "B" to describe the same type of nouns as two independent entities. In this case, the corresponding features defined with "A" and "B" are used only to distinguish between similar entities and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features.

[0028] In the description of this invention, the expression “A and / or B” (where A and B are used to formally represent specific features) will be used. The corresponding expression includes the following three combinations: only A, only B, and a combination of A and B.

[0029] As used in this invention, “about,” “approximately,” or “approximately” includes the stated value and the average value within an acceptable range of deviation from a particular value, wherein the acceptable range of deviation is determined by those skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the particular quantity (i.e., the limitations of the measurement system).

[0030] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0031] Example 1: This embodiment provides a vibration compensation optimization method suitable for atomic gravimeters, such as... Figure 1 As shown, the method flow includes: In step 101, at least one parameter to be optimized is set, and multiple parameter values ​​are generated within a preset range of each parameter to be optimized. The parameter values ​​corresponding to all parameters to be optimized are randomly combined to obtain multiple initial parameter samples.

[0032] This embodiment applies to the vibration compensation optimization of an atomic gravimeter. During actual vibration monitoring, the seismometer and atomic gravimeter are mounted at different positions on the support structure, resulting in asymmetrical transmission characteristics of the support structure. This causes a delay in the transmission of vibration information from the seismometer to the atomic gravimeter, leading to amplitude attenuation and time delay. Consequently, the reference velocity in the three-dimensional direction detected by the seismometer differs from the reference velocity in the three-dimensional direction of the atomic gravimeter. The reference velocity in the three-dimensional direction refers to the decomposition velocity of the detected vibration in the three-dimensional direction. Therefore, it is necessary to compensate for the velocity difference and delay between the seismometer and the atomic gravimeter to ensure the accuracy of the atomic gravimeter during use. The parameters used in the above compensation optimization process are the parameters to be optimized. The more reasonable the selected parameter values, the better the compensation optimization effect for the atomic gravimeter.

[0033] The preset range is set by those skilled in the art according to the actual situation. Each parameter to be optimized is randomly generated within its corresponding preset range. All parameter types are provided with a randomly generated value to form an initial parameter sample. A single initial parameter sample is applied to the atomic gravimeter to complete the optimization compensation calculation. However, the effect of the corresponding optimization compensation depends on the quality of the initial parameter sample itself.

[0034] In step 102, the objective function value corresponding to each initial parameter sample is obtained, and all initial parameter samples and their corresponding objective function values ​​are used together as the observation dataset.

[0035] In this embodiment, a complete atomic interferometry experiment is performed for each initial parameter sample to obtain complete interference fringe data. The objective function value corresponding to the respective initial parameter sample can be obtained from the interference fringe data. The objective function value represents the excellence of the corresponding initial parameter sample; the better the optimization and compensation effect of the initial parameter sample for the atomic gravimeter, the smaller the corresponding objective function value. The observation dataset is as follows: ,in, For the i-th initial parameter sample, Let be the objective function value corresponding to the i-th initial parameter sample.

[0036] In step 103, each parameter to be optimized is optimized through multiple rounds of iteration. In each round of iteration, multiple parameter values ​​are randomly generated within a preset range for each parameter to be optimized. The parameter values ​​corresponding to all parameters to be optimized are randomly combined to obtain multiple new parameter samples as unevaluated samples. The local optimal sample among all unevaluated samples is obtained based on the observation dataset. The local optimal sample and its corresponding objective function value are added to the observation dataset, and the updated observation dataset is applied to the next round of iteration.

[0037] In this embodiment, unevaluated samples are: other samples that can be generated besides the initial parameter samples that have been generated. Since the objective function value has not been calculated in the previous steps, the quality of these samples is not yet clear, so they are called unevaluated samples.

[0038] In this embodiment, a Gaussian model is obtained based on the observation dataset, and the predicted mean and standard deviation of each unevaluated sample are calculated using the Gaussian model. The expected improvement value of each unevaluated sample generated in this iteration is obtained based on the predicted mean and standard deviation. The expected improvement value can be used to represent the excellence of the corresponding unevaluated sample. Therefore, the local optimal sample in this iteration can be screened based on the expected improvement values ​​of all unevaluated samples. The local optimal sample is applied to the atomic gravimeter to conduct atomic interference experiments, and the objective function value corresponding to the local optimal sample is calculated. The local optimal sample and the corresponding objective function value are added to the observation dataset, which is the updated observation dataset. Therefore, the local optimal sample in this iteration is added to the observation dataset in each iteration and used in the next iteration.

[0039] In step 104, after the iteration is completed, the parameter sample with the smallest objective function value in the observation dataset is selected as the global optimal sample, and the global optimal sample is applied to the vibration compensation optimization.

[0040] In this embodiment, the criterion for determining the end of the iteration can be: a maximum number of iterations is preset, and before each iteration, the number of iterations in the current iteration is compared with the maximum number of iterations. If the current number of iterations is less than the maximum number of iterations, the iteration is determined to be not yet over. If the current number of iterations is greater than or equal to the maximum number of iterations, the iteration ends.

[0041] In this embodiment, the parameters to be optimized for the atomic gravimeter are set and multiple initial parameter samples are generated. These samples are then applied to the atomic gravimeter for atomic interferometry experiments to obtain the corresponding objective function values. All initial parameter samples and their corresponding objective function values ​​are used together as the observation dataset for multiple iterations. In each iteration, the newly generated unevaluated samples are evaluated based on the observation dataset to obtain the locally optimal samples for that iteration. The locally optimal samples and their corresponding objective function values ​​are then added to the observation dataset. Through multiple iterations, the globally optimal samples can be obtained and applied to the atomic gravimeter to achieve a relatively optimal compensation effect and improve the accuracy of the atomic gravimeter. Since only the initial parameter samples and locally optimal samples need to undergo atomic interferometry experiments to obtain the objective function values ​​among all generated samples, the efficiency of achieving ideal compensation accuracy is greatly improved.

[0042] Furthermore, in this embodiment, at least one parameter to be optimized is set, such as... Figure 2 As shown, the method flow includes: In step 201, a three-dimensional compensation model is constructed to compensate for the three-dimensional directional reference velocity, and a delay compensation model is constructed to compensate for the time delay.

[0043] In this embodiment, the setting of the parameters to be optimized needs to consider the actual compensation of the atomic gravimeter. One aspect is compensation for environmental vibration, i.e., a three-dimensional compensation model. This three-dimensional compensation model is used to correct directional coupling errors introduced by inaccurate alignment between the seismometer's three-axis sensitive directions and the atomic gravimeter's measurement axes, as well as inconsistencies in sensitivity or non-ideal orthogonality between the three-axis channels. Therefore, the three-dimensional compensation model introduces… and the three-dimensional compensation coefficient vector The original triaxial velocities of the seismometer are linearly combined to obtain the corrected effective velocity. , where m x and m y Both are used to correct horizontal sensitivity deviations and cross-coupling, m z Used to correct the gain in the vertical direction. The original velocity along the X-axis in the horizontal direction of the seismometer is [value missing]. The initial velocity along the y-axis in the horizontal direction of the seismometer is [value missing]. The original velocity in the vertical direction of the seismometer.

[0044] On the other hand, there is delay compensation, namely the delay compensation model. This model corrects for amplitude attenuation and time delay caused by the different physical installation positions of the Raman mirrors in the seismometer and atomic gravimeter, as well as the asymmetry of the transmission characteristics of the support structure. This model converts the reference velocity corrected by the three-dimensional compensation model into the best estimate of the true velocity of the Raman mirror. The expression is: Where A is the equivalent amplitude gain, used to compensate for amplitude attenuation along the transmission path. This is a time delay used to compensate for phase lag in the transmission path.

[0045] In step 202, the three-dimensional compensation parameters are obtained based on the three-dimensional compensation model, and the delay compensation parameters are obtained based on the delay compensation model.

[0046] The three-dimensional compensation model essentially optimizes the initial reference velocity using three-dimensional compensation parameters to obtain the reference velocity after three-dimensional compensation. Therefore, the expression of the three-dimensional compensation model can be obtained by decomposing it. Similarly, the delay compensation model essentially optimizes the reference velocity after three-dimensional compensation using delay compensation parameters to obtain the reference velocity after delay compensation. Therefore, the delay compensation parameters can be obtained by decomposing the expression of the three-dimensional compensation model.

[0047] In step 203, both the three-dimensional compensation parameters and the delay compensation parameters are used as parameters to be optimized.

[0048] The expression for the three-dimensional compensation model involved in this step is: ; in, The reference velocity after three-dimensional compensation. For three-dimensional compensation parameters, This is the original reference speed.

[0049] Through the expression of the three-dimensional compensation model It can be seen that the parameter used for environmental vibration compensation is m, therefore m is a three-dimensional compensation parameter.

[0050] The expression for the delay compensation model involved in this step is: ; in, The reference speed after delay compensation. The reference velocity after three-dimensional compensation. For equivalent amplitude gain, This is a time delay.

[0051] Through the expression of the delay compensation model It can be seen that the parameters used for delay compensation are: and ,therefore and Together they serve as parameters for delay compensation.

[0052] Based on the confirmation of the above-mentioned parameters to be optimized, the preset ranges corresponding to each parameter to be optimized can be: , , , , Each parameter type generates parameters randomly within its corresponding preset range. Combining the individual parameters generated by each parameter type forms a parameter sample.

[0053] Furthermore, once the parameters to be optimized are determined, a corresponding method is needed to determine the excellence of the parameters to be optimized corresponding to different initial parameter samples. Therefore, it is necessary to obtain the objective function value corresponding to the corresponding initial parameter sample. The corresponding design is as follows: Obtain the objective function value corresponding to each initial parameter sample, such as... Figure 3 As shown, the method flow includes: In step 301, the corresponding initial parameter samples are applied to the vibration compensation optimization, and the atomic gravimeter outputs the optimized interference fringe results.

[0054] In this embodiment, the interference fringe result refers to the periodic change signal of the atomic population (i.e. the number of atoms in a specific quantum state) caused by the phase difference between different paths after undergoing the process of beam splitting, propagation and recombining, utilizing the material wave characteristics of atoms.

[0055] In step 302, the theoretical fitting values ​​corresponding to the initial parameter samples are obtained based on the interference fringe results.

[0056] In this embodiment, the interference phase corresponding to the initial parameter sample is obtained based on the interference fringe results; the theoretical fitting value corresponding to the initial parameter sample is obtained based on the interference phase.

[0057] The expression for the theoretical fitted value involved in this step is: ; Where P is the theoretical fitted value, and a, b, and c are all fitted parameters. This is the interference phase.

[0058] In step 303, the objective function value corresponding to the initial parameter sample is obtained based on the theoretical fitting value corresponding to the initial parameter sample and the atomic population probability of each scanning point on each interference fringe result.

[0059] In this embodiment, the expression for the objective function value is: ; in, Initial parameter samples The objective function value, N represents the total number of interference fringes in the interference fringe results. k The number of scan points for the k-th interference fringe. Let be the atomic population probability of the k-th interference fringe at the i-th scan point. These are the theoretical fitted values ​​corresponding to the initial parameter samples.

[0060] As can be seen from the above steps, for an atomic gravimeter, when calculating the objective function value of each initial parameter sample, it is necessary to apply the corresponding initial parameter sample to the atomic gravimeter to conduct an atomic interference experiment in order to obtain the corresponding interference fringe results, which are then used to calculate the objective function value. If all parameter samples are evaluated for their excellence in order to find the relatively optimal parameter sample, the computational load is too large, the process is cumbersome and inefficient.

[0061] Furthermore, in this embodiment, to avoid evaluating the excellence of parameter samples by directly calculating the objective function value, a multi-round iterative approach is used. In each round of iteration, a relatively better parameter sample is found and included in the observation dataset until the end of the multi-round iteration, and the globally optimal sample is found. The corresponding design is as follows: Based on the observation dataset, the locally optimal sample among all unevaluated samples is obtained, such as... Figure 4 As shown, the method flow includes: In step 401, the kernel matrix is ​​obtained based on the updated observation dataset corresponding to the previous iteration.

[0062] It is important to note that in each iteration, parameters need to be randomly generated within a preset range for each parameter type to form new samples whose excellence has not been evaluated, i.e., unevaluated samples. In each iteration, a specified number of unevaluated samples need to be randomly generated, and relatively better samples need to be found among these unevaluated samples.

[0063] In this embodiment, the updated observation dataset corresponding to the previous iteration is the observation dataset that includes the locally optimal samples from the previous iteration and their corresponding objective function values. A Gaussian process surrogate model is constructed to obtain the kernel matrix, where: a corresponding covariance matrix is ​​established based on the n sets of data in the updated observation dataset corresponding to the previous iteration; and a cross-covariance matrix is ​​established by selecting m unevaluated samples within a preset interval. And construct the covariance matrix within the m parameter samples to be predicted. ,in, , and Together they form the kernel matrix K.

[0064] In step 402, the joint probability distribution is performed based on the kernel matrix, the updated observation dataset corresponding to the previous iteration, and all unevaluated samples generated in the current iteration to obtain the prediction mean and standard deviation corresponding to the corresponding iteration round.

[0065] In step 403, the excellence values ​​of all unevaluated samples in the current iteration are obtained based on the objective function value corresponding to the optimal parameter sample up to the current iteration, the predicted mean and standard deviation corresponding to the current iteration.

[0066] In this embodiment, the optimal parameter sample is the parameter sample with the smallest objective function value in the observation dataset up to the end of this iteration.

[0067] The expression for the excellence value is: ; in, Unevaluated sample The excellence value, This is the predicted mean for this iteration. This represents the standard deviation corresponding to this iteration. This represents the objective function value corresponding to the optimal parameter sample up to the current iteration round. For balancing parameters, Let be the cumulative distribution function of the standard normal distribution. Let Z be the probability density function of the standard normal distribution; where Z is... .

[0068] In step 404, the local optimal sample among all unevaluated samples in the current iteration is obtained based on the excellence values ​​of all unevaluated samples.

[0069] In this embodiment, the expression for the locally optimal sample is: ; in, For local optimal samples, Unevaluated sample The excellence value.

[0070] After reaching the maximum number of iterations, the parameter sample with the smallest objective function in the observation dataset updated in the last iteration is taken as the global optimal sample. The parameters of the global optimal sample are then substituted into the compensation process of the atomic gravimeter to achieve high-precision gravity measurement.

[0071] Furthermore, in this embodiment, the Gaussian process surrogate model can be replaced with other regression models, such as the Random Forest model or the Tree-structured Parzen Estimator (TPE), especially when the parameter dimension is higher or the data volume is larger, the replacement of the above models is more applicable.

[0072] On the other hand, in this embodiment, the expected improvement acquisition function is used to calculate the excellence value. Alternatively, for example, the upper confidence bound (UCB) or probability of improvement (PI) can be used for calculation.

[0073] On the other hand, in this embodiment, three-dimensional compensation and delay compensation are processed as two independent optimization tasks. As an alternative, the three parameters (m) of three-dimensional compensation can also be processed separately. x ,m y ,m z The two parameters (A, τ) for delay compensation are used together as a five-dimensional parameter space for joint Bayesian optimization.

[0074] In summary, the vibration compensation optimization method presented in this embodiment achieves ultra-high efficiency automatic parameter calibration, laying the foundation for real-time compensation. By constructing a Gaussian process proxy model and an intelligent acquisition strategy, it approximates the global optimum with extremely low experimental costs, requiring only a few parameter evaluations to complete the high-precision calibration of the parameters to be optimized. Compared to the tens of thousands of evaluations required by the traditional grid search method, this method improves efficiency by more than three orders of magnitude. This breakthrough solves the bottleneck problem of unacceptable calibration costs in high-dimensional parameter spaces, making it possible to perform online or near real-time adaptive tuning of vibration compensation parameters on mobile platforms or in dynamic environments, greatly expanding the application scenarios of atomic gravimeters.

[0075] After compensation using the parameters calibrated using this method, the uncertainty of gravity measurement is significantly reduced, the gain delay compensation is significantly optimized, and the final parameter solutions are highly consistent. Based on the efficient acquisition of high-precision parameters, the fidelity of vibration phase reconstruction is greatly improved. The entire calibration process requires no manual intervention and is fully automated. The system corresponding to this method is far more sensitive to time delay τ than to amplitude gain A, highlighting the decisive role of high-precision time synchronization in the vibration compensation link and providing a clear optimization direction for subsequent instrument structure and electronic design.

[0076] Those skilled in the art will understand that the vibration compensation optimization method of the present invention is not limited to atomic gravimeters, but can also be applied to other quantum precision measurement devices that require high-precision parameter calibration, such as atomic clocks and atomic interferometers, without limitation.

[0077] Example 2: like Figure 5 The diagram shown is a structural schematic of a vibration compensation optimization device for an atomic gravimeter according to an embodiment of the present invention. The vibration compensation optimization device for an atomic gravimeter in this embodiment includes one or more processors 41 and a memory 42.

[0078] Processor 41 and memory 42 can be connected via a bus or other means. Figure 5 Taking the example of a connection between China and Israel via a bus.

[0079] The memory 42, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs and non-volatile computer-executable programs, such as the vibration compensation optimization method for an atomic gravimeter in the above embodiments. The processor 41 executes the vibration compensation optimization method for an atomic gravimeter by running the non-volatile software program and instructions stored in the memory 42.

[0080] Memory 42 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 42 may optionally include memory remotely located relative to processor 41, which can be connected to processor 41 via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0081] The program instructions / modules are stored in memory 42 and, when executed by one or more processors 41, perform the vibration compensation optimization method for atomic gravimeters described in the above embodiments.

[0082] This invention also provides a computer storage medium storing computer program instructions; when the computer program instructions are executed by one or more processors, the processors execute the vibration compensation optimization method for atomic gravimeters provided in this invention.

[0083] Those skilled in the art will readily understand that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A vibration compensation optimization method suitable for atomic gravimeters, characterized in that, include: At least one parameter to be optimized is set, which is related to the vibration compensation optimization of the atomic gravimeter; In this process, multiple parameter values ​​are generated within a preset range for each of the parameters to be optimized, and the parameter values ​​corresponding to all the parameters to be optimized are randomly combined to obtain multiple initial parameter samples. Obtain the objective function value corresponding to each initial parameter sample, and use all the initial parameter samples and their corresponding objective function values ​​as the observation dataset; The optimization of each parameter to be optimized is carried out through multiple rounds of iteration. In each round of iteration, multiple parameter values ​​are generated within the preset range of each parameter to be optimized. The parameter values ​​corresponding to all the parameters to be optimized are randomly combined to obtain multiple new parameter samples as unevaluated samples. The local optimal sample among all the unevaluated samples is obtained according to the observation dataset. The local optimal sample and its corresponding objective function value are added to the observation dataset, and the updated observation dataset is applied to the next round of iteration. After the iteration is completed, the parameter sample with the smallest objective function value in the observation dataset is selected as the global optimal sample, and the global optimal sample is applied to the vibration compensation optimization.

2. The vibration compensation optimization method for atomic gravimeters according to claim 1, characterized in that, At least one of the parameters to be optimized is specifically included in the following: Construct a three-dimensional compensation model for compensating for reference velocities in three-dimensional directions, and construct a delay compensation model for compensating for time delays; The three-dimensional compensation parameters are obtained based on the three-dimensional compensation model, and the delay compensation parameters are obtained based on the delay compensation model. Both the three-dimensional compensation parameter and the delay compensation parameter are used as the parameters to be optimized.

3. The vibration compensation optimization method for atomic gravimeters according to claim 2, characterized in that, The expression for the three-dimensional compensation model is: ； in, The reference velocity after three-dimensional compensation. For three-dimensional compensation parameters, This is the original reference speed.

4. The vibration compensation optimization method for atomic gravimeters according to claim 2, characterized in that, The expression for the delay compensation model is: ； in, The reference speed after delay compensation. The reference velocity after three-dimensional compensation. For equivalent amplitude gain, This is a time delay.

5. The vibration compensation optimization method for atomic gravimeters according to claim 1, characterized in that, The step of obtaining the objective function value corresponding to each initial parameter sample specifically includes: The corresponding initial parameter samples are applied to the vibration compensation optimization, and the atomic gravimeter outputs the optimized interference fringe results. Based on the interference fringe results, the theoretical fitting values ​​corresponding to the initial parameter samples are obtained; The objective function value corresponding to the initial parameter sample is obtained based on the theoretical fitting value corresponding to the initial parameter sample and the atomic population probability of each scanning point on each interference fringe result.

6. The vibration compensation optimization method for atomic gravimeters according to claim 5, characterized in that, The step of obtaining the theoretical fitting value corresponding to the initial parameter sample based on the interference fringe result specifically includes: The interference phase corresponding to the initial parameter sample is obtained based on the interference fringe results. The theoretical fitting value corresponding to the initial parameter sample is obtained based on the interference phase.

7. The vibration compensation optimization method for atomic gravimeters according to claim 6, characterized in that, The expression for the theoretical fitted value is: ； Where P is the theoretical fitted value, and a, b, and c are all fitting parameters. The interference phase is denoted as .

8. The vibration compensation optimization method for atomic gravimeters according to claim 5, characterized in that, The step of obtaining the objective function value corresponding to the initial parameter sample based on the theoretical fitting value corresponding to the initial parameter sample and the atomic population probability of each scanning point on each interference fringe result specifically includes: The expression for the objective function value is: ； in, Initial parameter samples The objective function value, N represents the total number of interference fringes in the interference fringe results. k The number of scan points for the k-th interference fringe. Let be the atomic population probability of the k-th interference fringe at the i-th scan point. The theoretical fitted value corresponding to the initial parameter sample.

9. The vibration compensation optimization method for atomic gravimeters according to claim 1, characterized in that, The step of obtaining the locally optimal sample among all the unevaluated samples based on the observed dataset specifically includes: The kernel matrix is ​​obtained based on the updated observation dataset corresponding to the previous iteration. Based on the kernel matrix, the updated observation dataset corresponding to the previous iteration, and all the unevaluated samples generated in the current iteration, a joint probability distribution is performed to obtain the prediction mean and standard deviation for the corresponding iteration round. The excellence values ​​of all unevaluated samples in the current iteration are obtained based on the objective function value corresponding to the optimal parameter sample up to the current iteration, the predicted mean and the standard deviation corresponding to the current iteration. The local optimum sample in the current iteration is obtained from the excellence value of all the unevaluated samples.

10. The vibration compensation optimization method for atomic gravimeters according to claim 9, characterized in that, The step of obtaining the excellence value of all unevaluated samples in the current iteration based on the objective function value corresponding to the optimal parameter sample up to the current iteration, the predicted mean and the standard deviation corresponding to the current iteration, specifically includes: The expression for the excellence value is: ； in, Unevaluated sample The excellence value, This is the predicted mean value corresponding to this iteration. The standard deviation corresponding to this iteration is... The objective function value corresponding to the optimal parameter sample up to the current iteration round. For balancing parameters, Let be the cumulative distribution function of the standard normal distribution. Let be the probability density function of a standard normal distribution, and when the standard deviation is 0, the excellence value is set to 0.

11. A vibration compensation optimization device suitable for atomic gravimeters, characterized in that, The method includes at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor for performing the vibration compensation optimization method for an atomic gravimeter as described in any one of claims 1-10.

12. A non-volatile computer storage medium, characterized in that, The computer storage medium stores computer program instructions that, when executed by one or more processors, cause the processors to perform the vibration compensation optimization method for an atomic gravimeter as described in any one of claims 1-10.

Images