Clock error drift modeling method and system based on statistical likelihood optimization

Abstract

The application discloses a clock error drift modeling method and system based on statistical likelihood optimization. The method first acquires clock error data in a clock synchronization system and performs standardization processing; then, clock speed and clock drift are calculated, and the clock drift is modeled as a random process conforming to a Gaussian process; next, a neural network model is constructed, the negative log-likelihood of the clock drift is taken as a loss function for model training, and the loss function is optimized to minimize the clock error drift error; finally, the system generates correction instructions by using the trained model, and high-precision clock synchronization is maintained when GNSS signals are missing. The application effectively improves the robustness and precision of clock synchronization, and is particularly suitable for environments where GNSS signals are unstable or unavailable.

Description

Technical Field

[0001] This invention relates to the field of navigation and timing technology, and in particular to a clock drift modeling method and system based on statistical likelihood optimization. Background Technology

[0002] High-precision clock synchronization is the cornerstone of many modern key technologies. With the widespread adoption and application of Global Navigation Satellite Systems (GNSS), accurate time references have become particularly important. Ideally, receiving equipment can calibrate its local clock and maintain synchronization with the system time by continuously receiving GNSS signals. However, in real-world applications, GNSS signals are susceptible to interference from various factors, such as building obstruction, tunnel environments, severe weather, or malicious interference, leading to signal quality degradation or even complete interruption.

[0003] Traditional clock drift modeling methods typically employ differential constraints based on physical models, which fail to adequately consider the stochastic characteristics of clock drift and the impact of system noise, resulting in low clock synchronization accuracy in complex environments. While these traditional methods perform well in stable signal environments, they often struggle to capture the nonlinear and random components of clock drift. In complex dynamic environments or situations with prolonged GNSS signal interruptions, traditional models are insufficient in fitting the dynamic characteristics of clock bias, leading to increased prediction errors and a significant decrease in synchronization accuracy.

[0004] Therefore, there is an urgent need to develop a robust modeling method that can better characterize the random clock drift characteristics and maintain high-precision synchronization even when GNSS signals are missing. This invention aims to solve this technical challenge by introducing statistical likelihood optimization and Gaussian process modeling to provide a more accurate and reliable clock drift compensation scheme. Summary of the Invention

[0005] The purpose of this invention is to provide a clock drift modeling method based on statistical likelihood optimization. By modeling clock drift changes as a Gaussian process and using negative log-likelihood as a loss function, the clock synchronization process is optimized, overcoming the limitations of traditional methods in complex environments and ensuring high-precision and high-reliability clock synchronization of the system.

[0006] To solve the above-mentioned technical problems, the technical solution proposed in this application is as follows:

[0007] This invention provides a clock drift modeling method based on statistical likelihood optimization, comprising the following steps:

[0008] S1. Obtain clock difference data from the clock synchronization system, obtain clock difference information for each time step, and standardize the data.

[0009] S2. Calculate the clock speed based on the clock difference data. And Zhong Piao ;

[0010] S3, the clock will drift The process is modeled as a Gaussian stochastic process with a mean of zero and a variance of . ;

[0011] S4. Construct a neural network model capable of predicting clock errors. The neural network model adopts a multilayer perceptron structure, with historical clock error data as input and predicted clock error sequence as output. The model parameters are then initialized.

[0012] S5. Calculate the negative log-likelihood of clock drift and use the negative log-likelihood as the loss function of the model. By optimizing the loss function during the training process, the parameters of the clock drift model are dynamically adjusted.

[0013] S6. Using the pre-trained clock drift model, generate and send correction commands to the atomic clock based on the predicted clock drift.

[0014] Preferably, the data standardization process in step S1 specifically involves: performing Z-score standardization on the clock difference data, using the formula...

[0015]

[0016] Data for each clock difference Convert to standardized data ,in, The mean of the clock difference data. represents the standard deviation of the clock error data.

[0017] Preferably, the calculation of clock speed in step S2 is... And Zhong Piao Specifically, this is achieved through the following formula:

[0018]

[0019]

[0020] Preferably, the Gaussian process in step S3 is expressed as:

[0021]

[0022] Preferably, the loss function described in step S5 Defined as:

[0023]

[0024] in, Output the clock drift value of the sequence for the model.

[0025] Preferably, the dynamic adjustment of clock drift model parameters in step S5 includes the following sub-steps:

[0026] S51. Employ a gradient descent-based optimization algorithm to minimize the loss function;

[0027] S52. Adjusting hyperparameters during the optimization process;

[0028] S53. The training process is dynamically adjusted based on the clock synchronization accuracy of the validation set.

[0029] On the other hand, this application also claims protection for a clock drift modeling system based on statistical likelihood optimization, comprising:

[0030] The data acquisition and standardization module is used to acquire clock difference data from the clock synchronization system, obtain clock difference information for each time step, and perform standardization processing on the data.

[0031] The clock speed and clock drift calculation module is used to calculate the clock speed based on the clock difference data. And Zhong Piao ;

[0032] Gaussian process modeling module, used to model clock drift The process is modeled as a Gaussian stochastic process with a mean of zero and a variance of . ;

[0033] The neural network module, employing a multilayer perceptron structure, is used to receive historical clock difference data and output predicted clock difference sequences.

[0034] The training optimization module is used to calculate the negative log-likelihood of clock drift and use it as the loss function. And by optimizing the loss function, the clock drift model parameters are dynamically adjusted;

[0035] The calibration command generation module is used to generate and send calibration commands to the atomic clock based on the predicted clock drift using a pre-trained clock drift model.

[0036] Preferably, the data acquisition and standardization module is specifically used for:

[0037] The clock error data is Z-score standardized using formula (3).

[0038]

[0039] Data for each clock difference Convert to standardized data ,in, The mean of the clock difference data. represents the standard deviation of the clock error data.

[0040] Furthermore, this application also claims protection for an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, performs the steps of the method as described in any of the foregoing.

[0041] Furthermore, this application also claims protection for a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described in any of the foregoing.

[0042] Compared with the prior art, the present invention achieves the following beneficial technical effects:

[0043] This invention significantly improves the accuracy and robustness of clock synchronization by modeling clock drift as a Gaussian process and using negative log-likelihood optimization for neural network training. This method can autonomously maintain high-precision time synchronization even in environments with unstable or completely absent GNSS signals, effectively overcoming the excessive dependence on signal quality inherent in traditional methods. It is suitable for high-precision timing scenarios such as satellite navigation and autonomous driving, ensuring the continuity and reliability of clock synchronization. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 The flowchart illustrates a clock drift modeling method based on statistical likelihood optimization, as provided in an embodiment of the present invention.

[0046] Figure 2 This is a schematic diagram of a clock drift modeling system based on statistical likelihood optimization, provided as an embodiment of the present invention.

[0047] Figure 3 The data processing flowchart of a clock drift modeling system based on statistical likelihood optimization provided in this embodiment of the invention shows the entire process from raw clock drift data to final log-likelihood loss calculation, including data preprocessing, clock rate calculation, and Gaussian modeling. Detailed Implementation

[0048] 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 with reference to the accompanying drawings. 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.

[0049] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0050] Clock speed refers to the rate of change of clock bias over time, i.e., the first derivative of clock bias. It describes the deviation of the local clock frequency from the reference clock frequency.

[0051] Clock drift refers to the rate of change of clock speed over time, i.e., the second derivative of clock error. It describes the stability of the clock frequency itself and is a key intrinsic factor leading to the accumulation or variation of clock error. The core of this document is to model and optimize this aspect.

[0052] like Figure 1 and Figure 3 As shown in the figure, a clock drift modeling method based on statistical likelihood optimization according to an embodiment of this application includes the following steps:

[0053] S1. Obtain clock difference data from the clock synchronization system, obtain clock difference information for each time step, and standardize the data.

[0054] S2. Calculate the clock speed based on the clock difference data. And Zhong Piao ;

[0055] S3, the clock will drift The process is modeled as a Gaussian stochastic process with a mean of zero and a variance of . ;

[0056] S4. Construct a neural network model capable of predicting clock errors. The neural network model adopts a multilayer perceptron structure, with historical clock error data as input and predicted clock error sequence as output. The model parameters are then initialized.

[0057] S5. Calculate the negative log-likelihood of clock drift and use the negative log-likelihood as the loss function of the model. By optimizing the loss function during the training process, the parameters of the clock drift model are dynamically adjusted.

[0058] S6. Using the pre-trained clock drift model, generate and send correction commands to the atomic clock based on the predicted clock drift.

[0059] The data standardization process described in step S1 specifically involves performing Z-score standardization on the clock difference data using the formula...

[0060]

[0061] Data for each clock difference Convert to standardized data ,in, The mean of the clock difference data. represents the standard deviation of the clock error data.

[0062] The calculation of clock speed in step S2 And Zhong Piao Specifically, this is achieved through the following formula:

[0063]

[0064]

[0065] In step S3, the Gaussian process is represented as:

[0066]

[0067] In step S5, the loss function Defined as:

[0068]

[0069] in, Output the clock drift value of the sequence for the model.

[0070] In step S5, the dynamic adjustment of the clock drift model parameters includes the following sub-steps:

[0071] S51. Employ a gradient descent-based optimization algorithm to minimize the loss function;

[0072] S52. Adjusting hyperparameters during the optimization process;

[0073] S53. The training process is dynamically adjusted based on the clock synchronization accuracy of the validation set.

[0074] The following is a preferred embodiment of this application:

[0075] Example 1

[0076] S1. Obtain clock difference data from the clock synchronization system to obtain effective clock difference drift information for each time step, and standardize the data to reduce the impact of dimensional differences on optimization.

[0077] Data standardization includes:

[0078] Acquire valid clock difference data in the clock synchronization system, record the clock difference drift information at each time step, and denote it as... , ,… The clock difference data is Z-score standardized, that is, each clock difference data point is standardized using the following formula. Convert to standardized data :

[0079]

[0080] in, For standardized clock bias data, The mean of the clock difference data. The standard deviation of the clock bias data is used to eliminate dimensional differences and scale effects, making the subsequent optimization process more stable.

[0081] S2. Calculate the clock speed based on the aforementioned clock difference data. And Zhong Piao ;

[0082] S3, the clock will drift Model it as a stochastic process that conforms to Gaussian process, as shown in formula (1).

[0083]

[0084] Its mean is zero and its variance is This process model exhibits strong fitting capabilities, particularly in modeling time series data, capturing the nonlinearity and randomness of clock bias variations. This is achieved by adjusting the variance parameters of the Gaussian process. Optimizations enable precise capture of the dynamic characteristics of clock drift and maintain high robustness even when GNSS signals are missing or noise is high.

[0085] S4. Construct a neural network model capable of predicting clock errors. This model employs a multilayer perceptron structure, taking historical clock error data as input and outputting a predicted clock error sequence over a future period. Initialize the model parameters.

[0086] S5. Calculate the negative log-likelihood of clock drift and use it as the loss function of the model. This loss function... The definition is as follows:

[0087]

[0088] This function outputs the clock drift value for the model's sequence. It reduces prediction error by maximizing the negative log-likelihood and minimizing the clock drift error. Through loss function optimization during training, the clock parameters in the system are dynamically adjusted, ultimately improving the system's clock synchronization accuracy to the ideal value.

[0089] The above-mentioned dynamic adjustment of clock parameters in the system to improve the system's clock synchronization accuracy to the ideal value includes the following steps:

[0090] S51. A gradient descent-based optimization algorithm is used to minimize the loss function. By calculating the gradient of the loss function and adjusting system parameters, the clock synchronization accuracy is gradually improved.

[0091] S52. During the optimization process, the hyperparameters should be adjusted according to the specific application scenario to ensure that the model can converge quickly and avoid overfitting.

[0092] S53. By dynamically adjusting the clock synchronization accuracy of the validation set, the training process is ensured to maintain high synchronization accuracy when GNSS signals are lost.

[0093] S6. The system utilizes a pre-trained clock drift model to continuously send correction commands to the atomic clock based on the predicted clock drift, thereby maintaining high-precision clock tracking and time synchronization without external GNSS signal input.

[0094] like Figure 2 As shown, this application also provides a clock drift modeling system based on statistical likelihood optimization, used to implement the aforementioned clock drift modeling method based on statistical likelihood optimization, including:

[0095] The data acquisition and standardization module is used to acquire clock difference data from the clock synchronization system, obtain clock difference information for each time step, and perform standardization processing on the data.

[0096] The clock speed and clock drift calculation module is used to calculate the clock speed based on the clock difference data. And Zhong Piao ;

[0097] Gaussian process modeling module, used to model clock drift The process is modeled as a Gaussian stochastic process with a mean of zero and a variance of . ;

[0098] The neural network module, employing a multilayer perceptron structure, is used to receive historical clock difference data and output predicted clock difference sequences.

[0099] The training optimization module is used to calculate the negative log-likelihood of clock drift and use it as the loss function. And by optimizing the loss function, the clock drift model parameters are dynamically adjusted;

[0100] The calibration command generation module is used to generate and send calibration commands to the atomic clock based on the predicted clock drift using a pre-trained clock drift model.

[0101] The data acquisition and standardization module is specifically used for:

[0102] The clock error data is Z-score standardized using the following formula:

[0103]

[0104] Data for each clock difference Convert to standardized data ,in, The mean of the clock difference data. represents the standard deviation of the clock error data.

[0105] This application also provides an electronic device for implementing the above method, comprising:

[0106] At least one processor; and

[0107] A memory that is communicatively connected to the at least one processor;

[0108] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method.

[0109] The electronic device may include one or more processors, memory, and transmission devices, and the terminal may also include input / output devices.

[0110] Those skilled in the art will understand that the terminal can be a smartphone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile internet device (MID), a PAD, and other terminal devices.

[0111] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. This program can be stored in a computer-readable storage medium, which may include: a flash drive, read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk, etc. The storage medium stores a computer program that, when executed by a processor, implements the steps of the methods described above.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A clock drift modeling method based on statistical likelihood optimization, characterized in that, Includes the following steps: S1. Obtain clock difference data from the clock synchronization system, obtain clock difference information for each time step, and standardize the data. S2. Calculate the clock speed based on the clock difference data. And Zhong Piao ; S3, the clock will drift The model is a Gaussian-compliant stochastic process with a mean of zero and a variance of . ; S4. Construct a neural network model capable of predicting clock errors. The neural network model adopts a multilayer perceptron structure, with historical clock error data as input and predicted clock error sequence as output. The model parameters are then initialized. S5. Calculate the negative log-likelihood of clock drift and use the negative log-likelihood as the loss function of the model. The clock drift model parameters are dynamically adjusted through loss function optimization during training; the loss function mentioned in step S5 Defined as: in, Output the clock drift values ​​of the sequence for the model; S6. Using the pre-trained clock drift model, generate and send correction commands to the atomic clock based on the predicted clock drift.

2. The method according to claim 1, characterized in that, The data standardization process described in step S1 specifically involves performing Z-score standardization on the clock difference data using the formula... Data for each clock difference Convert to standardized data ,in, The mean of the clock difference data. represents the standard deviation of the clock error data.

3. The method according to claim 1, characterized in that, The calculation of clock speed in step S2 And Zhong Piao Specifically, this is achieved through the following formula: 。 4. The method according to claim 1, characterized in that, The Gaussian process described in step S3 is represented as follows: 。 5. The method according to claim 1, characterized in that, The dynamic adjustment of clock drift model parameters in step S5 includes the following sub-steps: S51. Employ a gradient descent-based optimization algorithm to minimize the loss function; S52. Adjusting hyperparameters during the optimization process; S53. The training process is dynamically adjusted based on the clock synchronization accuracy of the validation set.

6. A clock drift modeling system based on statistical likelihood optimization, characterized in that, include: The data acquisition and standardization module is used to acquire clock difference data from the clock synchronization system, obtain clock difference information for each time step, and perform standardization processing on the data. The clock speed and clock drift calculation module is used to calculate the clock speed based on the clock difference data. And Zhong Piao ; Gaussian process modeling module, used to model clock drift The model is a Gaussian-compliant stochastic process with a mean of zero and a variance of . ; The neural network module, employing a multilayer perceptron structure, is used to receive historical clock difference data and output predicted clock difference sequences. The training optimization module is used to calculate the negative log-likelihood of clock drift and use it as the loss function. The clock drift model parameters are dynamically adjusted by optimizing the loss function; the loss function Defined as: in, Output the clock drift values ​​of the sequence for the model; The calibration command generation module is used to generate and send calibration commands to the atomic clock based on the predicted clock drift using a pre-trained clock drift model.

7. The system according to claim 6, characterized in that, The data acquisition and standardization module is specifically used for: The clock error data is Z-score standardized using formula (3). Data for each clock difference Convert to standardized data ,in, The mean of the clock difference data. represents the standard deviation of the clock error data.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images