A fiber optic gyroscope temperature compensation method and system

Abstract

The present application discloses a method and system for temperature compensation of an optical fiber gyroscope, and relates to the field of temperature compensation. The method comprises: introducing chaotic disturbance and Gaussian variation mechanism into a sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm; optimizing a BP neural network using the improved sand cat swarm optimization algorithm to obtain a BP neural network optimization model; constructing a temperature error model based on the BP neural network optimization model; obtaining the temperature, temperature change rate and temperature-varying zero bias measurement value of an optical fiber gyroscope under variable temperature conditions; inputting the temperature and temperature change rate of the optical fiber gyroscope into a temperature error model to obtain a temperature error prediction value; and obtaining a temperature-compensated optical fiber gyroscope zero bias based on the temperature-varying zero bias measurement value and the temperature error prediction value. The present application can improve the accuracy of temperature compensation of an optical fiber gyroscope.

Description

Technical Field

[0001] The present application relates to the field of temperature compensation, and in particular to a fiber optic gyroscope temperature compensation method and system. Background Art

[0002] Fiber Optic Gyroscope (FOG) is an angular velocity sensor based on the Sagnac effect. It has many advantages such as high reliability, long service life, small size, light weight, zero start-up and no mechanical moving parts. It has been widely used in sea, land, air and other fields. Since the fiber optic gyroscope senses angular velocity based on the principle of light interference, the optical path and circuit parts are more sensitive to external temperature changes. The temperature drift of the fiber optic gyroscope caused by temperature has become a key issue limiting its measurement accuracy. Especially in high-precision application scenarios, temperature changes will significantly affect the performance parameters of the fiber optic gyroscope, thereby causing drift problems and limiting its use in complex environments. The polynomial compensation model used in the traditional fiber optic gyroscope temperature error is relatively simple and has poor adaptability, and cannot accurately achieve the compensation of the fiber optic gyroscope temperature. Summary of the invention

[0003] The purpose of the present application is to provide a fiber optic gyroscope temperature compensation method and system, which can improve the accuracy of fiber optic gyroscope temperature compensation.

[0004] To achieve the above objectives, this application provides the following solutions:

[0005] In a first aspect, the present application provides a fiber optic gyroscope temperature compensation method, comprising:

[0006] Chaotic perturbation and Gaussian mutation mechanism are introduced into the sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm.

[0007] The BP neural network is optimized by using the improved sand cat swarm optimization algorithm to obtain the BP neural network optimization model.

[0008] Constructing a temperature error model based on the BP neural network optimization model;

[0009] Under variable temperature conditions, obtain the temperature, temperature change rate and variable temperature zero bias measurement values ​​of the fiber optic gyroscope;

[0010] Inputting the temperature and temperature change rate of the fiber optic gyroscope into the temperature error model to obtain a temperature error prediction value;

[0011] The temperature-compensated fiber optic gyroscope zero bias is obtained based on the variable-temperature zero bias measurement value and the temperature error prediction value.

[0012] Optionally, an improved sand cat swarm optimization algorithm is used to optimize the BP neural network to obtain a BP neural network optimization model, including:

[0013] Initializing the structure of the BP neural network to obtain a parameter combination; the parameter combination includes: a weight matrix from the input layer to the hidden layer, a bias vector of the hidden layer, a weight matrix from the hidden layer to the output layer, and a bias vector of the output layer;

[0014] Initializing the position of the sand cat group based on the parameter combination to obtain an original definition domain, and determining individual fitness values ​​and average fitness values ​​of all individuals in the original definition domain;

[0015] Performing chaotic perturbations on individuals whose individual fitness values ​​are greater than or equal to the average fitness value, performing Gaussian mutations on individuals whose individual fitness values ​​are less than the average fitness value, and performing individual iterative updates until a maximum number of iterations or a convergence condition is reached, obtaining an individual with the best fitness, and using the individual with the best fitness to update the individual positions in the sand cat group to obtain a global optimal solution;

[0016] The global optimal solution is used as the optimal structural parameter of the BP neural network to obtain an optimized BP neural network, and the optimized BP neural network is used as the BP neural network optimization model.

[0017] Optionally, a new individual obtained by performing chaotic perturbation on the individuals whose individual fitness values ​​are greater than or equal to the average fitness value is expressed as:

[0018] ;

[0019] In the formula, represents the original domain, and are the upper and lower bounds of the original domain, represents the chaotic mapping function, Represents a new individual, Indicates an individual.

[0020] Optionally, a new individual obtained by performing Gaussian mutation on the individual whose individual fitness value is less than the average fitness value is expressed as:

[0021] ;

[0022] In the formula, Represents a new individual, represents a random number generated by Gaussian distribution, represents the optimal individual.

[0023] Optionally, initializing the position of the sand cat group based on the parameter combination to obtain an original definition domain includes:

[0024] The chaotic mapping method is adopted to initialize the positions of the sand cat group based on the parameter combination to obtain the original definition domain.

[0025] Optionally, building a temperature error model based on the BP neural network optimization model includes:

[0026] Obtain training sample sets and test sample sets;

[0027] The training sample set is used to train the BP neural network optimization model, and during the training process, the test sample set is used to test the trained BP neural network optimization model until the prediction error of the trained BP neural network optimization model reaches the set target, and the trained BP neural network optimization model is used as the temperature error model.

[0028] Optionally, the prediction error of the trained BP neural network optimization model is expressed as:

[0029] ;

[0030] In the formula, represents the prediction error, represents the number of samples, Indicates The temperature error prediction value corresponding to the sample is: Indicates The actual temperature error measurement value corresponding to the samples.

[0031] Optionally, under variable temperature conditions, obtaining the temperature, temperature change rate, and variable temperature zero bias measurement value of the fiber optic gyroscope includes:

[0032] Select multiple fiber optic gyroscopes as test samples, and perform zero bias output tests under variable temperature and constant temperature conditions to obtain the temperature, variable temperature zero bias output, and constant temperature zero bias output; wherein variable temperature means that the temperature increases or decreases according to a set value, and constant temperature means that the temperature is kept at a set value;

[0033] Obtaining the variable temperature zero bias measurement value based on the variable temperature zero bias output and the constant temperature zero bias output;

[0034] The temperature change rate is determined based on the temperature.

[0035] Optionally, the variable temperature zero bias measurement value is expressed as:

[0036] ;

[0037] In the formula, is the variable temperature zero bias measurement value, is the sampling number, is the scaling factor, is the Earth's rotation angular velocity, is the geographical latitude of the test site, Indicates The actual temperature error measurement value corresponding to the samples.

[0038] In a second aspect, the present application provides a fiber optic gyroscope temperature compensation system for implementing the fiber optic gyroscope temperature compensation method provided above; the fiber optic gyroscope temperature compensation system comprises:

[0039] Algorithm optimization module, used to introduce chaotic perturbation and Gaussian mutation mechanism into the sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm;

[0040] The network optimization module is used to optimize the BP neural network by using the improved sand cat swarm optimization algorithm to obtain the BP neural network optimization model;

[0041] A model building module, used for building a temperature error model based on the BP neural network optimization model;

[0042] A data acquisition module is used to obtain the temperature, temperature change rate and temperature-varying zero bias measurement value of the fiber optic gyroscope under variable temperature conditions;

[0043] An error prediction module, used for inputting the temperature and temperature change rate of the fiber optic gyroscope into the temperature error model to obtain a temperature error prediction value;

[0044] The temperature compensation module is used to obtain the temperature-compensated fiber optic gyroscope zero bias based on the variable temperature zero bias measurement value and the temperature error prediction value.

[0045] According to the specific embodiments provided in this application, this application has the following technical effects:

[0046] The present application provides a method and system for temperature compensation of a fiber optic gyroscope. In the process of improving the sand cat swarm optimization algorithm, by introducing chaotic disturbance, a population distribution with higher diversity and ergodicity can be generated in the process of optimizing the BP neural network using the improved sand cat swarm optimization algorithm. By introducing a Gaussian mutation mechanism, the global search capability of the algorithm can be improved and premature convergence of the network can be avoided in the process of optimizing the BP neural network using the improved sand cat swarm optimization algorithm. Based on this, by establishing a black box model (i.e., a temperature error model) of the temperature characteristics of the fiber optic gyroscope based on the BP neural network optimization model (i.e., the optimized BP neural network) to achieve temperature error prediction, the accuracy of temperature compensation of the fiber optic gyroscope can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 A schematic flow chart of a fiber optic gyroscope temperature compensation method provided in one embodiment of the present application;

[0049] Figure 2 A schematic diagram of the topological structure of a gyro temperature error compensation model based on a BP neural network provided in an embodiment of the present application;

[0050] Figure 3 A flow chart of a sand cat colony optimization algorithm provided in one embodiment of the present application;

[0051] Figure 4 A schematic diagram of chaos values ​​and dimensions of a SPM chaos map provided in an embodiment of the present application;

[0052] Figure 5 A schematic diagram of chaos values ​​and frequencies of SPM chaos mapping provided in an embodiment of the present application;

[0053] Figure 6 A schematic diagram of an ISCSO process provided in an embodiment of the present application;

[0054] Figure 7 A schematic diagram of a process of optimizing a BP neural network using ISCSO according to an embodiment of the present application;

[0055] Figure 8 A schematic diagram of pulse and zero bias corresponding data provided in an embodiment of the present application;

[0056] Fig. 9 A schematic diagram of gyroscope output under variable temperature conditions provided by an embodiment of the present application;

[0057] Fig.10 A schematic diagram of gyroscope output under constant temperature conditions provided in an embodiment of the present application;

[0058] Fig.11 A schematic diagram of a physical test setup provided in an embodiment of the present application;

[0059] Fig.12 A schematic diagram showing the comparison between the predicted value and the measured value of the temperature error model provided in one embodiment of the present application;

[0060] Fig.13 A schematic diagram of gyro bias comparison after temperature compensation error provided in an embodiment of the present application. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0062] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0063] In an exemplary embodiment, the present application provides a method for temperature compensation of a fiber optic gyroscope, which is executed by a computer device, specifically, a computer device such as a terminal or a server, or a terminal and a server. In the embodiment of the present application, the method is applied to a server as an example for explanation. Figure 1 As shown, the method includes:

[0064] Step 100: Chaotic perturbation and Gaussian mutation mechanism are introduced into the sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm.

[0065] Step 101: Use the improved sand cat swarm optimization algorithm to optimize the BP neural network and obtain the BP neural network optimization model. The structure of the BP neural network is as follows: Figure 2 shown. and Represents the connection weight threshold matrix between layers, is the hidden layer neuron.

[0066] The number of hidden layer nodes will affect the prediction accuracy of BP neural network. The formula for determining is:

[0067] .

[0068] In the formula, is the number of input layer nodes, for example . is the number of output layer nodes, for example . An integer from 1 to 10.

[0069] For example, if the number of input nodes is 2, the number of hidden layer nodes is 8, and the number of output nodes is 1, then Figure 2 The topology shown is used to establish the BP neural network finally used.

[0070] Step 102: Construct a temperature error model based on the BP neural network optimization model.

[0071] Step 103: Under variable temperature conditions, obtain the temperature, temperature change rate, and variable temperature zero bias measurement value of the fiber optic gyroscope.

[0072] Step 104: Input the temperature and temperature change rate of the fiber optic gyroscope into a temperature error model to obtain a temperature error prediction value.

[0073] Step 105: Obtain the temperature-compensated FOG zero bias based on the variable temperature zero bias measurement value and the temperature error prediction value.

[0074] Furthermore, the Sand Cat Swarm Optimization (SCSO) algorithm simulates the two main behaviors of sand cats in foraging: searching for prey and attacking prey. This strategy of imitating the natural behavior of sand cats makes the algorithm more efficient in the global search process, helps to avoid falling into the local optimal solution, and compared with the traditional BP neural network, the BP neural network optimized by SCSO will converge to a satisfactory solution faster during the iteration process, which can reduce training time and improve learning efficiency.

[0075] For example, in In the dimensional optimization problem, the sand cat is a The dimension array representing the solution to the problem is defined as:

[0076] .

[0077] In the formula, For the Sand cat, each variable value is a floating point number, and each Must be between the upper and lower bounds.

[0078] like Figure 3 As shown below, the main steps of the sand cat swarm optimization algorithm are:

[0079] (1) Randomly initialize the positions of the sand cat colony.

[0080] Depending on the size of the problem , , create an initialization matrix using the sand cat population, with:

[0081] .

[0082] In the formula, is the population matrix of the sand cat colony. For the A sand cat. For the The first dimensional coordinates, The fitness cost of each sand cat is obtained by evaluating the defined fitness function. The fitness function of the sand cat population is :

[0083] .

[0084] In the formula, For the The value obtained by calculating the fitness function of a sand cat, Represents the use of fitness function Population matrix of sand cat colonies Solve for fitness, To use the fitness function For The location coordinates of the sand cat ( ) to solve for fitness.

[0085] (2) Calculate the location of the sand cat colony and update the global optimal location information.

[0086] (3) Conversion factor .

[0087] After the sand cat group is initialized, it begins to search for or attack prey, that is, the exploration or exploitation stage. The main parameter controlling the conversion between the exploration and exploitation stages is the conversion coefficient , conversion factor The calculation formula is as follows:

[0088] .

[0089] In the formula, Represents a random number between 0 and 1. is the sensitivity coefficient of the sand cat colony, used to guide the conversion coefficient To achieve inter-stage transfer control, the calculation formula is:

[0090] .

[0091] In the formula, is the current iteration number, is the maximum number of iterations, is the hearing coefficient of the sand cat group, The value is 2.

[0092] (3) Searching for prey.

[0093] When the conversion factor , the sand cats enter the exploration phase. At this time, the first Sand cats according to current location , best candidate position And the sensitivity range of this cat , update the position according to the following formula to find other possible best prey positions.

[0094] .

[0095] .

[0096] In the formula, It means the number of iterations is The best candidate position when It means the number of iterations is The position at that time. It means the number of iterations is The position at that time.

[0097] (4) Attacking prey.

[0098] When the conversion factor , SandCat has entered the development phase, and in the current iteration , the first of the sand cat groups The sand cats are based on the best candidate locations With current location Generate a random location , and its calculation formula is:

[0099] .

[0100] Assuming that the sensitivity range of the sand cat is a circle, a random angle is randomly selected for each sand cat using the roulette wheel selection algorithm. , each cat in the group can move in different circular directions in the search space. The cat updates its position according to the following formula to attack the prey.

[0101] .

[0102] The Sand Cat Swarm Optimization Algorithm improves the global search capability of the neural network by balancing the exploration phase and the development phase, searching in multiple dimensions in space, and dynamically adjusting the search strategy. It reduces training errors and improves the network's ability to fit complex problems by optimizing weights and biases, adjusting activation function parameters, and adjusting network parameters such as learning rate, making network training more stable and efficient.

[0103] (5) Determine whether to output the optimal threshold.

[0104] Determine whether the number of iterations has reached the maximum number of iterations. If it has reached the maximum number of iterations, output the optimal threshold. If it has not reached the maximum number of iterations, calculate the position of the sand cat group and update the global optimal position information.

[0105] In SCSO, no matter the sand cat group completes the position update through the exploration phase or the development phase, there may still be some inferior solutions. Since the algorithm does not design a population mutation mechanism, these inferior solutions may cause the algorithm to fall into a local optimum and it is difficult to jump out. Therefore, in order to improve the global search ability of the algorithm and avoid premature convergence, it is necessary to introduce a mutation mechanism in SCSO to maintain the diversity of the sand cat group and ensure that better solutions can be explored and optimized in a complex search space.

[0106] The distribution of population initialization directly affects the convergence speed and solution accuracy of SCSO. By using chaotic sequences instead of traditional random initialization methods, a population distribution with higher diversity and ergodicity can be generated in the initial stage of SCSO, thereby expanding the coverage of SCSO in the search space. By applying chaotic perturbations to the optimal individuals, it can effectively help SCSO to jump out of the local optimum when the local search stagnates and continue the global search. This strategy not only improves the quality of the solution, but also significantly improves the efficiency and robustness of SCSO in solving complex problems. Based on this, in this embodiment, the present application introduces chaotic perturbations and Gaussian mutation mechanisms into the sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm. Among them, the process of improving the sand cat swarm optimization algorithm includes:

[0107] (1) Replace random initialization with chaotic sequence.

[0108] SCSO initializes the population through a random function. This method will result in insufficient diversity of individuals in the sand cat group and uneven distribution in the search space, thus affecting the global search capability. Chaotic mapping (Statistical Parametric Mapping, SPM) has a very uniform distribution function and good pseudo-random characteristics. Using SPM mapping to generate chaotic variables can make the population resources more evenly distributed in the search space, effectively avoid repeated initialization, maintain population diversity, and effectively prevent SCSO from falling into local optimality, expand the search space, and improve the global search capability. The iterative method of chaotic mapping is expressed as:

[0109] .

[0110] In the formula, , , is a random number between 0 and 1. It means the number of iterations is The variable value at It means the number of iterations is The variable value at that time.

[0111] when , When , the SPM mapping chaos value and frequency distribution are as follows Figure 4 and Figure 5 As shown. Figure 4 and Figure 5 It can be seen that the sequence generated by SPM has better ergodicity and randomness, is evenly distributed, and has excellent results. Therefore, this application chooses SPM to initialize SCSO. Due to its nonlinear and ergodic characteristics, the chaotic sequence can effectively avoid the algorithm from falling into the local optimum and expand the search space.

[0112] (2) SPM chaotic perturbation and Gaussian variation mixed perturbation strategy.

[0113] In order to avoid the generation of inferior solutions and ensure the diversity of the sand cat population, chaotic perturbation and Gaussian mutation mechanisms are introduced into SCSO to perturb the position of the currently found optimal solution, that is, to re-search the optimal individuals of the sand cat population on a global scale to ensure that a better solution can be found. When the new position of the sand cat generated by chaotic perturbation or Gaussian mutation is better than the original position, the current optimal position is replaced. Otherwise, the original position is retained. The sand cat in the current optimal position (i.e. the optimal individual) is replaced. ) introduces chaotic perturbation, perturbs the position of the optimal individual through chaotic mapping, and restores it to the original definition domain to obtain a new position ,have:

[0114] .

[0115] .

[0116] The current optimal individual From its original domain Mapping to a standardized interval Get individual This approach helps unify the scale of each variable, so that the entire search space can be explored more effectively when applying chaotic sequences for perturbations. Calling the SPM chaotic mapping function A chaotic sequence is generated to perturb the current optimal individual. The result after the chaotic sequence perturbation is restored to the original definition domain, ensuring that the new position is still within the specified range. Due to its nonlinear and ergodic characteristics, the chaotic sequence can effectively prevent the algorithm from falling into the local optimum and expand the search space.

[0117] Gaussian mutation of the sand cat population is to generate relatively wide range of random numbers through Gaussian distribution The positions of the sand cat colony are disturbed. The Gaussian distribution and the position update process after disturbance are expressed as:

[0118] .

[0119] .

[0120] In the formula, is a random number between 0 and 1. is the standard deviation of the Gaussian distribution.

[0121] Chaotic perturbation makes a slight perturbation to the optimal individual to jump out of the local optimal solution. Gaussian mutation introduces more randomness, and through random mutation, it can increase its exploration range in the search space. Combining the advantages of chaotic perturbation and Gaussian mutation, a mixed perturbation strategy of SPM and Gaussian mutation is formulated to mutate SCSO to find the current optimal position perturbation. Calculate the average fitness of all individuals in the population , this value is used to distinguish which individuals need chaotic perturbation and which individuals need Gaussian mutation. Chaotic perturbation is suitable for better individuals, and the purpose is to jump out of the local optimum through perturbation. Gaussian mutation is a random number that follows a normal distribution. to change the position of the individual. If the fitness of the current optimal sand cat individual Greater than or equal to the average fitness of the population , then the individual is optimized using chaotic perturbation. Otherwise, if the individual's fitness Below average , then Gaussian mutation is performed on it to introduce randomness to increase the diversity of individuals and ensure the comprehensiveness of the search process.

[0122] Whether it is the result of chaotic perturbation or Gaussian mutation, it must be ensured that it is within the definition domain. Calculate the new fitness , if the fitness is better than the current individual's fitness , then update the position and fitness of the individual. Among all the updated individuals, find the individual with the best fitness and update the global optimal solution.

[0123] The implementation process of the Improved Sand Cat Swarm Optimization (ISCSO) algorithm after replacing random initialization with chaotic sequence and optimizing the SPM and Gaussian mutation mixed perturbation strategy is as follows: Figure 6 shown.

[0124] By combining two different strategies, chaotic perturbation and Gaussian mutation, the algorithm maintains its global search capability while also enhancing its local development capability. Chaotic perturbation is suitable for individuals with better performance, and it escapes the local optimal trap by slightly perturbing the optimal individual. Gaussian mutation introduces more randomness, especially for individuals with poor performance. It increases their exploration range in the search space through random mutation, thereby improving the diversity of the population. This combined strategy enables the algorithm to avoid premature convergence while maintaining its global search capability, thus increasing the probability of finding the global optimal solution of the algorithm.

[0125] Based on the above description, the implementation process of step 101 of the present application includes:

[0126] Step 1: Initialize the structure of the BP neural network and set relevant parameters to obtain a parameter combination. The parameter combination includes: the weight matrix from the input layer to the hidden layer, the bias vector of the hidden layer, the weight matrix from the hidden layer to the output layer, and the bias vector of the output layer.

[0127] In this step, the number of neurons in the input layer, hidden layer and output layer of the BP neural network is set, and the parameters of the sand cat swarm optimization algorithm are defined, including the sand cat hearing coefficient , population size , maximum number of iterations , the upper and lower bounds of the search space and , parameter dimension , and the solution space (i.e. the original domain) , this solution space includes all weights and thresholds in the BP neural network, which represents the solution space of the optimization problem: In the formula, is the weight matrix from the input layer to the hidden layer, is the bias vector of the hidden layer, is the weight matrix from the hidden layer to the output layer, is the bias vector of the output layer.

[0128] Hidden layer output matrix And the output layer output matrix The calculation formula is: .

[0129] .

[0130] In the formula, For the logsig function, is a pureline function, represents the solution space vector, Parameter vector.

[0131] The total number of these parameters is the dimension of the search space of the Sand Cat Swarm Optimization algorithm. For example, if the input layer of the network has nodes, and the hidden layer has nodes, and the output layer has nodes, the total number of parameters for: .

[0132] Step 2: Using the chaotic mapping method, the position of the sand cat group is initialized based on the parameter combination to obtain the original domain, and the individual fitness value and the average fitness value of all individuals in the original domain are determined. That is, the position of the sand cat group is initialized by SPM chaotic mapping. Among them, the position of each sand cat in the population is randomly generated, and each position corresponds to a weight and threshold combination of the BP neural network. For each individual in the population, the fitness function Calculate its fitness value, which reflects the performance of the BP neural network corresponding to the individual on the training set, that is, the size of the output error of the BP neural network.

[0133] Step 3: Perform chaotic perturbations on individuals whose individual fitness values ​​are greater than or equal to the average fitness value, perform Gaussian mutations on individuals whose individual fitness values ​​are less than the average fitness value, and perform individual iterative updates until the maximum number of iterations or convergence conditions (such as the prediction error reaches the set target) are reached, and the individual with the best fitness is obtained. The individual position in the sand cat group is updated with the individual with the best fitness to obtain the global optimal solution.

[0134] Step 4: Use the global optimal solution as the optimal structural parameter of the BP neural network to obtain the optimized BP neural network, and use the optimized BP neural network as the BP neural network optimization model. According to the search mechanism of ISCSO, the sand cat group will continuously update its position to gradually approach the global optimal solution in the search space. In this process, the control parameters (i.e., the sensitivity coefficient of the sand cat group) are ) is used to guide the sensitivity range of the sand cat group, ensuring that the algorithm conducts extensive global search in the early stage and precise local development in the later stage. ISCSO deconstructs each sand cat position (i.e., the parameter combination of the BP neural network) into specific weights and thresholds, and then assigns these parameters to the BP neural network to obtain the BP neural network optimization model.

[0135] Based on the above description, in actual application, the specific operation process of ISCSO optimizing BP neural network is as follows: Figure 7 shown.

[0136] Furthermore, the implementation process of step 102 provided above in the present application can be described as follows:

[0137] 1) Obtain training sample set and test sample set.

[0138] 2) Use the training sample set to train the BP neural network optimization model, and use the test sample set to test the trained BP neural network optimization model during the training process, until the prediction error of the trained BP neural network optimization model reaches the set target, and then use the trained BP neural network optimization model as the temperature error model. For example, the temperature and temperature change rate of the fiber optic gyroscope under variable temperature conditions are input, and the gyroscope zero bias under the variable temperature conditions at the corresponding moment is output. A data sample set is constructed to train and test the neural network model. After the training and testing, the temperature error model is obtained. The variable temperature zero bias output is subtracted from the predicted value of the temperature error model to obtain the gyroscope zero bias output after temperature compensation.

[0139] Furthermore, the prediction error of the trained BP neural network optimization model is expressed as: In the formula, represents the prediction error, represents the number of samples, Indicates The temperature error prediction value corresponding to the sample is: Indicates The actual temperature error measurement value corresponding to the samples.

[0140] The fitness value is a key indicator in the sand cat swarm optimization algorithm, which is used to evaluate the performance of each solution (i.e., each sand cat) in solving a specific problem. Specifically, the fitness value is a value calculated by the fitness function, which represents the quality of the current solution. In the sand cat swarm optimization algorithm, each sand cat represents a potential solution, and they gradually find the optimal solution by iteratively updating their positions. The higher the fitness value, the closer the solution is to the optimal solution of the problem. Therefore, the fitness value is an important basis for guiding the algorithm's search direction and optimization process. The position of the sand cat refers to the specific coordinates or state of each sand cat in the solution space, which determines what solution the sand cat currently represents. During the algorithm iteration process, the sand cat will update its position based on the fitness value and the position information of other sand cats to explore the solution space and find a better solution.

[0141] In summary, the fitness value and the position of the sand cat play different roles in the sand cat swarm optimization algorithm. The fitness value is used to evaluate the quality of the solution, while the position of the sand cat determines the specific state of the current solution.

[0142] Furthermore, in one embodiment, the performance of the improved sand cat swarm optimization algorithm is tested. The Sand Cat Swarm Optimization Algorithm (SCSO), Dung Beetle Optimizer (DBO), Whale Optimization Algorithm (WOA), Zebra Optimization Algorithm (ZOA) and the improved sand cat swarm algorithm (ISCSO) are selected for comparison. The basic parameters of each algorithm are set as follows: the population size is set to 30, the maximum number of iterations is Seven test functions are selected to test the performance of each algorithm. The test function information is shown in Table 1. In Table 1, F1 is a low-dimensional unimodal test function, F2-F4 are low-dimensional multimodal test functions, F5 and F6 are high-dimensional unimodal test functions, and F7 is a high-dimensional multimodal test function.

[0143] Table 1 Test function information table

[0144]

[0145] In order to reduce the impact of randomness on the experiment, each algorithm was simulated 100 times, and the mean value Mean, standard deviation Std, convergence ratio Cr, and average termination iteration number Iter were recorded to evaluate the optimization ability of each algorithm. The experimental results are shown in Table 2.

[0146] Table 2 Test results of performance parameters of each algorithm

[0147]

[0148] (1) Horizontally compare the performance of each algorithm under the same test function.

[0149] Using the function F1 test, the ZOA algorithm has the best Std value on the Rosenbrock function (2.7383e-04), but ISCSO follows closely behind, and its performance on other performance indicators is more outstanding. ISCSO has the smallest average value of 5.2043e-04, indicating that its results are the most stable in multiple experiments. The convergence rate of ISCSO is 100%, which is superior to the ZOA algorithm and shows higher reliability. The average termination iteration number of ISCSO is 40.06, which is significantly lower than SCSO, DBO and ZOA, indicating its fast convergence ability on the Rosenbrock function.

[0150] Using the function F2 test, the Std value of ISCSO (0.0470) is slightly higher than that of DBO and WOA algorithms, but its average value is closest to the optimal value, indicating that it has better optimization ability. Although the average termination iteration number of ISCSO (457.83) is slightly higher than that of ZOA, it is slightly lower than that of other algorithms, and it still shows good convergence ability, indicating that it also has advantages in convergence speed. The Cr value performs best among all algorithms, and the convergence rate of ISCSO is 58%, which is higher than that of other algorithms. This reflects the strong adaptability of ISCSO to complex problems.

[0151] Using the function F3 test, the mean values ​​of each algorithm on the function F3 are close. ISCSO performs similarly to other algorithms, but it performs better on other indicators. The mean value of ISCSO is closest to the optimal value (0.9977), showing a high consistency of the results. Although all algorithms failed to fully converge on this function (Cr values ​​are all 0%), the stability of ISCSO makes it a reliable choice. All algorithms reached the maximum number of iterations of 500, showing the difficulty of this function, but the convergence advantage of ISCSO on other functions still makes it occupy an overall superior position.

[0152] Using the function F4 test, the ZOA algorithm performs slightly better on the Camel function, but the Mean value of ISCSO (-1.0313) is close to the Mean value of ZOA (-1.0316), showing a performance close to the global optimal. The standard deviation of ISCSO (2.7003e-04) is lower than that of other algorithms, showing significant stability on the Camel function. The convergence ratios of other algorithms such as ISCSO are all 100%, demonstrating its strong optimization ability in multidimensional space. The average termination iteration number of ISCSO is only 7.28, which is lower than that of other algorithms except the Cr value of the ZOA algorithm (6.92), indicating its fast convergence ability on the Camel function, further demonstrating its advantage as the optimal algorithm.

[0153] Using the function F5 test, the Mean value of ISCSO (5.4626e-04) is better than SCSO, ZOA, DBO and WOA, showing its strong optimization ability on high-dimensional unimodal test functions. ISCSO has the smallest standard deviation (2.8408e-04), indicating the high consistency of its results. The convergence rate of ISCSO is 100%, which is significantly better than other algorithms, further consolidating its position as the optimal algorithm. The average termination iteration number of ISCSO is 60.17, which is lower than other algorithms, showing its fast convergence ability on complex problems.

[0154] Using the benchmark function F6 test, the Mean value of ISCSO (4.5769e-04) is lower than that of other algorithms, indicating its optimal performance on this function. The standard deviation of ISCSO (2.8258e-04) is slightly higher than that of DBO, WOA, and ZOA, but it still maintains good stability. The convergence rate of ISCSO reaches 100%, which is not significantly different from other algorithms, but combined with other indicators, ISCSO's overall performance is the best. The average termination iteration number of ISCSO is 8.04, which is much lower than that of other algorithms, showing its extremely fast convergence speed on this function.

[0155] Using the function F7 test, the Mean value of ISCSO (2.4148e-04) is close to the global optimal value of 0, which is better than other algorithms, showing its strong optimization ability on multimodal functions. The standard deviation of ISCSO (2.5209e-04) is the smallest, indicating the high consistency and stability of its results. The convergence rate of ISCSO is 100%, the same as SCSO, which is better than other algorithms, showing its reliability in complex search spaces. The average termination iteration number of ISCSO is only 4.96, which is much lower than other algorithms, indicating its fast convergence performance on this function, further consolidating its position as the optimal algorithm.

[0156] (2) Vertically compare the performance of each algorithm on different test functions.

[0157] SCSO performs well on multiple functions, especially on functions F1, F4, F6 and F7, where the convergence rate reaches 100%. However, it performs poorly on more complex functions such as F2, F5 and F3, and fails to converge effectively, showing its limitations on complex problems. DBO performs relatively balanced on all test functions, but its convergence rate on functions F2, F5 and F7 is low, showing its shortcomings on complex and multimodal problems. WOA performs poorly on functions F1 and F5, with a high average termination iteration number, especially on functions F2, F5, F6 and F7, where the convergence speed is slow, indicating that its optimization efficiency is not as good as other algorithms. The convergence rate of ZOA on functions F2, F5, F6 and F7 does not reach 100%, and the average value, standard deviation and average convergence algebra are unstable. It fails to converge on function F3, indicating that it has limitations in complex search spaces. ISCSO performs well on all test functions, especially on functions F1, F2, F5, F6 and F7, where the Mean, Std and Cr values ​​are optimal or close to optimal, indicating its strong adaptability and stability in various optimization problems. In particular, the average termination iteration number of ISCSO is generally lower than that of other algorithms, showing its fast convergence ability on complex problems.

[0158] By comparing the performance of each algorithm on the seven functions in Table 2, it can be concluded that the improved sand cat swarm optimization algorithm (ISCSO) outperforms other algorithms in overall performance. ISCSO shows excellent optimization capabilities on a variety of test functions, especially on Rosenbrock, Michalewicz, Quartic, Schwefel's Problem 2.22 and Rastrigin functions.

[0159] First, ISCSO not only achieves the optimal or near-optimal Mean and Std values ​​on these functions, but also shows that it has extremely high stability in finding the global optimal solution and maintaining the consistency of the results. Especially on functions F1, F5, F6 and F7, the convergence ratio (Cr) and convergence speed (Iter) of ISCSO are significantly better than other algorithms, showing extremely strong global search capabilities and fast convergence characteristics.

[0160] Secondly, ISCSO performs particularly well on high-dimensional and complex problems. Compared with SCSO, DBO, WOA, and ZOA, ISCSO can not only quickly find the global optimal solution, but also significantly reduce the number of iterations, indicating that it is more efficient in dealing with complex search spaces. In addition, the stable performance of ISCSO on all test functions further proves its adaptability and robustness in dealing with multimodal and unimodal problems.

[0161] Finally, ISCSO introduces a variety of improved strategies such as chaotic map initialization, chaotic perturbation, and Gaussian mutation mixed perturbation strategy, which enables the algorithm to achieve a good balance between global search and local development, thereby performing well on a wide range of optimization problems. Overall, ISCSO not only significantly surpasses SCSO in various indicators, but also has a clear advantage in comparison with other optimization algorithms. Therefore, ISCSO can be regarded as an efficient, robust and adaptable optimization algorithm, which is suitable for solving a variety of complex optimization problems, making it one of the preferred tools for solving various complex optimization problems.

[0162] Furthermore, under variable temperature conditions, the process of obtaining the temperature, temperature change rate, and variable temperature zero bias measurement value of the fiber optic gyroscope may include:

[0163] 1. Select multiple fiber optic gyroscopes as test samples, and perform zero bias output tests under variable temperature and constant temperature conditions to obtain temperature, variable temperature zero bias output, and constant temperature zero bias output. Among them, variable temperature means that the temperature increases or decreases according to the set value, and constant temperature means that the temperature is kept at a set value.

[0164] 2. Obtain the variable temperature zero bias measurement value based on the variable temperature zero bias output and the constant temperature zero bias output. The collected variable temperature zero bias output and the constant temperature zero bias output of the fiber optic gyroscope are calibrated according to the relevant definitions. Based on this, the variable temperature zero bias measurement value is expressed as:

[0165] .

[0166] In the formula, is the variable temperature zero bias measurement value, is the sampling number, is the scaling factor, is the Earth's rotation angular velocity, is the geographical latitude of the test site, Indicates The actual temperature error measurement value corresponding to the samples.

[0167] The pulse is converted into gyro bias data to display data details, such as Figure 8 As shown, 2000s of continuous data in the heating experiment can be captured.

[0168] To show data details, such as Fig. 9 and Fig.10 As shown, the 14000s variable temperature zero bias output of the fiber optic gyroscope during the process of heating from -40℃ to 70℃ and the 12000s collected zero bias output in a constant temperature environment.

[0169] contrast Fig. 9 and Fig.10 It can be seen that the output of the fiber optic gyroscope is very sensitive to temperature changes. Under constant temperature conditions, the output of the gyroscope is relatively stable with a small fluctuation amplitude. Under variable temperature conditions, especially in the stage of rapid temperature increase, the gyroscope output has an obvious upward trend and significant irregular fluctuations. This shows that effective temperature compensation is crucial to improving the accuracy and stability of the fiber optic gyroscope, and reflects the necessity of establishing a temperature compensation model for the zero-bias output of the fiber optic gyroscope.

[0170] 3. Determine the temperature change rate based on the temperature.

[0171] Furthermore, the superiority of the temperature error model provided by the present application is verified through a fiber optic gyroscope temperature test.

[0172] The test object used in this embodiment is a certain high-precision fiber optic gyroscope. Four fiber optic gyroscopes produced in the same batch are selected and numbered FOG1, FOG2, FOG3 and FOG4. The zero bias stability of the fiber optic gyroscope numbered FOG1 at constant temperature is 0.0052° / h, and the zero bias stability at variable temperature is 0.0084° / h. Gyro zero bias is related to factors such as the angular velocity of the earth's rotation, the geographical latitude of the test site, the number of samplings, and the scale factor. FOG1, FOG2, FOG3, and FOG4 were subjected to temperature change tests in turn. The fiber optic gyroscopes were placed in a temperature box and kept at rest. The physical experiment process is set up as follows: Fig.11 As shown, the initial temperature of the four fiber optic gyroscopes is set to 20℃. The test process is carried out by cooling from 20℃ to -40℃ and keeping warm, heating from -40℃ to 70℃ to complete the heating process and keeping warm. After completion, the temperature is cooled from 70℃ to -40℃ and kept warm, and then heated from -40℃ to 70℃ and kept warm for a cycle. In the test, the high and low temperature insulation time of the fiber optic gyroscope is 150min, the temperature change rate is 1℃ / min, and the sampling frequency is 1Hz. The above temperature change test is repeated twice for each fiber optic gyroscope, and two groups of test data are obtained respectively, one group is used as the training sample of the neural network temperature compensation model, and the other group is used as the test sample of the temperature compensation model. The measured gyro zero bias output and the temperature and temperature change rate at the corresponding time are smoothed for 100s. The temperature and temperature change rate after 100 seconds of smoothing are used as the input of the temperature error model, and the measured value of the gyro temperature zero bias after 100 seconds of smoothing at the corresponding temperature is output. The temperature error model is constructed and trained, and the compensation effect of the temperature error model is tested using the test sample.

[0173] Taking FOG1 as an example, the effects of four models are compared: BPNN (Back Propagation Neural Network, i.e. BP Neural Network), DBO-BPNN (Back Propagation Neural Network based on Dung Beetle Optimizer), SCSO-BPNN (Back Propagation Neural Network based on Improved Sand Cat Swarm Optimization), and the temperature error model (ISCSO-BPNN) provided in this application. Among them, the comparison results of the measured zero-bias output data of FOG1 smoothed for 100s and the predicted value of the temperature error model are shown in the figure. Fig.12 The comparison results of gyro zero bias after temperature error compensation of four temperature compensation models are shown in Fig.13 shown.

[0174] It can be seen from Tables 3 to 7 that the ISCSO-BPNN provided in this application shows the best performance both in the test results of a single fiber optic gyroscope and the average compensation effect, and has important application value in the temperature compensation research of fiber optic gyroscopes.

[0175] Table 3 Comparison of zero bias stability of four FOG1 temperature compensation models after compensation

[0176]

[0177] Table 4 Comparison of zero bias stability of four FOG2 temperature compensation models after compensation

[0178]

[0179] Table 5 Comparison of zero bias stability of four FOG3 temperature compensation models after compensation

[0180]

[0181] Table 6 Comparison of zero bias stability of four FOG4 temperature compensation models after compensation

[0182]

[0183] Table 7 Comparison of zero bias stability improvement of four fiber optic gyroscopes after temperature compensation model compensation

[0184]

[0185] Based on the same inventive concept, the embodiment of the present application also provides a fiber optic gyroscope temperature compensation system for implementing the fiber optic gyroscope temperature compensation method involved above. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more fiber optic gyroscope temperature compensation system embodiments provided below can refer to the limitations of the fiber optic gyroscope temperature compensation method above, and will not be repeated here.

[0186] In an exemplary embodiment, a fiber optic gyro temperature compensation system is provided, comprising:

[0187] The algorithm optimization module is used to introduce chaotic perturbation and Gaussian mutation mechanism into the sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm.

[0188] The network optimization module is used to optimize the BP neural network by using the improved sand cat swarm optimization algorithm to obtain the BP neural network optimization model.

[0189] The model building module is used to build a temperature error model based on the BP neural network optimization model.

[0190] The data acquisition module is used to obtain the temperature, temperature change rate and variable temperature zero bias measurement value of the fiber optic gyroscope under variable temperature conditions.

[0191] The error prediction module is used to input the temperature and temperature change rate of the fiber optic gyroscope into the temperature error model to obtain the temperature error prediction value.

[0192] The temperature compensation module is used to obtain the temperature-compensated fiber optic gyroscope zero bias based on the variable temperature zero bias measurement value and the temperature error prediction value.

[0193] In an exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store fiber optic gyroscope temperature compensation data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a fiber optic gyroscope temperature compensation method is implemented.

[0194] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0195] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0196] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0197] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0198] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0199] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0200] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0201] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A fiber optic gyroscope temperature compensation method, characterized in that: The fiber optic gyro temperature compensation method comprises: Chaotic perturbation and Gaussian mutation mechanism are introduced into the sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm. The BP neural network is optimized by using the improved sand cat swarm optimization algorithm to obtain the BP neural network optimization model. Constructing a temperature error model based on the BP neural network optimization model; Under variable temperature conditions, obtain the temperature, temperature change rate and variable temperature zero bias measurement values ​​of the fiber optic gyroscope; Inputting the temperature and temperature change rate of the fiber optic gyroscope into the temperature error model to obtain a temperature error prediction value; Obtaining a temperature-compensated fiber optic gyroscope zero bias based on the variable temperature zero bias measurement value and the temperature error prediction value; Among them, the improved sand cat swarm optimization algorithm is used to optimize the BP neural network, and the BP neural network optimization model is obtained, including: Initializing the structure of the BP neural network to obtain a parameter combination; the parameter combination includes: a weight matrix from the input layer to the hidden layer, a bias vector of the hidden layer, a weight matrix from the hidden layer to the output layer, and a bias vector of the output layer; Initializing the position of the sand cat group based on the parameter combination to obtain an original definition domain, and determining individual fitness values ​​and average fitness values ​​of all individuals in the original definition domain; Performing chaotic perturbations on individuals whose individual fitness values ​​are greater than or equal to the average fitness value, performing Gaussian mutations on individuals whose individual fitness values ​​are less than the average fitness value, and performing individual iterative updates until a maximum number of iterations or a convergence condition is reached, obtaining an individual with the best fitness, and using the individual with the best fitness to update the individual positions in the sand cat group to obtain a global optimal solution; The global optimal solution is used as the optimal structural parameter of the BP neural network to obtain an optimized BP neural network, and the optimized BP neural network is used as the BP neural network optimization model.

2. The fiber optic gyro temperature compensation method according to claim 1, characterized in that: The new individuals obtained by performing chaotic perturbation on the individuals whose individual fitness values ​​are greater than or equal to the average fitness value are expressed as: ； In the formula, represents the original domain, and are the upper and lower bounds of the original domain, represents the chaotic mapping function, Represents a new individual, Indicates an individual.

3. The fiber optic gyro temperature compensation method according to claim 1, characterized in that: The new individual obtained by performing Gaussian mutation on the individual whose individual fitness value is less than the average fitness value is expressed as: ； In the formula, Represents a new individual, represents a random number generated by Gaussian distribution, represents the optimal individual.

4. The fiber optic gyro temperature compensation method according to claim 1, characterized in that: Initializing the position of the sand cat group based on the parameter combination to obtain the original definition domain includes: The chaotic mapping method is adopted to initialize the positions of the sand cat group based on the parameter combination to obtain the original definition domain.

5. The fiber optic gyro temperature compensation method according to claim 1, characterized in that: Building a temperature error model based on the BP neural network optimization model includes: Obtain training sample sets and test sample sets; The training sample set is used to train the BP neural network optimization model, and during the training process, the test sample set is used to test the trained BP neural network optimization model until the prediction error of the trained BP neural network optimization model reaches the set target, and the trained BP neural network optimization model is used as the temperature error model.

6. The fiber optic gyro temperature compensation method according to claim 5, characterized in that: The prediction error of the trained BP neural network optimization model is expressed as: ； In the formula, represents the prediction error, represents the number of samples, Indicates The temperature error prediction value corresponding to the sample is: Indicates The actual temperature error measurement value corresponding to the samples.

7. The fiber optic gyro temperature compensation method according to claim 1, characterized in that: Under variable temperature conditions, obtain the temperature, temperature change rate and variable temperature zero bias measurement values ​​of the fiber optic gyroscope, including: Select multiple fiber optic gyroscopes as test samples, and perform zero bias output tests under variable temperature and constant temperature conditions to obtain the temperature, variable temperature zero bias output, and constant temperature zero bias output; wherein variable temperature means that the temperature increases or decreases according to a set value, and constant temperature means that the temperature is kept at a set value; Obtaining the variable temperature zero bias measurement value based on the variable temperature zero bias output and the constant temperature zero bias output; The temperature change rate is determined based on the temperature.

8. The fiber optic gyro temperature compensation method according to claim 7, characterized in that: The variable temperature zero bias measurement value is expressed as: ； In the formula, is the variable temperature zero bias measurement value, is the sampling number, is the scaling factor, is the Earth's rotation angular velocity, is the geographical latitude of the test site, Indicates The actual temperature error measurement value corresponding to the samples.

9. A fiber optic gyro temperature compensation system, characterized in that: The fiber optic gyro temperature compensation system is used to implement the fiber optic gyro temperature compensation method according to any one of claims 1 to 8; the fiber optic gyro temperature compensation system comprises: Algorithm optimization module, used to introduce chaotic perturbation and Gaussian mutation mechanism into the sand cat swarm optimization algorithm to obtain an improved sand cat swarm optimization algorithm; The network optimization module is used to optimize the BP neural network by using the improved sand cat swarm optimization algorithm to obtain the BP neural network optimization model; A model building module, used for building a temperature error model based on the BP neural network optimization model; A data acquisition module is used to obtain the temperature, temperature change rate and temperature-varying zero bias measurement value of the fiber optic gyroscope under variable temperature conditions; An error prediction module, used for inputting the temperature and temperature change rate of the fiber optic gyroscope into the temperature error model to obtain a temperature error prediction value; The temperature compensation module is used to obtain the temperature-compensated fiber optic gyroscope zero bias based on the variable temperature zero bias measurement value and the temperature error prediction value.

Images