A method for calibrating an inertial navigation sensor of a test vehicle

Through technologies such as laser rangefinder and timing noise reduction autoencoder, an inertial navigation calibration model is built and sensor parameters are dynamically calibrated, which solves the problems of high cost and complex operation of traditional methods, and improves the accuracy and reliability of the inertial navigation sensor.

CN118816949BActive Publication Date: 2024-11-19HUNAN SHIBO TESTING TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411295939.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-18
Publication Date
2024-11-19
Estimated Expiration
2044-09-18

AI Technical Summary

Technical Problem

The existing inertial navigation sensor calibration methods rely on high-precision calibration platforms and professional laboratory environments. They are costly, complex and difficult to perform frequently, resulting in the drift of navigation solution results, affecting system accuracy and reliability.

Method used

A laser rangefinder is used for spatial geometric alignment, combined with a timing noise reduction autoencoder and extended Kalman filtering, an inertial navigation calibration model is built, sensor parameters are dynamically calibrated, errors are reduced and data accuracy is improved.

Benefits of technology

It realizes high-precision calibration of inertial navigation sensors, reduces system errors, improves long-term stability and reliability, reduces costs, simplifies operations, and is suitable for frequent use.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118816949B_ABST
    Figure CN118816949B_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of inertial navigation calibration, and discloses a method for calibrating an inertial navigation sensor of a test vehicle. The method comprises: using a laser rangefinder to calibrate the installation position of an inertial navigation sensor on a test vehicle; collecting vehicle motion data and performing noise elimination to obtain denoised vehicle motion data, wherein the vehicle motion data comprises speed, acceleration and displacement; constructing an inertial navigation calibration model to calibrate the dynamic error of the inertial navigation sensor, using the denoised vehicle motion data as input data of the model, and obtaining calibrated inertial navigation sensor parameters; and using the dynamically calibrated sensor parameters to achieve data measurement to ensure data accuracy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of inertial navigation calibration, and in particular to a method for calibrating an inertial navigation sensor of a test vehicle. Background Art

[0002] The inertial navigation system is one of the core components of modern test vehicle positioning and navigation. It estimates the vehicle's position, velocity and attitude by measuring the vehicle's acceleration and angular velocity. In test vehicles, a high-precision inertial navigation system is essential for achieving precise positioning, autonomous driving, and various advanced driver assistance systems. The inertial navigation system is mainly composed of inertial sensors such as accelerometers and gyroscopes. However, these sensors often have various errors in actual use, such as zero bias error, scale factor error, installation error, etc. These errors will accumulate over time, causing the navigation solution results to drift, seriously affecting the accuracy and reliability of the system. Therefore, accurate calibration of the inertial navigation sensor is a key prerequisite for ensuring high-precision positioning and navigation of the test vehicle. Traditional inertial navigation sensor calibration methods mainly rely on high-precision calibration platforms and professional laboratory environments. Although this method has high accuracy, it is expensive, complex to operate, and difficult to perform frequently. Summary of the invention

[0003] In view of this, the present invention proposes a test vehicle inertial navigation sensor calibration method to reduce system errors and improve the long-term stability and reliability of the inertial navigation system.

[0004] To achieve the above object, the present invention provides a method for calibrating an inertial navigation sensor of a test vehicle, comprising the following steps:

[0005] S1: Use a laser rangefinder to calibrate the installation position of the inertial navigation sensor on the test vehicle, where spatial geometric alignment is the implementation method of inertial navigation sensor calibration;

[0006] S2: collecting vehicle motion data and performing noise elimination to obtain denoised vehicle motion data, wherein the vehicle motion data includes speed, acceleration and displacement, wherein a time series denoising autoencoder is an implementation method of the noise elimination;

[0007] S3: constructing an inertial navigation calibration model to correct the dynamic error of the inertial navigation sensor, wherein the inertial navigation calibration model takes the denoised vehicle motion data as input and takes the corrected inertial navigation sensor parameters as output, wherein the extended Kalman filter is an implementation method of the inertial navigation calibration model;

[0008] S4: Use dynamically calibrated sensor parameters to measure data to ensure data accuracy.

[0009] As a further improvement method of the present invention:

[0010] Optionally, in step S1, using a laser rangefinder to calibrate the installation position of the inertial navigation sensor on the test vehicle includes:

[0011] S11: Place a laser rangefinder at a known position, randomly select three non-collinear reference points on the test vehicle, use the laser rangefinder to measure the distance between the sensor and each reference point, and calculate the position of the sensor in the vehicle coordinate system by trilateration using the distance measured by the laser rangefinder and the known reference point positions. The distance calculation formula is:

[0012] ;

[0013] in:

[0014] Respectively represent the position coordinates of three non-collinear reference points on the test vehicle in the vehicle coordinate system;

[0015] Respectively represent the measured distance between the sensor and each reference point;

[0016] Represents the position coordinates of the inertial navigation sensor in the vehicle coordinate system;

[0017] S12: Establish the conversion relationship between the vehicle coordinate system and the world coordinate system. The calculation formula is:

[0018] ;

[0019] in:

[0020] The position vector of the sensor in the world coordinate system;

[0021] Represents the rotation matrix of the vehicle coordinate system relative to the world coordinate system;

[0022] Represents the translation vector of the vehicle coordinate system in the world coordinate system;

[0023] Represents the position vector of the inertial navigation sensor in the vehicle coordinate system;

[0024] S13: Calculate the position coordinates of the sensor in the world coordinate system according to the coordinate system conversion relationship, and calculate the deviation with the measured data. The deviation calculation formula is:

[0025] ;

[0026] in:

[0027] Represents the position vector of the measured sensor in the world coordinate system;

[0028] S14: Correct the position of the sensor according to the calculated deviation. The calculation formula is:

[0029] ;

[0030] in:

[0031] Represents the corrected sensor position coordinates;

[0032] Indicates the position error correction step size;

[0033] S15: Re-measure the installation position of the sensor to ensure that the deviation is within the preset range and complete the installation calibration.

[0034] Optionally, in step S2, collecting vehicle motion data and performing noise elimination to obtain denoised vehicle motion data includes:

[0035] S21: normalizing the collected vehicle motion data to obtain normalized vehicle motion data;

[0036] S22: constructing a vehicle motion data denoising autoencoder model, wherein the vehicle motion data denoising autoencoder model structure includes an input layer, a composite noise layer, an encoder layer, a decoder layer and an output layer, wherein:

[0037] The input layer receives a shape of The normalized vehicle motion data is taken as input, where Indicates the number of vehicle motion data records, Indicates the length of vehicle motion data record;

[0038] The composite noise layer adds composite noise to each item in the input data, and the calculation formula is:

[0039] ;

[0040] in:

[0041] represents noise that follows a composite distribution;

[0042] represents the normalized vehicle motion data;

[0043] Represents the vehicle motion data after adding noise;

[0044] The composite distribution is composed of Gaussian distribution, Poisson distribution and exponential distribution weighted;

[0045] The encoder layer contains 3 fully connected layers, and the calculation formula is:

[0046] ;

[0047] ;

[0048] ;

[0049] in:

[0050] , , Represents the weight matrices of the first, second, and third fully connected layers in the encoder layer respectively

[0051] , , Respectively represent the bias vectors of the first, second, and third fully connected layers in the encoder layer;

[0052] represents the activation function;

[0053] Represents the output vector of the first fully connected layer of the encoder;

[0054] represents the output vector of the second fully connected layer of the encoder;

[0055] represents the output vector of the third fully connected layer of the encoder;

[0056] The decoder layer contains 3 fully connected layers, and the calculation formula is:

[0057] ;

[0058] ;

[0059] ;

[0060] in:

[0061] Represents the weight matrices of the first, second, and third fully connected layers in the decoder layer;

[0062] Respectively represent the bias vectors of the first, second, and third fully connected layers in the decoder layer;

[0063] represents the output vector of the first fully connected layer of the decoder;

[0064] represents the output vector of the second fully connected layer of the decoder;

[0065] represents the output vector of the third fully connected layer of the decoder;

[0066] The output layer directly outputs the decoder layer output vector as the denoised vehicle motion data;

[0067] S23: Optimize and solve the constructed vehicle motion data denoising autoencoder model to obtain model parameters, instantiate the model according to the model parameters, and use the instantiated model to perform vehicle motion data denoising.

[0068] Optionally, in step S23, optimizing and solving the constructed vehicle motion data denoising autoencoder model to obtain model parameters includes:

[0069] S23.1: Determine the loss function according to the constructed vehicle motion data denoising autoencoder model, and the loss function calculation formula is:

[0070] ;

[0071] in:

[0072] represents the i-th vehicle motion data record;

[0073] represents the i-th vehicle motion data record after noise reduction;

[0074] n represents the number of samples;

[0075] S23.2: Randomly initialize the model parameter vector , set the initial search direction ;

[0076] in: Represents the target loss function exist The gradient at the location;

[0077] S23.3: Update model parameters ;

[0078] in:

[0079] represents the update step size;

[0080] Indicates the search direction of the kth iteration;

[0081] represents the model parameters of the kth iteration;

[0082] S23.4: Calculate the target loss function in Gradient at position ;

[0083] S23.5: Update calculation search direction ,in: represents the conjugate coefficient of the kth iteration;

[0084] S23.6: Iterate step S23.3 until the maximum number of iterations is reached and the output model parameters are the optimal model parameters.

[0085] Optionally, in step S3, constructing an inertial navigation calibration model to correct a dynamic error of an inertial navigation sensor includes:

[0086] S31: construct an inertial navigation calibration model based on the extended Kalman filter, wherein the state vector of the inertial navigation calibration model includes the vehicle displacement, velocity and acceleration obtained by the inertial navigation sensor, and the observation vector of the model includes the vehicle position, velocity and acceleration obtained by GPS. The model state transfer process calculation formula is:

[0087] ;

[0088] in:

[0089] A nonlinear function representing the state transition process;

[0090] is the control input;

[0091] is the process noise;

[0092] Represents the state vector of the kth iteration;

[0093] Represents the state vector of the k+1th iteration;

[0094] The calculation formula of the state observation transfer process of the inertial navigation calibration model is:

[0095] ;

[0096] in:

[0097] represents the observation noise;

[0098] represents the observation vector of the kth iteration;

[0099] A nonlinear function representing the state observation transfer process;

[0100] S32: Randomly initialize the state vector , the covariance matrix and the observation covariance matrix ;

[0101] S33: Predict the state vector using the model state transition process based on the current round state vector , using the model state observation transfer process to calculate the observation vector ;

[0102] S34: Calculate the Kalman gain of the current round , the calculation formula is:

[0103] ;

[0104] in:

[0105] represents the observation matrix;

[0106] represents the covariance matrix of the kth round;

[0107] S35: Based on the calculated Kalman gain and observation value Update the state vector and covariance matrix,

[0108] The state vector update calculation formula is:

[0109] ;

[0110] The covariance matrix update calculation formula is:

[0111] ;

[0112] in:

[0113] represents the observation vector predicted based on the k-th round state vector;

[0114] S36: Iterate step S33 until the maximum number of iterations is reached, and correct the parameters of the inertial navigation sensor according to the last output state vector.

[0115] Optionally, in step S36, correcting the parameters of the inertial navigation sensor according to the last output state vector includes:

[0116] S36.1: concatenate the state vector output by the inertial navigation calibration model and the observation vector of the inertial navigation sensor to obtain a fusion vector;

[0117] S36.2: Use deep neural networks to establish nonlinear mapping relationships between fusion vectors and inertial navigation sensor parameters;

[0118] S36.3: Use the nonlinear mapping relationship between the fusion vector and the inertial navigation sensor parameters to calibrate the inertial navigation sensor parameters.

[0119] Use dynamically calibrated sensor parameters to measure data and calibrate the test vehicle's inertial navigation sensor to ensure data accuracy.

[0120] In order to solve the above problem, the present invention provides an electronic device, the electronic device comprising:

[0121] A memory storing at least one instruction;

[0122] Communication interface, enabling electronic equipment to communicate; and

[0123] The processor executes the instructions stored in the memory to implement the above-mentioned test vehicle inertial navigation sensor calibration method.

[0124] In order to solve the above problem, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned test vehicle inertial navigation sensor calibration method.

[0125] Compared with the prior art, the present invention proposes a method for calibrating an inertial navigation sensor of a test vehicle, which has the following advantages:

[0126] (1) This solution uses a laser rangefinder to calibrate the installation position of the inertial navigation sensor on the test vehicle to achieve spatial geometric alignment, ensure the installation position of the sensor is accurate, and improve the accuracy of the calibration;

[0127] (2) This scheme proposes a noise removal method based on a time series denoising autoencoder. This method collects vehicle motion data and uses a time series denoising autoencoder to remove noise, thereby obtaining denoised vehicle motion data. This method can effectively remove noise from motion data and improve data quality and accuracy.

[0128] (3) This scheme proposes an inertial navigation calibration scheme that can calibrate the inertial navigation sensor parameters of vehicle motion data to improve the accuracy of the calibration;

[0129] (4) This scheme uses a combination of methods such as spatial geometric alignment, temporal denoising autoencoder, extended Kalman filter and dynamic calibration to achieve precise correction of inertial navigation sensors and improve the performance of sensors in navigation and positioning applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0130] Figure 1 A schematic diagram of a flow chart of a method for calibrating an inertial navigation sensor of a test vehicle provided by an embodiment of the present invention;

[0131] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0132] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0133] The embodiment of the present application provides a method for calibrating an inertial navigation sensor of a test vehicle. The execution subject of the method for calibrating an inertial navigation sensor of a test vehicle includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for calibrating an inertial navigation sensor of a test vehicle can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0134] Embodiment 1:

[0135] S1: Use a laser rangefinder to calibrate the installation position of the inertial navigation sensor on the test vehicle, where spatial geometric alignment is the implementation method of inertial navigation sensor calibration.

[0136] In the step S1, a laser rangefinder is used to calibrate the installation position of the inertial navigation sensor on the test vehicle, including:

[0137] S11: Place a laser rangefinder at a known position, randomly select three non-collinear reference points on the test vehicle, use the laser rangefinder to measure the distance between the sensor and each reference point, and calculate the position of the sensor in the vehicle coordinate system by trilateration using the distance measured by the laser rangefinder and the known reference point positions. The distance calculation formula is:

[0138] ;

[0139] in:

[0140] Respectively represent the position coordinates of three non-collinear reference points on the test vehicle in the vehicle coordinate system;

[0141] Respectively represent the measured distance between the sensor and each reference point;

[0142] Represents the position coordinates of the inertial navigation sensor in the vehicle coordinate system;

[0143] S12: Establish the conversion relationship between the vehicle coordinate system and the world coordinate system. The calculation formula is:

[0144] ;

[0145] in:

[0146] The position vector of the sensor in the world coordinate system;

[0147] Represents the rotation matrix of the vehicle coordinate system relative to the world coordinate system;

[0148] Represents the translation vector of the vehicle coordinate system in the world coordinate system;

[0149] Represents the position vector of the inertial navigation sensor in the vehicle coordinate system;

[0150] S13: Calculate the position coordinates of the sensor in the world coordinate system according to the coordinate system conversion relationship, and calculate the deviation with the measured data. The deviation calculation formula is:

[0151] ;

[0152] in:

[0153] Represents the position vector of the measured sensor in the world coordinate system;

[0154] S14: Correct the position of the sensor according to the calculated deviation. The calculation formula is:

[0155] ;

[0156] in:

[0157] Represents the corrected sensor position coordinates;

[0158] Indicates the position error correction step size;

[0159] S15: Re-measure the installation position of the sensor to ensure that the deviation is within the preset range and complete the installation calibration.

[0160] S2: Collect vehicle motion data and perform noise elimination to obtain denoised vehicle motion data, wherein the vehicle motion data includes speed, acceleration and displacement, wherein a time series denoising autoencoder is an implementation method of the noise elimination.

[0161] In step S2, the vehicle motion data is collected and noise is eliminated to obtain the denoised vehicle motion data, including:

[0162] S21: normalizing the collected vehicle motion data to obtain normalized vehicle motion data;

[0163] S22: construct a vehicle motion data denoising autoencoder model, wherein the model introduces composite noise on the basis of a standard autoencoder to improve the model denoising effect, and the vehicle motion data denoising autoencoder model structure includes an input layer, a composite noise layer, an encoder layer, a decoder layer and an output layer, wherein:

[0164] The input layer receives a shape of The normalized vehicle motion data is taken as input, where Indicates the number of vehicle motion data records, Indicates the length of vehicle motion data record;

[0165] The composite noise layer adds composite noise to each item in the input data, and the calculation formula is:

[0166] ;

[0167] in:

[0168] represents noise that follows a composite distribution;

[0169] represents the normalized vehicle motion data;

[0170] Represents the vehicle motion data after adding noise;

[0171] The composite distribution is composed of Gaussian distribution, Poisson distribution and exponential distribution weighted;

[0172] The encoder layer contains 3 fully connected layers, and the calculation formula is:

[0173] ;

[0174] ;

[0175] ;

[0176] in:

[0177] , , Represent the weight matrices of the first, second, and third fully connected layers in the encoder layer respectively;

[0178] , , Respectively represent the bias vectors of the first, second, and third fully connected layers in the encoder layer;

[0179] represents the activation function;

[0180] Represents the output vector of the first fully connected layer of the encoder;

[0181] represents the output vector of the second fully connected layer of the encoder;

[0182] represents the output vector of the third fully connected layer of the encoder;

[0183] The decoder layer contains 3 fully connected layers, and the calculation formula is:

[0184] ;

[0185] ;

[0186] ;

[0187] in:

[0188] Represents the weight matrices of the first, second, and third fully connected layers in the decoder layer;

[0189] Respectively represent the bias vectors of the first, second, and third fully connected layers in the decoder layer;

[0190] represents the output vector of the first fully connected layer of the decoder;

[0191] represents the output vector of the second fully connected layer of the decoder;

[0192] represents the output vector of the third fully connected layer of the decoder;

[0193] The output layer directly outputs the decoder layer output vector as the denoised vehicle motion data;

[0194] S23: Optimize and solve the constructed vehicle motion data denoising autoencoder model to obtain model parameters, instantiate the model according to the model parameters, and use the instantiated model to perform vehicle motion data denoising.

[0195] In the step S23, the constructed vehicle motion data denoising autoencoder model is optimized and solved to obtain model parameters, including:

[0196] S23.1: Determine the loss function according to the constructed vehicle motion data denoising autoencoder model, and the loss function calculation formula is:

[0197] ;

[0198] in:

[0199] represents the i-th vehicle motion data record;

[0200] represents the i-th vehicle motion data record after noise reduction;

[0201] n represents the number of samples;

[0202] S23.2: Randomly initialize the model parameter vector , set the initial search direction ;

[0203] in: Represents the target loss function exist The gradient at the location;

[0204] S23.3: Update model parameters ;

[0205] in:

[0206] represents the update step size;

[0207] Indicates the search direction of the kth iteration;

[0208] represents the model parameters of the kth iteration;

[0209] S23.4: Calculate the target loss function in Gradient at position ;

[0210] S23.5: Update calculation search direction ,in: represents the conjugate coefficient of the kth iteration;

[0211] S23.6: Iterate step S23.3 until the maximum number of iterations is reached and the output model parameters are the optimal model parameters.

[0212] S3: Construct an inertial navigation calibration model to correct the dynamic error of the inertial navigation sensor, wherein the inertial navigation calibration model takes the denoised vehicle motion data as input and the corrected inertial navigation sensor parameters as output, wherein the extended Kalman filter is an implementation method of the inertial navigation calibration model.

[0213] In step S3, an inertial navigation calibration model is constructed to correct the dynamic error of the inertial navigation sensor, including:

[0214] S31: construct an inertial navigation calibration model based on the extended Kalman filter, wherein the state vector of the inertial navigation calibration model includes the vehicle displacement, velocity and acceleration obtained by the inertial navigation sensor, and the observation vector of the model includes the vehicle position, velocity and acceleration obtained by GPS. The model state transfer process calculation formula is:

[0215] ;

[0216] in:

[0217] A nonlinear function representing the state transition process;

[0218] is the control input;

[0219] is the process noise;

[0220] Represents the state vector of the kth iteration;

[0221] Represents the state vector of the k+1th iteration;

[0222] The calculation formula of the state observation transfer process of the inertial navigation calibration model is:

[0223] ;

[0224] in:

[0225] represents the observation noise;

[0226] represents the observation vector of the kth iteration;

[0227] A nonlinear function representing the state observation transfer process;

[0228] S32: Randomly initialize the state vector , the covariance matrix and the observation covariance matrix ;

[0229] S33: Predict the state vector using the model state transition process based on the current round state vector , using the model state observation transfer process to calculate the observation vector ;

[0230] S34: Calculate the Kalman gain of the current round , the calculation formula is:

[0231] ;

[0232] in:

[0233] represents the observation matrix;

[0234] represents the covariance matrix of the kth round;

[0235] S35: Based on the calculated Kalman gain and observation value Update the state vector and covariance matrix,

[0236] The state vector update calculation formula is:

[0237] ;

[0238] The covariance matrix update calculation formula is:

[0239] ;

[0240] in:

[0241] represents the observation vector predicted based on the k-th round state vector;

[0242] S36: Iterate step S33 until the maximum number of iterations is reached, and correct the parameters of the inertial navigation sensor according to the last output state vector.

[0243] In the step S36, the parameters of the inertial navigation sensor are corrected according to the state vector outputted last, including:

[0244] S36.1: concatenate the state vector output by the inertial navigation calibration model and the observation vector of the inertial navigation sensor to obtain a fusion vector;

[0245] S36.2: Use deep neural networks to establish nonlinear mapping relationships between fusion vectors and inertial navigation sensor parameters;

[0246] S36.3: calibrate inertial navigation sensor parameters using a nonlinear mapping relationship between fusion vectors and inertial navigation sensor parameters;

[0247] Data measurement is performed using dynamically calibrated inertial navigation sensor parameters to ensure data accuracy.

[0248] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0249] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0250] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for calibrating an inertial navigation sensor of a test vehicle, It is characterized in that The method comprises: S1: Use a laser rangefinder to calibrate the installation position of the inertial navigation sensor on the test vehicle, where spatial geometric alignment is the implementation method of inertial navigation sensor calibration; S2: collecting vehicle motion data and performing noise elimination to obtain denoised vehicle motion data, wherein the vehicle motion data includes speed, acceleration and displacement, wherein a time series denoising autoencoder is an implementation method of the noise elimination; S3: constructing an inertial navigation calibration model to correct the dynamic error of the inertial navigation sensor, wherein the inertial navigation calibration model takes the denoised vehicle motion data as input and takes the corrected inertial navigation sensor parameters as output, wherein the extended Kalman filter is an implementation method of the inertial navigation calibration model; Construct an inertial navigation calibration model to correct the dynamic error of the inertial navigation sensor, including: S31: constructing an inertial navigation calibration model based on an extended Kalman filter, wherein a state vector of the inertial navigation calibration model includes a vehicle displacement, velocity and acceleration obtained by an inertial navigation sensor, and an observation vector of the inertial navigation calibration model includes a vehicle position, velocity and acceleration obtained by a GPS; S32: Randomly initialize the state vector, covariance matrix and observation covariance matrix; S33: predicting the state vector of the next round and the observation vector of the current round according to the state vector of the current round; S34: Calculate the Kalman gain of the current round according to the covariance matrix; S35: Update the state vector and covariance matrix of the next round according to the calculated Kalman gain and observation vector; S36: iteratively executing step S33 until the maximum number of iterations is reached, and correcting the parameters of the inertial navigation sensor according to the state vector output last; S4: Use dynamically calibrated sensor parameters to measure data to ensure data accuracy.

2. A test vehicle inertial navigation sensor calibration method as claimed in claim 1, It is characterized in that In the step S1, a laser rangefinder is used to calibrate the installation position of the inertial navigation sensor on the test vehicle, including: S11: Place a laser rangefinder at a known position, randomly select three non-collinear reference points on the test vehicle, use the laser rangefinder to measure the distance between the sensor and each reference point, and calculate the position of the sensor in the vehicle coordinate system by trilateration using the distance measured by the laser rangefinder and the known reference point positions. The distance calculation formula is: ; in: Respectively represent the position coordinates of three non-collinear reference points on the test vehicle in the vehicle coordinate system; Respectively represent the measured distance between the sensor and each reference point; Represents the position coordinates of the inertial navigation sensor in the vehicle coordinate system; S12: Establish a conversion relationship between the vehicle coordinate system and the world coordinate system, and calculate the position coordinates of the sensor in the world coordinate system according to the coordinate system conversion relationship, calculate the deviation with the measured data, correct the position of the sensor, and complete the installation calibration.

3. A test vehicle inertial navigation sensor calibration method as claimed in claim 1, It is characterized in that In step S2, the vehicle motion data is collected and noise is eliminated to obtain the denoised vehicle motion data, including: S21: normalizing the collected vehicle motion data to obtain normalized vehicle motion data; S22: Construct a vehicle motion data denoising autoencoder model, wherein the vehicle motion data denoising autoencoder model structure includes an input layer, a composite noise layer, an encoder layer, a decoder layer and an output layer, wherein the input layer receives a shape of The normalized vehicle motion data is taken as input; in: Indicates the number of vehicle motion data records; Indicates the length of vehicle motion data record; The composite noise layer adds noise to each item in the input data, the encoder layer includes 3 fully connected layers, the decoder layer includes 3 fully connected layers, and the output layer directly outputs the decoder layer output vector as the vehicle motion data after noise reduction; S23: Optimize and solve the constructed vehicle motion data denoising autoencoder model to obtain model parameters, instantiate the model according to the model parameters, and use the instantiated model to perform vehicle motion data denoising.

4. A test vehicle inertial navigation sensor calibration method as claimed in claim 3, It is characterized in that In the step S23, the constructed vehicle motion data denoising autoencoder model is optimized and solved to obtain model parameters, including: S23.1: Determine the loss function according to the constructed vehicle motion data denoising autoencoder model, and the loss function calculation formula is: ; in: represents the i-th vehicle motion data record; represents the i-th vehicle motion data record after noise reduction; n represents the number of samples; S23.2: Randomly initialize the model parameter vector , set the initial search direction ; in: Represents the target loss function exist The gradient at the location; S23.3: Update model parameters ; in: represents the update step size; Indicates the search direction of the kth iteration; represents the model parameters of the kth iteration; S23.4: Calculate the target loss function in Gradient at position ; S23.5: Update calculation search direction ; in: represents the conjugate coefficient of the kth iteration; S23.6: Iterate step S23.3 until the maximum number of iterations is reached and the output model parameters are the optimal model parameters.

5. A test vehicle inertial navigation sensor calibration method as claimed in claim 1, It is characterized in that In the step S36, the parameters of the inertial navigation sensor are corrected according to the state vector outputted last, including: S36.1: concatenate the state vector output by the inertial navigation calibration model and the observation vector of the inertial navigation sensor to obtain a fusion vector; S36.2: Use deep neural networks to establish nonlinear mapping relationships between fusion vectors and inertial navigation sensor parameters; S36.3: Use the nonlinear mapping relationship between the fusion vector and the inertial navigation sensor parameters to calibrate the inertial navigation sensor parameters.