Robust estimation PNT multi-source information fusion method based on adaptive critical value improved IGGIII weight function

Abstract

The embodiment of the invention provides a robust estimation method of an improved IGGIII weight function based on an adaptive critical value, and relates to the technical field of positioning. According to the method, the critical value is adaptively adjusted, so that the positioning precision in a complex environment is effectively improved. The method comprises the following steps: carrying out a fixed point high-precision 5G positioning simulation experiment in an indoor open office scene, and verifying the effectiveness of an improved IGGIII weight function; on this basis, an IGGIII weight function based on an adaptive critical value is provided, the function calculates a residual confidence interval under a certain confidence level by using a robust mahalanobis distance statistical magnitude median residual, and a new equivalent full function critical value is constructed by using the boundary of the interval; compared with a traditional algorithm for setting a critical value through experience, the positioning precision of the method is improved by 2.9%. According to the robust estimation method, the critical value is adjusted in a self-adaptive mode, the positioning precision in a complex environment is effectively improved, and the robust estimation method has remarkable practicability and popularization value.

Description

Technical Field

[0001] This invention relates to the field of positioning technology, and in particular to a robust estimation method based on an adaptive critical value-based improved IGGⅢ weight function, which is used to improve the positioning accuracy of PNT (positioning, navigation, and timing) multi-source information fusion in complex environments. Background Technology

[0002] In complex environments, such as chemical plant areas, traditional positioning technologies are often severely affected by electrical equipment, radiation sources, and structurally complex obstacles, leading to the obstruction and attenuation of sensor measurement signals, thus affecting positioning accuracy. Furthermore, harsh environmental factors, including high temperatures, high pressures, and the presence of harmful gases, can weaken sensor performance, causing abnormal deviations in measurement results. Traditional robust estimation methods, such as Huber weight functions and L1-L2 methods, while possessing some robustness in handling outliers, still rely on empirically set fixed thresholds, which can lead to unstable estimation results in certain situations. Summary of the Invention

[0003] The purpose of this invention is to provide an adaptive critical value-based improved IGGⅢ weight function robust estimation algorithm to enhance positioning accuracy and stability in complex chemical plant areas. The specific solution is as follows:

[0004] In a first aspect, embodiments of the present invention provide an improved IGGⅢ weight function, the method comprising:

[0005] By using two harmonic coefficients, the three-segment information utilization rule of the traditional IGGⅢ weight function is transformed into a four-segment information utilization rule;

[0006] A simulation experiment was conducted using an indoor open-plan office environment to perform high-precision 5G positioning at fixed points.

[0007] In one embodiment of the present invention, the simulation experiment verifying the robustness effect specifically includes:

[0008] A simulation experiment of fixed-point high-precision 5G positioning was conducted in an indoor open-plan office setting. In this test, errors were artificially introduced into the observation data, which were then corrected using the least squares method. Robustness was evaluated using both the IGGⅢ method and an optimized scheme. The robustness of each scheme was assessed by comparing the evaluation results with the ideal estimates under error-free conditions—smaller differences indicate stronger robustness, while larger differences indicate weaker robustness.

[0009] In one embodiment of the present invention, the gross error addition scheme is specifically as follows:

[0010] Add a gross error of 0.020 ns to the observation at position 7;

[0011] Gross errors of -0.015 ns and 0.020 ns were added to the observations at positions 2 and 7, respectively.

[0012] Gross errors of -0.015ns, 0.020ns, and 0.018ns were added to the observations at positions 2, 7, and 8, respectively.

[0013] Secondly, embodiments of the present invention provide an adaptive critical value improved IGGⅢ weight function robust estimation localization algorithm to overcome the problem of decreased robustness efficiency of traditional empirically set fixed critical value weight functions. The method includes:

[0014] An equivalent weight function based on robust Mahalanobis distance statistics is constructed. The robust Mahalanobis distance statistics are built using the statistical regularities of adjacent innovation sequences, and then combined with RMDS. i The statistical distribution is determined, and at a given significance level, key thresholds and corresponding harmonic coefficients are calculated. State equations are constructed using IMU observations based on a strapdown algorithm, and observation equations are constructed based on INS-estimated positions and position data output from GNSS, LiDAR, and VINS sensors. A Kalman filter state prediction process is executed, robust state updates are performed, and a multi-source fusion localization solution is output.

[0015] Beneficial effects of the embodiments of the present invention:

[0016] This invention provides an improved adaptive critical value-based IGGⅢ weight function robust estimation precise positioning algorithm, contributing an important technical tool to the development of positioning services in complex factory areas. By utilizing the median residual based on robust Mahalanobis distance statistics to calculate the residual confidence interval and constructing the critical value of the equivalent weight function to achieve adaptive weighted matching, this method significantly improves the longitude and stability of positioning in complex factory scenarios compared to related technologies. Attached Figure Description

[0017] The accompanying drawings of this invention are provided to further understand the invention and constitute a part of this invention. They exemplify embodiments of the invention and their descriptions, and are used to explain the principles of the invention.

[0018] Figure 1 A flowchart of the PNT multi-source information fusion construction process based on the improved IGGⅢ weight function with adaptive critical values ​​is presented.

[0019] Figure 2 The simulation layout diagram shows the fixed-point high-precision 5G positioning of an indoor open office scenario.

[0020] Figure 3 A comparison chart of positioning errors under the ENU coordinate system is shown, illustrating the differences between different robust estimation methods. Detailed Implementation

[0021] The following describes in detail the specific implementation methods according to embodiments of the present invention. It should be understood that the described embodiments are for illustrative and explanatory purposes only and are not intended to limit the scope of the invention. Various changes and modifications can be made to the specific embodiments of the present invention based on the accompanying drawings without departing from the spirit and scope of the invention.

[0022] Traditional empirical robust estimation using IGGⅢ weight functions with fixed thresholds based on normal distribution statistics can lead to estimation results that fluctuate or collapse, resulting in reduced estimation efficiency and robustness. This causes problems such as decreased positioning accuracy and poor stability, severely impacting the performance of location services in complex factory areas. Therefore, the multi-source fusion positioning method using robust estimation using IGGⅢ weight functions with fixed thresholds based on normal distribution statistics faces the following problems: existing robust estimation methods using IGGⅢ weight functions are not smooth at ±k0, preventing the available information from fully utilizing its performance in robust estimation data processing; normal distribution statistics are easily affected by state estimates, and abnormal observation characteristics in complex factory environments cannot be known in advance, thus limiting the robustness efficiency of this statistic and making it difficult to obtain accurate matching results.

[0023] To address the aforementioned problems, this invention provides a robust estimation method based on an improved IGGⅢ weight function with adaptive critical values, which will be described in detail below.

[0024] First, a robust estimation method based on an improved IGGⅢ weight function with adaptive critical value, provided by an embodiment of the present invention, will be described.

[0025] See Figure 1 The present invention provides a flowchart of a robust estimation method based on an adaptive critical value improved IGGⅢ weight function. The method includes the following steps S101 to S104.

[0026] Step S101: Obtain INS / GNSS / LiDAR / VINS multi-source fusion vehicle data packets in a typical complex industrial plant environment, and construct precise positioning-related experiments using data collected by the vehicle platform equipped with the four navigation sensors.

[0027] The reference trajectory for the experimental test was provided by the high-precision Novatel SPAN-CPT navigation system. The GNSS positioning solution used pseudorange single-point positioning, the VINS positioning solution was based on the VINS-MONO algorithm to process camera and INS data, and the LiDAR positioning solution was calculated using normal distribution transformation.

[0028] Step S102: Add the IGGⅢ weight function algorithm module based on adaptive critical value improvement to construct an equivalent weight function based on robust Mahalanobis distance statistics. Calculate the critical thresholds corresponding to each statistical value at a given significance level.

[0029] Specifically, the two harmonic coefficients transform the three-segment information utilization rule of the traditional IGGⅢ weight function into a four-segment information utilization rule. The formula for the improved IGGⅢ weight function is as follows:

[0030]

[0031] Where k0 and k1 are harmonic coefficients. It is the standardized residual.

[0032] In practical applications, the harmonic coefficients of the traditional IGGⅢ weight function are fixed values. In order to improve the rationality of the threshold for determining whether an observation is reliable or should be eliminated, an adaptive critical value selection scheme based on robust Mahalanobis distance statistics is designed.

[0033] Specifically, a robust Mahalanobis distance statistic is constructed using the statistical regularities of adjacent innovation sequences, combined with RMDS. i The statistical distribution of the key threshold, at a given significance level, is defined by the following formula:

[0034]

[0035] Based on the above, the formula for the corresponding harmonic coefficient in the improved IGGⅢ weight function based on adaptive critical value is as follows:

[0036]

[0037] Step S103: Construct state equations based on strapdown algorithm using IMU observations and construct observation equations based on INS position calculations and position quantities output by three observation sensors: GNSS, LiDAR, and VINS.

[0038] Step S104: Execute the Kalman filter state prediction process, perform robust state updates, and output the multi-source fusion localization solution. Specifically, execute the Kalman filter state prediction process to obtain the predicted state and its error covariance matrix. Calculate the innovation and its covariance matrix, standardize adjacent innovation sequences, compare the statistical values ​​with the key thresholds calculated in step S102 one by one, calculate the weights of the corresponding observations based on the equivalent weight function, and construct the variance inflation matrix. Combine the state prediction with robust state updates. Output a reliable multi-source fusion localization solution and use the obtained state as the input for the next epoch.

Claims

1. A robust estimation method for PNT multi-source information fusion based on an adaptive critical value improved IGGⅢ weight function, characterized in that, The method includes: The improved IGGⅢ weight function addresses the problem of insufficient data processing in robust estimation; The confidence interval of the residuals at a certain confidence level is calculated using the median residuals based on the robust Mahalanobis distance statistic. Construct a new critical value for the equivalent weight function based on the boundary of the residual confidence interval; Using this critical value for robust estimation improves positioning accuracy.

2. The method according to claim 1, characterized in that, The median residual is obtained by constructing a robust Mahalanobis distance statistic based on the news sequence of adjacent epochs.

3. A PNT multi-source information fusion system based on the method of claim 1, characterized in that, The system includes: A multi-source data acquisition module is used to collect data from sensors such as GNSS, IMU, camera, and lidar. The data processing module is used to execute the robust estimation method described in claim 1; The positioning result output module is used to output the positioning results with improved accuracy.

4. The system according to claim 3, characterized in that, The data processing module further includes: An outlier detection unit is used to identify and mark outlier observations; An adaptive critical value adjustment unit is used to adjust the critical value of the equivalent weight function based on the outlier detection results.

5. A method for verifying positioning accuracy using the method or system described in any one of claims 1 to 4, characterized in that, The method includes: A simulation experiment of fixed-point high-precision 5G positioning was conducted in an indoor open office setting. The experimental results were compared with the best estimate under normal conditions to verify the robustness.

6. An improved IGGⅢ weight function, characterized in that, This weighting function divides the standardized residuals into four segments—effective information, the first segment of usable information, the second segment of usable information, and harmful information—by adjusting the harmonic coefficients, thereby achieving adaptive processing of different types of information.

7. The weighting function according to claim 6, characterized in that, The harmonic coefficient is dynamically adjusted based on the magnitude of the residual.

8. A performance comparison method for robust estimation based on improved IGGⅢ weight function, characterized in that, The method includes the following steps: A simulation experiment of fixed-point high-precision 5G positioning was conducted in an indoor open office setting. Add gross errors to the observation data; Least square adjustment was performed, and robust estimation was conducted using the IGG III scheme and the improved scheme. Compare the results with the best valuation under normal conditions; the smaller the difference, the stronger the resilience, and vice versa.

9. The method according to claim 8, characterized in that, The aforementioned gross error inclusion scheme includes: Add a gross error of 0.020 ns to the observation at position 7; Gross errors of -0.015 ns and 0.020 ns were added to the observations at positions 2 and 7, respectively. Gross errors of -0.015ns, 0.020ns, and 0.018ns were added to the observations at positions 2, 7, and 8, respectively.

10. A PNT multi-source fusion precise localization method based on robust estimation of IGGⅢ weight function using improved adaptive critical values, characterized in that... This method utilizes multi-source data from complex factory environments for experiments, and analyzes and compares the positioning performance of four schemes: Huber equivalent weight function with empirically set fixed thresholds, standard variance expansion robust filtering algorithm, improved IGGⅢ equivalent weight function with adaptive thresholds proposed by our research group, and traditional EKF positioning solution.

Images