CN-120141462-B - Underground space unmanned system positioning method and system under multi-source interference

CN120141462BCN 120141462 BCN120141462 BCN 120141462BCN-120141462-B

Abstract

The invention discloses a method and a system for locating an unmanned system in an underground space under multi-source interference, which relate to the technical field of unmanned system locating, the invention firstly collects all motion data of the unmanned system at all collection moments, builds a system model, then defines a cost function in a statistical similarity framework, iterates a state vector mean value and an error covariance matrix at all collection moments, then utilizing the variational Bayesian learning to jointly infer a noise covariance matrix in a sliding window, finally outputting a position estimation result and an error covariance matrix of the unmanned system, the invention solves the problem of poor integrated navigation positioning effect of the unmanned system caused by non-Gaussian noise of the complex interference coupling sensor in the underground space, and improves the positioning precision and the anti-interference capability of the unmanned system in the underground space.

Inventors

WANG GUOQING
YU MINGLONG
Zhu Zhaolei
CHEN XINKAI
YANG CHUNYU
MA LEI
DAI WEI
MIAO YANZI

Assignees

中国矿业大学

Dates

Publication Date: 20260505
Application Date: 20250312

Claims (9)

1. The underground space unmanned system positioning method under the condition of multi-source interference is characterized by comprising the following steps: acquiring data, namely acquiring navigation information of an unmanned system in a subsurface space by utilizing a combined navigation system of a UWB tag and a micro-inertia measurement unit which are assembled on the unmanned system and a preset UWB ranging base station; s11, constructing a state transition matrix and a measurement matrix, and simultaneously acquiring measurement noise by using a sensor; S12, acquiring target values of motion data of the unmanned system at each acquisition time, and recording the target values as state vectors Then Representing unmanned system in the first State vectors at each acquisition time; S13, acquiring measurement values of motion data of the unmanned system at each acquisition time, and recording the measurement values as measurement vectors Then Representing unmanned system in the first Measuring vectors at each acquisition time; S14, constructing a linear state space model with control inputs of process noise, measurement noise and non-Gaussian noise: , , In the formula, Representing unmanned system in the first A state transition matrix for each acquisition instant, In order to measure the vector quantity, In order to measure the dimension of the vector, Is that The dimension of the real number is the same, , Is that The dimension of the real number is the same, , Is that The dimension of the real number is the same, As a system state vector of the system, As a dimension of the system state vector, Is that The dimension of the real number is the same, Representing unmanned system in the first The process noise at the time of the acquisition, , Representing unmanned system in the first The measurement noise at the time of the acquisition, , And All represent unmanned systems in the first Non-gaussian interference noise at the individual acquisition instants, Representing unmanned system in the first The probability that the individual acquisition instants are affected by non-gaussian noise, Representing the measured variable at the first Probability of being affected by non-Gaussian noise at each acquisition time; Step three, updating filtering parameters, namely acquiring prior state vectors and covariance matrixes of a system, defining a cost function in a framework of statistical similarity, and iterating the state vector mean value and the error covariance matrix at each acquisition time to obtain posterior state vectors and error covariance matrixes; Based on the obtained posterior state estimation, utilizing the variable dB leaf learning in a sliding window to jointly infer a noise covariance matrix; and fifthly, outputting the position estimation result and the covariance matrix of the unmanned system after the loop iteration is completed.
2. The method for positioning an underground space unmanned system under multi-source interference according to claim 1, wherein the construction of the state transition matrix and the measurement matrix comprises the following steps: , , In the middle of Representing the duration of a predetermined time interval, , Representative of And (5) a dimensional identity matrix.
3. The method for positioning an underground space unmanned system under multi-source interference according to claim 2, wherein the filtering parameter is updated as follows: S21, solving a priori state vector of the system according to a Kalman filtering method Covariance matrix ; S22, under the statistical similarity framework, maximizing a cost function defined by using a system equation and covariance information Will be Approximately gaussian, i.e. In the following Representing state variables at the first The average value of the individual acquisition moments is, Representing unmanned system in the first An estimated error covariance matrix at each acquisition time; S23, iterating the state vector mean value and the estimated error covariance matrix in the cost function at each acquisition time, judging whether the result after each iteration meets the iteration termination condition, if the result after a certain iteration is not met, continuing to iterate until the result after the iteration meets the iteration termination condition, taking the state vector mean value after the iteration as the state vector estimated value, and outputting the state vector estimated value and the estimated error covariance matrix after the iteration.
4. The method for positioning an underground space unmanned system under multi-source interference according to claim 3, wherein the iteration is performed on the state vector mean and the estimation error covariance matrix in the cost function at each acquisition time, and the specific process is as follows: s31, acquiring a measurement variable of the unmanned system at each acquisition time and a measurement noise covariance matrix at each acquisition time; s32, will The maximization problem of (2) is converted into the related And Is a major problem of the (c) in the (c), And Representation by The optimal posterior probability density function obtained by iterative approximation of the secondary fixed point can be obtained according to a maximum criterion: , In the formula, Representing unmanned system in the first The a priori state vectors at the individual acquisition instants, Represents the first The first acquisition time The Kalman filter gain in the iteration is calculated as follows: , In the formula, Representing unmanned system in the first The first acquisition time The corrected one-step prediction error in the second iteration estimates the covariance matrix, Representing unmanned system in the first The first acquisition time Measuring a noise covariance matrix by corrected one-step prediction in the secondary iteration; , In the formula, Represents the first Measuring a noise covariance matrix at each acquisition moment; In the first place The first acquisition time Modified one-step prediction noise covariance matrix in multiple iterations And The calculation is as follows: , In the formula, Represents the first The one-step prediction error at each acquisition instant estimates the covariance matrix, 、 Represents the first The first acquisition time The auxiliary variables used for the iterations are calculated as follows: , In the formula, Representing the parameters of the degree of freedom, Representing the bandwidth of the core, And Represents the first The first acquisition time The auxiliary variables used for the iterations are calculated as follows: , 。
5. the method for positioning an underground space unmanned system under multi-source interference according to claim 3, wherein the determining whether the result after each iteration satisfies the iteration termination condition comprises the following specific steps: When each iteration is carried out at each acquisition time, recording the number of iterations and the result after each iteration, and then: , In the middle of Represents a termination threshold when the above-mentioned inequality is established or passed The sub-loop, when the inequality is not satisfied and does not pass And in the secondary loop, the iteration termination condition is not met.
6. The method for locating an unmanned underground space system under multi-source interference according to claim 4, wherein the joint inferred noise covariance matrix comprises the following steps: s41, setting a time interval, which is recorded as Then Defining a state transition probability density function and a measurement likelihood probability density function in a time interval: , In the formula, Representing that the unmanned system is in A state transition probability density function over the acquisition time, Representing that the unmanned system is in A function of the probability density of the measured likelihood over the time of acquisition, 、 All represent noise covariance matrices; s42, constructing a noise covariance matrix, and the noise covariance matrix And Is modeled using an inverse weisat distribution as: , In the formula, Representation of A set of time-of-day measurement, 、、 And Is that And The degree of freedom parameter and the inverse scale matrix of the model, 、、 And (3) with For its corresponding The posterior variable of the moment of time, As a forgetting factor, 、、、 ; S43, correcting the scale matrix to obtain a corrected scale matrix And Can be expressed as: , In the formula, 、、、 Respectively represent unmanned systems in the first The first acquisition time Posterior variable of the second iteration; s44, smoothing the estimated vector according to an RTS smoothing algorithm Covariance matrix Smoothing gain : , , , And for posterior variables 、、 And Updating; S45, judging whether the results of each iteration at each sample acquisition moment meet the iteration termination condition, if the results of a certain iteration at a certain sample acquisition moment do not meet the iteration termination condition, continuing iteration until the results of the iteration meet the iteration termination condition, and outputting the obtained state vector estimated value Estimation error covariance matrix corresponding to the same 。
7. The method for locating an unmanned underground space system under multi-source interference according to claim 6, wherein said sum vs. a posterior variable 、、 And The updating is carried out by the following specific processes: , , , 。
8. The method for positioning an underground space unmanned system under multi-source interference according to claim 6, wherein the specific process is as follows: when each iteration is performed at each sample acquisition time, recording the times of each iteration at each sample acquisition time and the results after each iteration, and then: , When the above inequality is established or passed The sub-loop, when the inequality is not satisfied and does not pass And in the secondary loop, the iteration termination condition is not met.
9. An underground space unmanned system positioning system that performs the underground space unmanned system positioning method under multi-source interference of any one of claims 1-8, comprising: The data acquisition module is used for acquiring navigation information of the unmanned system in an underground space by utilizing the combined navigation system of the UWB tag and the micro-inertia measurement unit which are assembled on the unmanned system and a preset UWB ranging base station; S11, constructing a state transition matrix and a measurement matrix, and simultaneously acquiring measurement noise by using a sensor; S12, acquiring target values of motion data of the unmanned system at each acquisition time, and recording the target values as state vectors Then Representing unmanned system in the first State vectors at each acquisition time; S13, acquiring measurement values of motion data of the unmanned system at each acquisition time, and recording the measurement values as measurement vectors Then Representing unmanned system in the first Measuring vectors at each acquisition time; S14, constructing a linear state space model with control inputs of process noise, measurement noise and non-Gaussian noise: , , In the formula, Representing unmanned system in the first A state transition matrix for each acquisition instant, In order to measure the vector quantity, In order to measure the dimension of the vector, Is that The dimension of the real number is the same, , Is that The dimension of the real number is the same, , Is that The dimension of the real number is the same, As a system state vector of the system, As a dimension of the system state vector, Is that The dimension of the real number is the same, Representing unmanned system in the first The process noise at the time of the acquisition, , Representing unmanned system in the first The measurement noise at the time of the acquisition, , And All represent unmanned systems in the first Non-gaussian interference noise at the individual acquisition instants, Representing unmanned system in the first The probability that the individual acquisition instants are affected by non-gaussian noise, Representing the measured variable at the first Probability of being affected by non-Gaussian noise at each acquisition time; The filtering parameter updating module is used for acquiring prior state vectors and covariance matrixes of the system, defining a cost function in a framework of statistical similarity, and iterating the state vector mean value and the error covariance matrix at each acquisition time to obtain posterior state vectors and error covariance matrixes; the joint inference noise covariance matrix module utilizes the variable decibel leaf-based learning in a sliding window based on the obtained posterior state estimation, and the joint inference noise covariance matrix is obtained; The output module is used for completing loop iteration and outputting a position estimation result and a covariance matrix of the unmanned system.

Description

Underground space unmanned system positioning method and system under multi-source interference Technical Field The invention relates to the technical field of unmanned system positioning, in particular to a method and a system for positioning an underground space unmanned system under multi-source interference. Background In the process of positioning an unmanned system in an underground space under multi-source interference, various interference factors such as electromagnetic interference, impact vibration, dust water mist, network attack and the like are faced, the interference can obviously influence the data acquisition of a wireless positioning signal and a micro inertial sensor, so that the error characteristic of the positioning sensor presents complex non-Gaussian distribution, the positioning sensor of the unmanned system is influenced by non-Gaussian noise and unknown interference from different sources, and the position estimation information of the unmanned system can not be accurately acquired, so that the positioning error is increased, and the operation task of the system is influenced, therefore, the design of a high-robustness combined positioning method and system becomes the key for improving the autonomous operation capability of the unmanned system in the underground space under the complex environment with multi-source interference. According to the method, GNSS/UWB fusion IMU positioning is adopted in a well, UWB fusion IMU positioning is adopted in the well, a well-underground interaction area is positioned through confidence fusion, the position information of the GNSS or UWB and the IMU is input into a Kalman filter for calibration, and the original resolving position of the system is corrected. In the prior art, as disclosed in the patent application of the invention with the publication number of CN104156564A, a method for simplifying the detection rule of underground equipment based on discrete Kalman filtering is disclosed. Aiming at the scheme, the applicant discovers that the technology at least has the following technical problems that 1, the method only considers the combined positioning under the ideal Gaussian distribution error characteristic, and when the sensor error characteristic shows obvious non-Gaussian characteristic or other unknown interference influence, the positioning precision is obviously reduced, and the positioning precision cannot be ensured. 2. The data after the calibration of the Kalman filter has certain errors, and the corrected data is not estimated and updated after the calibration of the Kalman filter, so that the accuracy of the corrected data cannot be ensured. Disclosure of Invention Aiming at the technical defects, the invention aims to provide an underground space unmanned system positioning method and system under multi-source interference. In order to solve the technical problems, the invention adopts the following technical scheme: in a first aspect, the present invention provides a method for positioning an unmanned system in an underground space under multi-source interference, comprising the steps of: step one, data acquisition, namely acquiring navigation information of the unmanned system in a subsurface space by utilizing a combined navigation system of UWB labels and micro-inertia measurement units which are assembled on the unmanned system and a preset UWB ranging base station. And step two, constructing a system model, namely acquiring various motion data of the unmanned system and building the system model. And step three, updating filtering parameters, namely acquiring prior state vectors and covariance matrixes of the system, defining a cost function in a framework of statistical similarity, and iterating the state vector mean value and the error covariance matrix at each acquisition time to obtain posterior state vectors and error covariance matrixes. And step four, jointly deducing a noise covariance matrix, namely, based on the obtained posterior state estimation, utilizing the variable dB leaf learning in a sliding window to jointly deduce the noise covariance matrix. And fifthly, outputting the position estimation result and the covariance matrix of the unmanned system after the loop iteration is completed. Preferably, the system model is constructed by the following steps of S11, constructing a state transition matrix and a measurement matrix, and simultaneously acquiring measurement noise by using a sensor. S12, acquiring target values of motion data of the unmanned system at each acquisition time, and referring to the target values as state vectors, and marking the state vectors as x, wherein x k represents the state vector of the unmanned system at the kth acquisition time. S13, acquiring measurement values of all motion data of the unmanned system at all acquisition moments, and referring to the measurement values as measurement vectors, and marking the measurement vectors as y, wherein y k represents a measuremen