A tunnel azimuth angle determination method based on a gyroscope and a gravity plumb line

Abstract

The present application relates to the technical field of underground engineering surveying, and particularly relates to a tunnel azimuth angle determination method based on a gyro and a gravity plumb line, which comprises the following steps: continuously outputting an absolute azimuth reference by switching a star-based and optical gyro dual reference of a ground multi-source data reference station; generating a differential compensation sequence with a microsecond-level time stamp by differentiating a homologous gyroscope; combining a gravity plumb line deviation and a Laplace equation to complete real-time reduction of astronomical azimuth to geodetic true azimuth; and completing error compensation by time alignment and dynamic extended Kalman filtering, and outputting a high-precision geodetic true azimuth. The present application eliminates MEMS gyro zero bias, random walk and temperature drift error by differential processing of a ground reference station and a homologous reference gyro underground. Astronomical azimuth is converted to a geodetic coordinate system by combining gravity plumb line deviation correction and Laplace equation reduction, and underground equipment is accurately and real-timely oriented by star-based gyro redundancy and time alignment compensation.

Description

Technical Field

[0001] This invention relates to the field of underground engineering surveying technology, specifically a method for determining the azimuth angle of a tunnel based on a gyroscope and a gravity plumb line. Background Technology

[0002] In tunnel excavation, especially in long-distance, deep-buried tunnel projects, accurate azimuth measurement is crucial for achieving high-precision breakthrough. However, traditional measurement methods have many limitations due to the complete failure of GPS signals inside the tunnel.

[0003] Chinese patent CN114370877A discloses a tunnel measurement method based on traverse lines and gyroscope orientation. This invention corrects accumulated errors in traverse measurements by setting up traverse points both on the ground and underground, and using a gyrotheodolite to measure the gyro azimuth angle of specific underground "gyro edges." While this method improves tunnel orientation accuracy to some extent, it typically relies on transporting high-precision gyroscopes from the ground to the mine for point-by-point, edge-by-edge static measurements. This process is cumbersome, time-consuming, and severely disrupts construction progress, making real-time, continuous dynamic orientation difficult to achieve. Its correction capability is limited by the finite number of gyro edges, and it cannot effectively suppress the random drift accumulated over time by low-cost sensors such as underground MEMS gyroscopes, or the systematic errors caused by complex environments.

[0004] Chinese patent CN116358536A discloses a method for determining the heading angle in IMU-GPS integrated navigation. Although it discloses a scheme that uses GPS and IMU to determine the heading angle and fuses them through Kalman filtering, it is mainly applied to vehicle or robot navigation in open environments and does not consider special working conditions such as GPS signal loss and gravity field anomalies (i.e., vertical deviation) in tunnels. This method directly uses the azimuth angle provided by GPS as the observation, but in underground engineering, there is a vertical deviation between the "astronomical azimuth angle" directly measured by the gyroscope and the "true azimuth angle in the geodetic coordinate system" required by the project due to the uneven distribution of mass inside the earth. This deviation can reach several seconds or even arcseconds in complex terrain areas such as mountain tunnels, which is a non-negligible source of error for high-precision tunneling.

[0005] Therefore, existing technologies lack a comprehensive solution that can both eliminate sensor drift in real time and correct systematic biases caused by geophysical fields. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method for determining the azimuth angle of a tunnel based on a gyroscope and the vertical line of gravity.

[0007] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a method for determining the azimuth angle of a tunnel based on a gyroscope and the vertical line of gravity, comprising: S1: Deploy a multi-source data reference station in the tunnel ground entrance area. The multi-source data reference station includes a satellite-based observation system and a static high-precision optical gyroscope. The initial true north azimuth angle is obtained in real time through the multi-source data reference station. Based on the measurement signal quality index of the satellite-based observation data, the reference source is switched between the satellite-based observation system and the static high-precision optical gyroscope to achieve continuous output of the absolute azimuth reference. S2: Deploy a reference gyroscope of the same origin as the underground flow measurement system at the ground reference station, perform differential processing on the initial true north azimuth and the real-time azimuth measurement value of the reference gyroscope to generate a differential compensation sequence for compensating for the error of the underground sensor, and send the differential compensation sequence to the underground tunneling area after marking it with an absolute timestamp. S3: The instrument astronomical azimuth and measured gravitational acceleration vector of the current measuring point are collected through the underground flow measurement system. Based on the initial calculated coordinates of the current underground measuring point and combined with the Earth's gravity field model, the equivalent gravity vertical deviation of the current measuring point is calculated. Based on the Laplace azimuth equation, the instrument astronomical azimuth is reduced to the true geodetic azimuth in the geodetic coordinate system. S4: Based on the absolute timestamp, align the differential compensation sequence sent from the ground with the geodetic true azimuth observation data after underground calculation to address the communication delay caused by the transmission of the differential compensation sequence. Perform dynamic joint adjustment through the error state extended Kalman filter model to correct the systematic and random errors of the underground sensor, and output the geodetic true azimuth after error compensation.

[0008] Furthermore, in step S1, the multi-source data reference station includes a satellite-based observation system, a static high-precision optical gyroscope, and environmental sensors; The measurement signal quality index of real-time acquisition of satellite-based observation data, when the index is higher than the preset threshold, the initial true north azimuth angle is obtained through the satellite-based observation system, and the static high-precision optical gyroscope is simultaneously calibrated and zero-biased. When this indicator falls below a preset threshold, the satellite-based observation data is cut off, and the system switches to the calibrated static high-precision optical gyroscope to output the initial true north azimuth.

[0009] Furthermore, in step S2, the reference gyroscope that is from the same source as the underground flow measurement system is a MEMS gyroscope of the same model and batch as the measuring gyroscope used in the underground flow measurement system. During differential processing, the error sequence between the reference gyroscope measurement and the initial true north azimuth is extracted. The error sequence includes the common random walk error and system drift characteristics of similar sensors. Combined with meteorological environmental data collected by environmental sensors, a differential compensation sequence is generated.

[0010] Furthermore, in step S2, the generated differential compensation sequence is stamped with an absolute timestamp at the microsecond level using the IEEE 1588PTP precise time protocol, and the timestamped differential compensation sequence is sent to the underground flow measurement system via a communication link.

[0011] Furthermore, in step S3, the calculation steps for the equivalent gravity vertical deviation are as follows: Based on the initial calculated coordinates of the underground measuring points, a high-order Earth gravity field model is used to calculate the theoretical normal gravity vector of the measuring points. By comparing the measured gravitational acceleration vector with the theoretical normal gravity vector, the meridional and trochanteric components of the gravity vertical deviation are calculated, and the equivalent gravity vertical deviation at the current measuring point is obtained.

[0012] Furthermore, in step S4, time alignment is performed using an error-state extended Kalman filter model to address the communication delay caused by differential compensation sequence transmission. The underground flow measurement system uses high-frequency historical sliding window data cached internally to enable the filter to backtrack to the timestamp corresponding to the differential compensation sequence for differential state update, and then reintegrates forward to the current time, achieving strict alignment of the time axis between the surface and underground data.

[0013] Furthermore, the satellite-based observation system is a GNSS multi-antenna direction-finding array or an astronomical telescope; The environmental sensors include meteorological sensors and vibration sensors, used to collect temperature, air pressure, and environmental vibration data.

[0014] Furthermore, the measurement signal quality indicators of the satellite-based observation data include GNSS signal-to-noise ratio and spatial geometric accuracy factor; The preset threshold is a pre-calibrated critical value that ensures the accuracy of the azimuth angle of satellite-based observation meets the benchmark requirements.

[0015] Furthermore, the underground flow measurement system includes a MEMS gyroscope array, a high-precision triaxial accelerometer, and underground environment sensors; The instrument's astronomical azimuth angle is obtained through high-frequency attitude measurement using a MEMS gyroscope array; the actual gravitational acceleration vector is measured using a high-precision triaxial accelerometer under static or uniform motion conditions; and temperature and vibration data of underground stations are collected using underground environmental sensors.

[0016] Furthermore, the higher-order Earth gravity field model adopts the EGM2008 Earth gravity field model.

[0017] Furthermore, when a communication interruption occurs, resulting in the inability to receive the differential compensation sequence transmitted from the ground, the main control unit of the underground flow measurement system activates the built-in environmental disturbance prediction model based on the Long Short-Term Memory (LSTM) network. The environmental disturbance prediction model extracts the historical differential compensation sequence received within a pre-set time period before the communication interruption. It utilizes the nonlinear mapping relationship between underground environmental physical quantities and sensor drift learned under the previous normal communication state, and combines it with the temperature and vibration data collected in real time by the underground environmental sensor to autonomously extrapolate and generate a virtual differential compensation sequence with prediction variance, so as to maintain the continuous update of underground error compensation.

[0018] Furthermore, in step S4, the error state extended Kalman filter model introduces a dynamic estimation mechanism. When processing data, it reads the vibration variance collected by the underground environment sensor in the underground flow measurement system in real time, and combines it with the predicted variance added by the differential compensation sequence sent from the ground. It then dynamically adjusts the measurement noise covariance matrix to perform optimal weight allocation and corrects the random walk error caused by vibration.

[0019] Furthermore, the error sequence is extracted by the processor of the ground reference station using the following formula: , in, for Error sequence extracted at time step The azimuth angle measured by a ground reference gyroscope of the same model at time t. The absolute true north azimuth of the satellite-based observation system at time t is given. The units of the error sequence, the measured azimuth, and the absolute true north azimuth are all degrees or radians.

[0020] Furthermore, the specific mathematical model of the Laplace azimuth equation is as follows: , in, This is the true azimuth in the reduced geodetic coordinate system. This refers to the astronomical azimuth angle of the instrument. For the calculated deviation components of the Maoyou circle, The latitude of the underground measuring point is where it is located, and The range of values ​​satisfies .

[0021] Furthermore, the method is applied to the scenarios of azimuth measurement and orientation control for tunneling and breakthrough in mountain tunnels, urban underground utility tunnels, mine roadways, and water conservancy tunnels; The underground mobile measurement system is mounted on a tunnel boring machine, a gyro theodolite, a total station, an underground mobile measurement robot, and a portable manual measurement terminal.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention effectively eliminates systematic and random errors such as zero bias, random walk, and temperature drift common to similar sensors like MEMS gyroscopes by using differential processing between a ground-based reference station and an underground co-source reference gyroscope. Simultaneously, by introducing gravity vertical deviation correction, the Laplace equation is used to accurately convert the astronomical azimuth angle measured by the instrument to a practical geodetic coordinate system, fundamentally solving the azimuth distortion problem caused by anomalies in the Earth's gravitational field. Compared to traditional methods that rely solely on single gyroscope orientation or simple combined navigation, this significantly improves accuracy.

[0023] By setting up a dual-mode redundant switching mechanism of satellite-based / gyroscope at the reference station, the continuous output of the absolute azimuth reference is ensured, avoiding reference interruption caused by satellite signal blockage or interference. Combined with the differential compensation sequence with precise timestamp and the time alignment backtracking filtering algorithm of the underground flow system, the underground tunneling equipment can obtain the compensated high-precision azimuth angle in real time, meeting the real-time control requirements of dynamic construction.

[0024] When the communication link between the surface and underground is interrupted, the LSTM-based environmental disturbance prediction model built into the underground system can use historical data and real-time environmental information to autonomously extrapolate the virtual differential compensation sequence and maintain the continuity of error correction.

[0025] Furthermore, the dynamic Kalman filter mechanism can dynamically adjust the weights based on the prediction variance of vibration sensor data and differential data, effectively suppressing measurement noise under harsh working conditions such as strong vibrations, and greatly improving the robustness and reliability of the system in complex construction environments. Attached Figure Description

[0026] Figure 1 This is a flowchart of the steps of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Example 1 This embodiment is applied to a mountain railway tunnel project. The tunnel is 18km long with a maximum burial depth of 800m. Two hard rock TBMs are used to excavate from opposite directions at the tunnel entrance and exit. The directional accuracy requirement is to meet the high-speed railway engineering specification requirement of ≤50mm for the lateral deviation of the tunnel breakthrough. The specific implementation steps are as follows: Please see Figure 1 This invention provides a method for determining the azimuth angle of a tunnel based on a gyroscope and the vertical line of gravity, comprising: S1: Deploy a multi-source data reference station in the tunnel ground entrance area. The multi-source data reference station includes a satellite-based observation system and a static high-precision optical gyroscope. The initial true north azimuth angle is obtained in real time through the multi-source data reference station. Based on the measurement signal quality index of the satellite-based observation data, the reference source is switched between the satellite-based observation system and the static high-precision optical gyroscope to achieve continuous output of the absolute azimuth reference. First, a location with a wide field of vision, stable foundation, and no strong electromagnetic interference is selected at the tunnel entrance. Millimeter-precision 2000 geodetic coordinates of this location are obtained through 72 hours of static GNSS measurements, serving as the coordinate reference for the entire orientation system. A multi-source data reference station is deployed at this location. The hardware system consists of three parts: a satellite-based observation system, a static high-precision optical gyroscope, and environmental sensors. The satellite-based observation system uses a Trimble BD982 dual-antenna GNSS direction-finding array. The baseline length of the two receiving antennas is set to 2m. The baseline direction is calibrated using a 0.5″-level Leica TS60 total station with a calibration accuracy better than 1mm. It can simultaneously receive full-frequency signals from four major satellite navigation systems: GPS, BeiDou, GLONASS, and Galileo, with a direction-finding accuracy better than 0.03°. The static high-precision optical gyroscope uses a high-performance fiber optic gyroscope with zero bias stability of 0.0005° / h, which is installed on an active vibration-isolated optical platform with a vibration isolation frequency range of 0.5Hz-200Hz. After installation, it is horizontally leveled with a leveling accuracy better than 5″ to avoid the influence of environmental vibration on the measurement accuracy of the gyroscope. The environmental sensors include a PT100 temperature sensor, a bar pressure sensor, and a piezoelectric triaxial vibration sensor, all with a sampling frequency of 10Hz. They collect temperature, bar pressure, and vibration data of the environment where the base station is located, respectively, for subsequent environmental disturbance correction of sensor errors. The pre-set threshold is a critical value to ensure that the accuracy of the azimuth angle of the satellite-based observation meets the benchmark requirements. The calibration is completed through full-condition on-site testing: After the benchmark station is deployed, a continuous 168-hour full-condition test is conducted, and the GNSS signal-to-noise ratio, spatial geometric precision factor (PDOP) value, satellite-based direction finding azimuth angle results, and the reference azimuth angle true value output by the high-precision fiber optic gyroscope are collected simultaneously. The deviation between the satellite-based direction finding results and the true value is no more than 0.08° as the benchmark accuracy qualification condition. The signal-to-noise ratio and PDOP values ​​corresponding to all test data that meet the accuracy requirements are counted. The lower limit of the signal-to-noise ratio of 36dB-Hz and the upper limit of the PDOP of 2.2 under the 95% confidence interval are taken as the critical threshold and written into the industrial control host program of the benchmark station. After the system is powered on, it acquires measurement signal quality indicators of satellite-based observation data in real time at a frequency of 30Hz, including GNSS carrier signal-to-noise ratio and PDOP value, and performs threshold comparison and reference switching operations: When the real-time signal-to-noise ratio is ≥36dB-Hz and PDOP≤2.2, the quality of the satellite-based observation data is determined to meet the benchmark requirements. The initial true north azimuth angle is obtained through a GNSS dual-antenna array and used as the absolute azimuth reference of the system. At the same time, the static high-precision optical gyroscope is calibrated online with zero bias every 10 minutes. The calibration method is as follows: with the true north azimuth angle output by the satellite as a reference, the static output data of the optical gyroscope is continuously collected for 10 minutes. The zero bias value is calculated by fitting the data using the least squares method and written into the calibration register of the gyroscope. After calibration, the zero bias stability of the fiber optic gyroscope can be improved to within 0.0003° / h. When the real-time signal-to-noise ratio is <36dB-Hz or PDOP>2.2, the quality of the satellite-based observation data is determined to be unsatisfactory. The host immediately cuts off the reference input of the satellite-based observation data and completes a smooth switch within 10ms. The initial true north azimuth angle is output using a fiber optic gyroscope that has completed zero-bias calibration. During the switch, the azimuth angle jump is less than 0.01°, maintaining the continuous output of the absolute azimuth reference and avoiding reference interruption caused by the failure of satellite-based signals due to mountain obstruction or heavy rain. Through the above steps, a 7×24-hour continuous and stable output of the absolute azimuth reference is achieved. Even under conditions of continuous 72 hours of rainy weather and complete failure of GNSS signal, it can still provide an absolute azimuth reference with an accuracy better than 0.02°, thus ensuring the reference reliability of the entire orientation system from the source. Additional explanation: For deep-buried tunnel projects with ultra-long distance and ultra-high precision requirements, the satellite-based observation system can be replaced by a 200mm aperture refracting astronomical telescope. By observing the astronomical coordinates of celestial bodies such as Polaris and the Sun, the initial true north azimuth angle is measured using the astronomical meridian method, with a measurement accuracy better than 0.005°. In scenarios where the satellite-based signal is unavailable for a long period of time, the absolute true north azimuth angle can be obtained through regular observations using the astronomical telescope, and the fiber optic gyroscope can be calibrated to ensure the long-term stability of the absolute azimuth reference. The environmental sensor uses an integrated digital weather station and a triaxial piezoelectric vibration sensor. The integrated weather station has a sampling frequency of 1Hz and can simultaneously collect temperature, humidity, air pressure and wind speed data for atmospheric refraction correction and gyroscope temperature drift compensation. The vibration sensor has a sampling frequency of 1000Hz and collects environmental vibration data in the range of 0.1Hz to 1000Hz to correct the impact of vibration on the measurement accuracy of gyroscope and accelerometer. S2: Deploy a reference gyroscope of the same origin as the underground flow measurement system at the ground reference station, perform differential processing on the initial true north azimuth and the real-time azimuth measurement value of the reference gyroscope, generate a differential compensation sequence for compensating for the error of the underground sensor, and send the differential compensation sequence to the underground tunneling area after marking it with an absolute timestamp. Inside the temperature-controlled and vibration-isolated enclosure of the ground reference station, a reference gyroscope of the same origin as the underground flow measurement system is deployed. Specifically, it is an ADIS16488MEMS inertial measurement unit of the same model, batch, and calibration process as the underground equipment. The hardware characteristics, error distribution, and temperature response characteristics of the two are completely identical, maximizing the extraction of common error features of similar sensors. The installation attitude, three-axis orthogonal installation method, and horizontal leveling accuracy of the reference gyroscope are completely identical to those of the underground equipment. The active temperature control module inside the enclosure controls the temperature at 25℃±0.5℃, which is consistent with the operating temperature range of the underground equipment. The common error sequence extraction and differential compensation sequence generation of the ground reference station's industrial-grade ARM processor, at a synchronization frequency of 100Hz, simultaneously acquire real-time azimuth measurements from the reference gyroscope. and the initial true north azimuth angle output in step S1. To ensure that the timestamps of the two data sets are completely consistent, with a time synchronization error of less than 10μs, all angular measurements are uniformly taken in radians (SI) to avoid calculation errors caused by unit conversion. The error sequence is extracted using the following formula: , in, for Error sequence extracted at time step For the same model of ground reference gyroscope The azimuth angle measured at a given time. For satellite-based observation systems The absolute true north azimuth at time t is used to determine the error sequence. This error sequence contains several common error characteristics inherent in this batch of MEMS gyroscopes, including zero-bias instability, angle random walk, and rate random walk. The original error sequence is filtered by a moving average with a window length of 10 sampling points to remove the influence of high-frequency noise and obtain a smoothed error sequence. By combining temperature, air pressure, and vibration data collected by environmental sensors, environmental disturbance correction is performed on the error sequence: the additional errors caused by environmental changes are eliminated through a pre-calibrated temperature-zero bias third-order polynomial model and a vibration-noise linear model. The temperature-zero bias model is obtained by fitting the gyroscope zero bias data with temperature changes within a temperature range of -40℃ to 85℃. The vibration-noise model is obtained by fitting the linear relationship between the gyroscope random walk error and vibration acceleration within a vibration acceleration range of 0.1g to 5g. A differential compensation sequence containing common error correction and environmental disturbance correction is generated, which can be directly used for error compensation of the same type of gyroscope underground, with a compensation efficiency of over 92%. Both the ground reference station and the underground flow measurement system are equipped with industrial-grade Ethernet switches that support the IEEE 1588 PTPv2 protocol for timestamp marking and data transmission. The ground reference station is set as the PTP master clock, equipped with an OCXO temperature-compensated crystal oscillator, with a daily stability better than 1e-9. The underground flow measurement system is set as the PTP slave clock, equipped with a TCXO temperature-compensated crystal oscillator. The two are connected through a single-mode fiber optic link deployed in the tunnel, with a master-slave clock synchronization accuracy better than 500ns. After the ground processor generates the differential compensation sequence, it immediately obtains the current absolute timestamp with a resolution of 1ns through the PTP hardware clock. The absolute timestamp and the differential compensation sequence are encapsulated into a 64-byte standard UDP data packet, of which 16 bytes store the time identifier, 16 bytes store the error value, 16 bytes store the ambient temperature and air pressure data, and 16 bytes store the vibration amplitude and check value. The data packet is then transmitted in real time to the underground flow measurement system mounted on the TBM in the tunnel at a frequency of 100Hz through the fiber optic communication link. Additional explanation: For mine roadways and small utility tunnel projects without fiber optic communication, the timestamped differential compensation sequence can be transmitted underground via a wireless LoRa communication link. The communication distance can reach 10km, meeting the communication needs of small and medium-sized underground projects. The timestamp resolution of the differential compensation sequence can be adjusted according to the accuracy requirements of the project, with a minimum of 1ms to ensure the accuracy requirements of time alignment. S3: The instrument astronomical azimuth and measured gravitational acceleration vector of the current measuring point are collected through the underground flow measurement system. Based on the initial calculated coordinates of the current underground measuring point and combined with the Earth's gravity field model, the equivalent gravity vertical deviation of the current measuring point is calculated. Based on the Laplace azimuth equation, the instrument astronomical azimuth is reduced to the true geodetic azimuth in the geodetic coordinate system. The underground flow measurement system, which collects data from underground measuring points, is directly integrated into the automatic guidance system of the TBM. The system has built-in ADIS16488 MEMS gyroscope array, a quartz flexible triaxial accelerometer with a bias stability of 5μg, a PT1000 temperature sensor, and a MEMS triaxial vibration sensor, all from the same batch as the ground reference gyroscope. It moves synchronously with the TBM and synchronizes with the PTP clock of the ground base station after power-on. During TBM constant-speed tunneling or in a stopped state, the system initiates measurements: High-frequency attitude measurement is performed at 200Hz using three orthogonally mounted MEMS gyroscope arrays. The astronomical azimuth angle of the current measuring point is calculated using quaternion attitude calculation. Three sets of readings are continuously observed and averaged to reduce the impact of random noise. When the TBM is stopped, it remains stationary for 30 seconds. The measured gravity acceleration vector of the current measuring point is acquired using a triaxial accelerometer. After removing construction vibration noise through a sliding mean filter, a smoothed measured gravity vector is obtained. Simultaneously, temperature and vibration data of the measuring point are collected in real time by underground environmental sensors for subsequent error correction. The calculation of equivalent gravity vertical deviation first obtains the rough coordinates of the current underground measuring point. These coordinates are obtained through the geodetic coordinates of the tunnel design axis corresponding to the current excavation mileage of the TBM. The coordinate accuracy is 50m, which is sufficient to meet the calculation requirements. The geodetic coordinates (longitude L, latitude B, geodetic height H) of the measuring point are converted into geocentric rectangular coordinates. Based on the converted coordinates, the EGM2008 high-order Earth gravity field model stored locally in the system is called. The order of this model is extended to 2190, and the spatial resolution is about 5km. The theoretical normal gravity vector of the measuring point is calculated by expanding the spherical harmonic function, and the calculation accuracy is better than 10μGal. By comparing the measured gravitational acceleration vector with the theoretical normal gravity vector, the meridional component ξ (north-south direction) and the ramusoidal component η (east-west direction) of the gravity vertical deviation are calculated through vector decomposition. The equivalent gravity vertical deviation of the current measuring point is obtained with a calculation accuracy better than 0.5″. This step only requires a rough coordinate with an accuracy of 100m to complete the calculation, without the need for the high-precision measuring point coordinates at the decimeter or even centimeter level required by existing technologies, which greatly reduces the accuracy requirements for the coordinates of underground measuring points. Azimuth reduction based on the Laplace azimuth equation: Based on the Laplace azimuth equation, the astronomical azimuth angle of the instrument is reduced to the true geodetic azimuth angle in the geodetic coordinate system. The specific formula is as follows: , in, This is the true azimuth in the reduced geodetic coordinate system. This refers to the astronomical azimuth angle of the instrument. For the calculated deviation components of the Maoyou circle, The latitude of the underground measuring point is where it is located, and The range of values ​​satisfies To avoid calculation anomalies such as infinitely large tangent values, all angle measurements are uniformly measured in radians to ensure calculation accuracy. In this embodiment, we assume the instrument's astronomical azimuth angle at a certain measuring point. , deviation components of the Mao-You circle Geographical latitude , Calculated True azimuth after reduction If this deviation is not corrected, it will result in a lateral deviation of about 33mm in a 10km long tunnel, which exceeds the accuracy requirements for the tunnel breakthrough of high-speed railway tunnel projects. Through the correction in this step, this system error can be completely eliminated. By calculating the gravity vertical deviation using a high-order Earth gravity field model and combining it with the Laplace azimuth equation, the astronomical azimuth measured by the underground gyroscope is converted to the true geodetic azimuth under the national geodetic coordinate system, eliminating the systematic azimuth error caused by the gravity vertical deviation and ensuring the consistency between underground measurement results and the ground coordinate system. Additional notes: The underground mobile measurement system can be designed as a portable underground directional terminal with overall dimensions of 200mm×150mm×100mm, weight ≤2kg, and protection rating of IP67. It can be adapted to harsh environments with high humidity and dust in underground mines. It can be mounted on drilling rigs and excavating equipment, or it can be manually used by surveyors for fixed-point measurements. The built-in MEMS gyroscope array adopts a three-axis orthogonal redundant installation design to achieve redundancy fault tolerance for single gyroscope failures. At the same time, the attitude measurement accuracy is improved through multi-gyroscope data fusion. S4: Based on the absolute timestamp, the differential compensation sequence transmitted from the ground and the geodetic true azimuth observation data after underground calculation are aligned to address the communication delay caused by the transmission of the differential compensation sequence. Dynamic joint adjustment is performed through the error state extended Kalman filter model to correct the systematic and random errors of the underground sensor, and the geodetic true azimuth after error compensation is output.

[0029] The STM32H743 main control unit of the underground flow measurement system has an 8MB SRAM cache and a high-frequency historical sliding window with a length of 2s and a sampling frequency of 200Hz. It caches gyroscope angle increments, accelerometer specific force, attitude calculation results, and reduced geodetic true azimuth data in real time. The window adopts a first-in-first-out circular caching mechanism.

[0030] To address the approximately 100ms communication delay caused by fiber optic transmission within the tunnel, when the underground system receives a timestamped differential compensation sequence, it first parses the absolute timestamp of the differential sequence, matches the sampling point index corresponding to that timestamp within the historical sliding window, and obtains the raw inertial data and attitude observation data corresponding to that moment. The control error state extended Kalman filter is then traced back to the moment corresponding to that timestamp, and the Kalman filter update for that moment is completed using the differential compensation sequence as the observation, yielding the optimal state estimate for that moment. Based on this optimal state estimate, combined with the raw gyroscope and accelerometer data from that moment to the current moment within the historical sliding window, the quaternion Runge-Kutta method is used for forward reintegration, recalculating the attitude and azimuth results within that time period to obtain the optimal azimuth estimate for the current moment. This achieves strict time axis alignment between the ground differential data and the underground observation data, with a time alignment accuracy better than 5ms, completely eliminating the data time misalignment problem caused by communication delay. An error-state extended Kalman filter model is used to dynamically adjust the aligned differential compensation sequence and the ground-normalized geodetic true azimuth observation data. The state vector of the filter model is 15-dimensional, including 3-dimensional position error, 3-dimensional velocity error, 3-dimensional attitude angle error, including azimuth angle error, 3-dimensional gyroscope zero bias error, and 3-dimensional accelerometer zero bias error. The observations consist of two parts: first, the azimuth error correction amount corresponding to the differential compensation sequence sent from the ground; and second, the reduced geodetic true azimuth observation value. The filtering model introduces a dynamic estimation mechanism to calculate the variance of vibration data collected by underground environmental sensors within a 1-second window in real time. Simultaneously, it reads the predicted variance added by the differential compensation sequence. During normal communication, this variance is calculated by the ground reference station; during communication interruption, it is predicted by the LSTM model. Based on these two variance values, the measurement noise covariance matrix R is dynamically adjusted. When the variance of underground vibration increases, it indicates that the accelerometer observations are severely affected by vibration. Accordingly, the measurement noise variance of the accelerometer observations should be increased and its weight reduced. When the prediction variance of the differential compensation sequence increases, it indicates that the reliability of the differential correction amount decreases. Accordingly, the measurement noise variance of the differential observations should be increased and its weight reduced. Through the above dynamic adjustment, the optimal weight allocation of the observations under different working conditions is achieved, suppressing the random walk error caused by underground construction vibration. Under strong vibration conditions, the azimuth accuracy after filtering can be improved by more than 40%. The system monitors the status of the ground communication link in real time. When no data is received from the ground within three consecutive differential compensation sequence reception periods, it is determined that the communication is interrupted and the built-in environmental disturbance prediction model based on the Long Short-Term Memory (LSTM) network is immediately activated. The LSTM model employs a three-layer network structure with 64 neurons in the hidden layer. The input layer consists of historical differential compensation sequences, temperature data, and vibration data. The output layer contains the predicted differential compensation sequence and its corresponding prediction variance. The model was pre-trained before tunnel construction. The training dataset included drift data from a MEMS gyroscope of the same model under full operating conditions within a temperature range of -20℃ to 60℃ and a vibration acceleration range of 0.1g to 5g, along with the corresponding differential compensation sequences. The dataset contained over 1 million samples, and the model's prediction error after training was less than 0.03° / h. After the model starts, it first extracts the historical differential compensation sequence and the corresponding temperature and vibration data within the 30 minutes before the communication interruption to initialize the hidden layer state of the model. Combined with the temperature and vibration data collected in real time by the current underground environment sensors, it autonomously extrapolates at a frequency of 1Hz to generate a virtual differential compensation sequence with prediction variance. This sequence is then input into the filter model to maintain continuous updates of underground error compensation. When communication is restored, the system automatically switches back to the differential compensation sequence sent from the ground. At the same time, it uses the ground data during the communication interruption to perform online incremental learning on the LSTM model to optimize the model's prediction accuracy. With this model, even if communication is interrupted for 2 hours, the orientation accuracy can still be maintained better than 0.1°, which fully meets the orientation requirements of tunnel construction and avoids construction stagnation caused by communication interruption. After the above-mentioned time alignment, filtering adjustment, and error compensation, the underground flow measurement system outputs the true azimuth of the earth after error compensation at a frequency of 50Hz in real time, which is directly input into the TBM automatic guidance system to realize real-time closed-loop control of the TBM tunneling azimuth. In this embodiment, the system has an equivalent orientation accuracy better than 0.08° / h throughout the entire process of tunneling the 18km ultra-long tunnel, and the final tunnel breakthrough lateral deviation is 28mm, which fully meets the requirements of high-speed railway engineering specifications. Additional notes: The underground mobile surveying system can interact with multiple devices, including tunnel boring machines, gyro theodolites, total stations, and underground mobile surveying robots, via standard RS485 and Ethernet interfaces. It directly outputs high-precision geodetic azimuth angles to the control systems of the corresponding devices without requiring large-scale modifications to existing equipment, demonstrating strong adaptability. For scenarios involving underground mobile surveying robots, azimuth angle measurement results can be linked with the robot's SLAM system to achieve high-precision positioning and mapping in GNSS-free underground environments. The core of this step is to solve the problem of data time misalignment caused by communication delay. Dynamic joint adjustment of multi-source data is completed through dynamic extended Kalman filtering. At the same time, continuous operation is achieved under communication interruption conditions through LSTM model, and high-precision geodetic true azimuth is output.

[0031] Example 2 This embodiment is applied to the breakthrough directional operation of deep mining roadways in metal mines. The project background is the main transport roadway project of an iron mine from the -800m level to the -1200m level. The roadway is 6.2km long and adopts two middle sections for counter-excavation. The core pain points of the operation scenario are: many branches in the underground roadway, severe electromagnetic interference, frequent blasting operations leading to easy interruption of communication links, and severe vibration in the underground environment. Traditional optical gyro theodolites are complicated to operate and take more than 30 minutes for a single measurement, which cannot meet the rapid orientation requirements after blasting and tunneling. In addition, GNSS signals are completely ineffective in the kilometer-deep underground.

[0032] Specifically, an intrinsically safe multi-source data reference station for mining is deployed at the main ventilation shaft opening of the mine's surface industrial area. The satellite-based observation system adopts a GNSS multi-antenna direction-finding array that supports BeiDou-3 full-frequency signals, adapting to the complex terrain obstructed by surrounding mountains. The static high-precision optical gyroscope uses a vibration-isolated fiber optic gyroscope to cope with environmental vibrations caused by blasting operations on the mine surface. The ground reference gyroscope and the underground mobile measurement system use the same batch of intrinsically safe industrial-grade shock-resistant MEMS gyroscopes for mining, adapting to the harsh working conditions of underground blasting vibrations. The underground mobile measurement system is integrated into an intrinsically safe portable measurement terminal for mining, meeting the explosion-proof requirements of underground gas environments. The terminal can be directly mounted on underground loaders and drilling rigs, and also supports manual fixed-point measurements by hand.

[0033] During system operation, the ground reference station ensures continuous 24 / 7 output of the absolute azimuth reference at the wellhead through intelligent switching between GNSS and fiber optic gyroscopes. Even if the GNSS signal is completely lost due to obstruction by surrounding mountains or heavy rain, the reference accuracy can still be maintained through the calibrated fiber optic gyroscopes. The differential compensation sequence generated on the ground is transmitted in real time through the underground industrial ring network. When blasting operations cause a temporary interruption of the communication link, the underground terminal immediately activates the built-in LSTM environmental disturbance prediction model, extracts the historical differential data from the 30 minutes before the interruption, and combines it with the temperature and vibration data collected in real time underground to autonomously extrapolate the virtual differential compensation sequence, maintaining the continuity of directional operations. The underground terminal measures the gravity vector using a triaxial accelerometer and calculates the gravity vertical deviation of the measuring point in real time using the EGM2008 Earth gravity field model, completing the conversion from the instrument's astronomical azimuth to the true geodetic azimuth. At the same time, it corrects the random walk error caused by blasting vibration through a dynamic filtering mechanism, and outputs the high-precision true geodetic azimuth of the tunnel excavation in real time.

[0034] In a GNSS-free environment at a depth of -1200m, this method achieves a directional accuracy better than 0.12° / h, with a single directional operation taking less than 5 minutes, resulting in an operational efficiency more than 6 times higher than that of traditional gyro theodolites. Even in the event of a 2-hour continuous communication interruption due to blasting operations, the directional accuracy can still meet the requirements for tunnel breakthrough. The final lateral deviation of the tunnel breakthrough is less than 80mm, fully meeting the breakthrough accuracy requirements for deep mining tunnels in metal mines.

[0035] Example 3 This embodiment is applied to the directional tunneling operation of a deep-buried water conveyance tunnel in a cross-basin water diversion project. The project background is a water conveyance project supporting the Yangtze River-Huaihe River water diversion project in a certain province. The single deep-buried water conveyance tunnel is 28.7km long with a maximum burial depth of 1280m. Two hard rock TBMs are used to excavate from opposite directions at the tunnel entrance and exit. The core pain points of the operation scenario are: the tunnel passes through multiple geological fault zones, underground water inrush, severe rock breaking vibration, high humidity and dust in the tunnel environment, large transmission delay and fluctuation of long-distance fiber optic communication, and the error of traditional inertial orientation method accumulates rapidly with the tunneling distance. It cannot meet the millimeter-level breakthrough accuracy requirements of a 28.7km ultra-long tunnel, nor can it adapt to the real-time guidance requirements of continuous dynamic tunneling of TBMs.

[0036] Specifically, multi-source data reference stations are deployed at the entrances and exits of the tunnel's construction adits. The satellite-based observation system employs a redundant configuration of a GNSS dual-antenna direction-finding array and an astronomical telescope. In cases of continuous rainy weather or mountain obstruction leading to substandard GNSS signal quality, the initial true north azimuth angle can be obtained through the Polaris observation method using an astronomical telescope, further enhancing the reliability of the reference under all operating conditions. The static high-precision optical gyroscope uses an ultra-high-performance fiber optic gyroscope with a zero-bias stability of 0.0003° / h, installed in a temperature-controlled and vibration-isolated room at the tunnel entrance, achieving long-term stable absolute azimuth reference output. The ground reference gyroscope and the underground flow measurement system mounted on the TBM inside the tunnel use the same batch of high-precision MEMS gyroscope arrays. The ground reference station and the system inside the tunnel achieve precise IEEE 1588PTP clock synchronization through 28km of single-mode fiber laid inside the tunnel, ensuring microsecond-level timestamp accuracy across the entire line.

[0037] The underground flow measurement system is directly integrated into the TBM's automatic guidance system host, moving synchronously with the TBM. It collects high-frequency attitude data and measured gravity acceleration vectors of the TBM in real time, and combines them with the coarse coordinates corresponding to the mileage of the tunnel's design axis. Using the locally stored EGM2008 Earth gravity field model, it calculates the gravity vertical deviation of the measurement points in real time, completing the real-time conversion from the instrument's astronomical azimuth to the true geodetic azimuth. To address the approximately 150ms communication delay caused by long-distance fiber optic transmission, a backtracking reintegration mechanism using the error state extended Kalman filter is employed to achieve strict time alignment between the ground differential compensation sequence and the observation data inside the tunnel. To address the strong vibration interference caused by the TBM breaking through hard rock, a dynamic estimation mechanism is used to read the vibration variance in real time and dynamically adjust the measurement noise covariance matrix of the filter model to correct the random walk error caused by vibration. When the tunnel crosses a fault zone, causing fiber optic communication to be interrupted, the LSTM environmental disturbance prediction model is immediately activated to extrapolate and generate a virtual differential compensation sequence, maintaining the continuous and stable operation of the TBM's automatic guidance system. This method enables continuous real-time dynamic orientation of a 28.7km ultra-long deep-buried tunnel, with an equivalent orientation accuracy better than 0.08° / h. The azimuth angle is updated in real time at a frequency of 50Hz during TBM excavation, fully meeting the automatic guidance requirements for continuous dynamic excavation of hard rock TBMs. Finally, the lateral deviation of the tunnel breakthrough is less than 30mm, which is far better than the 100mm breakthrough deviation limit required by water conservancy engineering specifications, ensuring the high-precision and smooth breakthrough of the inter-basin water diversion project.

[0038] In summary, this invention achieves long-term continuous and stable output of the absolute azimuth reference through a dual-reference redundancy switching mechanism with multiple reference stations; it extracts common errors of MEMS gyroscopes and generates differential compensation sequences through differential processing of homogeneous gyroscopes, fundamentally suppressing the accumulation of gyroscope drift errors over time; it achieves accurate conversion from astronomical azimuth to geodetic true azimuth through the EGM2008 gravity field model and the Laplace azimuth equation, eliminating systematic errors caused by gravity vertical deviation; it solves the time misalignment problem caused by communication delay through a backtracking reintegration time alignment mechanism and dynamic Kalman filtering, improving orientation accuracy in complex vibration environments; and it achieves continuous orientation operations under communication interruption conditions through an LSTM environmental disturbance prediction model, significantly improving the system's environmental adaptability and operational continuity. This invention can be widely applied to high-precision orientation operations in various underground engineering projects, effectively ensuring the breakthrough accuracy of long-distance tunnels, roadways, and tunnels.

[0039] Furthermore, this method can be applied to various tunneling and breakthrough scenarios, including mountain tunnels, urban underground utility tunnels, mine roadways, and water conservancy tunnels, for azimuth measurement and orientation control. The underground mobile measurement system of this method is mounted on tunnel boring machines, gyro theodolites, total stations, underground mobile measurement robots, and portable manual measurement terminals.

[0040] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for determining the azimuth angle of a tunnel based on a gyroscope and the vertical line of gravity, characterized in that, Includes the following steps: S1: Deploy a multi-source data reference station in the tunnel ground entrance area. The multi-source data reference station includes a satellite-based observation system and a static high-precision optical gyroscope. The initial true north azimuth angle is obtained in real time through the multi-source data reference station. Based on the measurement signal quality index of the satellite-based observation data, the reference source is switched between the satellite-based observation system and the static high-precision optical gyroscope to achieve continuous output of the absolute azimuth reference. S2: Deploy a reference gyroscope of the same origin as the underground flow measurement system at the ground reference station, perform differential processing on the initial true north azimuth and the real-time azimuth measurement value of the reference gyroscope to generate a differential compensation sequence for compensating for the error of the underground sensor, and send the differential compensation sequence to the underground tunneling area after marking it with an absolute timestamp. S3: The instrument astronomical azimuth and measured gravitational acceleration vector of the current measuring point are collected through the underground flow measurement system. Based on the initial calculated coordinates of the current underground measuring point and combined with the Earth's gravity field model, the equivalent gravity vertical deviation of the current measuring point is calculated. Based on the Laplace azimuth equation, the instrument astronomical azimuth is reduced to the true geodetic azimuth in the geodetic coordinate system. S4: Based on the absolute timestamp, align the differential compensation sequence sent from the ground with the geodetic true azimuth observation data after underground calculation to address the communication delay caused by the transmission of the differential compensation sequence. Perform dynamic joint adjustment through the error state extended Kalman filter model to correct the systematic and random errors of the underground sensor, and output the geodetic true azimuth after error compensation.

2. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 1, characterized in that, In step S1, the multi-source data reference station includes a satellite-based observation system, a static high-precision optical gyroscope, and an environmental sensor. The measurement signal quality index of real-time acquisition of satellite-based observation data, when the index is higher than the preset threshold, the initial true north azimuth angle is obtained through the satellite-based observation system, and the static high-precision optical gyroscope is simultaneously calibrated and zero-biased. When this indicator falls below a preset threshold, the satellite-based observation data is cut off, and the system switches to the calibrated static high-precision optical gyroscope to output the initial true north azimuth.

3. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 2, characterized in that, In step S2, the reference gyroscope that is from the same source as the underground flow measurement system is a MEMS gyroscope of the same model and batch as the measuring gyroscope used in the underground flow measurement system. During differential processing, the error sequence between the reference gyroscope measurement and the initial true north azimuth is extracted. The error sequence includes the common random walk error and system drift characteristics of similar sensors. Combined with meteorological environmental data collected by environmental sensors, a differential compensation sequence is generated.

4. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 1, characterized in that, In step S2, the generated differential compensation sequence is stamped with an absolute timestamp at the microsecond level using the IEEE 1588PTP precise time protocol, and the timestamped differential compensation sequence is sent to the underground flow measurement system via a communication link.

5. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 1, characterized in that, In step S3, the calculation steps for the equivalent gravity vertical deviation are as follows: Based on the initial calculated coordinates of the underground measuring points, a high-order Earth gravity field model is used to calculate the theoretical normal gravity vector of the measuring points. By comparing the measured gravitational acceleration vector with the theoretical normal gravity vector, the meridional and trochanteric components of the gravity vertical deviation are calculated, and the equivalent gravity vertical deviation at the current measuring point is obtained.

6. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 1, characterized in that, In step S4, time alignment is performed using an error-state extended Kalman filter model to address the communication delay caused by the transmission of the differential compensation sequence. The underground flow measurement system uses high-frequency historical sliding window data cached internally to enable the filter to backtrack to the timestamp corresponding to the differential compensation sequence for differential state update, and then reintegrates forward to the current time, achieving strict alignment of the time axis between the surface and underground data.

7. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 2, characterized in that, The satellite-based observation system is a GNSS multi-antenna direction-finding array or an astronomical telescope; The environmental sensors include meteorological sensors and vibration sensors, used to collect temperature, air pressure, and environmental vibration data.

8. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 2, characterized in that, The measurement signal quality indicators of the satellite-based observation data include GNSS signal-to-noise ratio and spatial geometric accuracy factor. The preset threshold is a pre-calibrated critical value that ensures the accuracy of the azimuth angle of satellite-based observation meets the benchmark requirements.

9. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 6, characterized in that, The underground flow measurement system includes a MEMS gyroscope array, a high-precision triaxial accelerometer, and an underground environment sensor. The instrument's astronomical azimuth angle is obtained through high-frequency attitude measurement using a MEMS gyroscope array; the actual gravitational acceleration vector is measured using a high-precision triaxial accelerometer under static or uniform motion conditions; and temperature and vibration data of underground stations are collected using underground environmental sensors.

10. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 5, characterized in that, The higher-order Earth gravity field model adopted is the EGM2008 Earth gravity field model.

11. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 9, characterized in that, When a communication interruption occurs, preventing the reception of differential compensation sequences transmitted from the ground, the main control unit of the underground flow measurement system activates the built-in environmental disturbance prediction model based on a long short-term memory network (LSTM). The environmental disturbance prediction model extracts the historical differential compensation sequence received within a pre-set time period before the communication interruption. It utilizes the nonlinear mapping relationship between underground environmental physical quantities and sensor drift learned under the previous normal communication state, and combines it with the temperature and vibration data collected in real time by the underground environmental sensor to autonomously extrapolate and generate a virtual differential compensation sequence with prediction variance, so as to maintain the continuous update of underground error compensation.

12. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 9, characterized in that, In step S4, the error state extended Kalman filter model introduces a dynamic estimation mechanism. When processing data, it reads the vibration variance collected by the underground environment sensor in the underground flow measurement system in real time, and combines it with the predicted variance added by the differential compensation sequence sent from the ground. It then dynamically adjusts the measurement noise covariance matrix to perform optimal weight allocation and corrects the random walk error caused by vibration.

13. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 3, characterized in that, The error sequence is extracted by the processor of the ground reference station using the following formula: ， in, for Error sequence extracted at time step The azimuth angle measured by a ground reference gyroscope of the same model at time t. The absolute true north azimuth of the satellite-based observation system at time t is given. The units of the error sequence, the measured azimuth, and the absolute true north azimuth are all degrees or radians.

14. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 1, characterized in that, The specific mathematical model of the Laplace azimuth equation is as follows: ， in, This is the true azimuth in the reduced geodetic coordinate system. This refers to the astronomical azimuth angle of the instrument. For the calculated deviation components of the Maoyou circle, The latitude of the underground measuring point is where it is located, and The range of values ​​satisfies .

15. The method for determining the tunnel azimuth angle based on a gyroscope and the vertical line of gravity according to claim 1, characterized in that, The method is applied to the azimuth measurement and orientation control scenarios of tunneling and breakthrough in mountain tunnels, urban underground utility tunnels, mine roadways, and water conservancy tunnels. The underground mobile measurement system is mounted on a tunnel boring machine, a gyro theodolite, a total station, an underground mobile measurement robot, and a portable manual measurement terminal.

Images