Intelligent deviation correction method for tunneling equipment based on inertial navigation

Abstract

The present invention discloses an intelligent deviation correction method for tunneling equipment based on inertial navigation, which relates to the technical field of dynamic stress measurement and includes the following steps: integrating piezoresistive strain microsensors on the surface of the support beam of the tunneling machine; in the present invention, dynamic stress self-sensing is realized by integrating piezoresistive sensors on the support beam, and an ANFIS model is used to construct a non-linear compensation relationship to break through the limitations of traditional linear calibration; a piezoelectric-magnetostrictive composite structure is designed to convert vibration energy into electrical energy, drive the inverse piezoelectric effect and feedback the calibration model to achieve self-powered dynamic compensation; a bionic self-repairing support beam is adopted, and microcracks are repaired in real time through a shape memory material and a nanoparticle filling structure, and a superhydrophobic coating is combined to enhance environmental adaptability; the passive vibration isolation idea is subverted, and a self-sensing-self-compensation-self-repair system is constructed to significantly improve the navigation accuracy, endurance and reliability of tunneling equipment under high-frequency vibration, providing key technical support for the intelligentization of underground engineering.

Description

Technical Field

[0001] The present invention relates to the technical field of dynamic stress measurement, and specifically to an intelligent deviation correction method for tunneling equipment based on inertial navigation. Background Art

[0002] In the field of navigation for underground engineering tunneling equipment, the intelligent deviation correction technology based on inertial navigation is the mainstream technical solution, which real-time calculates the pose of the equipment and adjusts the propulsion mechanism by integrating micro-electromechanical inertial sensors (such as accelerometers, gyroscopes). Its technical path mainly includes:

[0003] A. Linear model calibration: The sensors are statically calibrated at multiple positions according to relevant standards to obtain linear parameters such as zero bias and scale factor. There is a problem that the support structure is an ideal rigid body, ignoring the non-linear deformation under dynamic stress.

[0004] B. Filtering algorithm compensation: Linear filtering algorithms such as extended Kalman filter are used to suppress vibration noise and fuse multi-source data, but they lack the ability to handle non-linear distortion.

[0005] C. Mechanical vibration isolation design: Passive vibration isolation structures such as rubber pads and metal springs are used to attenuate the influence of low-frequency vibration (<100Hz) of the equipment on the sensors, but they cannot cope with the stress coupling problem in the high-frequency band (100 - 500Hz).

[0006] Core defects and engineering bottlenecks of the existing technology:

[0007] With the application of hard rock intelligent tunneling equipment, the following technical shortcomings have been exposed under high-frequency vibration conditions:

[0008] 1. Limitations of the linear model assumption: The existing inertial navigation system relies on the linear response assumption, but the support structure of the micro-electromechanical accelerometer will undergo micro-scale non-linear deformation under high-frequency vibration, triggering a two-way coupling of stress and deformation, resulting in the sensor output signal containing non-linear harmonic components, exceeding the allowable range of linear error, reducing the signal-to-noise ratio, and affecting the convergence efficiency of the filtering algorithm.

[0009] 2. Lack of technology for dynamic stress compensation: Traditional calibration methods only correct linear parameters, and there is a positioning deviation caused by the accumulation of non-linear errors during the navigation process, which further causes the failure of the equipment control strategy, affecting the tunneling accuracy and equipment stability.

[0010] 3. Challenges to sensor reliability: The rigid support structure design does not consider the drift of structural dynamics parameters under dynamic stress. High-frequency vibration is likely to cause material fatigue damage, exacerbating the negative impact of environmental factors on sensor performance, resulting in a decrease in the stability of the output signal and an increase in equipment maintenance costs.

[0011] With the increasing requirements for the tunneling accuracy, the traditional technology fails to solve the non-linear errors caused by micro-stresses, resulting in the cumulative positioning errors being unable to meet higher standards. The long-term operation of hard rock tunneling equipment poses higher requirements for the lifespan and stability of sensors, necessitating the improvement of the sensors' anti-vibration and environmental adaptability to reduce the downtime losses caused by sensor failures. There is a gap in the suppression of non-linear errors under high-frequency vibrations in the existing technology, and an active stress compensation scheme is needed to fill the technical gap in the suppression of dynamic non-linear errors and provide key support for the intelligent tunneling of underground projects.

[0012] In view of this, an intelligent deviation correction method for tunneling equipment based on inertial navigation is provided to overcome the above problems. Summary of the Invention

[0013] The purpose of the present invention is to provide an intelligent deviation correction method for tunneling equipment based on inertial navigation to solve the problems raised in the above background technology.

[0014] To solve the above technical problems, the intelligent deviation correction method for tunneling equipment based on inertial navigation provided by the present invention includes the following steps:

[0015] Integrate piezoresistive strain micro-sensors on the surface of the support beam of the tunneling machine, amplify the output signal of the sensors through a Wheatstone bridge, and establish a linear relationship between the voltage signal and the stress.

[0016] Collect the sensor data under different vibration frequencies and stresses, preprocess it through band-pass filtering, and then input it into the ANFIS model for training to construct a non-linear relationship of stress-deformation-output.

[0017] Based on finite element simulation, obtain the centroid offset under different stresses, obtain the mathematical model of centroid offset and stress through polynomial fitting, generate a compensation signal and superimpose it on the original output of the sensor.

[0018] Adopt a piezoelectric-magnetostrictive composite structure, use piezoelectric ceramics to capture and store vibration energy, and correct the ANFIS model through the inverse piezoelectric effect and the feedback signal of magnetostrictive materials.

[0019] Through a honeycomb microporous support beam filled with shape memory polymer and silver nanoparticles, monitor the strain gradient in real time and trigger a self-repair program to repair micro-cracks.

[0020] Furthermore, the support beam is made of single-crystal silicon material, with dimensions designed as 500×100×20, the sensor resolution reaches 0.001% strain, and the amplification factor of the Wheatstone bridge is 1000.

[0021] Furthermore, the stress-strain calibration process includes: applying dynamic stress to the support beam using a high-precision material testing machine, calculating the stress according to the bending theory of the beam, and obtaining the relationship between the voltage signal and the stress through linear fitting.

[0022] Furthermore, the data acquisition covers vibration frequencies of 150 - 300 Hz and stresses of 0 - 50 MPa. The duration of each data acquisition is 10 seconds, and the sensor acquisition frequency is 2000 Hz. The passband of the band - pass filter is set to 150 - 300 Hz.

[0023] Furthermore, in the input space of the ANFIS model, the stress is divided into 5 Gaussian - type fuzzy sets with central values of 5 MPa, 15 MPa, 25 MPa, 35 MPa, and 45 MPa respectively; the vibration frequency is divided into 3 trapezoidal - type fuzzy sets. The consequent parameters are optimized using the least - squares method, and the root - mean - square error RMSE ≤ 0.01 after training.

[0024] Furthermore, finite - element simulation is carried out using ANSYS Mechanical software, and the mesh size is set to 0.5 mm.

[0025] Furthermore, in the piezoelectric - piezomagnetic composite structure, PZT - 5H is selected for the piezoelectric ceramic, Terfenol - D is selected for the magnetostrictive material, and the support beam adopts a double - structure of main beam + auxiliary beam, which is connected by a flexible hinge. The thickness of the flexible hinge is 0.5 mm, and the stiffness is 50 N / m.

[0026] Furthermore, the strain calculation formula for the inverse piezoelectric effect is , where is the electric field strength, and the PZT generates compensating strain by controlling the voltage; the magnetic - field change signal of the magnetostrictive material is fed back to the ANFIS model.

[0027] Furthermore, the bionic self - healing support beam uses a shape - memory polymer with a glass transition temperature of 50 °C. The micropores are filled with 50 - nm silver nanoparticles with a volume ratio of 10%; the surface is coated with a polytetrafluoroethylene super - hydrophobic coating with a contact angle of 165° ± 2°.

[0028] Furthermore, the self - healing trigger mechanism is as follows: when the carbon nanotube strain sensor on the surface of the support beam detects a strain gradient > 10% / mm, the SMP is heated using the waste heat of the device, causing the silver nanoparticles to migrate and repair the microcracks, and the repair time ≤ 10 minutes.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] 1. Subversive breakthrough in non - linear deformation self - sensing compensation technology:

[0031] Self - calibration is achieved by reversely utilizing deformation energy:

[0032] Breaking through the traditional rigid support + linear calibration mode, a high-resolution piezoresistive strain microsensor is integrated on the surface of the support beam of the MEMS accelerometer. Utilizing the piezoresistive effect of single-crystalline silicon material, the bending stress of the support beam is converted into an electrical signal in real time, constructing a direct mapping of deformation energy - electrical signal, without relying on external calibration equipment, and solving the problem of real-time monitoring of non-linear deformation of the support structure under high-frequency vibration.

[0033] Non-linear coupling precise modeling and dynamic compensation:

[0034] An adaptive neuro-fuzzy inference system is used to online train the non-linear mapping model of stress, vibration frequency and sensor output, breaking through the linearization assumption of the coupling relationship by traditional calibration methods, and realizing dynamic feed-forward compensation for non-linear distortion. The measured second harmonic suppression rate reaches 92%, the signal-to-noise ratio is increased from 30 dB to 45 dB, which is 15 dB higher than that of traditional linear filtering technology, significantly improving the accuracy of inertial navigation signal calculation, the overbreak is reduced from 25% to 8%, and the roadway support cost is reduced by more than 60%.

[0035] 2. Self-powered energy capture - compensation integration:

[0036] Resource utilization and closed-loop compensation of vibration energy:

[0037] A piezoelectric - piezomagnetic composite structure is designed. Through the direct piezoelectric effect, vibration energy is converted into electrical energy and stored in a supercapacitor. At the same time, the piezomagnetic effect of Terfenol-D is used to generate a magnetic field signal proportional to the stress, forming an energy capture - stress sensing - deformation compensation closed loop. Without external power supply, the device battery life is extended by 50%, solving the pain point that traditional active compensation technology relies on external power supply.

[0038] Dual-domain feedback to improve dynamic response ability:

[0039] The captured electrical energy is used to drive the PZT to generate the inverse piezoelectric effect, offsetting 85% of the bending deformation of the support beam; at the same time, the parameters of the ANFIS model are calibrated in real time through the magnetic field signal, shortening the model update period from 10 seconds to 2 seconds, and realizing the real-time response to stress changes under high-frequency vibration. The support beam adopts a double structure of main beam + auxiliary beam, while increasing the energy conversion efficiency by 32%, ensuring that the natural frequency drift of the main beam is <0.8%, taking into account both energy capture and inertial measurement accuracy.

[0040] 3. Reliability of bionic self-healing support structure:

[0041] Self-healing and self-cleaning design inspired by biological tendons

[0042] A honeycomb microporous support beam is prepared using shape memory polymer, filled with 50nm silver nanoparticles, and mimics the self-healing mechanism of biological tendons. When the strain gradient > 10% / mm triggers microcracks, the SMP is softened by heating with the waste heat of the device, and the silver nanoparticles migrate at a speed of 5μm / s to repair the cracks. After repair, the elastic modulus recovers to 95% of the original value, saving 8 hours of downtime compared to traditional sensor replacement, increasing the mean time between failures from 3000 hours to 8500 hours, and reducing the maintenance cost by 62%.

[0043] Enhanced adaptability to complex environments:

[0044] The surface of the support beam is coated with a polytetrafluoroethylene superhydrophobic coating, combined with the self-cleaning effect of the honeycomb micropores, reducing the dust adhesion by 92%; in an environment with a humidity of 98%RH and a dust concentration of 1500mg / m³, the output variance of the sensor decreases from 0.3g² to 0.09g², meeting the high-precision standard of MT / T989-2020, and breaking through the problems of fatigue failure and signal drift of traditional rigid support beams under high vibration and high dust conditions.

[0045] 4. Overall technological progress and industry value:

[0046] Subvert the traditional idea of eliminating vibration interference, convert high-frequency vibration energy into compensation energy, and reversely utilize the nonlinear deformation energy of the support structure as a feedback signal, breaking the inherent perception that rigid support is more reliable. Achieve self-healing through a bionic flexible structure, demonstrating a new thinking of non-traditional mechanical design.

[0047] Integrate technologies in multiple fields such as materials science, dynamics, and intelligent control to form an integrated solution of self-sensing - self-compensation - self-healing, providing key support for the intelligent tunneling of underground engineering.

[0048] The present invention solves the long-existing problems of nonlinear error, energy dependence, and reliability in inertial navigation systems under high-frequency vibration, realizes the technological leap from passive vibration isolation to active intelligence, and provides a new paradigm for the intelligentization of underground engineering equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the intelligent deviation correction method for tunneling equipment based on inertial navigation of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0051] Please refer to Figure 1 , the present invention provides a technical solution:

[0052] Refer to Figure 1 As shown, an embodiment of the intelligent deviation correction method for tunneling equipment based on inertial navigation:

[0053] I. Application scenario and equipment configuration:

[0054] This embodiment is applied to a certain type of cantilever hard rock tunneling machine (model: EBZ260H), which operates in a hard rock roadway of a coal mine in Shanxi. The rock hardness of the roadway is f = 12 - 14, the vibration frequency range generated during cutting is 150 - 300 Hz, and the vibration acceleration reaches 8 - 12 g. The equipment is equipped with a MEMS accelerometer (model: ADXL355, range ±50 g, noise density 20 μg / √Hz) for pose monitoring of inertial navigation.

[0055] II. Implementation steps:

[0056] (I) Integration of piezoresistive strain microsensors and stress monitoring:

[0057] 1. Sensor selection and integration:

[0058] Basis for material selection: The support beam is made of single-crystalline silicon material, which has good mechanical and electrical properties. From a mechanical perspective, the elastic modulus of single-crystalline silicon , can ensure that the support beam maintains a certain structural stability under high-frequency vibration; from an electrical perspective, its piezoresistive coefficient , meets the requirement of converting tiny strain into a detectable electrical signal.

[0059] The size of the support beam is designed to be 500 ×100 ×20 , this size maximizes the strain transfer efficiency while ensuring the structural strength.

[0060] Sensor integration process: Integrate piezoresistive strain microsensors on the surface of the support beam through microelectronic processes. The resolution of this sensor reaches 0.001% strain. Use a Wheatstone bridge to amplify the output of the sensor, and the amplification factor is 1000. The working principle of the Wheatstone bridge is based on the condition of bridge balance. When the support beam generates strain, resulting in a change in the sensor resistance, the bridge becomes unbalanced and outputs a voltage signal proportional to the resistance change. Let the power supply voltage of the bridge be , and the arm resistances are respectively , , , .

[0061] Initially , when the sensor resistance changes to When it is, the output voltage of the bridge

[0062] ;

[0063] Due to the change in resistance being related to the strain and through subsequent calculations, the output voltage related to the stress can be obtained.

[0064] 2. Stress-strain calibration:

[0065] Calibration experiment process: Apply dynamic stress to the support beam using a high-precision material testing machine (accuracy 0.1%FS). According to the bending theory of the beam, the stress:

[0066] ,

[0067] where is the bending moment, is the distance from the neutral axis, is the moment of inertia of the cross-section. For the rectangular cross-section of the support beam (width , height ), . In the experiment, gradually apply of dynamic stress and synchronously record the output voltage signal of the piezoresistive strain microsensor amplified by the Wheatstone bridge and the actual stress value.

[0068] Data processing and results: Perform linear fitting on the collected data to obtain the relationship between the voltage signal and the stress as:

[0069] ,

[0070] where and the linearity . This indicates a high linear correlation between the voltage signal and the stress, verifying the accuracy of the sensor for stress monitoring. This linear relationship enables the accurate acquisition of the stress on the support beam by measuring the voltage signal in practical applications, providing reliable data for subsequent self-sensing compensation.

[0071] (2) ANFIS non-linear model training:

[0072] 1. Data collection and preprocessing:

[0073] Data collection: Conduct data collection under the actual working conditions of the roadheader cutting hard rock. For different vibration frequencies (covering the range of ) and stress ( Range), the acquisition duration of each group of data is set to 10 seconds. Since the sensor acquisition frequency is , each group of data contains 20,000 sampling points. Through such long-term and high-frequency sampling, the dynamic characteristics of the sensor output under different working conditions can be fully captured.

[0074] Preprocessing: The acquired data is first processed by band-pass filtering. The passband of the filter is set to . This is because low-frequency noise (such as low-frequency mechanical vibrations of equipment, power supply interference, etc.) and high-frequency noise (such as high-frequency clutter in the circuit) will affect the accuracy of the sensor output. By band-pass filtering, these noises can be effectively removed, improving the data quality and enabling subsequent model training to be based on a cleaner signal.

[0075] 2. Construction of the fuzzy inference system:

[0076] Basis for dividing the input space: For stress , it is divided into 5 fuzzy sets: extremely low, low, medium, high, and extremely high. A Gaussian membership function is selected, and its expression is:

[0077] ;

[0078] where is the center of the membership function, and the corresponding stress values are taken as 5 MPa, 15 MPa, 25 MPa, 35 MPa, and 45 MPa respectively; is the standard deviation, and empirical values are taken to ensure reasonable overlap and discrimination between each fuzzy set. For the vibration frequency , it is divided into 3 fuzzy sets: low frequency, medium frequency, and high frequency. A trapezoidal membership function is adopted, and its expression is:

[0079] ;

[0080] where, for the low-frequency fuzzy set, the corresponding parameters are , ; for the medium-frequency fuzzy set , ; for the high-frequency fuzzy set , . This division method is based on the analysis of the frequency and stress distribution in the actual working conditions, and can reasonably cover the entire input space, providing an accurate basis for fuzzy inference.

[0081] Model training process: The least squares method is used to optimize the consequent parameters of the 60 fuzzy rules of the ANFIS. The goal of the least squares method is to minimize the root mean square error (RMSE) between the model output and the actual output. The calculation formula of

[0082] ;

[0083] where is the number of samples, is the model output value, is the actual measured value. After training, , meeting the requirements for model accuracy, indicating that the trained ANFIS model can accurately construct the non-linear relationship between stress-deformation-output.

[0084] (III) Generation of dynamic error compensation signal:

[0085] 1. Calculation of centroid offset:

[0086] Finite element simulation: Use the finite element simulation software ANSYS Mechanical to simulate the situation of the MEMS accelerometer under different stresses. In the simulation model, accurately set the material properties (such as the elastic modulus and Poisson's ratio of single crystal silicon) and geometric dimensions of the support beam and mass block. By applying different magnitudes of stress , simulate the deformation of the mass block to obtain the centroid offset . To ensure the simulation accuracy, the mesh size is set to , and after multiple verifications, this mesh size can accurately capture the stress-strain distribution of the structure.

[0087] Data fitting process: Perform polynomial fitting on the centroid offset data obtained from the simulation under different stresses. Let the fitting function be:

[0088] ,

[0089] Obtained by least squares fitting: , , (unit: m), fitting error < 0.5%, . This indicates that the polynomial function obtained by fitting can accurately describe the relationship between the centroid offset and stress, providing a reliable mathematical model for compensation signal generation.

[0090] 2. Generation of compensation signal:

[0091] Determination of inertial force coefficient: The inertial force coefficient is determined by the formula , where is the mass of the mass block, is the acceleration, is the centroid offset. In this embodiment, through measurement and calculation, .

[0092] Calculation and effect of compensation signal: According to the real-time stress , the compensation signal is calculated through the formula . After the compensation signal is converted by DA (16-bit precision), it is superimposed on the original output of the sensor. In actual tests, the Fourier transform is used to analyze the signal spectrum, and the measured second harmonic suppression rate reaches 92%. At the same time, the signal-to-noise ratio is increased from 30 dB to 45 dB, effectively canceling the non-linear distortion caused by high-frequency vibration and significantly improving the accuracy of the sensor output.

[0093] (4) Implementation of the piezoelectric-magnetostrictive composite structure:

[0094] 1. Integrated design of energy capture and compensation:

[0095] Material selection: The piezoelectric ceramic PZT-5H is selected, and its d31 coefficient is , and the power density can reach , which can efficiently convert mechanical energy into electrical energy under vibration. The magnetostrictive material Terfenol-D is selected, and its magnetostrictive coefficient . According to the piezomagnetic effect, the generated magnetic field change can be used to provide a magnetic field signal proportional to the stress.

[0096] Double-beam structure design and effect: The support beam adopts a double structure of main beam + auxiliary beam. The main beam is used for inertial measurement, and the auxiliary beam is specially used for energy capture. The two are connected by a flexible hinge. The thickness of the flexible hinge is designed to be , and the stiffness is . This design can not only ensure the effective transfer of the vibration energy of the auxiliary beam to the piezoelectric ceramic, but also minimize the impact on the inertial measurement accuracy of the main beam. Through actual tests, the energy conversion efficiency is increased by 32%, and the natural frequency drift of the main beam is only 0.8% < 1%, meeting the design requirements.

[0097] 2. Double-domain feedback control:

[0098] Energy capture and inverse piezoelectric effect: During vibration, PZT generates electricity through the direct piezoelectric effect, and the generated electrical energy is stored in a supercapacitor (capacity 10 mF). When stress compensation is required, the stored electrical energy is used to drive PZT to generate the inverse piezoelectric effect. According to the principle of the inverse piezoelectric effect, PZT generates strain under the action of an electric field, and its strain calculation formula is , where is the electric field strength. By controlling the voltage, the compensation strain generated by PZT is measured, and this strain can cancel 85% of the bending deformation of the support beam.

[0099] Magnetic field signal and model calibration: The magnetic field change signal generated by Terfenol-DReal-time feedback is sent to the ANFIS model to correct the stress parameters in the model. Since the magnetic field change is proportional to the stress, it can quickly and accurately reflect the stress change. Through this feedback mechanism, the model update cycle is shortened from 10 seconds to 2 seconds, greatly improving the response speed and accuracy of the model to real-time working condition changes.

[0100] (V) Implementation of the bionic self-healing support beam:

[0101] 1. Preparation of the support beam:

[0102] Materials and structure: The glass transition temperature of the shape memory polymer (SMP) is 50 °C, and a honeycomb microporous structure is formed by 3D printing technology. The pore diameter of the honeycomb microporous structure is designed to be , and the pore spacing is . This structural design increases the flexibility and self-cleaning ability of the material while ensuring a certain strength of the support beam. 50 nm silver nanoparticles are filled in the micropores, with a volume ratio of 10%. The silver nanoparticles can migrate to form a conductive path during the softening of SMP for repairing microcracks. The surface is coated with a polytetrafluoroethylene superhydrophobic coating, and the contact angle of the coating is 165° ± 2°, effectively reducing dust adhesion.

[0103] Mechanical property test: After the support beam is prepared, its mechanical properties are tested. The elastic modulus before repair is 2 GPa. When simulated microcracks are generated and repaired, the elastic modulus recovers to 1.9 GPa, reaching 95% of the original value. The dust adhesion amount is measured by the weighing method. Compared with the support beam without the superhydrophobic coating, the dust adhesion amount is reduced by 92% after coating, verifying the effectiveness of the structure and coating design.

[0104] 2. Self-healing triggering and execution:

[0105] Self-healing triggering mechanism: Carbon nanotube strain sensors with a resolution of 0.005% strain are integrated on the surface of the support beam. The strain gradient of the support beam is monitored in real time. When the strain gradient > 10% / mm, the self-healing program is triggered. This triggering threshold is determined based on the research on the strain change when microcracks occur in the support beam under actual working conditions, and it can detect the occurrence of microcracks in a timely and accurate manner.

[0106] Self-healing execution process: When the self-healing program is triggered, the SMP is heated using the waste heat of the equipment (the waste heat of the hydraulic system, temperature 60 °C) to reach the softening point. During the softening process of the SMP, the silver nanoparticles start to migrate under the action of an electric field, and the measured migration speed is 5 μm / s. For microcracks with a length ≤ 50 μm, the repair can be completed within 10 minutes. Compared with the traditional replacement of sensors, 8 hours of downtime is saved, greatly improving the operation efficiency and reliability of the equipment.

[0107] III. Control Module and Engineering Verification Data:

[0108] 1. Propulsion Mechanism Control:

[0109] Connection of the control module and algorithm: The compensated signal output by ANFIS is connected to the PLC control module (model: S7-1200). In the PLC, the stroke of the propulsion oil cylinder is adjusted in real time according to the compensated sensor signal, and the control algorithm period is set to 20 ms. By precisely controlling the stroke of the propulsion oil cylinder, the accurate adjustment of the tunneling machine pose is realized.

[0110] Overexcavation comparison data: In actual engineering applications, the overexcavation is statistically analyzed. Before adopting this intelligent deviation correction method, the overexcavation was 25%; after adopting this method, the overexcavation decreased to 8%, and the overexcavation was reduced by more than 60% compared with the traditional method, significantly improving the tunneling accuracy and reducing resource waste and subsequent roadway support costs.

[0111] 2. Long-term Operation Test:

[0112] Data on reliability improvement: Through long-term operation monitoring, the mean time between failures (MTBF) increased from 3000 hours to 8500 hours, and the maintenance cost decreased by 62%. This shows that this intelligent deviation correction method effectively improves the reliability of the equipment and reduces the downtime and maintenance costs caused by equipment failures.

[0113] Data on environmental adaptability: Under the harsh environment of 98% RH humidity and 1500 mg / m³ dust concentration, the output variance of the sensor is tested. Before adopting this method, the output variance was 0.3 g²; after adopting this method, the output variance decreased to 0.09 g², meeting the MT / T 989-2020 standard, fully verifying the adaptability and stability of this method in complex environments.

[0114] IV. Summary:

[0115] In the self-sensing compensation technology, the support structure of the MEMS sensor is combined with the piezoresistive strain sensing technology, and the piezoresistive effect of single-crystal silicon is used to realize stress monitoring, breaking through the inherent mode of traditional "rigid support + linear calibration". In the vibration energy capture-compensation integrated structure, the piezoelectric effect, magnetostrictive effect and structural dynamics are integrated, and the vibration energy is converted from the interference source into compensation energy and stress sensing signals.

[0116] Regarding the non - linear deformation of the support structure, traditional technologies regard it as interference and attempt to eliminate it. In contrast, this method reversely utilizes the deformation energy. By integrating piezoresistive strain microsensors, the deformation energy is converted into electrical signals for self - sensing compensation. This reverse thinking mode provides a brand - new idea for solving the non - linear error problem. In the design of bionic self - repairing support beams, mimicking the self - repairing mechanism of biological tendons, a structure filled with shape - memory polymers and silver nanoparticles is adopted, solving the fatigue crack problem of traditional rigid support beams.

[0117] Through this intelligent deviation correction method, significant improvements have been achieved in multiple performance indicators. The signal - to - noise ratio has increased by 15 dB, effectively improving the quality of sensor signals and enabling the inertial navigation system to calculate the device pose more accurately. The battery life has been extended by 50%, solving the pain point of existing active compensation technologies relying on external power supply and improving the autonomous operation ability of the device. The MTBF has increased by 167%, greatly enhancing the reliability of the device, reducing equipment maintenance costs and downtime. The over - excavation volume has been reduced by more than 60% compared with traditional methods, improving the tunneling accuracy, meeting the high - precision requirements of MT / T989 - 2020, and bringing significant economic benefits and improved project quality to underground engineering construction.

[0118] This method has successfully solved the non - linear error problem of the inertial navigation system under high - frequency vibration, which has long restricted the intelligent development of underground engineering tunneling equipment. Through technologies such as self - sensing compensation, vibration energy capture - compensation integration, and bionic self - repair, self - calibration of sensor output, dynamic stress compensation, and self - repair and self - cleaning of the structure have been realized, providing key support for intelligent tunneling in underground engineering and promoting the progress of industry technologies.

[0119] When dealing with high - frequency vibration, traditional technologies mainly adopt passive vibration isolation technologies, attempting to attenuate the impact of vibration on sensors through structures such as rubber pads and metal springs. In contrast, this method breaks through this traditional idea, regards vibration energy as an exploitable resource, and realizes energy capture and stress compensation through a piezoelectric - piezomagnetic composite structure, subverting the traditional concept of "eliminating vibration interference" and providing new ideas and methods for vibration control and utilization. Traditional vibration control means simply isolate or weaken vibration, without fully exploring the energy value contained in vibration and its potential connection with system performance. This method not only effectively reduces the interference of high - frequency vibration on the inertial navigation system, but also cleverly converts vibration energy into electrical energy required for system operation, realizing the reuse of energy. At the same time, through the synergistic effect with stress compensation and model calibration, the stability and accuracy of the system are further improved.

[0120] Breaking the inherent perception that "rigid supports are more reliable": In traditional mechanical design concepts, it is generally believed that rigid support structures can provide more reliable stability. However, in this method, the design of the bionic flexible support beam, inspired by biological tendon structures, adopts a honeycomb microporous structure filled with shape memory polymers and silver nanoparticles, giving full play to the nonlinear characteristics of the material. When stress causes microcracks to occur, self-healing can be achieved. This flexible support structure shows higher reliability and durability than traditional rigid supports while ensuring a certain strength. Experimental data shows that the mean time between failures of the support beam using this method has been significantly improved, and the maintenance cost has been significantly reduced, strongly proving the superiority of the flexible support structure in specific application scenarios, breaking the long-standing inherent perception, and providing a new direction for mechanical structure design.

[0121] Promoting the development of the industry: The application of this intelligent deviation correction method for tunneling equipment based on inertial navigation not only significantly improves the operation performance and reliability of roadheader in complex working conditions but also injects new vitality into the intelligent development of the entire underground engineering field. In actual engineering, high-precision tunneling operations can reduce over-excavation, lower roadway support costs, and improve construction efficiency and safety. At the same time, this method is also expected to provide reference for the technological upgrading and innovation of other related fields, promote the development of multi-disciplinary cross-integration, and drive the overall improvement of the industry's technical level.

[0122] In summary, through the application of a series of innovative technologies and unique design concepts, this method solves many problems in the existing technology, achieves unexpected technical effects, has prominent substantive features and significant progress, has important theoretical significance and practical application value, meets the creative requirements of invention patents, and brings a new solution and development opportunity to the field of intelligent deviation correction of tunneling equipment based on inertial navigation.

Claims

1. An intelligent deviation correction method for tunneling equipment based on inertial navigation, characterized in that: The following steps are involved: A piezoresistive strain microsensor is integrated on the surface of the support beam of the tunnel boring machine, and the sensor output signal is amplified by a Wheatstone bridge to establish a linear relationship between the voltage signal and the stress. The sensor data under different vibration frequencies and stresses are collected, pre-processed by bandpass filtering, and then input into the ANFIS model for training to construct the nonlinear relationship between stress, deformation and output; Based on finite element simulation, the center of mass offset under different stresses is obtained, and the mathematical model of center of mass offset and stress is obtained through polynomial fitting. The compensation signal is generated and superimposed on the original output of the sensor. The piezoelectric-piezomagnetic composite structure is adopted, piezoelectric ceramics are used to capture and store vibration energy, and the ANFIS model is corrected through the feedback signal of the inverse piezoelectric effect and magnetostrictive materials. Through the honeycomb microporous support beam filled with shape memory polymer and nanosilver particles, the strain gradient is monitored in real time to trigger the self-healing program to repair microcracks.

2. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: The support beam is made of single crystal silicon material, with a size of 500×100×20. The sensor resolution reaches 0.001% strain and the Wheatstone bridge magnification is 1000.

3. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: The stress-strain calibration process includes: using a high-precision material testing machine to apply dynamic stress to the support beam, calculating the stress based on the beam bending theory, and obtaining the relationship between the voltage signal and the stress through linear fitting.

4. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: The data collection covers vibration frequency of 150-300 Hz and stress of 0-50 MPa. Each set of data collection lasts 10 seconds, and the sensor collection frequency is 2000 Hz. The passband of the bandpass filter is set to 150-300 Hz.

5. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: In the input space of the ANFIS model, the stress is divided into five Gaussian fuzzy sets, with central values ​​of 5MPa, 15MPa, 25MPa, 35MPa, and 45MPa respectively; the vibration frequency is divided into three trapezoidal fuzzy sets, and the least squares method is used to optimize the consequent parameters. After training, the root mean square error RMSE ≤ 0.

01.

6. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: The finite element simulation was performed using ANSYS Mechanical software, and the mesh size was set to 0.5 mm.

7. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: In the piezoelectric-piezomagnetic composite structure, the piezoelectric ceramic is PZT-5H, the magnetostrictive material is Terfenol-D, and the support beam adopts a main beam + sub-beam double structure, which is connected by a flexible hinge. The thickness of the flexible hinge is 0.5mm and the stiffness is 50N / m.

8. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: The strain calculation formula for the inverse piezoelectric effect is: ,in is the electric field strength, and the PZT produces compensatory strain by controlling the voltage; the magnetic field change signal of the magnetostrictive material is fed back to the ANFIS model.

9. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: The bionic self-repairing support beam uses shape memory polymer with a glass transition temperature of 50°C. The micropores are filled with 50nm nanosilver particles, accounting for 10% of the volume; the surface is coated with a polytetrafluoroethylene superhydrophobic coating with a contact angle of 165°±2°.

10. The intelligent deviation correction method for tunneling equipment based on inertial navigation according to claim 1, characterized in that: The self-repair trigger mechanism is: when the carbon nanotube strain sensor on the surface of the support beam detects a strain gradient of >10% / mm, the waste heat from the equipment is used to heat the SMP, allowing the nanosilver particles to migrate and repair microcracks, with a repair time of ≤10 minutes.

Images