Cable tunnel inspection unmanned aerial vehicle system and control method

Abstract

The invention discloses a cable tunnel inspection unmanned aerial vehicle system which comprises a layered anti-electromagnetic interference machine body structure which sequentially comprises a carbon fiber composite material layer, a permalloy magnetic shielding layer and a copper-plated electronic cabin from outside to inside. Wherein the outer layer is used for shielding high-frequency electromagnetic interference; the middle layer is used for inhibiting a low-frequency power-frequency magnetic field; and the inner layer forms a Faraday cage structure for isolating residual radiation noise. The invention further discloses a control method which comprises the steps that the dynamic pressure difference and the attitude angular velocity are obtained; inputting the wind speed into the constructed LSTM neural network model, outputting a wind speed change trend in a future period, and calculating an optimal motor rotating speed adjustment amount in combination with a pre-calibrated nonlinear motor tension curve; and dynamically adjusting the deflection angle of the control surface according to the attitude angle error of the gyroscope by adopting a self-adaptive PID algorithm. According to the invention, the flight stability and control precision of the unmanned aerial vehicle in a strong disturbance environment are significantly improved.

Description

Technical Field

[0001] This invention mainly relates to the field of cable tunnel inspection technology, specifically to a cable tunnel inspection drone system and control method. Background Technology

[0002] High-voltage cable tunnels pose a significant challenge to autonomous drone inspections, with one of the core problems being intense electromagnetic interference. High-voltage cables within the tunnels radiate high-frequency electromagnetic waves ranging from 50Hz power frequency to MHz levels, causing distortion in drone flight control signal transmission, data jumps in onboard sensors, and even malfunctions. Existing electromagnetic interference mitigation solutions often employ a single shielding material, such as an all-metal casing or surface coating, which struggles to effectively cover a wide range of electromagnetic waves and suffers from impedance mismatch, a sharp drop in shielding effectiveness at specific frequencies, and a significant increase in airframe weight.

[0003] Another key challenge lies in the complex airflow field within tunnels, including stable longitudinal winds and random local eddies, which can easily induce attitude instability in UAVs. Current wind-resistant control methods mainly rely on feedback adjustment mechanisms based on gyroscopes and accelerometers. However, due to inherent sensing, computation, and execution delays in these systems, their response speed is insufficient to cope with sudden wind disturbances, resulting in lag in control effectiveness. Furthermore, existing UAV aerodynamic designs are typically optimized for open airspace and lack fuselage structural designs specifically for the unique flow fields within tunnels, further limiting their flight stability under strong disturbances. Summary of the Invention

[0004] In view of the technical problems existing in the prior art, the present invention provides a cable tunnel inspection drone system and control method that improves the flight stability of drones in strong disturbance environments.

[0005] To solve the above-mentioned technical problems, the technical solution proposed by this invention is as follows: A cable tunnel inspection drone system includes a layered electromagnetic interference-resistant airframe structure, comprising, from the outside to the inside, a carbon fiber composite material layer, a permalloy magnetic shielding layer, and a copper-plated electronic compartment; wherein the outer carbon fiber composite material layer is used to shield high-frequency electromagnetic interference; the middle permalloy magnetic shielding layer is used to suppress low-frequency power frequency magnetic fields; and the inner copper-plated electronic compartment forms a Faraday cage structure to isolate residual radiated noise.

[0006] Preferably, the layered electromagnetic interference-resistant body structure satisfies the impedance gradient condition, and the impedance of each layer satisfies the following relationship:

[0007] in The impedance of the carbon fiber composite layer; The impedance of the permalloy magnetic shielding layer; The impedance of the copper-plated electronic compartment.

[0008] Preferably, the thickness d of the permalloy layer satisfies:

[0009] Where λ is the wavelength of 50Hz.

[0010] Preferably, it also includes an anti-airflow disturbance mechanism, which includes aerodynamic rudders and vortex vanes; the aerodynamic rudders are installed at the end of the UAV arm; the vortex vanes are multiple pieces and are installed on the top of the fuselage to disrupt the separation flow generated by crosswinds.

[0011] This invention also discloses a control method for a cable tunnel inspection UAV system as described above, including feedforward-feedback composite wind-resistant control, specifically: Feedforward control: The dynamic pressure difference output by the airborne barometer and the attitude angular velocity measured by the IMU gyroscope are acquired as the raw input signals; The original input signal is fed into the constructed LSTM neural network model, which outputs a predicted value of the wind speed change trend in the future period. The controller calculates the optimal motor speed adjustment based on the predicted wind speed change trend and the pre-calibrated nonlinear motor tension curve. Feedback control: An adaptive PID algorithm is used to dynamically adjust the control surface deflection angle based on the gyroscope attitude angle error.

[0012] Preferably, the LSTM neural network model is trained through the following steps: Temporal features were extracted using a sliding window with a window length of 1 second and a step size of 0.1 second. Normalize the input data; The Huber loss function is used for model training.

[0013] Preferably, the process of obtaining the gyroscope attitude angle error is as follows: use a gyroscope to measure the current pitch angle of the UAV, and calculate the difference between the current pitch angle and the pitch angle under windless conditions to obtain the gyroscope attitude angle error.

[0014] Compared with the prior art, the advantages of the present invention are as follows: This invention achieves full-spectrum shielding by complementing the frequency bands of the materials in the layered anti-electromagnetic interference body structure, which can achieve full-band shielding from 50Hz to 1GHz (traditional solutions only cover a single frequency band, or there is impedance mismatch, and electromagnetic waves of specific frequencies are easily interfered with).

[0015] This invention employs a feedforward-feedback composite wind-resistant control strategy, forming a complete wind-resistant architecture of "feedforward prediction for coarse adjustment and feedback for real-time fine adjustment." This division of labor makes the power system operate more smoothly, while the faster-responding control surfaces are responsible for high-frequency micro-correction, ultimately significantly reducing the overall wind-resistant response delay of the system and substantially improving the flight stability and control accuracy of the UAV in strong disturbance environments. Attached Figure Description

[0016] Figure 1 This is a structural diagram of the layered electromagnetic interference-resistant body structure in an embodiment of the present invention.

[0017] Figure 2 This is a flowchart of the control method in an embodiment of the present invention.

[0018] Legend: 1. Carbon fiber composite material layer; 2. Permalloy magnetic shielding layer; 3. Copper-plated electronic compartment. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, the cable tunnel inspection drone system provided in this embodiment of the invention includes a layered anti-electromagnetic interference airframe structure, which, from the outside to the inside, includes a carbon fiber composite material layer 1, a permalloy magnetic shielding layer 2, and a copper-plated electronic compartment 3. The outer layer consists of a carbon fiber composite material layer 1 (0.5 mm thick) to shield against high-frequency interference (>1 MHz). Intermediate layer: Permalloy magnetic shielding layer 2 (μ>50,000), suppressing 50Hz power frequency magnetic field; Inner layer: Copper-plated electronic compartment 3 (Faraday cage structure) to isolate residual radiated noise.

[0021] By complementing the frequency bands of the above-mentioned materials, full-spectrum shielding can be achieved, enabling full-band shielding from 50Hz to 1GHz (traditional solutions only cover a single frequency band, or have impedance mismatch, making electromagnetic waves of specific frequencies extremely susceptible to interference).

[0022] Specifically, the wave impedance and operating principle of each of the above layers are shown in the table below:

[0023] When electromagnetic waves propagate through different materials, they encounter impedance mismatch (similar to the reflection of sound waves when they travel from air into water). The layered material design described above achieves a gradual impedance change, allowing the electromagnetic waves to "smoothly transition" rather than be abruptly reflected, thus reducing energy bounce. Specifically, the total impedance of each layer satisfies the following relationship:

[0024] in The impedance of the carbon fiber composite layer; The impedance of the permalloy magnetic shielding layer; The impedance of the copper-plated electronic compartment.

[0025] Specifically, impedance gradient is achieved through thickness optimization, where the thickness d of the permalloy layer satisfies:

[0026] Where λ is the wavelength of 50Hz, the calculated thickness of the permalloy layer is d≈0.3mm, and the thickness of the other carbon fiber composite layer 1 is 0.5mm.

[0027] Furthermore, the cable tunnel inspection drone system also includes an anti-airflow disturbance mechanism, which includes: Active control surfaces: Miniature pneumatic rudders (deflection range ±30°, conventional structure, not shown in the figure) are installed at the end of the arm and are adjusted in real time by a servo motor; Vortex generator array: Multiple sets (e.g., 6 sets) of miniature vortex vanes (5mm high, a conventional baffle structure, not shown in the figure) are arranged on the top of the fuselage to disrupt the separation flow generated by the crosswind.

[0028] like Figure 2 As shown, the control method for the cable tunnel inspection UAV system provided in this embodiment of the invention includes a dynamic electromagnetic interference response strategy, and the specific steps are as follows: The magnetic field strength is monitored in real time by an airborne triaxial magnetometer (accuracy 0.1μT). If the magnetic field strength exceeds the threshold (e.g. 50μT), redundant sensors are turned off (e.g., GPS is turned off), and only the lidar + IMU navigation is retained.

[0029] It also includes a feedforward-feedback composite wind resistance control strategy, specifically: The system acquires the dynamic pressure difference (ΔP) output by the airborne barometer and the attitude angular velocity ω measured by the IMU gyroscope at a sampling rate of 100Hz. x ,ω y As the original input signal; A lightweight LSTM neural network model is constructed. This model takes the time sequence of the aforementioned sensor data (such as a 1-second time window) as input, and after network calculation, outputs a predicted value of the wind speed change trend in the next 200ms (including dynamic pressure difference ΔPpred, attitude angular velocity ωxpred, ωypred).

[0030] Based on the predicted wind speed changes and a pre-calibrated nonlinear motor tension curve, the controller calculates the optimal motor speed adjustment. Specifically, for headwinds, the system will increase the speed to maintain attitude and position; for tailwinds, it may appropriately reduce the speed to avoid over-acceleration, thereby achieving precise energy control.

[0031] Implementation example: When the system predicts that it will encounter a longitudinal gust of 5 m / s, the feedforward controller will immediately calculate and instruct the motor speed to increase by about 15% to generate sufficient compensating pull before the main wind disturbance arrives.

[0032] Further feedback control: An adaptive PID algorithm is adopted to dynamically adjust the control surface deflection angle based on the gyroscope attitude angle error (the gyroscope is used to measure the current pitch angle of the UAV and calculate the difference between it and the pitch angle under windless conditions, and the control surface is triggered to rotate through feedback control).

[0033] For example, if the gyroscope detects a pitch angle deviation of 4°, the feedback control triggers a 20° deflection of the control surface to compensate for the deviation, thereby generating a precise aerodynamic torque to counteract wind disturbance.

[0034] By employing a feedforward-feedback composite wind-resistant control strategy, a complete wind-resistant architecture of "feedforward prediction coarse adjustment and feedback real-time fine adjustment" is formed. This division of labor makes the power system operate more smoothly, while the faster-responding control surfaces are responsible for high-frequency micro-correction, ultimately significantly reducing the overall wind-resistant response delay of the system and significantly improving the flight stability and control accuracy of the UAV in strong disturbance environments.

[0035] The LSTM wind speed prediction model uses Huber loss (δ=1.0) to balance the advantages of MSE and MAE. Its training data includes 20 sets of wind speed-attitude datasets collected in a simulated tunnel wind tunnel (covering step changes of 3-15 m / s), with Gaussian noise (σ=0.05) added to improve robustness. By converting the model to INT8 format (TensorRT tool), the volume is compressed by 75%, and the inference speed is improved by 2.1 times. Parallel computation using the GPU cores (384 CUDA cores) of the onboard Jetson Xavier NX on a drone achieves a latency of <15ms.

[0036] This invention innovatively fuses the dynamic pressure difference (ΔP) of the airborne barometer with the attitude angular velocity (ω) of the IMU in a time sequence as the model input. This positioning method is specifically designed for tunnel environments without GPS. Through a lightweight model design for embedded deployment, the described LSTM structure, quantization and acceleration scheme is a complete technical solution optimized to meet the stringent real-time requirements (latency <15ms) and computing power limitations of UAV airborne platforms.

[0037] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A cable tunnel inspection drone system, characterized in that, The structure includes a layered electromagnetic interference-resistant body structure, which consists of a carbon fiber composite material layer (1), a permalloy magnetic shielding layer (2), and a copper-plated electronic compartment (3) from the outside to the inside. The outer carbon fiber composite material layer (1) is used to shield high-frequency electromagnetic interference; the middle permalloy magnetic shielding layer (2) is used to suppress low-frequency power frequency magnetic fields; and the inner copper-plated electronic compartment (3) forms a Faraday cage structure to isolate residual radiated noise.

2. The cable tunnel inspection drone system according to claim 1, characterized in that, The layered electromagnetic interference-resistant body structure satisfies the impedance gradient condition, and the impedance of each layer satisfies the following relationship: in The impedance of the carbon fiber composite layer (1); The impedance of the permalloy magnetic shielding layer (2) is given; The impedance of the copper-plated electronic compartment (3).

3. The cable tunnel inspection drone system according to claim 2, characterized in that, The thickness d of the permalloy layer satisfies: Where λ is the wavelength of 50Hz.

4. The cable tunnel inspection drone system according to claim 1, 2, or 3, characterized in that, It also includes an anti-airflow disturbance mechanism, which includes aerodynamic rudders and vortex vanes; the aerodynamic rudders are installed at the end of the UAV arms; the vortex vanes are multiple pieces and are installed on the top of the fuselage to disrupt the separation flow generated by crosswinds.

5. A control method for a cable tunnel inspection drone system according to any one of claims 1-4, characterized in that, This includes feedforward-feedback composite wind resistance control, specifically: Feedforward control: The dynamic pressure difference output by the airborne barometer and the attitude angular velocity measured by the IMU gyroscope are acquired as the raw input signals; The original input signal is fed into the constructed LSTM neural network model, which outputs a predicted value of the wind speed change trend in the future period. The controller calculates the optimal motor speed adjustment based on the predicted wind speed change trend and the pre-calibrated nonlinear motor tension curve. Feedback control: An adaptive PID algorithm is used to dynamically adjust the control surface deflection angle based on the gyroscope attitude angle error.

6. The control method according to claim 5, characterized in that, The LSTM neural network model is trained through the following steps: Temporal features were extracted using a sliding window with a window length of 1 second and a step size of 0.1 second. Normalize the input data; The Huber loss function is used for model training.

7. The control method according to claim 5 or 6, characterized in that, The process of obtaining the gyroscope attitude angle error is as follows: use a gyroscope to measure the current pitch angle of the UAV, and calculate the difference between it and the pitch angle under windless conditions to obtain the gyroscope attitude angle error.

Images