Aircraft skin global stress sensing system based on edge intelligence and digital twinning

Abstract

The invention discloses an aircraft skin global stress sensing system based on edge intelligence and digital twinning, and relates to the technical field of intelligent operation and maintenance and digital engineering of aviation equipment, and the system comprises a parameter optimization processing module which is used for determining aircraft monitoring area division and conductor network topology, constructing a parameter optimization model, and carrying out parameter optimization processing; determining conductor network specification data of each area of the pre-embedded composite material skin; the intelligent acquisition module is used for arranging a core acquisition circuit of an acquisition substation in each area and is in butt joint with a sensing network; the edge processing sensing module is used for acquiring data of each acquisition substation based on a sensing network, carrying out airborne real-time stress alarm and feeding back the data to an aircraft administrator end; and the cloud processing module synchronizes the data of each acquisition substation and the alarm data to the digital twin platform of the aircraft to form a prediction maintenance report. According to the hierarchical network architecture provided by the invention, the number of monitoring points can be easily expanded to a ten thousand level, dead-corner-free monitoring of the skin of the whole aircraft is realized, and the flight safety and economy of the aircraft are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance and digital engineering technology for aviation equipment, specifically to an aircraft skin full-domain stress sensing system based on edge intelligence and digital twins. Background Technology

[0002] As aviation equipment develops towards longer endurance, higher reliability, and greater intelligence, the need for real-time, online, and predictive monitoring of the health status of aircraft structures, especially large-area skin, is becoming increasingly urgent. Traditional manual inspections and scheduled overhauls can no longer meet the stringent requirements of modern aviation operations for both safety and economy. Therefore, structural health monitoring technology has become crucial for ensuring flight safety and enabling condition-based maintenance.

[0003] Currently, structural health monitoring technologies applied in the aerospace field have the following main shortcomings:

[0004] Fiber optic sensing technology, represented by fiber optic grating sensors, involves complex system integration, requiring precise light sources, demodulation equipment, and a large amount of optical cables, resulting in high costs and a risk of breakage in high-strain regions. Piezoelectric sensor technology suffers from the fragility of the piezoelectric element itself, requiring adhesives for structural coupling, and its long-term reliability is greatly affected by the environment. Furthermore, its driving and acquisition systems are complex and energy-intensive. While wireless sensor network technology solves the wiring problem, its core sensor nodes still require battery power or energy harvesting, facing challenges in terms of power continuity, electromagnetic compatibility, and communication reliability in harsh airborne environments.

[0005] At the same time, current structural health monitoring technologies also share common shortcomings:

[0006] Various sensors are usually designed for specific types of damage or local areas, making it difficult to form a large-scale, unified sensor network covering the entire skin. Point-type or small array deployment methods cannot capture the overall stress field evolution and the spatiotemporal correlation of damage.

[0007] Furthermore, current structural health monitoring technology systems face enormous challenges in transmitting the massive, high-frequency data required for skin health monitoring, suffer from poor real-time performance, and data analysis is mostly limited to current state assessment and simple threshold alarms, lacking the ability to dynamically simulate and predict the initiation and propagation of damage. Summary of the Invention

[0008] The purpose of this invention is to provide an aircraft skin full-domain stress sensing system based on edge intelligence and digital twins, to solve the problems of fragmented sensing, delayed analysis, and passive maintenance in current aircraft structural health management. Specifically, it includes:

[0009] The parameter optimization processing module is used to determine the aircraft monitoring area division and conductor network topology, construct a parameter optimization model, and determine the conductor network specification data for each area pre-embedded in the composite material skin based on the parameter optimization model.

[0010] The intelligent acquisition module, based on a proven multi-channel solution, deploys the core acquisition circuit of acquisition substations in various regions and connects to the sensor network.

[0011] The edge processing and sensing module acquires data from each data acquisition substation based on the sensor network, performs real-time airborne alarms, and feeds them back to the aircraft administrator.

[0012] The cloud processing module synchronizes data and alarm data from each data collection substation to the aircraft digital twin platform to generate predictive maintenance reports.

[0013] According to the above technical solution, the method for determining the aircraft monitoring area division and conductor network topology includes:

[0014] The aircraft skin is divided into several monitoring zones, and a data acquisition substation is deployed in each monitoring zone. Each data acquisition substation manages all sensor loops within the monitoring zone.

[0015] According to the above technical solution, the construction of a parameter optimization model, and the determination of conductor network specifications for each region of the pre-embedded composite material skin based on the parameter optimization model, include:

[0016] The diameter, length, and spatial density of conductors are configured differently to suit the aerodynamic load and stress distribution characteristics of different parts of the aircraft.

[0017] In the specific deployment process, the choice between conductor diameter and length is a trade-off decision. Specifically, if priority is given to ensuring long-term stable monitoring and high reliability during the high-speed cruise phase, a larger diameter conductor (e.g., 0.2 mm) should be selected. This is because cruise is the aircraft's most important and longest-lasting state, and the aerodynamic loads are relatively stable but continuous. A large-diameter conductor can provide the most stable signal with the highest signal-to-noise ratio during this phase, which is beneficial for long-term trend analysis and fatigue damage monitoring. However, if priority is given to ensuring extreme sensitivity to weak abnormal signals during low-speed, high-maneuver phases such as takeoff and landing, a medium-diameter conductor (e.g., 0.1 mm) should be selected. This is because the structural load changes drastically during takeoff and landing, making it a damage-prone phase, and the signal is relatively weak. Choosing the diameter with optimal performance during this phase can maximize the probability of early damage identification. In terms of length, the length directly determines the "spatial average" range of the strain sensed by the sensor.

[0018] For fine-grained monitoring of localized areas (such as cracks or hole edges), shorter conductors (e.g., a few centimeters to tens of centimeters) are needed to match their dimensions to the monitored local features (such as stress concentration areas) to achieve high spatial resolution and capture local strain gradients. For regional / global average strain monitoring (such as panel tension or bending), longer conductors (e.g., tens of centimeters) can be used to cover a larger area and measure its overall or average strain state, suitable for load monitoring and overall deformation sensing. In practical operation, the required spatial resolution can be determined first based on the specific objectives of structural health monitoring (locating damage or sensing overall load) and the strain field characteristics obtained from finite element analysis, thus initially defining the range of conductor length.

[0019] In terms of spatial density, the conductor wiring density in low-stress areas of an aircraft should be higher than that in high-stress areas. High-stress areas are typically located on the lower surface of the wing (near the fuselage root), around the main landing gear bay, and at the junction of the fuselage and wing. They usually need to reach 60%-90% of the yield strength and adopt a sparse layout to minimize the weakening of the main load-bearing structure while acquiring critical signals. Low-stress areas are typically located on the top of the fuselage, the non-load-bearing area behind the nose radome, and the non-load-bearing skin at the rear of the tail. They are usually below 30% of the yield strength and adopt a higher density layout to form a sensor network, improving the ability to locate and identify the contours of random damage (such as impacts).

[0020] Based on the above technical solution, after the conductor is laid out, a quantitative relationship between the change in induced electromotive force and the skin strain is further established:

[0021] During flight, the conductor circuits within the wing skin of an aircraft continuously cut through the Earth's magnetic field, generating a motional electromotive force (EMF). When the wing deforms due to aerodynamic loads, the change in skin stress alters the resistance of the embedded conductors, thereby modulating the induced EMF signal. By precisely monitoring the changes in the induced EMF in multiple conductor circuits, the stress state and health of the wing structure can be inverted, enabling real-time detection and alarm of abnormal deformation. Based on this principle, when the aircraft skin deforms, the embedded conductors experience strain, and the rate of change in conductor resistance is calculated. When the conductor resistance changes slightly, the change in the voltage signal generated on the load is calculated, and the aircraft skin strain is then inverted based on the change in induced EMF. This approach transforms mechanical strain, which is difficult to measure directly, into easily measurable electrical signal changes, thus providing a novel monitoring paradigm.

[0022] Specific technical means include:

[0023] When a conductor loop is connected to a test circuit, its equivalent circuit is modeled as containing an induced electromotive force. Voltage source and resistor In series, the loop current... satisfy:

[0024]

[0025] in, This represents the input impedance of the measurement system, which is the change in conductor resistance caused by skin strain. When, resistance Initial resistance of the conductor and The sum of these two values ​​will modulate the output voltage as follows:

[0026]

[0027] in, Represents the change in induced electromotive force; initial resistance of the conductor The initial resistance of the guide body when it is stress-free.

[0028] The induced electromotive force V specifically refers to the electromotive force generated by the conductor circuit of the wing during aircraft flight at a velocity... When a conductor cuts through magnetic field lines of the Earth's magnetic field, the free charges within it move directionally under the influence of the Lorentz force, thus creating a potential difference across the conductor. Therefore, the conductor can be considered as being composed of several infinitesimal elements. Composition, each micro-element Each location has its own unique magnetic field vector. According to Faraday's law of electromagnetic induction, an induced electromotive force (EMF) will be generated in the conducting rod. The tiny induced EMF dV generated on each infinitesimal element is:

[0029]

[0030] The induced electromotive force V of the entire conductor is the integral of the infinitesimal induced electromotive forces generated on all infinitesimal elements:

[0031]

[0032] in: It is the velocity of the conducting rod moving in the magnetic field. Magnetic flux density The length of the conductor rod, for and The included angle, for and The included angle.

[0033] exist , , When they are perpendicular to each other, an induced electromotive force can be further generated. The calculation formula is:

[0034]

[0035] in: ω is the angular velocity.

[0036] Based on the above technical solution, since the branches are connected in parallel, the voltage is the same, and the current is inversely proportional to the resistance. When the resistance of a conductor changes due to deformation, the current in its branch will change accordingly: when the resistance increases, the current decreases; when the resistance decreases, the current increases. According to Ohm's law, the measurable voltage signal in the circuit is:

[0037]

[0038] Therefore, the aforementioned change in induced electromotive force is further formed:

[0039]

[0040] Based on the above technical solution, the change in conductor resistance can be obtained. The change in the resistance of the conductor It is related to the strain factor of the material, the initial resistance of the conductor, and the strain of the material, and is specifically expressed as follows:

[0041]

[0042] in, Indicates the strain factor of the material; This represents the initial resistance of a conductor when it is free from stress. It indicates the strain of the material.

[0043] Based on the above technical solution, the process of converting mechanical strain, which is difficult to measure directly, into electrical signal changes that are easy to measure accurately is completed.

[0044] In this application, a strain sensitivity coefficient of the system is also provided, which demonstrates the level of accuracy achieved in this application. However, it is not a limitation of this application. If a higher level of accuracy is achieved by adopting the method of this application, it should also be within the scope of protection of this application.

[0045] The strain sensitivity coefficient of the system is expressed as The unit is ;definition:

[0046]

[0047] The strain resolution of a system is equal to the ratio of its voltage resolution to its strain sensitivity coefficient. This system has a voltage resolution of 1 μV, therefore it can achieve... The strain resolution.

[0048] Furthermore, this application also includes a perception and early warning mechanism, in which the system configures a current sensor on the core acquisition circuit to monitor the current value in real time, and sets a current threshold judgment node on the edge processing perception module;

[0049] The formula for calculating the current threshold is:

[0050]

[0051] in: This is the signal-to-noise ratio coefficient (usually between 3 and 4). The standard deviation of current noise. This is the allowable stress variation. This is the current-stress sensitivity coefficient. This is the safety margin factor (usually 0.5);

[0052] Because each core acquisition circuit is connected independently in parallel, the sensor network can accurately locate the position of the deformed conductor rod and generate an alarm signal. The alarm signal includes two types of signals: one for judging the loop current at the current threshold judgment node. Exceeding the current threshold When the system is in a state of stress, a stress-sensing alarm signal is displayed; if no loop current data is received within a time threshold, a line alarm signal is displayed, wherein the time threshold is set by the system.

[0053] This application also provides a cloud processing module that acquires historical test data from the aircraft digital twin platform, and based on the historical test data and real-time data from various acquisition substations and alarm data, calls the system's internal prediction model to generate a regional prediction and maintenance report. Specific prediction models include, but are not limited to, time series prediction, regression analysis, machine learning prediction, and deep learning prediction.

[0054] Compared with the prior art, the beneficial effects of the present invention are:

[0055] 1. This application uses a massive dynamic electromotive force sensor network deployed on the aircraft body to sense skin stress in real time, transforming mechanical strain, which is difficult to measure directly, into electrical signal changes that are easy to measure accurately. At the same time, it completely eliminates the dependence on external power supply and achieves true passive monitoring.

[0056] 2. This application utilizes airborne edge intelligent nodes for localized analysis and collaborates with ground-based high-fidelity digital twins to ultimately achieve closed-loop predictive maintenance from global stress perception and real-time anomaly diagnosis to remaining life prediction.

[0057] 3. The network architecture proposed in this application allows the number of monitoring points to be easily expanded to tens of thousands, enabling full-skin monitoring without blind spots. This solves the sensor problems in the current structural health monitoring technology field and improves aircraft flight safety and economy. Attached Figure Description

[0058] Figure 1 This is a schematic diagram showing the overall technological evolution of the aircraft skin full-domain stress sensing system based on edge intelligence and digital twins, from principle verification to engineering system of the present invention.

[0059] Figure 2 This is a schematic diagram of edge and cloud data processing for the aircraft skin global stress sensing system based on edge intelligence and digital twins of the present invention.

[0060] Figure 3 This is a schematic diagram of the 32-channel scheme of the aircraft skin global stress sensing system based on edge intelligence and digital twin of the present invention;

[0061] Figure 4 This is a schematic diagram of a simulated component of the aircraft skin global stress sensing system based on edge intelligence and digital twins according to the present invention.

[0062] Figure 5 This is a schematic diagram of the rotating magnet device in the aircraft skin global stress sensing system based on edge intelligence and digital twin of the present invention.

[0063] Figure 6 This is a schematic diagram of the data of a 0.05mm diameter copper wire port at different rotation speeds in the aircraft skin global stress sensing system based on edge intelligence and digital twin of this invention;

[0064] Figure 7 This is a schematic diagram of the data of a 0.1mm diameter copper wire port at different rotation speeds in the aircraft skin global stress sensing system based on edge intelligence and digital twin of this invention;

[0065] Figure 8 This is a schematic diagram of the data of a 0.2mm diameter copper wire port at different rotation speeds in the aircraft skin global stress sensing system based on edge intelligence and digital twins of this invention. Detailed Implementation

[0066] The technical solutions of the present invention will now be described with reference to the accompanying drawings. In the embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the term "exemplary" is intended to present the concept in a concrete manner. Furthermore, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one. Figure 1 , Figure 2 As shown, this application provides an aircraft skin global stress sensing system based on edge intelligence and digital twins, which can convert mechanical strain, which is difficult to measure directly, into electrical signal changes that are easy to measure accurately. Specific details are as follows:

[0067] Example 1: This application deploys the core acquisition circuit of the acquisition substation in each region using a verified multi-channel scheme;

[0068] In this embodiment, a 32-channel scheme is used only as an example and is not intended to limit the scope of this application. Specifically, it includes:

[0069] A 32-channel synchronous acquisition system is constructed using multiple 16-bit high-precision ADC chips, achieving strict synchronous sampling with an external clock via an I2C bus. A preamplifier with low noise is used for programmable gain amplification of the signal, and the ADC is positioned as close as possible to the sensing electrodes to reduce transmission noise. An STM32 microcontroller is used to control the sampling timing, ensuring simultaneous acquisition of multiple data streams and guaranteeing signal consistency and reliability. Specific connections (e.g.) Figure 3 (As shown) includes:

[0070] The STM32 microcontroller is electrically connected to several ADC chips via the I2C bus. These ADC chips convert analog voltage signals to digital signals for processing by the STM32 microcontroller. The ADC chips are also electrically connected to signal amplifiers, which amplify small voltage signals. The STM32 microcontroller is also connected to a DC-DC converter for step-down voltage regulation to power the main controller. A communication interface is also connected for communication with the circuit board and the computer. A buzzer is connected for alarm prompts, and user buttons are connected to enable / disable the alarm function. The main control chip is an STM32F030C8T6, which is the core control unit. The development board uses a 32-bit ARM Cortex-M0 core, featuring high performance and low power consumption, suitable for multi-channel synchronous data acquisition and real-time processing tasks.

[0071] Example 2: In this application, to establish a complete theoretical chain from measurable electrical signals (changes in induced electromotive force) to target mechanical quantities (aircraft skin strain), and to derive the quantitative mapping relationship between the two, the simulation conditions are constructed as follows:

[0072] The system uses copper wires of different diameters to simulate different strain levels in the wing skin. A rotating magnet device simulates aircraft flight and the Earth's magnetic field. A uniformly magnetized ring-shaped permanent magnet is used to simulate the Earth's magnetic field. The uniform ring-shaped magnet is fixed to a high-speed angle grinder, and the simulated aircraft's rotation speed is precisely adjusted by controlling the grinder's speed. 0.4mm thick industrial-grade aluminum alloy sheets are selected and cut and shaped according to the geometric parameters of the domestically produced Type 919 wing to create an airfoil structure that conforms to aerodynamic characteristics. The entire simulation component is as follows: Figure 4 As shown;

[0073] The rotating magnet device comprises a high-performance neodymium iron boron permanent magnet with a circular hole (10cm in diameter and 10mm in aperture) as the excitation source. The magnet is fixed at the center of the angle grinder turntable, which is then mounted on the shaft of the infinitely variable speed angle grinder. This ensures the magnetic field direction is perpendicular to the plane of rotation, allowing the magnetic field lines to uniformly pass through the area containing the copper wire, creating a stable and repeatable magnetic excitation environment. To further adjust the distance between the magnet and the aluminum plate, an adjustable iron frame support system is used, with a glass plate added between the magnet and the aluminum plate as an insulating medium. This maintains the flatness of the aluminum plate surface and prevents deformation of the specimen due to magnet adsorption. The structure of the device is as follows: Figure 5 As shown.

[0074] In this embodiment, the simulation conditions also include adding a 0.05 mm thick permalloy shielding sheet between the microcontroller and the glass plate to solve the problem of external magnetic interference;

[0075] In this embodiment, the selection of conductor diameter and length adopts the principle of "optimizing the initial resistance of the conductor". With external resistor The impedance matching strategy enables and They are on the same order of magnitude.

[0076] The specific design process includes:

[0077] Based on the selected diameter Based on the initially determined length range, calculate The possible range of values ​​for .

[0078] Based on the typical input impedance of a preamplifier (such as an instrumentation amplifier) (Typically tens of kΩ to several MΩ), adjust the length ,make Falling on Within a comparable range (e.g., from several hundred Ω to several kΩ), to achieve better impedance matching.

[0079] The specific processing in this embodiment includes:

[0080] By keeping the conductor diameters (0.05mm, 0.1mm, 0.2mm) and spatial arrangement (sparse / dense) constant, the system's magnet rotation speed was varied (simulating different airspeeds). The average value and standard deviation of the induced electromotive force signal at each rotation speed were calculated, and their linear correlation and sensitivity were analyzed. Keeping the rotation speed constant, conductor loops with different diameters were replaced to explore the mapping relationship between conductor geometric parameters and the output induced electromotive force signal, providing a calibration basis for strain inversion. Specific data are as follows: Figures 6-8 As shown.

[0081] The simulation results in this embodiment have inherent errors and uncertainties. Therefore, it is necessary to analyze their sources and quantify their impact in order to identify key error sources and provide a basis for subsequent model optimization and accuracy improvement.

[0082] The error includes physical process error and measurement system error;

[0083] The physical process errors include magnetic field coupling errors and mechanical motion errors; the magnetic field coupling error refers to the error caused by the non-uniformity of the geomagnetic field and the rotating magnetic field; the mechanical motion error refers to the error caused by the fluctuation of rotational speed.

[0084] The measurement system errors include ADC quantization errors and amplifier and contact errors.

[0085] In this embodiment, it is specifically represented as follows:

[0086] Physical process error (magnetic field coupling error):

[0087] Taking the 1000 r / min operating condition as an example, the 10 sets of measured values ​​are 0.723, 0.697, 0.678, 0.702, 0.688, 0.649, 0.734, 0.686, 0.724, and 0.647 mV, respectively. The calculated average value is 0.6928 mV, and the standard deviation is 0.030 mV. The difference between the maximum and minimum values ​​is 0.734 - 0.647 = 0.087 mV. The proportion of this difference to the average value (i.e., the deviation caused by magnetic field non-uniformity) is:

[0088]

[0089] This deviation is mainly due to the 8μT / mm magnetic field gradient at the edge of the rotating magnet, which causes the magnetic field strength of the copper wire cutting at different positions to vary, which is completely consistent with the fluctuation trend of the data in the table.

[0090] Physical process error (mechanical motion error):

[0091] At a set speed of 2000 r / min, the measured speed fluctuation range was: 30 r / min, the relative fluctuation was calculated 1.5%. Combined Figure 6 (0.2mm copper wire) 1000r / min data: average value is 0.4426mV, standard deviation is 0.046mV, and the actual deviation of the effect of speed fluctuation on the induced electromotive force is:

[0092]

[0093] The difference between this value and the theoretical deviation of rotational speed fluctuation (1.5%) proves that the final measurement result is caused by the superposition and coupling of multiple error factors such as mechanical rotational speed fluctuation and spatial magnetic field inhomogeneity.

[0094] Measurement system error (ADC quantization error):

[0095] The system uses the ADS1115 (16-bit) ADC chip. Within the 0.001-3.3mV range, the quantization interval is calculated as follows:

[0096]

[0097] For the system's minimum detection signal, i.e., 1μV, the quantization error percentage is:

[0098]

[0099] Using data from 0.05mm copper wire at 1000 r / min (mean 0.0812mV, standard deviation 0.0037mV), the proportion of quantization error to the standard deviation of this data set is:

[0100]

[0101] It meets the system's measurement accuracy requirement of 1μV.

[0102] Measurement system errors (amplifier and contact errors):

[0103] Amplifier Noise: The COSINA333 instrumentation amplifier used has an input offset voltage of 25μV. At a 1μV range, the theoretical noise percentage is... However, in actual measurements, the noise was significantly suppressed due to the use of a filtering circuit. Using data from a 0.2mm copper wire at 500 r / min (average 0.3159mV, standard deviation 0.0169mV), the actual noise percentage was:

[0104]

[0105] Contact resistance error: The calculated resistance of the 0.1mm copper wire is 1.284Ω. The contact resistance between the copper wire and the microcontroller pin is 0.5Ω ± 0.1Ω. The relative deviation of the total resistance is:

[0106]

[0107] The average induced electromotive force under the condition of 500 r / min with 0.2 mm copper wire is 0.3159. The deviation of the induced electromotive force signal caused by contact resistance is approximately:

[0108]

[0109] In error propagation and composite calculations, Type A uncertainty is:

[0110] Taking a 0.05mm copper wire at 1500r / min as an example, with a sample size of n=10, the standard uncertainty is calculated as follows:

[0111]

[0112] The relative Type A uncertainty is:

[0113]

[0114] Type B uncertainty:

[0115] ADC quantization uncertainty:

[0116]

[0117] The relative percentage is 0.029%;

[0118] Uncertainty regarding magnetic field inhomogeneity:

[0119]

[0120] The relative proportion is 7.2%;

[0121] Contact resistance uncertainty:

[0122]

[0123] The relative proportion is 4.5%;

[0124] Amplifier noise uncertainty:

[0125]

[0126] The relative percentage is 3.1%.

[0127] The total relative uncertainty of Type B is:

[0128]

[0129] Combined uncertainty

[0130]

[0131] Taking confidence factor k=2, the expanded uncertainty is:

[0132]

[0133] Example 3: Based on conductor resistance variation Analysis and processing to determine material strain:

[0134] When the wing deforms due to airflow impact, the conductor rod attached to the wing panel will bend or stretch accordingly. According to the resistance formula:

[0135]

[0136] in: The resistivity of the material The effective length of the conductor rod. It represents the cross-sectional area.

[0137] If the conductor rod is stretched, the length Increase and cross-sectional area Decrease, resistance Increase. If the conductor rod is compressed, the length... Reduce and cross-sectional area Increase, resistance Reduced. According to existing technology, the resistivity of copper wire is... Taking a copper wire with a diameter of 0.1mm and a folded length of 30cm as an example, the resistance is:

[0138]

[0139] When a conductor is subjected to stress, its length Cross-sectional area and resistivity Every minute change will occur, resulting in a change in resistance. .

[0140] In this embodiment, the resistance change is determined. Relationship with strain:

[0141]

[0142] Among them, strain The following formula, which relates to length, is adopted:

[0143]

[0144] in: The initial resistance of the conductor when there is no stress. The strain factor of the material is typically about 2 for metallic conductors. This refers to the strain of the material. The stress ultimately manifests as the resistance of the conductor circuit itself. Changes ,Right now .

[0145] Taking a 30cm long, 0.1mm diameter copper wire under 5000 micro-strains (the maximum capacity an aircraft structure must withstand without failure) as an example, calculations show that:

[0146]

[0147] in: This is the original length of the copper wire. This refers to the maximum capacity that the aircraft structure must be able to withstand without damage. This refers to the amount of copper wire drawn. Given the length of the stretched copper wire, calculate the diameter after stretching based on the law of conservation of volume. :

[0148]

[0149] Further simplification yields:

[0150]

[0151] in: The original diameter of the copper wire is 0.1mm.

[0152] According to the resistance formula , :

[0153]

[0154]

[0155]

[0156]

[0157] According to the formula Substitute the numerical values ​​into the calculation:

[0158]

[0159]

[0160] Calculated With metallic conductors The theoretical value is close, and the deviation originates from the conductor resistivity during actual deformation. The minute changes caused by the piezoresistive effect.

[0161] Example 4: Demonstrating the system's strain resolution based on the system's strain sensitivity coefficient:

[0162] In this embodiment, we still use a 0.1mm diameter copper wire as an example: its initial resistance for strain factor of materials It is 2. (Design of a high input impedance amplifier), reference induced electromotive force (Data taken from 2000 r / min experiments);

[0163] The sensitivity coefficient is calculated (at this time) = ):

[0164]

[0165] The above indicates that when the skin is produced The system should theoretically be able to detect approximately [amount] under minute strain. The voltage signal change. When the system has a voltage resolution of 1 μV, its theoretical strain resolution is:

[0166]

[0167] Therefore, at the system optimization level, the approach of "enhancing" can be adopted. The two core directions, "suppressing noise" and "suppressing sound noise," are specifically demonstrated in this embodiment through the following optimization methods:

[0168] promote aspect:

[0169] Strategy 1: Select high strain factor conductor materials and test specialized resistance strain gauge wires such as constantan or Karma alloy. The strain factor of these materials... It typically ranges from 2.0 to 4.0 (copper is only 2.0), and exhibits excellent stability over a wide temperature range.

[0170] Strategy 2: Optimize the initial resistance of the conductor With external resistor Impedance matching (as described in Example 2) can be divided into the following three steps:

[0171] System modeling: establishing With initial resistance With external resistor A changing mathematical model or simulation.

[0172] optimization By selecting thinner (e.g., 0.05 mm), longer, or wires with specific resistivity, Adjust to Similar magnitude. The goal is to make It will be between several hundred ohms and several thousand ohms, instead of the current 1.28 ohms.

[0173] optimization At the front end of the signal conditioning circuit, a matching resistor is artificially and precisely connected in series. For example, if the optimized... If the impedance is 500Ω, then a 500Ω precision matching resistor can be designed and connected in series with the conductor to achieve the optimal power transfer state between the total source impedance and the input impedance of the subsequent amplifier circuit.

[0174] Regarding noise suppression:

[0175] Strategy 1: Upgrade the front-end amplifier: Select a low-temperature drift, ultra-low-noise instrumentation amplifier (such as ADI's AD8237), whose input noise density can be as low as... The performance level is far superior to general-purpose operational amplifiers. Implement composite filtering: Design a "bandpass filter" in the signal path. Set the high-pass cutoff frequency at 1-10Hz to filter out extremely low-frequency temperature drift and jitter noise; set the low-pass cutoff frequency at 10-100Hz to filter out high-frequency switching noise and radio frequency interference (RFI). This greatly suppresses out-of-band noise. Optimize the power supply: Use a low-noise LDO linear regulator for the analog circuitry and add... A high-performance filter network completely eliminates ripple noise from the switching power supply. Enhanced shielding and grounding: Critical analog circuit components are encased in a single permalloy shield, and a single-point grounding strategy is employed to prevent noise from being introduced through ground loops.

[0176] Strategy 2: Oversampling and Digital Averaging: Increasing the ADC sampling rate to well above the signal bandwidth (e.g., 1kHz), and then digitally averaging a large sample, can effectively improve the voltage resolution to well below 1μV. Adaptive Digital Filtering: Implementing digital lock-in amplification or adaptive filtering algorithms in the microcontroller. Using a reference signal synchronized with the magnet's rotational speed, an extremely narrow "signal passband" can be created, allowing only strain-related frequency components to pass through, suppressing in-band noise to the extreme. Dynamic Baseline Correction: Setting up a "reference sensor"—that is, an unstressed, identical conductive loop. Using it to monitor environmental background noise (such as magnetic field fluctuations, temperature drift) in real time, and then digitally subtracting this background from the signal of the working loop to achieve dynamic compensation.

[0177] Example 5: Generating a regional prediction and maintenance report based on the system's internal prediction model. This example uses the gradient descent model as an example for a brief demonstration:

[0178] The data input consists of regional skin strain data and the number of regional alarm data points within each time period. The data output is the predicted time of skin failure. Several sets of sample data are selected for fitting to form an initial function. This initial function can be a fitted regression function, where the predicted time of skin failure decreases as the regional skin strain data gradually increases, and the predicted time of skin failure decreases as the number of regional alarm data points within each time period increases. A loss function is set based on the initial function, and the weak learner corresponding to the minimum loss function is used as the initial weak learner for the initial training set. A negative gradient processing formula is constructed, and for each set of data input i, a negative gradient is formed. :

[0179]

[0180] in, For each set of input data values, for The corresponding loss function; The model from the previous learner is used; t represents the current iteration number; Differential;

[0181] Based on the negative gradient, a regression tree is formed, and the leaf node region of the t-th regression tree is denoted as . By using regression trees to fit the data and obtaining the best fit value, a strong learner for the weak learner is constructed based on the best fit value.

[0182]

[0183] in, This represents the strong learner obtained in the t-th iteration; The learner representing the previous round; Represents the best-fit value; j and J represent the leaf node and leaf region on the regression tree, respectively; I represents the value closest to the best-fit value. Combination, representing the decision tree fitting function for this round;

[0184] The initial function is replaced by the strong learner to generate a new prediction output. The predicted time of skin failure is obtained based on the new prediction output, and a regional prediction maintenance report is generated and fed back to the administrator.

[0185] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An aircraft skin full-domain stress sensing system based on edge intelligence and digital twin, characterized in that: The system includes: The parameter optimization processing module is used to determine the aircraft monitoring area division and conductor network topology, construct a parameter optimization model, and determine the conductor network specification data for each area pre-embedded in the composite material skin based on the parameter optimization model. The intelligent acquisition module, based on a proven multi-channel solution, deploys the core acquisition circuit of acquisition substations in various regions and connects to the sensor network. The edge processing and sensing module acquires data from each data acquisition substation based on the sensor network, performs real-time airborne alarms, and feeds them back to the aircraft administrator. The cloud processing module synchronizes data and alarm data from each data collection substation to the aircraft digital twin platform to generate predictive maintenance reports.

2. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 1, characterized in that: The methods for determining the aircraft monitoring area division and conductor network topology include: The aircraft skin is divided into several monitoring zones, and a data acquisition substation is deployed in each monitoring zone. Each data acquisition substation manages all sensor loops within the monitoring zone.

3. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 2, characterized in that: The construction of the parameter optimization model, and the determination of conductor network specifications for each region of the pre-embedded composite material skin based on the parameter optimization model, include: The diameter, length, and spatial density of conductors are configured differently to suit the aerodynamic load and stress distribution characteristics of different parts of the aircraft. The conductor wiring density in low-stress areas of an aircraft should be higher than that in high-stress areas.

4. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 1, characterized in that: Also includes: When the aircraft skin deforms, the embedded conductors will experience strain. Calculate the change in conductor resistance. When the resistance of a conductor changes, the change in the voltage signal generated on the load is calculated, thus forming a quantitative relationship between the change in induced electromotive force and the skin strain.

5. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 4, characterized in that: Also includes: When a conductor loop is connected to a test circuit, its equivalent circuit is modeled as a voltage source containing an induced electromotive force V connected in series with a resistor R. Then the loop current I satisfies: in, This represents the input impedance of the measurement system, which is the change in conductor resistance caused by skin strain. When, resistance Initial resistance of the conductor and The sum of these factors means that the change in current will modulate the output voltage, specifically: in, Represents the change in induced electromotive force; initial resistance of the conductor The initial resistance of the guide body when it is stress-free.

6. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 5, characterized in that: The induced electromotive force The calculation methods include: The conductor is considered to be composed of several infinitesimal elements. Composition, each micro-element Each location has its own unique magnetic field vector. According to Faraday's law of electromagnetic induction, an induced electromotive force (EMF) will be generated in the conducting rod. The tiny induced EMF dV generated on each infinitesimal element is: The induced electromotive force V of the entire conductor is the integral of the infinitesimal induced electromotive forces generated on all infinitesimal elements: in: It is the velocity of the conducting rod moving in the magnetic field. Magnetic flux density The length of the conductor rod, for and The included angle, for and The included angle.

7. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 5, characterized in that: The change in conductor resistance Related to the strain factor of the material and the initial resistance of the conductor And it is related to the strain of the material, specifically expressed as: in, Indicates the strain factor of the material; It indicates the strain of the material.

8. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 7, characterized in that: Also includes: Will Substituting into the formula for calculating the change in induced electromotive force, the strain sensitivity coefficient of the system is defined as follows: The unit is ; Among them, the definition = ; The strain resolution of the system is equal to the ratio of the system's voltage resolution to the system's strain sensitivity coefficient. This system can achieve... The strain resolution.

9. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 5, characterized in that: Also includes: The system is equipped with a current sensor on the core acquisition circuit to monitor the current value in real time, and a current threshold judgment node is set on the edge processing sensing module. Current threshold The calculation formula is: in: The signal-to-noise ratio coefficient. The standard deviation of current noise. This is the allowable stress variation. This is the current-stress sensitivity coefficient. This refers to the safety margin factor. Because each core acquisition circuit is connected independently in parallel, the sensor network can accurately locate the position of the deformed conductor rod and generate an alarm signal. The alarm signal includes two types of signals: one is used at the current threshold judgment node to determine if the loop current I exceeds the current threshold. When the system is in a state of stress, a stress-sensing alarm signal is displayed; if no loop current data is received within a time threshold, a line alarm signal is displayed, wherein the time threshold is set by the system.

10. The aircraft skin global stress sensing system based on edge intelligence and digital twin as described in claim 1, characterized in that: The cloud processing module specifically includes: Historical test data is acquired from the aircraft digital twin platform. Based on the historical test data and the real-time data and alarm data from each acquisition substation, the system's internal prediction model is invoked to generate a regional predictive maintenance report.

Images