Fiber bragg grating strain measurement method and system based on deep learning

Abstract

The invention discloses a fiber bragg grating strain measurement method and system based on deep learning, designs a double-grating structure and a deep learning analysis module, and separates temperature and strain components in grating wavelengths in real time. The system comprises a fiber grating sensor module, a data acquisition module, a deep learning analysis module and a display module, high-precision separation and real-time compensation of strain and temperature are realized through high-precision data acquisition and deep learning model processing, and the system is widely applied to the field of health monitoring of engineering structures such as bridges and tunnels.

Description

Technical Field

[0001] The present invention relates to the technical field of fiber grating sensors, and in particular to a fiber grating strain measurement method and system based on deep learning. Background Art

[0002] Fiber Bragg Grating (FBG) sensors are widely used in civil engineering, aerospace, energy and other fields due to their high sensitivity, corrosion resistance and anti-electromagnetic interference capabilities. However, FBG sensors are sensitive to both temperature and strain, which increases the complexity of measurement, especially in complex environments where temperature changes and mechanical vibrations can affect measurement accuracy.

[0003] Traditional temperature compensation technology (such as the dual grating method) has alleviated this problem to a certain extent, but it still faces the challenges of insufficient accuracy and poor real-time performance under the interference of multiple factors. With the rapid development of deep learning technology, the application of deep learning to the signal processing and decoupling of fiber Bragg gratings can more accurately separate the temperature and strain components, and improve the measurement accuracy and environmental adaptability. Summary of the invention

[0004] The present invention proposes a fiber Bragg grating strain measurement method and system based on deep learning. By combining the hardware design of the fiber Bragg grating with the deep learning algorithm, real-time decoupling and high-precision compensation of temperature and strain components are achieved. Specifically, the following innovations are included: Dual grating design: one grating is used to measure temperature and the other grating is used to measure strain, and the functional separation of the two is achieved through packaging and structural design.

[0005] Deep learning temperature compensation: Use multi-layer neural networks (such as dual-head CNN) to extract the features of fiber Bragg grating signals and separate the temperature and strain components.

[0006] Hardware optimization: Design multi-layer packaging for complex environments to enhance the system's anti-interference capabilities.

[0007] Data collection and processing: High-precision data collection system combined with edge computing module to ensure real-time and low latency.

[0008] Environmental adaptability: The system is optimized to maintain stable measurement in high temperature, low temperature, high humidity and high vibration environments.

[0009] To achieve the above object, the present invention is implemented by adopting the following technical solutions: In a first aspect, the present invention provides a packaging material and structure suitable for a temperature grating, comprising: Select polyimide (PI) as a material suitable for packaging temperature gratings, meeting the following requirements: High temperature sensitivity: The material's coefficient of thermal expansion (CTE) needs to be high to amplify the effect of temperature on the grating wavelength.

[0010] Low strain sensitivity: The elastic modulus of the material needs to be high to reduce the deformation effect of external stress on the optical fiber.

[0011] Environmental adaptability: It has good corrosion resistance and aging resistance, and is suitable for complex environments.

[0012] Mechanical protection: Protects optical fiber from mechanical damage and humidity.

[0013] Furthermore, the packaging design of the temperature grating needs to shield the transmission of external strain to the optical fiber, while amplifying the effect of temperature on the wavelength of the optical fiber. A layered structure of "hard shell + viscoelastic material + optical fiber fixation" is adopted; Furthermore, the temperature response is amplified: temperature changes lead to thermal expansion of the packaging shell, and the stress of the expansion of the hard shell is transmitted to the fiber Bragg grating through the intermediate viscoelastic layer, thereby amplifying the wavelength drift signal of the grating;

[0014]

[0015]

[0016]

[0017]

[0018] Finally, by choosing polyimide as the main packaging material and combining it with the layered design of "hard shell + viscoelastic material", the temperature grating achieved high sensitivity to temperature and insensitivity to strain, providing a reliable hardware foundation for the precise separation of temperature and strain.

[0019] In the second aspect, the present invention provides a high-precision data acquisition system to achieve real-time capture and high-precision measurement of the wavelength change of the double grating. The system includes the following components: a tunable narrow-band laser, a photodetector, and a data acquisition card.

[0020] Tunable narrowband laser: Tunable narrowband laser (Tunable Narrowband Laser) can emit continuous narrowband laser by adjusting the length or refractive index of the resonant cavity to change the central wavelength of the output light within a certain range; the wavelength scanning range and accuracy of the tunable narrowband laser ensure full coverage of the Bragg grating reflection wavelength, and can detect wavelength drift with picometer resolution; Photodetector: Photodetector uses the photoelectric effect to convert the received light signal into an electrical signal. The intensity change of the electrical signal reflects the reflection intensity of the spectrum; the high gain and low noise characteristics of the photodetector enable the weak reflection signal to be amplified, ensuring excellent signal quality; Data acquisition card: The data acquisition card (DAQ) receives the electrical signal output by the photodetector, converts it into a digital signal through analog-to-digital conversion (ADC) and transmits it to the computer; the data acquisition card's million-level sampling rate and 16-bit resolution ensure the accurate capture and complete storage of the reflected signal, providing a reliable data source for the deep learning module.

[0021] The third aspect is to develop a temperature compensation algorithm based on deep learning, which can separate the temperature and strain components from the fiber grating wavelength change in real time to achieve high-precision measurement. The content includes three links: data acquisition, model design and training, system deployment and real-time analysis; by completing experimental data acquisition, deep learning model design and training, and system deployment step by step, the developed temperature compensation algorithm achieves high-precision real-time separation of temperature and strain components; In a fourth aspect, the present invention provides a method for optimizing the hardware and software of a fiber grating sensor system, which can improve accuracy and computational efficiency.

[0022] Compared with the prior art, the present invention has the following beneficial effects: By choosing polyimide as the main packaging material and combining the layered design of "hard shell + viscoelastic material", the temperature grating achieves high sensitivity to temperature and insensitivity to strain, providing a reliable hardware foundation for the precise separation of temperature and strain. By selecting high-performance tunable narrow-band lasers, low-noise photodetectors and high-speed and high-resolution data acquisition cards, the designed data acquisition system can achieve high-precision and real-time capture of the wavelength changes of the dual gratings, meeting the needs of fiber Bragg grating strain and temperature measurement in complex environments; By completing experimental data collection, deep learning model design and training, and system deployment step by step, the developed temperature compensation algorithm achieves high-precision real-time separation of temperature and strain components, which is suitable for fiber Bragg grating sensing monitoring in complex environments; By optimizing the experimental design, packaging method, data acquisition frequency and deep learning model parameters, the system's accuracy and stability in complex environments such as high temperature, low temperature, high humidity, high vibration and electromagnetic interference are significantly improved to meet the needs of engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings: Figure 1 A flowchart of a fiber Bragg grating strain measurement method and system based on deep learning is provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0024] Grating writing and packaging The Bragg grating is inscribed in the optical fiber using laser writing technology. The temperature grating is encapsulated in a hard shell and a viscoelastic layer is used to isolate the strain.

[0025] Data collection and transmission The laser scans periodically, and the photodetector converts the reflected signal into an electrical signal, which is then processed by the acquisition card and transmitted to the computing module. Training and deployment of deep learning models By programming the thermostat and strain loading machine to collect large-scale temperature and strain combination data, the model training set is constructed. After optimization, the model is deployed to the edge node in a lightweight framework.

[0026] Real-time signal processing and feedback The system uses the deep learning module to infer signals in real time, outputs the results through the display module, and provides status warnings.

[0027] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.

[0028] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present invention.

[0029] The following are several implementation examples designed based on the uploaded patent document content and combined with the practical application of fiber Bragg grating and deep learning: Example 1: Bridge structure health monitoring system Target Design a bridge health monitoring system based on deep learning and fiber Bragg grating to collect bridge vibration and strain data in real time and monitor the operating status of the bridge.

[0030] System Structure 1. Sensor module: Multiple groups of fiber Bragg grating (FBG) sensors are deployed to monitor the strains of key parts of the bridge (such as supports, main beams and suspension cables).

[0031] Each set of sensors includes a temperature grating and a strain grating, and the anti-interference ability is improved through packaging design.

[0032] Data acquisition module: Equipped with high-precision tunable narrow-band laser, the scanning range covers the grating reflection wavelength.

[0033] Use photodetectors and data acquisition cards to achieve high-frequency data capture (100 Hz).

[0034] Deep Learning Modules: A two-head convolutional neural network (CNN) is used to separate temperature and strain signals in real time.

[0035] Embed the model into edge computing devices to ensure real-time performance.

[0036] Implementation steps Install sensor modules on the bridge surface.

[0037] The real-time reflection signal is acquired using an acquisition system and transmitted to an analysis device.

[0038] Analyze wavelength changes based on deep learning models and separate temperature and strain.

[0039] The analysis results are transmitted to the display and alarm system for management to make decisions.

[0040] Advantages High real-time performance: data analysis and feedback delay is less than 50 ms.

[0041] High precision: strain error is less than 2 µε.

[0042] Adapt to complex environments such as high vibration and high humidity.

[0043] Example 2: Long-term monitoring of tunnel structure Target Design a system for long-term tunnel health monitoring that can operate reliably in high humidity environments and provide accurate data on temperature and strain.

[0044] System Design Sensor packaging optimization: Polyimide material is used as the outer layer protection to enhance moisture resistance and corrosion resistance.

[0045] The middle layer is filled with silicone to isolate mechanical stress, and the inner layer of UV glue protects the fiber grating.

[0046] Deep Learning Models: Taking into account the characteristics of high humidity environment, the model input includes wavelength data after noise compensation.

[0047] The residual network module is introduced to enhance the robustness of the model in high noise environments.

[0048] Data collection and processing: The sampling frequency was optimized to 1 Hz to reduce the amount of data in long-term monitoring.

[0049] Implementation process Install packaged sensors at key locations in the tunnel, such as the vault and side walls.

[0050] The data acquisition system collects and uploads wavelength change data on a regular basis every day.

[0051] The deep learning model analyzes wavelength changes and provides long-term temperature and strain trend reports.

[0052] Advantages Low power consumption: suitable for long-term monitoring.

[0053] It has strong adaptability to high humidity environment and its data stability is better than traditional methods.

[0054] Example 3: Industrial Equipment Vibration Monitoring Target In high vibration environments, monitor the operating status of equipment and stress distribution of mechanical components to provide early warning.

[0055] System Structure Sensor arrangement: Install fiber gratings on key equipment components to monitor mechanical stress and vibration.

[0056] It adopts reinforced packaging, and the inner layer uses low modulus silicone to absorb vibration and shock.

[0057] Real-time analytics: High frequency sampling (200 Hz) to capture rapidly changing vibration signals.

[0058] The LSTM deep learning model is used to analyze vibration signal sequences and predict equipment failures.

[0059] Advantages Strong vibration resistance: Hardware and software are combined and optimized to reduce signal distortion by 90%.

[0060] Predictive analysis: detect equipment anomalies in advance based on historical data.

[0061] Example 4: Water conservancy project monitoring Target Monitor the health of dam and reservoir structures, and measure temperature and strain in high humidity and corrosive environments.

[0062] System Optimization Package design: Use multi-layer packaging, with the outer layer made of corrosion-resistant material (such as fluorinated polymer) and the middle layer using waterproof silicone.

[0063] Deep Learning: Online calibration mechanism: Regularly input known calibration data for adjustment.

[0064] Data collection: The sampling frequency was adjusted to 10 Hz to balance real-time performance and data volume.

[0065] Advantages Strong long-term stability.

[0066] Under high humidity conditions, the signal attenuation rate is reduced by 50%.

[0067] These embodiments combine specific scenarios, closely integrate hardware optimization with deep learning algorithms, and provide a systematic solution for structural health monitoring in complex environments.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A fiber Bragg grating strain measurement method based on deep learning, characterized in that: include: The fiber Bragg grating sensor module is used to collect the wavelength change signal of the dual gratings, where the strain grating is sensitive to strain and temperature, and the temperature grating is only sensitive to temperature; the wavelength change data is collected in real time through the data acquisition system and transmitted to the deep learning analysis module; The input data is analyzed using a deep learning model to separate the temperature and strain components. The strain signal is temperature compensated based on the separation results to generate corrected strain data.

2. The fiber Bragg grating strain measurement method based on deep learning according to claim 1, characterized in that: The data acquisition system comprises a tunable narrow-band laser, a photoelectric detector and a data acquisition card, and is used for capturing the reflection wavelength signal of the double grating and performing digital processing.

3. The fiber Bragg grating strain measurement method based on deep learning according to claim 1 is characterized in that: The deep learning model uses a two-headed convolutional neural network (CNN) to predict temperature change and strain change respectively, and its loss function includes a multi-task weighted optimization formula: ， 4. The fiber Bragg grating strain measurement method based on deep learning according to claim 3 is characterized in that: The model is deployed on edge computing nodes and uses a lightweight deep learning framework to achieve real-time reasoning with an inference time of less than 10 milliseconds.

5. The fiber Bragg grating strain measurement method based on deep learning according to claim 1 is characterized in that: The fiber grating sensor module adopts a double-layer packaging design, including: an outer layer made of polyimide for providing high temperature sensitivity and environmental adaptability; and a middle layer filled with silica gel for absorbing vibration and shielding mechanical strain.

6. The fiber Bragg grating strain measurement method based on deep learning according to claim 1 is characterized in that: The system regularly adjusts the parameters of the deep learning model through an online learning mechanism to adapt to changes in complex environments and ensure measurement accuracy.

7. The fiber Bragg grating strain measurement method based on deep learning according to claim 1 is characterized in that: The system uses optimized hardware and deep learning algorithms to improve measurement accuracy and real-time performance under high temperature, low temperature, high humidity, high vibration and electromagnetic interference environments.

8. A fiber Bragg grating strain measurement system based on deep learning, characterized in that: include: Fiber Bragg grating sensor module, used to collect wavelength change signals of dual gratings; Data acquisition module, used to collect and digitize the signals of the dual grating; A deep learning analysis module to separate temperature and strain components and implement real-time temperature compensation; Display module, used to output corrected strain data and provide status alarm.

9. The fiber Bragg grating strain measurement system based on deep learning according to claim 8, characterized in that: The data acquisition module comprises: Tunable narrow-band laser with high-resolution wavelength scanning capability; Photodetector, supporting high gain and low noise characteristics; Data acquisition card with megapixel sampling rate and high resolution.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images