Hydrogen energy equipment hydrogen loss detection system and method based on customized ultrasonic signal

By using a customized ultrasonic signal generator and multimodal probe array, combined with machine learning algorithms, the problem of identifying material property differences in hydrogen energy equipment was solved, enabling real-time and accurate detection of hydrogen damage and improving detection efficiency and reliability.

CN120847257APending Publication Date: 2025-10-28WUHU INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510997798.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-19
Publication Date
2025-10-28

AI Technical Summary

Technical Problem

Existing detection technologies have difficulty taking into account the differences in properties of different materials, and are unable to effectively identify hydrogen-induced microcracks at the interface between carbon fiber layers and the metal liner. There are detection blind spots, and the level of intelligent signal processing is low, which cannot meet the real-time detection needs of hydrogen energy equipment in service.

Method used

A customized ultrasonic signal generator is used to generate a joint modulation signal in the time and frequency domains. Combined with a multimodal probe array and a dynamic analysis module, a machine learning algorithm is integrated for signal processing to achieve multi-source feature fusion analysis of hydrogen energy equipment.

Benefits of technology

It realizes comprehensive, accurate and real-time detection of hydrogen damage status of hydrogen energy equipment, improves detection efficiency and reliability, reduces missed detection and misjudgment rates, and adapts to complex structures and environmental changes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120847257A_ABST
    Figure CN120847257A_ABST
Patent Text Reader

Abstract

The invention discloses a hydrogen energy equipment hydrogen loss detection system and method based on a customized ultrasonic signal, and the system comprises an ultrasonic signal generator which generates a customized ultrasonic signal according to the acoustic impedance characteristic of a material, and achieves the differential recognition of hydrogen-induced microcracks of a carbon fiber and a metal liner through time-frequency domain joint modulation and optimization; the ultrasonic signal detector is used for receiving reflection, transmission and scattering signals by adopting a multi-mode probe array so as to realize accurate positioning and detection of hydrogen damage of different parts; the dynamic analysis module integrates a machine learning algorithm to process signal characteristics and outputs a damage evaluation result; the method comprises the steps of generating and transmitting a customized ultrasonic signal to equipment, receiving a response signal, performing noise reduction and feature extraction, identifying a hydrogen damage state based on multi-source signal fusion and feature database matching, visually displaying a result and providing maintenance decision support. Through combination of customized signals, multi-modal detection and intelligent analysis, the detection efficiency and reliability are remarkably improved, and the method is used for safety maintenance of hydrogen energy equipment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention specifically relates to a hydrogen loss detection system and method for hydrogen energy equipment based on customized ultrasonic signals, belonging to the field of hydrogen energy equipment detection technology. Background Technology

[0002] As a clean and efficient energy source, the safety of hydrogen-powered equipment is of paramount importance. Hydrogen damage (such as hydrogen-induced cracking and interfacial debonding) significantly affects the reliability of hydrogen-powered equipment constructed from carbon fiber composites and metal liner. Existing detection technologies face multiple challenges: current conventional ultrasonic testing methods struggle to account for the differences in material properties and lack the ability to identify hydrogen-induced microcrack scattering signals at the carbon fiber interlayer interface and metal liner, easily leading to missed detections or misjudgments; traditional single-probe testing modes have blind spots and poor adaptability to curved components and materials of varying thicknesses; signal processing relies on manual feature extraction and lacks intelligent fusion analysis of multi-source features in the time, frequency, and nonlinear domains, making it difficult to meet the real-time testing requirements of in-service hydrogen-powered equipment in terms of detection efficiency and accuracy.

[0003] Therefore, overcoming the problems of insufficient differential detection capability for hydrogen damage in multi-material structures and low accuracy of multi-modal signal fusion analysis in existing technologies, and realizing comprehensive, accurate and real-time detection of hydrogen damage status of hydrogen energy equipment, is a key technical problem that urgently needs to be solved in the field of hydrogen energy equipment safety detection. Specifically, there are: (1) Insufficient differential detection capability for multi-material structures: Conventional ultrasonic testing is difficult to distinguish the hydrogen-induced microcrack scattering response between the carbon fiber interlayer interface and the metal liner, which is easy to miss or misjudge; (2) Limitations of single probe detection mode: Traditional single probes have blind spots for curved components (such as small curvature pressure vessels) and materials with cross-scale thickness (0.5-20mm), and have poor adaptability; (3) Low level of intelligent signal processing: It relies on manual feature extraction and lacks intelligent fusion analysis of multi-source features in the time domain, frequency domain and nonlinear domain, which cannot meet the real-time detection needs in service. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention provides a hydrogen loss detection system for hydrogen energy equipment based on customized ultrasonic signals, comprising:

[0005] Ultrasonic signal generator: Based on the acoustic impedance characteristics of hydrogen energy equipment materials, it generates customized ultrasonic signals and uses time-frequency domain joint modulation technology to optimize the signal waveform through wavelet packet decomposition, effectively solving the problem of identifying the differential scattering response of hydrogen-induced microcracks between carbon fiber interlayer interfaces and metal liner.

[0006] Ultrasonic signal detector: Constructed as a multi-modal probe array, including phased array probes, through-type probes and surface wave probes, to achieve accurate detection of hydrogen damage in different parts of hydrogen energy equipment, receive the reflection, transmission and scattering signals generated by customized ultrasonic signals in the equipment, effectively eliminate detection blind spots, and adapt to components with different curvatures and thicknesses;

[0007] Dynamic Analysis Module: Integrates machine learning algorithms to process the time-domain, frequency-domain, and nonlinear-domain characteristic parameters of scattered signals in real time, output hydrogen damage status assessment results, and realize intelligent fusion analysis of multi-source signals.

[0008] The entire detection system process begins with an ultrasonic signal generator producing a customized ultrasonic signal, which is then received by a detector from the equipment's response signal. This response signal is then analyzed in detail by a dynamic analysis module. These three modules complement each other functionally, together forming a complete hydrogen loss detection system for hydrogen energy equipment, ensuring both accuracy and efficiency in the detection process.

[0009] Preferably, the ultrasonic signal generator includes:

[0010] Signal source module: Generates initial electrical signals based on programmable logic devices, supporting high-precision waveform adjustment;

[0011] Power amplifier module: Connected to the signal source module, it provides dynamic gain adjustment to meet the ultrasonic excitation energy requirements of metallic materials;

[0012] Ultrasonic transducer module: It adopts a flexible array of piezoelectric composite materials to realize curved surface fitting detection and multi-channel independent excitation, converting electrical signals into customized ultrasonic signals and transmitting them into hydrogen energy equipment.

[0013] The signal source module generates an initial electrical signal, which is then amplified by the power amplifier module to meet the operational requirements of the ultrasonic transducer module. The ultrasonic transducer module then converts the electrical signal into an ultrasonic signal and transmits it into the hydrogen energy equipment. These three modules are interconnected and support each other, jointly fulfilling the function of the ultrasonic signal generator and providing a reliable ultrasonic signal source for hydrogen loss detection in the hydrogen energy equipment.

[0014] Preferably, the ultrasonic signal detector comprises:

[0015] Ultrasonic probes, acting as the signal receiving front-end, are responsible for capturing customized ultrasonic signals reflected or transmitted within the equipment and converting them into electrical signals. These electrical signals contain information about the internal structure of the hydrogen energy equipment, particularly important data regarding hydrogen damage (such as hydrogen-induced cracking and interfacial debonding). Different types of ultrasonic probes, such as phased array probes, transmission probes, and surface wave probes, are designed and functionally suited for efficient detection of different parts of the equipment and types of damage, ensuring the comprehensiveness and accuracy of the signals.

[0016] Signal processing module: This module performs further preprocessing on the electrical signals received from the probe. This process includes signal amplification, filtering, and analog-to-digital conversion, aiming to improve signal quality, reduce noise interference, and provide a clear and reliable signal foundation for subsequent data analysis. Optimization of the signal processing module enables more effective extraction of hydrogen damage-related feature information, providing strong support for subsequent damage assessment.

[0017] The data analysis module, after preprocessing the electrical signals, utilizes machine learning algorithms to deeply mine and analyze the time-domain, frequency-domain, and nonlinear-domain feature parameters. This module not only extracts relevant features of hydrogen damage but also accurately identifies and assesses the hydrogen damage state by comparing and matching it with a pre-established feature database. The output of the data analysis module includes information on the degree, location, and type of hydrogen damage, providing crucial data for generating inspection reports and making maintenance decisions.

[0018] As a signal receiving front-end, the ultrasonic probe captures customized ultrasonic signals reflected or transmitted from the equipment and converts them into electrical signals. These electrical signals carry important data on the internal structure of the hydrogen energy equipment and hydrogen damage (such as hydrogen-induced cracking and interface debonding). The acquired electrical signals are transmitted to the signal processing module, which performs preprocessing such as amplification, filtering, and analog-to-digital conversion to improve signal quality and reduce noise interference, providing a clear and reliable signal foundation for subsequent data analysis. The preprocessed signal is then transmitted to the data analysis module, which uses machine learning algorithms to deeply analyze the time-domain, frequency-domain, and nonlinear-domain characteristic parameters of the signal. By comparing and matching with a pre-established feature database, it identifies the hydrogen damage state and outputs information such as the degree, location, and type of hydrogen damage, providing crucial information for generating inspection reports and maintenance decisions. During this process, the dynamic analysis module transmits parameter commands to the ultrasonic signal generator. The three modules are electrically connected to form a collaborative working system: the ultrasonic probe completes signal acquisition, the signal processing module optimizes the signal, and the data analysis module performs intelligent evaluation, jointly ensuring the accuracy and efficiency of hydrogen damage detection.

[0019] The detection method steps are as follows:

[0020] S401: Generates customized ultrasonic signals via an ultrasonic signal generator and transmits them to hydrogen energy equipment;

[0021] S402: Receive the response signal of a customized ultrasonic signal in a hydrogen energy equipment using an ultrasonic signal detector;

[0022] S403: Processes and analyzes the response signal, uses wavelet transform and time-frequency analysis algorithms to suppress high-frequency noise, extracts hydrogen damage-related feature parameters from the preprocessed signal, and fuses multi-source features based on DS evidence theory to output hydrogen damage assessment results;

[0023] S404: Based on the feature information, collect and organize the feature parameter data of different hydrogen damage states and form a feature database. Compare and match the extracted feature information with the feature database to identify the hydrogen damage state, and then assess the degree, location and type of hydrogen damage to generate a detection report.

[0024] Preferably, before generating a customized ultrasonic signal, the signal parameters are determined and the ultrasonic signal generator is set according to the materials, structure, and expected detection sensitivity of the hydrogen energy equipment.

[0025] Preferably, after determining the hydrogen damage state of the hydrogen energy equipment, the method further includes visualizing the detection results, generating images and charts, and providing users with maintenance suggestions and decision support based on the detection results.

[0026] Beneficial effects

[0027] By generating customized ultrasonic signals using an ultrasonic signal generator, achieving multimodal precision detection using an ultrasonic signal detector, and performing intelligent signal processing using a dynamic analysis module, combined with corresponding detection methods, the shortcomings of existing technologies are effectively overcome. This enables comprehensive, accurate, and real-time detection and assessment of the hydrogen damage state of hydrogen energy equipment, significantly improving detection efficiency and reliability, and ensuring the safe operation of hydrogen energy equipment.

[0028] The tasks that this invention can accomplish include accurately detecting hydrogen-induced microcracks, delamination, debonding, and other damage in carbon fiber composite interlayer interfaces and metal liner components of hydrogen energy equipment, with a positioning accuracy of ±0.5mm, and the ability to detect defects ≥0.5mm in size; through a multi-modal probe array composed of phased array, through-type, and surface wave probes, it covers curved components (minimum radius of curvature 5mm) and materials of varying thicknesses (0.5-20mm), eliminating detection blind spots; it processes the time domain, frequency domain, and nonlinear domain characteristics of signals in real time, combines them with feature database matching, outputs the damage type (crack / delamination / bulge), location, and level (1-5), and generates a detection report; it visualizes the results using pseudo-color images and 3D models, automatically generates monitoring, repair, or replacement suggestions based on the damage level, and supports remote data upload.

[0029] Its advantages include strong multi-material differential detection capability, which can distinguish hydrogen-induced damage between carbon fiber interlayer and metal liner, reducing the rate of missed detection / false judgment; excellent detection adaptability and comprehensiveness, with multi-modal probe array reducing detection blind zone by more than 90% and adapting to complex structures; intelligent and efficient signal processing, which realizes automated analysis based on machine learning and DS evidence theory, with processing latency <50ms and efficiency improved by more than 80% compared to manual processing; high real-time performance and reliability, supporting online real-time detection, outputting evaluation results with confidence ≥92%, and adapting to harsh environments (-40℃-85℃, IP68 protection, 100g vibration load).

[0030] The advantages stem from the use of customized ultrasonic signal technology, which generates 2-10MHz composite ultrasonic signals based on the acoustic impedance characteristics of materials. Waveform optimization is achieved through time-frequency domain joint modulation (wavelet packet decomposition), enhancing the identification of scattering responses at different material interfaces. Furthermore, a flexible array of piezoelectric composite materials (32 independently excited channels) enables surface-fit detection. The multi-modal probe array design is highly targeted; phased array probes use electronic focusing to achieve ±70° beam deflection for composite material delamination detection; a penetrating probe with a dual-crystal structure and path-compensating wedges detects microcracks in thick-walled metals; and a surface wave probe with an adaptive curvature coupling layer detects surface and near-surface defects. Intelligent signal processing algorithms integrate 15-dimensional feature parameters (such as dominant frequency offset, fractal dimension, and nonlinear coefficients), using K-means clustering for dimensionality reduction and improved Jeffrey divergence to calculate evidence conflict, thus enhancing reliability. System hardware is co-optimized; the signal generator (FPGA + broadband power amplifier) ​​and detector (high-speed ADC + multiple probes) are connected via armored cables, and standard test blocks are used to calibrate sound velocity, gain, and focusing depth (error ≤1%-3%), ensuring detection accuracy and stability. Attached Figure Description

[0031] Figure 1 The flowchart shows a hydrogen loss detection method based on customized ultrasonic signals.

[0032] Figure 2 This is a diagram illustrating the architecture of a hydrogen loss detection system for hydrogen energy equipment based on customized ultrasonic signals. Detailed Implementation

[0033] A method for hydrogen loss detection system of hydrogen energy equipment based on customized ultrasonic signals according to the present invention is described in [reference needed]. Figure 1 See also: A hydrogen loss detection system for hydrogen energy equipment based on customized ultrasonic signals. Figure 2 The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0034] The detection system consists of an ultrasonic signal generator, an ultrasonic signal detector, and a dynamic analysis module connected by electrical connections to form a collaborative working system.

[0035] An ultrasonic signal generator generates customized ultrasonic signals based on the acoustic impedance characteristics of the material. It uses time-frequency domain joint modulation technology (wavelet packet decomposition) to optimize the waveform and solve the problem of differentiating hydrogen-induced microcracks in carbon fiber and metal lining.

[0036] The specific components are as follows:

[0037] Signal source module: Generates initial electrical signals based on programmable logic devices (such as Xilinx Zynq UltraScale+ series FPGA chips), supporting high-precision waveform adjustment from 100MHz to 10GHz.

[0038] Power amplifier modules: such as the TGA2591-EPU broadband power amplifier, which provides 10-2000MHz bandwidth and 30dB dynamic gain adjustment to meet the excitation energy requirements of metal materials.

[0039] Ultrasonic transducer module: It adopts a PZT-5H piezoelectric composite material flexible array (50mm×30mm×0.5mm, 32-channel independent excitation unit), and is equipped with MEMS microlenses to achieve curved surface fitting (minimum curvature radius 5mm) to emit 2-10MHz ultrasonic signals.

[0040] Ultrasonic signal detector

[0041] Multimodal probe arrays (phased array, transmission-type, surface wave probes) receive reflected / transmitted / scattered signals, covering hydrogen damage detection in different areas:

[0042] Phased array probe: 16-element linear array, center frequency 5MHz, element spacing 1.5mm, electronic focusing ±70° beam deflection, for detecting delamination defects in carbon fiber composite laminates (0.5-10mm thickness).

[0043] Penetration probe: Dual crystal transceiver structure (2.5MHz transmitting end, 5MHz receiving end center frequency), equipped with sound path compensation wedge, to detect microcracks in metal inner liner (1-20mm wall thickness).

[0044] Surface wave probe: 4 independent probes, center frequency 10MHz, curvature adaptive coupling layer (0.3mm thickness), detects surface and near-surface defects (depth ≤2mm).

[0045] Signal processing module: Uses AD9288 12-bit high-speed ADC chip, 100MS / s sampling rate, and anti-aliasing filter (cutoff frequency 15MHz) to preprocess the signal.

[0046] Dynamic Analysis Module: Integrates the NVIDIA Jetson AGX Orin computing platform to run multi-scale feature fusion algorithms.

[0047] A total of 15 characteristic parameters were extracted, including those in the time domain (amplitude decay rate, rise time), frequency domain (harmonic distortion index), and nonlinear domain (Hilbert-Huang transform relative nonlinear coefficients).

[0048] A decision-making model is constructed based on the DS evidence theory. An improved Jeffrey divergence is introduced to calculate the degree of evidence conflict. When the confidence level is ≥92%, the damage assessment results (defect type, location ±0.5mm, level 1-5) are output.

[0049] Each module employs standardized interfaces in its hardware design, transmitting signals via armored coaxial cables. It boasts an IP68 protection rating, adaptable to temperatures ranging from -40℃ to 85℃ and vibration loads up to 100g, ensuring stability during in-service testing. The ultrasonic signal generator, based on the acoustic impedance characteristics of hydrogen energy equipment materials, utilizes programmable logic devices to generate an initial electrical signal. After dynamic gain adjustment by the power amplification module, it is converted into a customized ultrasonic signal by an ultrasonic transducer module composed of a flexible array of piezoelectric composite materials. It employs time-frequency domain joint modulation technology, optimizing the waveform through wavelet packet decomposition, effectively solving the challenge of identifying the differentiated scattering response of hydrogen-induced microcracks between carbon fiber interlayer interfaces and the metal liner. Specifically, the signal source module uses a specific FPGA chip, supporting high-precision waveform adjustment and outputting composite ultrasonic signals; the power amplification module uses a broadband power amplifier adapted to the excitation energy requirements of metallic materials; and the flexible array design of the ultrasonic transducer module enables curved surface fitting detection and multi-channel independent excitation, ensuring the accuracy of signal transmission.

[0050] The ultrasonic signal detector employs a multi-modal probe array design, including a phased array probe, a through-beam probe, and a surface wave probe, which respectively receive reflected, transmitted, and scattered signals from within the equipment. The phased array probe uses electronic focusing to detect delamination defects in composite materials, suitable for components with varying curvatures. The through-beam probe utilizes a bicrystalline structure and a path compensation wedge to address beam attenuation issues during the detection of microcracks in thick-walled metals. The surface wave probe captures surface and near-surface defect signals through a curvature-adaptive coupling layer. The received electrical signals are amplified, filtered, and preprocessed by the signal processing module using analog-to-digital conversion before being transmitted to the data analysis module. Machine learning algorithms are used to extract time-domain, frequency-domain, and nonlinear-domain feature parameters, which are then compared and matched with a feature database to identify hydrogen damage status. The signal processing module uses a high-speed ADC chip and an anti-aliasing filter to improve signal quality; the data analysis module integrates a computing platform to perform deep feature mining on the signals, extracting multi-dimensional feature parameters and performing dimensionality reduction processing to provide data support for damage assessment.

[0051] The dynamic analysis module integrates machine learning algorithms to process multi-source signal features in real time. Based on DS evidence theory, it fuses feature parameters and outputs assessment results for damage degree, location, and type. This module constructs a feature database, collecting feature parameters of different hydrogen damage states. Through similarity calculation and evidence conflict analysis, it achieves accurate damage identification and level assessment. When the evidence conflict is below a threshold, the assessment result is output; otherwise, a secondary scan is triggered to ensure detection reliability.

[0052] The detection method based on the above system is as follows:

[0053] First, a customized ultrasonic signal is emitted to the equipment by an ultrasonic signal generator. After the detector receives the response signal, it uses wavelet transform and time-frequency analysis algorithms to reduce noise and extract hydrogen damage-related features. Through multi-source feature fusion and feature database matching, a detection report is generated and the results are visualized, while also providing maintenance decision support. The system calibration process uses standard test blocks of the same material to calibrate sound velocity, probe gain, and focusing depth. Customized signal generation is based on genetic algorithm-optimized parameters to adapt to the equipment's material and structural characteristics. Multimodal signal acquisition achieves data alignment through a synchronization mechanism, and preprocessing algorithms suppress noise and retain effective features. Damage assessment combines clustering algorithms and evidence theory to ensure accuracy. The visualization and maintenance suggestion module generates differentiated strategies based on damage levels and supports remote data upload.

[0054] Details to be explained in the implementation scheme: (1) System calibration: use the same material standard test block (carbon fiber laminate / aluminum alloy inner liner) to calibrate the sound velocity (error ≤1%), probe gain (echo amplitude difference ≤3dB), and focusing depth (0.1-50mm accuracy ±0.2mm).

[0055] (2) Custom signal generation: Based on the material (CFRP / Al-6061), structural curvature (≥5mm), and sensitivity (defect ≥0.5mm), a composite ultrasonic signal (Chirp pulse 2-8MHz sweep frequency + Burst pulse 5-cycle 5MHz sine wave) with a peak-to-peak value of 200Vpp is generated through genetic algorithm optimization.

[0056] (3) Multimodal signal acquisition: Phased array probe performs electronic scanning at a frame rate of 50Hz to generate a 1024×1024 pixel C-scan image; Transmission probe synchronously acquires transmitted wave amplitude; Surface wave probe acquires 0-10μs surface wave signal, focusing on detecting stress concentration areas (such as welds).

[0057] (4) Signal processing and evaluation, preprocessing: sym8 wavelet basis 5-layer decomposition to suppress noise above 100kHz, and improve signal-to-noise ratio by 15dB;

[0058] Feature extraction: K-means clustering is used to reduce the dimensionality of 2000+ standard test blocks; when making fusion decisions, if the conflict degree is <0.3, the result is output; otherwise, a second scan is performed.

[0059] Results output visualization: pseudo-color C-scan image (0.5mm×0.5mm resolution), 3D defect reconstruction model, OpenCV edge detection and size measurement.

[0060] Maintenance recommendations: Generate strategies based on damage levels (Level 1-2 monitoring, Level 3 local repair, Level 4-5 replacement warning) and upload them to the remote platform via Modbus TCP protocol.

[0061] Key technical parameters and implementation details of this embodiment:

[0062] a. Hardware protection: The signal generator and detector are connected by an armored coaxial cable with IP68 protection rating, adaptable to temperatures of -40℃ to 85℃ and vibration loads of 100g.

[0063] b. Algorithm training: The dynamic analysis module uses 100,000 sets of noisy signal data to pre-train the neural network offline, with an online inference latency of <50ms.

[0064] c. Feature Database: Collect feature parameters (such as dominant frequency offset, fractal dimension, and nonlinear coefficient) of different hydrogen damage states to form a standard feature database for matching and identification.

Claims

1. A hydrogen loss detection system for hydrogen energy equipment based on customized ultrasonic signals, characterized in that, include: Ultrasonic signal generator: Based on the acoustic impedance characteristics of hydrogen energy equipment materials, it generates customized ultrasonic signals and uses time-frequency domain joint modulation technology to optimize the signal waveform through wavelet packet decomposition, effectively solving the problem of identifying the differential scattering response of hydrogen-induced microcracks between carbon fiber interlayer interfaces and metal liner. Ultrasonic signal detector: Constructed as a multi-modal probe array, including phased array probes, through-type probes and surface wave probes, to achieve accurate detection of hydrogen damage in different parts of hydrogen energy equipment, receive the reflection, transmission and scattering signals generated by customized ultrasonic signals in the equipment, effectively eliminate detection blind spots, and adapt to components with different curvatures and thicknesses; Dynamic Analysis Module: Integrates machine learning algorithms to process the time-domain, frequency-domain, and nonlinear-domain characteristic parameters of scattered signals in real time, output hydrogen damage status assessment results, and realize intelligent fusion analysis of multi-source signals; The ultrasonic signal generator, ultrasonic signal detector, and dynamic analysis module are electrically connected to form a collaborative working system.

2. The detection system as described in claim 1, characterized in that, The ultrasonic signal generator includes: Signal source module: Generates initial electrical signals based on programmable logic devices, supporting high-precision waveform adjustment; Power amplifier module: electrically connected to the signal source module, providing dynamic gain adjustment to meet the ultrasonic excitation energy requirements of metallic materials; Ultrasonic transducer module: It adopts a flexible array of piezoelectric composite materials to realize curved surface fitting detection and multi-channel independent excitation, converting electrical signals into customized ultrasonic signals and transmitting them into hydrogen energy equipment; The signal source module generates an initial electrical signal, which is then amplified by the power amplifier module to meet the working requirements of the ultrasonic transducer module. The ultrasonic transducer module is responsible for converting the electrical signal into an ultrasonic signal and transmitting it into the hydrogen energy equipment. These three modules are interconnected and support each other, together completing the function of the ultrasonic signal generator and providing an ultrasonic signal source for hydrogen loss detection in the hydrogen energy equipment.

3. The detection system as described in claim 1, characterized in that, The ultrasonic signal detector includes: Ultrasonic probe: As the front end of signal reception, it is responsible for capturing customized ultrasonic signals reflected or transmitted in the equipment and converting them into electrical signals. The electrical signals include important data on hydrogen damage, hydrogen-induced cracks, and interface debonding. Different types of ultrasonic probes are used to detect different parts of the equipment and different types of damage. Signal processing module: Further preprocesses the electrical signals received from the probe. This process includes signal amplification, filtering, and analog-to-digital conversion steps, aiming to improve signal quality, reduce noise interference, and provide a clear and reliable signal basis for subsequent data analysis. Through the optimization of the signal processing module, feature information related to hydrogen damage is extracted to support subsequent damage assessment. Data Analysis Module: After preprocessing the electrical signal, machine learning algorithms are used to deeply mine and analyze the time-domain, frequency-domain, and nonlinear-domain feature parameters of the signal. By comparing and matching with a pre-established feature database, the module can accurately identify and assess the hydrogen damage status. The output results include the degree, location, and type of hydrogen damage, providing an important basis for generating detection reports and maintenance decisions.

4. A method for detecting hydrogen loss in hydrogen energy equipment based on customized ultrasonic signals, characterized in that, The steps are as follows: The S401 generates customized ultrasonic signals via an ultrasonic signal generator and transmits them to the hydrogen energy equipment. S402 uses an ultrasonic signal detector to receive the response signal of a customized ultrasonic signal in a hydrogen energy equipment. The S403 processes and analyzes the response signal, uses wavelet transform and time-frequency analysis algorithms to suppress high-frequency noise, extracts hydrogen damage-related feature parameters from the preprocessed signal, and fuses multi-source features based on DS evidence theory to output hydrogen damage assessment results. Based on the feature information, S404 collects and organizes feature parameter data of different hydrogen damage states and forms a feature database. It then compares and matches the extracted feature information with the feature database to identify the hydrogen damage state, and further assesses the degree, location and type of hydrogen damage to generate a detection report.

5. The detection method as described in claim 4, characterized in that, In step S401, a customized ultrasonic signal is generated by an ultrasonic signal generator and transmitted to the hydrogen energy equipment. Before generating the customized ultrasonic signal, the signal parameters are determined and the ultrasonic signal generator is set according to the materials, structure and expected detection sensitivity of the hydrogen energy equipment.

6. The detection method as described in claim 4, characterized in that, After S403 determines the hydrogen damage status of hydrogen-powered equipment, it also includes visualizing the detection results, generating images and charts, and providing users with maintenance suggestions and decisions based on the detection results.