CN-122021302-A - Fuel cell leakage safety simulation early warning method and system based on acoustic spectrum recognition

Abstract

The invention provides a fuel cell leakage safety simulation early warning method and system based on sound wave spectrum recognition, wherein the method comprises the steps of establishing a model, integrating a fluid dynamics calculation module and a sound wave propagation simulation module, generating a sound print library, setting leakage parameter combination in a simulation environment, running transient acoustic simulation, obtaining sound pressure data of virtual monitoring points, extracting sound print feature vectors, constructing a leakage sound print feature library, monitoring in real time, continuously collecting sound wave signals, carrying out signal conditioning and fast Fourier transformation, outputting real-time sound print features, carrying out pattern matching on the real-time sound print features and the leakage sound print feature library, identifying leakage states, dynamically evaluating safety risk levels, and carrying out parallel operation according to the risk levels by responding control.

Inventors

• SUN YONGMING
• WU KUN

Assignees

• 华中科技大学

Dates

Publication Date

20260512

Application Date

20260128

Claims (10)

1. 1. The fuel cell leakage safety simulation early warning method based on the sound wave spectrum identification is characterized by comprising the following steps of: S1, building a model, namely building a three-dimensional digital model of a fuel cell system and a closed space through a multi-physical-field coupling simulation platform, integrating a fluid dynamics calculation module and an acoustic wave propagation simulation module, wherein the fluid dynamics module is used for solving leakage port gas jet parameters, and the acoustic module is used for predicting attenuation rules of acoustic waves in the space; s2, generating a voiceprint library, namely setting leakage parameter combinations in a simulation environment, running transient acoustic simulation, acquiring sound pressure data of virtual monitoring points, extracting voiceprint feature vectors after frequency domain analysis, and constructing the leakage voiceprint feature library; s3, real-time monitoring, namely deploying an acoustic sensor array at a corresponding position of a real system, continuously collecting acoustic signals, and outputting real-time voiceprint characteristics through signal conditioning and fast Fourier transformation; S4, intelligent diagnosis, namely performing pattern matching on the real-time voiceprint features and the leakage voiceprint feature library, identifying leakage states through a pre-trained classification model, and dynamically evaluating safety risk levels based on feature amplitude values; and S5, responding to control, namely triggering a grading alarm according to the risk grade, and executing ventilation, power reduction or shutdown operation in parallel.
2. 2. The method of claim 1, wherein in step S2, the leakage parameter combination comprises at least two variables of leakage aperture, leakage location, system pressure; the voiceprint feature vector includes a primary resonant frequency, a particular band energy duty cycle, and a sound pressure level difference for a plurality of frequency bands.
3. 3. The method of claim 1, wherein in step S3, the signal conditioning includes bandpass filtering, noise reduction and gain adjustment, and wherein the fast fourier transform has a frequency domain coverage of 20kHz to 100kHz to focus the ultrasonic frequency band.
4. 4. The method according to claim 1, wherein in step S4, the pre-trained classification model is a machine learning model trained based on simulation data, the output of which includes three states of no leakage, micro leakage and severe leakage, and the security risk level is classified into low, medium and high levels according to the leakage type and the sound pressure level amplitude.
5. 5. The method of claim 4, wherein the machine learning model employs a support vector machine or neural network algorithm, and wherein training data is derived from multi-condition simulation results in the voiceprint feature library.
6. 6. The method of claim 1, wherein in step S5, the hierarchical alarm includes a visual alert, an audible alert, and a text prompt, and wherein the coordinated operation prioritizes the performance of enhanced ventilation and triggers an emergency shutdown at high risk.
7. 7. A fuel cell leakage safety warning system for implementing the method of any one of claims 1 to 6, comprising: the simulation modeling module is used for constructing and running the multi-physical field coupling simulation model and outputting a voiceprint feature library; the signal acquisition and processing module is connected with the ultrasonic sensor array and used for capturing and preprocessing the acoustic wave signals in real time and extracting frequency domain characteristics; the leakage diagnosis module is embedded with a classification model and is used for matching real-time characteristics with a characteristic library and outputting leakage identification results and risk grades; and the early warning execution module is used for issuing an alarm according to the risk level and sending a control instruction to an external executor.
8. 8. The system of claim 7, wherein the leak diagnosis unit is integrated with a model update module for optimizing classification model parameters based on real-time data.
9. 9. The system of claim 7, further comprising a human-machine interface module for dynamically displaying the acoustic spectrogram, the risk level change curve, and the historical pre-warning record.
10. 10. The system of claim 7, wherein the placement location of the ultrasonic sensor array is determined based on optimization of acoustic energy distribution in the simulation model, the sensor frequency response range covering 20kHz to 100kHz.

Description

Fuel cell leakage safety simulation early warning method and system based on acoustic spectrum recognition Technical Field The invention relates to the technical field of fuel cell safety, in particular to a fuel cell leakage safety simulation early warning method and system based on sound wave spectrum identification. Background The fuel cell, especially proton exchange membrane fuel cell, has the advantages of high energy conversion efficiency, environmental friendliness and the like, and has great application potential in closed environment power systems such as aerospace, submarine navigation and the like. However, the presence of high pressure hydrogen in fuel cell systems makes leakage risk one of its most significant safety threats. In the airtight space where ventilation is limited, even if a small amount of hydrogen leaks and accumulates, an explosive atmosphere may be formed, causing serious accidents. Existing fuel cell leak detection methods rely primarily on contact physical sensors, such as electrochemical hydrogen concentration sensors, pressure sensors, and the like. The method has obvious limitations that the response of the hydrogen concentration sensor has lag, the distribution number is limited, the full coverage monitoring of the whole space is difficult to realize, the pressure drop rule is generally used for the system tightness test, and the real-time online monitoring in the running state is difficult to realize. Acoustic detection, which is a non-invasive detection technique, has been applied to leak detection of pressure vessels and pipes, and its basic principle is that when high pressure gas leaks from a narrow gap, a broadband acoustic signal is generated due to turbulence, vortex, etc., which contains audible sound and ultrasonic components. However, applying the acoustic detection technology to a complex fuel cell system and realizing accurate leakage identification and safety assessment still faces great challenges, namely that firstly, the fuel cell system body (such as an air compressor and a cooling pump) can generate strong background noise when in operation to cause serious interference to leakage sound wave signals, secondly, the characteristics (such as frequency and amplitude) of the leakage sound waves are closely related to leakage aperture, leakage pressure, propagation path and environment structure, the relationship is complex, and the judgment is difficult to be carried out through a simple threshold value, and furthermore, an effective means for carrying out the correlation assessment on the acoustic simulation and the safety state of the fuel cell system is lacking. Therefore, a new method for early, accurate, intelligent diagnosis and early warning of fuel cell leakage in a closed environment is urgently needed to overcome the above-mentioned drawbacks. Disclosure of Invention The invention aims to provide a fuel cell leakage safety simulation early warning method and system based on sound wave spectrum recognition, which are used for solving the problem that the prior art cannot quickly, accurately and noninvasively recognize early micro leakage of a fuel cell. The technical scheme for solving the technical problems is as follows: in a first aspect, the invention provides a fuel cell leakage safety simulation early warning method based on acoustic spectrum recognition, which comprises the following steps: S1, building a model, namely building a three-dimensional digital model of a fuel cell system and a closed space through a multi-physical-field coupling simulation platform, integrating a fluid dynamics calculation module and an acoustic wave propagation simulation module, wherein the fluid dynamics module is used for solving leakage port gas jet parameters, and the acoustic module is used for predicting attenuation rules of acoustic waves in the space; s2, generating a voiceprint library, namely setting leakage parameter combinations in a simulation environment, running transient acoustic simulation, acquiring sound pressure data of virtual monitoring points, extracting voiceprint feature vectors after frequency domain analysis, and constructing the leakage voiceprint feature library; s3, real-time monitoring, namely deploying an acoustic sensor array at a corresponding position of a real system, continuously collecting acoustic signals, and outputting real-time voiceprint characteristics through signal conditioning and fast Fourier transformation; S4, intelligent diagnosis, namely performing pattern matching on the real-time voiceprint features and the leakage voiceprint feature library, identifying leakage states through a pre-trained classification model, and dynamically evaluating safety risk levels based on feature amplitude values; and S5, responding to control, namely triggering a grading alarm according to the risk grade, and executing ventilation, power reduction or shutdown operation in parallel. Preferably, in step S2, the leakage parameter com