Fluent simulation-based pipeline hydrogen-doped leakage real-time monitoring and early warning method and monitoring and early warning system

Abstract

The invention relates to the technical field of pipeline safety detection, in particular to a pipeline hydrogen-doped leakage real-time monitoring and early warning method and system based on Fluent simulation, and the method comprises the steps: obtaining leakage diffusion data of gas in a pipeline detected by multiple sensors, and carrying out the preprocessing of the data; extracting data features: carrying out conjoint analysis on a time-frequency domain to obtain a time-frequency matrix; by calculating mutual information evaluation correlation of different sensor data, feature statistical correlation is realized, and features with non-significant spatial correlation are removed; the pressure data, the flow data and the concentration data are subjected to sub-channel filtering and fusion processing based on federated Kalman filtering (FKF); establishing a pipeline model based on Fluent, and simulating a pipeline hydrogen-doped leakage process by taking the extracted features as input conditions; a risk model is constructed based on real-time sensor data, simulation prediction data and environmental parameters, and a hierarchical early warning mechanism is established. Through cooperative monitoring of multiple sensors, leakage characteristic data can be accurately captured, and safety and reliability of the system are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline safety detection, and in particular to a real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation and a monitoring and early warning system. Background Art

[0002] Hydrogen energy is currently a relatively advanced clean energy source. Therefore, considering hydrogen-blended transportation on the basis of the existing natural gas pipeline network and deeply integrating natural gas and hydrogen will help accelerate deep decarbonization in industries such as industry and construction, and help improve the transportation efficiency and safety of natural gas pipelines. However, during the transportation process, factors such as pipeline corrosion, material aging, and external force damage may cause leakage of hydrogen-blended natural gas. Therefore, safety control after leakage of hydrogen-blended natural gas pipelines is very important. Through technical means, early detection, accurate assessment and rapid response of leakage risks can be achieved to ensure energy transportation safety, protect the ecological environment, reduce economic losses, and provide safety guarantees for the large-scale application of hydrogen energy infrastructure.

[0003] Existing safety control methods for hydrogen-blended natural gas pipeline leaks rely on real-time detection of hydrogen and methane within the integrated pipeline corridor to determine gas concentrations. These methods then calculate explosion and alarm concentration thresholds based on the gas transported by the hydrogen-blended natural gas pipeline. These gas concentrations are then compared with the explosion and alarm concentration thresholds to determine the leak level, and different safety control measures are implemented for each leak level. However, this safety control method and system for hydrogen-blended natural gas pipeline leaks only detects hydrogen and methane concentration data alone, failing to integrate a multi-sensor collaborative monitoring network to accurately capture leak characteristics. Furthermore, these methods and systems are unable to dynamically simulate hydrogen concentration field distribution and improve operational and maintenance decision-making efficiency through simulation modeling and interactive visualization. Therefore, a real-time monitoring and early warning system for hydrogen-blended pipeline leaks based on Fluent simulation is proposed. Summary of the Invention

[0004] The technical problem to be solved by the present invention is that the existing hydrogen blending pipeline leakage detection only considers the gas concentration problem, has low detection accuracy and poor effect.

[0005] To this end, the present invention provides a real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation and a monitoring and early warning system.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] A real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation includes the following steps:

[0008] Step 1: Acquire the leakage and diffusion data of the gas inside the multi-sensor detection pipeline and pre-process the data;

[0009] Step 2: extracting data features. Step 2 specifically includes:

[0010] S2.1 extracts time domain features based on signal fluctuation characteristics, extracts frequency domain features based on FFT spectrum energy distribution, and jointly analyzes the time and frequency domains to obtain a time-frequency matrix;

[0011] S2.2 evaluates the correlation by calculating the mutual information of different sensor data, realizes the statistical association of features, and eliminates features with insignificant spatial correlation;

[0012] S2.3 performs channel filtering and fusion processing on pressure, flow and concentration data based on the federated Kalman filter FKF;

[0013] Step 3: Establish a pipeline model based on Fluent, use the extracted features as boundary conditions and fluid conditions, and simulate the hydrogen leakage process in the pipeline;

[0014] Step 4: Build a risk model based on real-time sensor data, simulation prediction data, and environmental parameters, and establish a graded early warning mechanism.

[0015] Furthermore, the time domain feature extraction process in step S2.1 specifically includes: calculating the signal mean and signal variance; capturing the pressure drop rate according to the first-order difference, calculating according to the zero-crossing rate formula, and detecting the high-frequency components of the signal.

[0016] Furthermore, in step S2.1, the frequency domain feature extraction process includes: converting the time domain signal into a frequency domain energy distribution through Fourier transform; calculating the energy proportion of a specific frequency band in the frequency domain energy distribution, wherein for the main frequency component, extracting the frequency and amplitude corresponding to the spectrum peak, and calculating the frequency domain entropy value based on the power spectrum entropy to measure the spectrum complexity and calculate the frequency domain features.

[0017] Furthermore, in step S2.1, the joint analysis of the time-frequency domain specifically includes: generating a time-frequency spectrum according to the short-time Fourier transform STFT to capture the transient frequency characteristics of the leakage signal, determining relevant parameters, including the original signal, window function, window length, number of overlapping points, sampling frequency and number of Fourier points, and performing a fast Fourier transform; converting the original signal and the window function into a column vector, calculating the signal length, and calculating the number of window sliding times n according to the signal length nx, the window length WinLen and the number of overlapping points noverlap, that is, the number of columns of the signal when the source signal is divided into columns, expressed as n=fix((nx-noverlap) / (WinLen-noverlap)); expressing the signal selected by each window function sliding as a column, determining the value of each column, and obtaining a matrix Fig with n columns and WinLen rows; expanding the window function converted into a column vector into a matrix w with n columns, and performing a dot product on the matrix Fig and w, and performing a fast Fourier transform on the result of the dot product to obtain a time-frequency matrix, and outputting a spectrum according to the time-frequency matrix.

[0018] Furthermore, the step S2.2 specifically includes:

[0019] S2.2.1 Establish a local coordinate system for the pipeline and record the three-dimensional coordinates of each sensor. Construct an adjacency matrix based on the topological relationship and implement sensor spatial coordinate mapping by determining the distance between two sensors.

[0020] S2.2.2 Slice the time domain and frequency domain features of each sensor into time windows to construct a spatiotemporal aligned dataset;

[0021] S2.2.3 Apply a time lag to the time domain characteristics of the downstream sensor based on the pressure wave propagation time;

[0022] S2.2.4 Calculate the cross-correlation analysis including calculating the cross-correlation coefficient of the pressure change rate of sensors i and j and calculating the Pearson correlation coefficient of the high-frequency energy between sensors;

[0023] S2.2.5 Construct a spatial covariance matrix to evaluate the correlation between different sensor data and eliminate features with insignificant spatial correlation.

[0024] Furthermore, the step S2.3 specifically includes:

[0025] S2.3.1 Define the optimal state estimate X g And P is the pipeline pressure, Q is the mass flow rate, and C is the hydrogen concentration;

[0026] S2.3.2 Design local filters for pressure, flow, and concentration channels;

[0027] S2.3.3 Calculate the channel-wise Kalman filter;

[0028] S2.3.4 performs data fusion processing after filtering and calculates the information allocation factors of pressure, flow, and concentration based on the sensor confidence level to achieve information allocation and weight calculation, which can be expressed as:

[0029] where β P +β Q +β C =1,tr(P i ) is the trace of the error covariance matrix of the i-th local filter. After the information is distributed, the information form is fused, which is expressed as: where β P , β Q , β C are the information distribution factors for pressure, flow and concentration respectively, P P 、P Q 、P C are the state estimation error covariances of each local filter (pressure, flow, concentration) to characterize the uncertainty of the estimation results, X P 、X Q 、X C are the state quantities estimated independently by each local filter, P g is the global state estimation error covariance after fusion, which reflects the overall uncertainty of the fusion result, X g is the optimal state estimate after fusion;

[0030] The optimal state estimate X g Feedback is given to each local filter and its initial state is reset to achieve feedback reset, which is expressed as:

[0031] Furthermore, the step 2 also includes S2.4, simulating a 5mm aperture leakage in a real pipeline to inject a known leakage signal, recording the sensor data as a benchmark, calculating the positioning error MAE and the leakage error RMSE of the fusion result respectively, and verifying the target positioning error after data fusion.

[0032] Furthermore, in step 4, the risk model is expressed as: L = α·concentration+β·diffusion rate+γ·population exposure, where α, β, and γ represent the weights of each indicator, respectively, and α = 0.6, β = 0.3, and γ = 0.1 are set.

[0033] A real-time monitoring and early warning system for implementing the above-mentioned real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation includes:

[0034] A monitoring and management module, comprising a plurality of sensors arranged in the pipeline;

[0035] A data acquisition module, wherein the data acquisition module is connected to the monitoring and management module;

[0036] A simulation analysis module connected to the data acquisition module;

[0037] A leakage warning module, which is connected to the simulation analysis module and the data acquisition module;

[0038] A visualization interaction module is connected to the leakage warning module and the simulation analysis module.

[0039] Furthermore, the various sensors include temperature sensors, pressure sensors, gas flow meters, external concentration detection sensors and acoustic sensors. The temperature sensor monitors abnormal pipeline temperature data based on detecting pipeline temperature parameters; the pressure sensor is used to detect the air pressure inside the pipeline; the gas flow meter is used to detect the gas flow rate parameters inside the pipeline; the external concentration detection sensor is used to detect the hydrogen mixing ratio inside the pipeline; the acoustic sensor is used to detect pipeline leakage data. The pressure sensor, infrared concentration detection sensor and acoustic sensor are equidistantly spaced around high-risk points for pipeline leakage.

[0040] The beneficial effect of the present invention is that the present application has the advantage of further improving the dynamic early warning decision-making and emergency response capabilities when hydrogen-blended pipelines leak based on simulation modeling and visual interaction, and solves the problems of the existing hydrogen-blended natural gas pipeline leakage safety control methods and systems in the above-mentioned background technology in improving leakage detection accuracy and early warning response speed, as well as ensuring system safety and reliability.

[0041] Specifically, the real-time monitoring and early warning system for pipeline hydrogen leakage based on Fluent simulation can accurately capture leakage characteristic data through a multi-sensor collaborative monitoring network combined with data filtering and fusion processing, and construct a leakage diffusion model based on pipeline three-dimensional modeling and Fluent simulation analysis to facilitate dynamic simulation of hydrogen concentration field distribution; by proposing a dynamic adjustment mechanism for risk thresholds, it can adapt to changes in different complex environments, adopt a fuzzy logic comprehensive evaluation model to facilitate the construction of a comprehensive risk assessment dimension, and establish a false alarm suppression algorithm to effectively reduce the system false alarm rate; by constructing a pipeline digital twin model, the operation and maintenance decision-making efficiency is improved based on visualization capabilities, and combined with the risk threshold mechanism, a multi-level cascade control strategy is adopted to achieve efficient emergency response. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The present invention will be further described below with reference to the accompanying drawings and examples.

[0043] Figure 1 This is a schematic diagram of the modules of the real-time monitoring and early warning system for pipeline hydrogen leakage of the present invention.

[0044] Figure 2The figure is a flow chart of the real-time monitoring and early warning method for pipeline hydrogen leakage of the present invention.

[0045] Figure 3 It is the pressure cloud diagram of the pipeline of the present invention under different leakage apertures.

[0046] Figure 4 This is a velocity cloud diagram of the leakage gas at different leakage apertures of the present invention.

[0047] Figure 5 Schematic diagram of pressure changes in the pipeline of the present invention at different times.

[0048] Figure 6 Schematic diagram of the velocity change of leaked gas at different times according to the present invention. DETAILED DESCRIPTION

[0049] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0050] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, features defined as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0051] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0052] A real-time monitoring and early warning system for pipeline hydrogen leakage based on Fluent simulation includes a monitoring and management module, a data acquisition module, a simulation and analysis module, a visual interaction module, and a leakage early warning module.

[0053] The monitoring and management module includes multiple sensors installed in the pipeline. The monitoring and management module monitors the pipeline operation status and parameter data based on multiple sensors; the data acquisition module obtains the leakage and diffusion data of the gas inside the pipeline detected by multiple sensors and processes the data; the simulation analysis module is connected to the data acquisition module to obtain the processed data, and is used to build a pipeline gas leakage simulation model, and verify the model simulation effect according to the Fluent simulation software; the leakage warning module is connected to the simulation analysis module and the data acquisition module to build a hierarchical warning mechanism and provide emergency response decisions based on the risk level threshold; the visualization interaction module is connected to the leakage warning module and the simulation analysis module to realize the visualization architecture design in combination with the digital twin model.

[0054] The various sensors include temperature sensors, pressure sensors, gas flow meters, external concentration detection sensors, and acoustic sensors. The temperature sensor detects pipeline temperature parameters to monitor abnormal pipeline temperature data; the pressure sensor is used to detect internal pipeline pressure; the gas flow meter is used to detect gas flow rate parameters within the pipeline; the external concentration detection sensor is used to detect the hydrogen content within the pipeline; and the acoustic sensor is used to detect pipeline leakage data. Specifically, based on the high-risk points of pipeline leakage, a group of pressure sensors, infrared concentration detection sensors, and acoustic sensors are deployed every 10 meters around them. The pressure sensor and gas flow meter are connected to the pipeline flange interface; the acoustic sensor is fixed to the outside of the pipeline with a bracket; and the infrared concentration detection sensor is fixed to two symmetrical sides of the pipeline using a transmission installation method.

[0055] It should be noted that the installation of multiple sensors requires consideration of pipeline operating parameters for sensor selection: pipeline operating parameters include pipeline material, pipe diameter, pipeline temperature, internal pressure flow rate, and hydrogen blending ratio. The layout of sensor installation locations also requires analysis of potential risk points for pipeline leakage, including welds, valves, elbows, and weak corrosion surfaces.

[0056] Through the analysis of potential risk points and targeted sensor selection and layout, abnormal conditions such as pipeline leakage can be discovered in a timely manner, providing effective risk prevention and control measures for the safe operation of the pipeline and reducing the possibility of accidents such as pipeline leakage. A real-time monitoring method for pipeline hydrogen leakage based on Fluent simulation,

[0057] Step 1: Data acquisition and preprocessing

[0058] S1.1 Obtain leakage and diffusion data of gas inside the pipeline through the monitoring management module and the data acquisition module. S1.1.1 Configure an embedded control chassis at the pipeline distribution section. The chassis integrates a signal conditioning module, a built-in power supply module, and a surge protector to receive monitoring data from multiple sensor modes.

[0059] S1.1.2 For acoustic sensors, use a high-frequency continuous sampling mode and store data in a ring buffer. For pressure sensors, temperature sensors, and infrared concentration detection sensors, use an event-triggered sampling mode. Set the system to quickly start the high-speed acquisition mode when the pressure, temperature, and concentration data exceed the set thresholds.

[0060] S1.2 Preprocessing of collected gas leakage and diffusion data

[0061] S1.2.1 Data alignment and synchronization:

[0062] A precise time protocol is used to achieve microsecond-level time alignment of sensor data, and cubic spline interpolation is performed on asynchronously sampled data to eliminate data timing misalignment caused by differences in sensor sampling times. A three-dimensional pipeline coordinate system is established based on the GIS model, and sensor locations are mapped to coordinate points, unifying the coordinate representation of the spatial locations of different sensors.

[0063] S1.2.2 Data cleaning:

[0064] Missing value processing: fill in missing data according to the mean interpolation method, expressed as

[0065] Outlier processing: Based on the 3σ principle, abnormal data are identified and the mean and standard deviation of the data are calculated. Standard deviation

[0066] The data is standardized and expressed as At this time, the mean is 0 and the standard deviation is 1. A normal distribution diagram based on the standardized data is constructed, and the intervals (μ-σ, μ+σ), (μ-2σ, μ+2σ), and (μ-3σ, μ+3σ) are set. At this time, outliers outside the interval (μ-3σ, μ+3σ) are eliminated;

[0067] S1.3 Data Noise Reduction

[0068] For acoustic signal data, the wavelet denoising method is used to decompose the signal into different frequency bands and remove the mechanical vibration noise in each frequency band. For pressure and temperature signal data, the sliding average filtering method is used to dynamically adjust the sliding window length according to the pipeline response time for filtering.

[0069] Step 2: Feature extraction and fusion

[0070] S2.1 extracts time domain features based on signal fluctuation characteristics and frequency domain features based on FFT spectrum energy distribution;

[0071] The signal fluctuation characteristics include: a. signal mean, which reflects the average deviation characteristics of the signal; b. signal variance, which measures the intensity of signal fluctuation; c. signal kurtosis, which detects the sharpness of the signal and determines whether there is leakage based on the increase in kurtosis data;

[0072] S2.1.1 The time domain feature extraction steps include:

[0073] S2.1.1.1 Calculation of statistical characteristics: The calculation formula for the signal mean and signal variance is:

[0074] Mean: Where n is the number of signal sample data, x i is the i-th signal sample data;

[0075] variance:

[0076] Kurtosis: Where μ4 is the fourth-order central moment.

[0077] S2.1.1.2 Dynamic Characteristics Capture: Capture the pressure drop rate based on the first-order difference, expressed as:

[0078] dP = np.diff(pressure_window) / dt, where dt is the sampling interval, pressure_window is the pressure signal sequence in the current time window, that is, a one-dimensional array, and np.diff() calculates the difference between adjacent elements in the array;

[0079] According to the zero-crossing rate formula, the high-frequency component of the detection signal is calculated as follows:

[0080]

[0081] where x i is the i-th sample value of the discrete-time signal, N is the total number of samples in the time window, which is used to determine the time resolution of the calculation, sgn() is a function that returns the sign of the sample value, and threshold is the threshold. When the signal fluctuation amplitude exceeds the threshold, it is recorded as a valid zero crossing to avoid false counts caused by small noise or quantization errors.

[0082] S2.1.2 Event-driven features:

[0083] Dynamically extract time-domain features based on predefined event trigger conditions (including threshold breakthroughs and signal mutations) to enhance real-time performance and optimize computational efficiency, including:

[0084] a. Peak detection: Use scipy.signal.find_peaks to detect local minima of the pressure signal and set a minimum height threshold to detect negative pressure waves caused by leaks;

[0085] b. Waveform indicators: Set the rise time as the time from 10% to 90% of the signal peak value, and set the decay coefficient as the exponential decay rate after the peak value.

[0086] S2.1.3 The time-frequency domain feature extraction steps include:

[0087] S2.1.3.1 Signal transformation: The time domain signal is transformed into a frequency domain energy distribution through Fourier transform, which is expressed as:

[0088]

[0089] Where X(f) is the frequency domain representation of the Fourier transform, and x(t) represents the time domain signal;

[0090] S2.1.3.2 Calculate the energy contribution of a specific frequency band in the frequency domain energy distribution, expressed as:

[0091]

[0092] For the main frequency component, the frequency and amplitude corresponding to the spectrum peak are extracted, and the frequency domain entropy value is calculated based on the power spectrum entropy to measure the spectrum complexity and calculate the frequency domain characteristics. The entropy value will decrease when leakage occurs, and the power spectrum entropy is expressed as: where p i is the normalized frequency spectrum density, which shows the proportion of the energy of the i-th frequency component to the total energy to reflect the importance of the frequency component; is the power spectrum entropy to characterize the degree of disorder of the signal frequency domain energy distribution;

[0093] S2.1.4 Joint analysis in time and frequency domains

[0094] Generate a time-frequency spectrum based on the short-time Fourier transform (STFT) to capture the transient frequency characteristics of the leakage signal, including:

[0095] S2.1.4.1 Determine relevant parameters, including the original signal, window function, window length, number of overlap points, sampling frequency, and number of Fourier points. The number of Fourier points is primarily used in the Fourier transform process. When the signal length is less than the number of Fourier points, the system automatically pads with zeros before performing the fast Fourier transform.

[0096] S2.1.4.2 Convert the original signal and the window function into column vectors, calculate the signal length, and calculate the number of window slides n based on the signal length nx, the window length WinLen, and the number of overlap points noverlap. This is the number of columns in the signal when the source signal is divided into columns, expressed as n = fix((nx - nverlap) / (WinLen - nverlap)).

[0097] S2.1.4.3 Represent the signal selected by each window function slide as columns, determine the value of each column, and obtain a matrix Fig with n columns and WinLen rows;

[0098] S2.1.4.4 Expand the window function converted into a column vector into an n-column matrix w, perform dot product on the matrix Fig and w, and perform fast Fourier transform on the result of the dot product to obtain a time-frequency matrix. Based on the time-frequency matrix, output the spectrum diagram.

[0099] S2.2 evaluates the correlation by calculating the mutual information of different sensor data, realizes the statistical association of features, and eliminates features with insignificant spatial correlation.

[0100] S2.2.1 Establish a local coordinate system for the pipeline and record the three-dimensional coordinates of each sensor. Construct an adjacency matrix based on the topological relationship. By judging that the distance between two sensors is ≤20m, mark them as adjacent to achieve sensor spatial coordinate mapping.

[0101] S2.2.2 Slice the time domain and frequency domain features of each sensor by time window, construct a spatiotemporal alignment dataset, and construct a time domain feature matrix T, where the matrix dimension is N sensors × M time domain features, expressed as:

[0102] where t ij represents the jth time domain feature of the i-th sensor;

[0103] Construct the frequency domain feature matrix F, where the matrix dimension is N sensors × K frequency domain features, expressed as:

[0104] where f ij represents the jth frequency domain feature of the i-th sensor.

[0105] S2.2.3 Signal propagation delay compensation: Based on the pressure wave propagation time, a time lag is applied to the time domain characteristics of the downstream sensor, expressed as:

[0106]

[0107] where x i 、x j are the axial coordinates of sensors i and j, respectively.

[0108] S2.2.4 Spatial correlation statistics:

[0109] Cross-correlation analysis includes calculating the cross-correlation coefficient of the pressure change rate of sensors i and j and calculating the Pearson correlation coefficient of the high-frequency energy between sensors:

[0110] S2.2.4.1 Calculate the correlation coefficient of the pressure change rate of sensors i and j and analyze the spatial propagation of the time domain characteristics, expressed as:

[0111]

[0112] By finding the time delay τ corresponding to the maximum correlation coefficient max , to infer the leakage direction; dP i (t) is the pressure change rate of sensor i at time t, dP j (t+τ) is the pressure change rate of sensor j at time t+τ, μ i 、μ j are the average values of the pressure change rates of sensors i and j in the time window, τ is the time delay used to capture the time difference of pressure wave propagation, σ i , σ j are the standard deviations of the pressure change rates of sensors i and j, respectively, to reflect data volatility;

[0113] S2.2.4.2 Calculate the Pearson correlation coefficient of high-frequency energy between sensors and analyze the frequency domain energy correlation, expressed as: where ρ ij is the Pearson correlation coefficient, which quantifies the linear correlation of energy changes between sensors; is the covariance of the frequency band energy between sensor i and sensor j, reflecting the coordination of energy changes; is the energy value of sensor i and j in the 50kHz frequency band, are the standard deviations of the frequency band energies of sensor i and sensor j, respectively, representing the amplitude of energy fluctuations; ρ ij The higher it is, the more it indicates that sensor i and sensor j are in the same leakage diffusion path. 50k means that sensors i and j are in the 50KHz frequency band;

[0114] S2.2.5 Spatial covariance matrix includes constructing a joint time-frequency domain covariance matrix and modeling the local covariance to evaluate the correlation between different sensor data, which is expressed as:

[0115] Among them, C TT is the covariance between time domain features, C TF is the covariance of time domain and frequency domain features, C FF is the covariance between frequency domain features, CFT is the covariance of frequency domain and time domain features, and Hotelling's T2 test is used to eliminate features with insignificant spatial correlation.

[0116] S2.3 performs channel filtering and fusion processing on the pressure, flow and concentration data based on the federated Kalman filter FKF.

[0117] S2.3.1 Define the optimal state estimate X g To include all physical quantities to be estimated, it is expressed as:

[0118]

[0119] Where P is the pipeline pressure, Q is the mass flow rate, and C is the hydrogen concentration;

[0120] S2.3.2 Design the channel-wise local filter, expressed as:

[0121] a. Pressure channel: The local state vector is Where P is the pressure value at the sensor location, is the rate of change of pressure over time, i.e., the pressure change rate;

[0122] Construct the dynamic model as a one-dimensional wave equation in is the spatial gradient of flow along the pipeline axis, reflecting the change in flow distribution caused by leakage; c is the hydrogen leakage wave velocity;

[0123] Its observation model is z P =P+v P , where z P is the actual measured value of the pressure sensor, v P is the observation noise caused by the pressure sensor error;

[0124] b. Traffic channel: The local state vector is Where Q is the flow rate of hydrogen per unit area of the pipeline, reflecting the fluid transport state;

[0125] Constructing a dynamic model as the mass conservation equation in is the rate of change of hydrogen flow in the pipeline over time, i.e., the flow rate change rate, is the spatial gradient of the flow rate along the pipeline axis within the unit area of the pipeline leakage port, It is the ratio of the effective area of the pipeline leakage port to the gas density, which reflects the influence of the geometric characteristics of the leakage port on the gas leakage flow rate; It is the spatial gradient of pressure along the pipeline axis, reflecting the change in pressure distribution caused by leakage;

[0126] Its observation model is z Q=Q+v Q , where z Q is the actual measured value of the flow sensor, v Q The observation noise generated by the flow sensor error;

[0127] c. Concentration channel: The local state vector is Where C is the volume concentration of hydrogen in the mixed gas, which directly reflects the degree of leakage; is the spatial gradient of concentration along the pipeline axis, which characterizes the leakage diffusion direction and rate;

[0128] Constructing a dynamic model as a convection-diffusion equation in is the rate of change of hydrogen concentration over time, i.e., the concentration change rate; u is the average flow rate of the gas in the pipeline, which determines the transport of hydrogen with the mainstream fluid; D is the molecular diffusion coefficient of hydrogen in the mixed gas, which reflects the diffusion rate caused by the concentration gradient;

[0129] Its observation model is z C =C+v C , where z C is the actual measurement value of the concentration sensor, reflecting the hydrogen concentration at the current location; v C is the observation noise caused by the concentration sensor error;

[0130] S2.3.3 Channel-wise Kalman Filtering: a. The local KF prediction of the pressure channel is expressed as:

[0131]

[0132] Among them A P is the propagation matrix based on the speed of sound, u P is the external input condition, B P is the control input matrix, reflecting the influence of external input conditions on pressure; X P is the local state vector, w P is the process noise, is the predicted value of the local state vector, and its updated expression is:

[0133] where K P is the Kalman gain, which weighs the confidence of the predicted value and the measured value; is the prediction error covariance matrix, which represents the uncertainty of state prediction; H P is the observation matrix, which maps the state vector to the sensor measurement value; is the transposed matrix of the observation matrix, which is used to back-propagate the measurement residual to the state space; R P To observe the noise covariance, quantify the measurement error of the pressure sensor;

[0134] is the updated state estimate, is the predicted state vector, z P is the actual measured value of pressure, is the pressure measurement value of the predicted state, is the measurement residual, i.e. the difference between the actual measured value and the predicted value;

[0135] b. Flow channel adaptive KF: Adaptive adjustment based on flow fluctuation amplitude to achieve dynamic noise adjustment, expressed as: Q Q =α·var(Q raw )·I, where Q Q is the process noise covariance matrix of the flow channel, which represents the uncertainty of the model prediction; α is the empirical coefficient and is set to 0.1, var(Q raw ) is the variance of the original data of the flow sensor, which reflects the fluctuation intensity of the flow signal; Q raw is the raw data of the flow sensor, I is the unit matrix, and the scalar variance is expanded into the covariance matrix;

[0136] c. Concentration channel extended KF: Construct a nonlinear observation model, expressed as:

[0137] h(X C )=C·e -βy

[0138] where h(X C ) is the nonlinear observation model describing the sensor theoretical output and state quantity, X C is the state vector, C is the hydrogen concentration, β is the sensor attenuation coefficient, and then the Jacobian matrix is calculated for filtering, which is expressed as:

[0139] The Jacobian matrix H C is a nonlinear observation model h(X C ) for the state vector X C The partial derivative matrix of ;

[0140] S2.3.4 Processing the filtered data by fusion, including:

[0141] The information allocation factor is calculated based on the sensor confidence to realize information allocation and weight calculation, which is expressed as: where β P +β Q +β C =1;

[0142] After the information is distributed, the information form is fused, which can be expressed as:

[0143] where βP , β Q , β C are the information distribution factors for pressure, flow and concentration respectively, P P 、P Q 、P C are the state estimation error covariances of each local filter (pressure, flow, concentration) to characterize the uncertainty of the estimation results, X P 、X Q 、X C are the state quantities estimated independently by each local filter, P g is the global state estimation error covariance after fusion, which reflects the overall uncertainty of the fusion result, X g is the optimal state estimate after fusion;

[0144] The optimal state estimate X g Feedback is given to each local filter and its initial state is reset to achieve feedback reset, which is expressed as:

[0145] S2.4 simulates a 5mm aperture leak in a real pipeline to inject a known leakage signal, records the sensor data as a benchmark, and calculates the positioning error MAE and leakage error RMSE of the fusion results, which are expressed as:

[0146]

[0147] where x sim is the leakage point location coordinate predicted by simulation, x real is the actual location coordinate of the leak, C sim is the simulated concentration value of the monitoring point, C real is the measured concentration value at the monitoring point, and n is the number of sampling samples;

[0148] After data fusion, the target positioning error must be less than 0.5m, and the target leakage concentration detection error must be less than 0.396vol. The deviation between the leakage point location predicted by the pipeline simulation model and the actual leakage location is measured based on the positioning error MAE. The deviation between the hydrogen concentration distribution predicted by the simulation model and the measured value is evaluated based on the leakage volume error RMSE.

[0149] Step three: simulate pipeline hydrogen leakage. Build a pipeline model in the simulation analysis module to simulate the pipeline hydrogen leakage process.

[0150] S3.1 Pipeline 3D Reconstruction

[0151] SolidWorks was used to establish a pipeline geometric model based on the pipeline diameter, wall thickness and material as pipeline parameters, and the flange, weld, elbow and valve sizes as component parameters. The leakage port was modeled and processed based on the definition of typical leakage morphology according to historical leakage data, and the aperture range was set to 0.1 to 50 mm.

[0152] The hydrogen doping volume parameter is set to 5% to 30%, and a mixed gas property library is established, which is expressed as: where ρ mix is the density of the mixed gas, x is the mole fraction of hydrogen, is the density of hydrogen component, is the density of methane gas component, μ mix is the viscosity of the mixed gas, is the dynamic viscosity of hydrogen, is the dynamic viscosity of methane gas.

[0153] Predefine the leakage state and build a leakage scenario library, including setting the leakage location distribution, gas diffusion direction and current environmental conditions, specifically:

[0154] 1) Leakage location: welds, valves, and curved and straight pipe sections; 2) Leakage direction: axial direction, radial direction, and oblique direction; 3) Environmental conditions: current wind speed, ambient temperature, etc.

[0155] S3.2 Model simulation establishment

[0156] According to the laws of conservation of mass, energy, momentum and components in pipeline gas leakage simulation, a k-ε turbulence model is established, which is expressed as:

[0157]

[0158] Where ρ is the density in kg / m 3 ; t is time, unit is s; μ is dynamic viscosity, unit is Pa·s, μ t is the viscosity coefficient and μ t =ρC μ k 2 / ε,C μ is the empirical coefficient, σ k , σ ε are the turbulent Prandtl numbers, G k is the term for the generation of turbulent kinetic energy caused by the mean velocity gradient, G b is the term for the generation of turbulent kinetic energy caused by buoyancy, Y M is the effect of compressible velocity flow pulsation expansion on the total dissipation rate, and C 1ε 、C 2ε 、C 3εare empirical coefficients, with values of 1.44, 1.92, and 0.09, respectively; k is the flow energy in J, and x is i =x, y, z, ε is the dissipation rate of flow energy.

[0159] Set the mass and momentum source terms at the leak port, expressed as: in Indicates the leakage mass flow rate, C d is the flow coefficient, which is 0.6 to 0.8; A is the effective area of the leakage port, ρ is the density of the mixed gas in the pipeline, P pipe is the pipeline pressure, i.e. the upstream pressure of the leak point; P amb The ambient pressure is the back pressure downstream of the leak, and the gravity option is activated based on the buoyancy effect g = 9.81 m / s 2 , setting the density ρ as a function related to temperature T, expressed as: R mix =∑y i R i , to construct a pipeline gas leakage diffusion model, where P op is the absolute pressure of the mixed gas in the pipeline, T is the absolute temperature of the mixed gas, R mix is the gas constant of the mixed gas, y i is the mole fraction of the i-th gas component, R i is the gas constant of the i-th gas component.

[0160] S3.3 Grid division

[0161] For the straight pipe section, a hexahedron structured grid is used with a size of 10 to 50 mm. For the complex area, a tetrahedron unstructured grid is used. The grid at the leak location is encrypted with a local encryption of 0.1 mm.

[0162] S3.4 Boundary Condition Setting

[0163] The soil particle size of the pipeline's external environment is set to 0.5 mm, the soil porosity is 0.25, the soil moisture content is 10%, and the ambient air humidity is 30%.

[0164] Considering only the effects of nitrogen, oxygen, carbon dioxide, and water vapor in the air, the calculated volume fractions of oxygen, carbon dioxide, and water vapor in the air are 21.000%, 0.040%, and 0.920%, respectively, while the volume fractions of oxygen, carbon dioxide, and water vapor in the soil are 18.900%, 0.036%, and 10.000%, respectively;

[0165] According to the k-ε turbulence model, the atmospheric outlet boundary is set as Outlet-pressure, and the inlet boundary of the leak port is set as Inletpressure. The pressure is set to 6.4 MPa, the temperature is set to 310.15 K, the atmospheric temperature is set to 293.15 K, the soil temperature is set to 281.15 K, the bottom of the soil layer is set as the Wall boundary, the interface between the atmosphere and the soil is set as the Interior boundary, and the boundaries on both sides of the soil layer are set as the Symmetry boundaries.

[0166] According to the Fluent simulation software combined with model simulation verification, the leakage mass flow rate is output by outputting the hydrogen gas integral fraction cloud map H2vol%, velocity vector, turbulent kinetic energy and pressure and concentration distribution curves along the center line of the pipeline as flow field data. The volume of the combustible cloud in the H2>4% area and the diffusion time of the warning line are used as leakage characteristic data.

[0167] Step 4: Build a hierarchical early warning mechanism

[0168] S4.1 uses real-time sensor data, simulation prediction data, and environmental parameters as input sources, specifically: a. Real-time sensor data: pressure drop rate, hydrogen concentration, and acoustic signal value; b. Simulation prediction data: the flammable cloud diffusion range and lower explosion limit arrival time output by Fluent simulation; c. Environmental parameters: wind speed, temperature, and surrounding population density data;

[0169] S4.2 uses fuzzy logic comprehensive evaluation based on input source data to construct a risk model, which is expressed as: L = α·concentration + β·diffusion rate + γ·population exposure, where α, β, and γ represent the weights of each indicator, respectively, and α = 0.6, β = 0.3, and γ = 0.1;

[0170] S4.3 Based on the risk model, determine the comprehensive risk indicators, build a risk level threshold hierarchy, compare the risk level threshold with the threshold to divide the risk level, and realize the risk factor fusion analysis, where the risk level threshold is expressed as:

[0171] a. When the hydrogen leakage concentration is less than 2% LEL and the diffusion time is greater than 10 minutes, it is judged to be the first level risk level, which is relatively low risk;

[0172] b. When the hydrogen leakage concentration is 2-4% LEL and the diffusion time is 5-10 minutes, it is judged to be the second-level risk level, which is relatively high risk;

[0173] c. When the hydrogen leakage concentration is greater than 4% LEL and the diffusion time is less than 5 minutes, the current risk level is determined to be Level 3, which is extremely high;

[0174] At the same time, the XGBoost model is trained based on historical leakage data to dynamically optimize the risk level threshold.

[0175] Step 5: Provide emergency response decisions based on risk level thresholds and output them to the visual interaction module for display.

[0176] 1) Multi-level alarm triggering: a. Install audible and visual alarms at the pipeline leak point and on the system console based on audible and visual devices; b. Pop up a full-screen red warning on the system HMI interface and display the leak point location in real time; c. Send structured warning information to the responsible person via a third-party platform;

[0177] 2) False alarm suppression: a. The alarm is confirmed when data sampling exceeds the set threshold for three consecutive times and lasts for more than 10 seconds, achieving false alarm filtering control; b. Based on ambient wind speed data, when the wind speed is greater than 10m / s and the concentration fluctuates greatly, the confirmation time is automatically extended to 30s;

[0178] 3) Emergency control execution: a. Close the electric ball valves upstream and downstream of the pipeline leak point and activate the vent valve pressure relief process; b. Start the nitrogen injection mode and cooperate with the spray system to release water mist to reduce the hydrogen concentration; c. Start the air curtain machine around the pipeline to form a gaseous barrier to inhibit gas diffusion; d. Combine with intelligent broadcasting to play directional voice alarms and control the LED indicator lights to automatically plan escape routes.

[0179] By comprehensively considering real-time sensor data, simulation prediction data, and environmental parameters, multi-dimensional data input is achieved, which can fully reflect the actual status and potential risks of pipeline operation, improve the reliability of the early warning mechanism, and use fuzzy logic comprehensive evaluation to build a risk model. Through comprehensive analysis of various indicators, the limitations of single indicator evaluation are avoided, thereby improving the objectivity and effectiveness of risk assessment.

[0180] By clearly dividing risk levels and setting detection thresholds based on key factors such as hydrogen leakage concentration and diffusion time, and dynamically optimizing risk level thresholds by training the XGBoost model based on historical leakage data, the accuracy and effectiveness of the early warning mechanism can be continuously improved, and the adaptability and self-learning ability of the early warning mechanism can be enhanced.

[0181] The multi-level alarm trigger mechanism, which includes on-site sound and light warnings, system interface reminders, and information push notifications to responsible personnel, ensures that relevant personnel at all levels are aware of leakage accidents in a timely manner, effectively improving the comprehensiveness and timeliness of alarm information transmission, and facilitating the rapid initiation of emergency response.

[0182] By setting strict alarm confirmation conditions and dynamically adjusting the confirmation time according to environmental factors, false alarms can be effectively suppressed and resource waste caused by false alarms can be avoided, thereby ensuring the reliability and stability of the system's emergency response. At the same time, the coordination between emergency control execution measures can effectively respond to pipeline gas leakage accidents and minimize the harm and losses caused by the accidents.

[0183] according to Figure 3 show, Figure 3 Figures a, b, and c are pressure cloud maps when the leakage hole diameter is 1, 2, and 5 mm, respectively. Combined with the data in the figure, we can see that there is a leakage hole in a 5-meter straight pipe. As the gas moves from left to right, a semicircular pressure drop is formed from the inlet to the area near the leakage point and gradually weakens. The closer to the leakage point, the smaller the pressure. After the gas passes through the leakage hole, the part close to the upper wall of the pipe instantly compresses the gas, and the pipe wall expands, causing the pressure to suddenly increase and reach a maximum value.

[0184] according to Figure 3 The data shows that when a leak occurs in a straight pipe, the pressure fluctuation changes almost vertically. When the medium is about to approach the leak hole, the force forms a semicircle and gradually weakens. When it passes through the leak hole, high pressure is instantly generated. At this time, the larger the leak hole diameter, the more medium leaks out of the pipeline, resulting in a lower overall pressure in the pipeline. The pressure drop near the leak hole is more stable, the lower the pressure is, the closer the low pressure is to the upper wall of the pipeline, and the higher the high pressure generated by compression after passing through the leak hole. As for external factors, experimental analysis shows that the length of the pipeline has no effect on the overall pressure of the pipeline during leakage.

[0185] according to Figure 4 The data in the figure shows that panels a, b, and c are velocity cloud maps when the leakage hole diameter is 1, 2, and 5 mm, respectively. According to the velocity cloud map, the fluid velocity in the straight pipe with a leakage hole is low in the area close to the upper pipe wall of the entire pipe section, and the velocity decreases as it gets closer to the wall.

[0186] When a leak hole appears in the pipeline, a very small part of the low-speed flow appears along the upper wall of the pipeline. The fluid speed gradually increases until it is ejected from the leak hole and reaches a maximum speed point. After passing the leak hole, the speed increases significantly downward along the upper wall of the pipeline until it returns to the original fluid speed in the pipeline.

[0187] Due to the fluidity of the fluid medium, the flow velocity of the medium before and after the leakage hole is basically stable, and the velocity change is concentrated in the range below the leakage hole. From a large amount of data analysis, it can be seen that the larger the leakage hole diameter, the faster the flow velocity of the medium in the pipeline, the more gradual the velocity increase, and the smaller the increase range. After passing through the leakage hole, the velocity increases significantly downward along the upper pipe wall.

[0188] according to Figure 5 show, Figure 5Panel a shows the pressure change in the pipeline at the first moment, panel b shows the pressure change at the second moment, panel c shows the pressure change at the third moment, and panel d shows the pressure change at the fourth moment. As the time gradient changes, the pressure in the pipeline also gradually changes. Comparing panels a and b in the figure, we can see that as time changes, the pressure distribution of the fluid in the pipeline moves from the two sides to the center, and the pressure gradient gradually widens. The two sides meet and intersect, eventually causing the pressure in the pipeline to stabilize.

[0189] When the pipeline is operating, the pressure flows from the left input to the right. When a leak occurs, the pressure is released and overflowed at the leak hole, causing the pressure flowing to the right to decrease. The pressure distribution around the leak hole is shown in parts c and d in the figure. The pressure shifts and diffuses to the left. As time goes by, the boundary line of the color change gradually blurs, and the pressure transition gradually decreases.

[0190] according to Figure 6 Chinese data show that Figure 6 Panel a shows the velocity change of the leaking gas at the first moment, panel b shows the velocity change of the leaking gas at the second moment, panel c shows the velocity change of the leaking gas at the third moment, and panel d shows the velocity change of the leaking gas at the fourth moment. Compared to pressure, velocity changes are more obvious. As can be seen in parts a and b of the figure, the colors on the sides are lighter, advancing towards the darker center, then blending and gradually becoming lighter. As shown in the icons, the lighter the color, the faster the velocity. When a leak occurs, a negative pressure wave is generated at the leak hole. During a leak, the negative pressure wave propagates from the leak hole to both sides, causing a collision with the normal fluid velocity, resulting in a faster velocity on both sides than in the center. Parts c and d of the figure show that the velocity distribution at the leak hole is also shifted to the left, and the velocity in the pipeline is stable. When a leak hole appears, the fluid sprays toward the leak, causing the velocity to accelerate.

[0191] In summary, this real-time monitoring and early warning system for hydrogen-doped pipeline leakage based on Fluent simulation can accurately capture leakage characteristic data through a multi-sensor collaborative monitoring network combined with data filtering and fusion processing. It also constructs a leakage diffusion model based on pipeline 3D modeling and Fluent simulation analysis, facilitating dynamic simulation of hydrogen concentration field distribution.

[0192] By proposing a dynamic adjustment mechanism for risk thresholds, it can adapt to changes in different complex environments. It uses a fuzzy logic comprehensive evaluation model to facilitate the construction of a comprehensive risk assessment dimension, and establishes a false alarm suppression algorithm to effectively reduce the system's false alarm rate.

[0193] By building a pipeline digital twin model, the efficiency of operation and maintenance decision-making is improved based on visualization capabilities, and combined with the risk threshold mechanism, a multi-level cascade control strategy is adopted to achieve efficient emergency response.

[0194] With the above-described preferred embodiments of the present invention as a guide, and with reference to the above description, relevant personnel are fully capable of making various changes and modifications without departing from the technical spirit of this invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation, characterized in that: The following steps are included: Step 1: Acquire the leakage and diffusion data of the gas inside the multi-sensor detection pipeline and pre-process the data; Step 2: extracting data features. Step 2 specifically includes: S2.1 extracts time domain features based on signal fluctuation characteristics, extracts frequency domain features based on FFT spectrum energy distribution, and jointly analyzes the time and frequency domains to obtain a time-frequency matrix; S2.2 evaluates the correlation by calculating the mutual information of different sensor data, realizes the statistical association of features, and eliminates features with insignificant spatial correlation; S2.3 performs channel filtering and fusion processing on pressure, flow and concentration data based on the federated Kalman filter FKF; Step 3: Establish a pipeline model based on Fluent, use the extracted features as boundary conditions and fluid conditions, and simulate the hydrogen leakage process in the pipeline; Step 4: Build a risk model based on real-time sensor data, simulation prediction data, and environmental parameters, and establish a graded early warning mechanism.

2. The real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation according to claim 1 is characterized in that: The step S2.1, the time domain feature extraction process specifically includes: calculating the signal mean and signal variance; capturing the pressure drop rate according to the first-order difference, calculating according to the zero-crossing rate formula, and detecting the high-frequency components of the signal.

3. The real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation according to claim 1 is characterized in that: In step S2.1, the frequency domain feature extraction process includes: converting the time domain signal into the frequency domain energy distribution through Fourier transform; calculating the energy proportion of a specific frequency band in the frequency domain energy distribution, wherein for the main frequency component, extracting the frequency and amplitude corresponding to the spectrum peak, and calculating the frequency domain entropy value based on the power spectrum entropy to measure the spectrum complexity and calculate the frequency domain features.

4. The real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation according to claim 1 is characterized in that: In step S2.1, the joint analysis of the time-frequency domain specifically includes: generating a time-frequency spectrum according to the short-time Fourier transform STFT to capture the transient frequency characteristics of the leakage signal, determining relevant parameters, including the original signal, window function, window length, number of overlapping points, sampling frequency and number of Fourier points, and performing fast Fourier transform; converting the original signal and the window function into a column vector, calculating the signal length, and calculating the number of window sliding times n according to the signal length nx, the window length WinLen and the number of overlapping points noverlap, that is, the number of columns of the signal when the source signal is divided into columns, expressed as n=fix((nx-noverlap) / (WinLen-noverlap)); expressing the signal selected by each window function sliding as a column, determining the value of each column, and obtaining a matrix Fig with n columns and WinLen rows; expanding the window function converted into a column vector into a matrix w with n columns, and performing dot product on the matrix Fig and w, and performing fast Fourier transform on the result of the dot product to obtain a time-frequency matrix, and outputting the spectrum according to the time-frequency matrix.

5. The real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation according to claim 1 is characterized in that: The step S2.2 specifically includes: S2.2.1 Establish a local coordinate system for the pipeline and record the three-dimensional coordinates of each sensor. Construct an adjacency matrix based on the topological relationship and implement sensor spatial coordinate mapping by determining the distance between two sensors. S2.2.2 Slice the time domain and frequency domain features of each sensor into time windows to construct a spatiotemporal aligned dataset; S2.2.3 Apply a time lag to the time domain characteristics of the downstream sensor based on the pressure wave propagation time; S2.2.4 Calculate the cross-correlation analysis including calculating the cross-correlation coefficient of the pressure change rate of sensors i and j and calculating the Pearson correlation coefficient of the high-frequency energy between sensors; S2.2.5 Construct a spatial covariance matrix to evaluate the correlation between different sensor data and eliminate features with insignificant spatial correlation.

6. The real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation according to claim 1 is characterized in that: The step S2.3 specifically includes: S2.3.1 Define the optimal state estimate X g And P is the pipeline pressure, Q is the mass flow rate, and C is the hydrogen concentration; S2.3.2 Design local filters for pressure, flow, and concentration channels; S2.3.3 Calculate the channel-wise Kalman filter; S2.3.4 performs data fusion processing after filtering and calculates the information allocation factors of pressure, flow, and concentration based on the sensor confidence level to achieve information allocation and weight calculation, which can be expressed as: where β P +β Q +β C =1,tr(P i ) is the trace of the error covariance matrix of the i-th local filter. After the information is distributed, the information form is fused, which is expressed as: where β P , β Q , β C are the information distribution factors for pressure, flow and concentration respectively, P P 、P Q 、P C are the state estimation error covariances of each local filter (pressure, flow, concentration) to characterize the uncertainty of the estimation results, X P 、X Q 、X C are the state quantities estimated independently by each local filter, P g is the global state estimation error covariance after fusion, which reflects the overall uncertainty of the fusion result, X g is the optimal state estimate after fusion; the optimal state estimate X g Feedback is given to each local filter and its initial state is reset to achieve feedback reset, which is expressed as:

7. The real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation according to claim 1 is characterized in that: The step 2 also includes S2.4, simulating a 5mm aperture leakage in a real pipeline to inject a known leakage signal, recording the sensor data as a benchmark, calculating the positioning error MAE and the leakage error RMSE of the fusion result respectively, and verifying the target positioning error after data fusion.

8. The real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation according to claim 1 is characterized in that: In step 4, the risk model is expressed as: L = α·concentration+β·diffusion rate+γ·population exposure, where α, β, and γ represent the weights of each indicator, and α = 0.6, β = 0.3, and γ = 0.1 are set.

9. A real-time monitoring and early warning system for implementing the real-time monitoring and early warning method for pipeline hydrogen leakage based on Fluent simulation as described in any one of claims 1 to 8, characterized in that: include, A monitoring and management module, comprising a plurality of sensors arranged in the pipeline; A data acquisition module, wherein the data acquisition module is connected to the monitoring and management module; A simulation analysis module connected to the data acquisition module; A leakage warning module, which is connected to the simulation analysis module and the data acquisition module; A visualization interaction module is connected to the leakage warning module and the simulation analysis module.

10. The real-time monitoring and early warning system for pipeline hydrogen leakage based on Fluent simulation according to claim 9 is characterized in that: The various sensors include temperature sensors, pressure sensors, gas flow meters, external concentration detection sensors and acoustic sensors. The temperature sensor monitors abnormal pipeline temperature data based on detecting pipeline temperature parameters; the pressure sensor is used to detect the air pressure inside the pipeline; the gas flow meter is used to detect the gas flow rate parameters inside the pipeline; the external concentration detection sensor is used to detect the hydrogen mixing ratio inside the pipeline; the acoustic sensor is used to detect pipeline leakage data. The pressure sensor, infrared concentration detection sensor and acoustic sensor are equidistantly spaced around high-risk points for pipeline leakage.

Images