A method and system for optimizing parameters of high-pressure gas test pressure.

Abstract

This invention belongs to the field of parameter optimization technology, specifically relating to a method and system for optimizing parameters in high-pressure gas detection and testing. The method includes: simultaneously collecting pressure and temperature data from upstream and downstream nodes of a pipeline to construct thermodynamic-pressure coupling characteristics, and extracting a spatiotemporal turbulence calibration factor using physical delay alignment; subsequently, calculating pipeline turbulence dissipation indices based on this calibration factor, and generating dynamic pressure compensation weights based on an explosion-proof safety boundary mechanism; finally, adaptively weighting the pressure sequence and fusing it with temperature features, inputting the results into a deep neural network to output the final optimized parameters. This invention can effectively eliminate spurious interference from spatially propagated waves, capture the true fluid state, prevent the risk of overpressure rupture, and improve the safety and accuracy of automated high-pressure detection and control systems.

Description

Technical Field

[0001] This invention relates to the field of parameter optimization technology. More specifically, this invention relates to a method and system for optimizing parameters in high-pressure gas testing. Background Technology

[0002] In actual industrial production sites, high-pressure gas testing is typically conducted within a closed, long-distance pipeline network. This real-world industrial testing environment differs significantly from the static pressure model of an ideal laboratory environment. In long-distance pipeline pressure testing scenarios, the pipeline not only experiences pressure attenuation due to resistance along the route, but is also susceptible to complex conditions such as vibrations from external equipment startup and shutdown, sudden drops in ambient temperature, or temperature rise due to airflow friction on the pipe walls. These factors trigger internal energy conversion in gas molecules, generating hidden additional thermodynamic expansion stresses. Furthermore, the normal operation of upstream valves inevitably generates spatial pressure operation waves that propagate towards the target testing node.

[0003] However, existing pressure testing control schemes typically rely solely on gauge pressure readings from a single node and employ traditional linear control algorithms such as PID for parameter adjustment. When using these existing technologies to address pressure testing control issues in complex industrial scenarios, the system often fails to effectively isolate spurious interference caused by the normal operation wave propagation of upstream valves. This interference can easily be misinterpreted as leakage or abnormal oscillation within the target pipeline. Especially under severe conditions such as extreme industrial gas shocks or when violent fluid oscillations occur within the pipeline, traditional linear compensation mechanisms can directly amplify these severe abnormal fluctuations linearly. This leads to the underlying control system outputting dangerous pressure testing commands that severely exceed the range, making it easy for pipeline overpressure damage or even rupture and explosion to occur during actual testing. Consequently, this approach fails to meet the explosion-proof safety and control stability requirements of automated high-pressure gas testing processes. Summary of the Invention

[0004] To address the technical problem of poor parameter optimization in high-pressure gas testing, the present invention provides solutions in the following aspects.

[0005] In a first aspect, the present invention provides a method for optimizing parameters of a high-pressure gas testing pressure, comprising: Real-time pressure and temperature of upstream and target nodes in the pipeline network are collected to form real-time pressure and temperature sequences. The relative rates of change of pressure and temperature at the corresponding nodes are calculated. The length of the spatial pipeline and the theoretical velocity of sound are obtained. The thermodynamic-pressure coupling characteristics are constructed based on the relative rates of change of pressure and temperature at the target node and the logarithm of the relative rate of change of pressure. The physical delay sampling steps are calculated based on the spatial pipeline length and the theoretical velocity of sound. The relative rates of change of pressure at the upstream nodes are time-delay aligned according to the physical delay sampling steps to obtain the spatial delay gradient alignment index. The relative rates of change of pressure at the target node and the spatial delay gradient are then used to align the spatial delay gradient. The system obtains the conduction wave decoupling ratio using a uniform index; extracts a spatiotemporal turbulence calibration factor based on a preset continuous time window and the conduction wave decoupling ratio; obtains a pipeline turbulence dissipation index by combining the dispersion of the spatiotemporal turbulence calibration factor with the thermodynamic pressure coupling characteristics; performs exponential asymptotic transformation and controlled logarithmic amplification on the pipeline turbulence dissipation index sequentially to obtain dynamic pressure compensation weights; weights the real-time pressure sequence of the target node using the dynamic pressure compensation weights to obtain a weighted compensation pressure sequence; concatenates the weighted compensation pressure sequence with the real-time temperature sequence to form a fusion input feature matrix, inputs it into a deep neural network model, and outputs optimized parameters for the test pressure.

[0006] In complex industrial closed long-distance pipeline scenarios, this invention can effectively capture the real thermodynamic expansion stress caused by high-frequency transient airflow impact and drastic temperature changes. At the same time, it accurately eliminates the false interference of spatial transmission waves caused by the normal operation of upstream valves. By replacing the traditional PID control algorithm with lag through a nonlinear deep neural network, it realizes intelligent testing of pressure parameters of high-pressure gas under complex and variable working conditions, and optimal reasoning and analysis without human intervention, thus ensuring the high efficiency and operational safety of the automated pipeline testing system.

[0007] Preferably, the construction of the thermodynamic pressure coupling feature includes: Calculate the absolute value of the real-time pressure difference between the target node at the current time and the previous time, divide it by the sum of the real-time pressure at the previous time and the preset first small positive value, and obtain the relative pressure change rate. The natural constant is summed with the relative rate of change of pressure, and the natural logarithm is taken to obtain the logarithmic decay term. Calculate the relative rate of temperature change of the target node, multiply it by the logarithmic decay term, and add 1 to obtain the adjustment coefficient; The thermodynamic pressure coupling feature is obtained by multiplying the real-time pressure at the current moment with the adjustment coefficient.

[0008] When facing application scenarios such as sudden drops in external environment or severe temperature rise due to friction of pipe walls, this invention can smooth out instantaneous data abrupt changes, truly reflect the additional expansion stress caused by the conversion of internal energy of gas molecules, and avoid dangerous pressure test control deviations caused by the system simply reading gauge pressure readings and masking the true pressure state.

[0009] Preferably, obtaining the conducted wave decoupling ratio includes: At each sampling moment of the continuous time window, the absolute value of the difference between the relative rate of change of pressure at the target node and the spatial delay gradient alignment index is calculated. Take the negative of the absolute value of the difference as the exponent, calculate the exponential function value with the natural constant as the base, and obtain the suppression response function; Subtract the suppression response function from 1 to obtain the decoupling ratio of the propagated wave at the target node at the corresponding sampling time.

[0010] When faced with the complex environment of intertwined sound waves and pressure waves in industrial pipeline networks, this invention can accurately distinguish whether the pressure fluctuation of the current node is due to the spatial transmission of normal operating waves from upstream or to the actual abnormal friction or leakage of fluid inside the pipeline, thereby effectively preventing the control system from generating false leakage or abnormal oscillation alarms during normal operation.

[0011] Preferably, the extraction of the spatiotemporal turbulence calibration factor includes: Calculate the difference between the decoupling ratio of the propagated wave at the current sampling time and the decoupling ratio of the propagated wave at the previous sampling time, and use it as the first-order differential rate of change. The square of the decoupling ratio of the conducted wave at the current sampling moment is summed with the square of the first-order differential rate of change to obtain the superposition value of the ground state and the oscillation. The spatiotemporal turbulence calibration factor of the target node at the current sampling time is extracted by taking the square root of the superposition value of the ground state and the oscillation.

[0012] This invention can capture the absolute intensity and high-frequency oscillation amplitude of the real turbulence within a continuous time window when the pipeline is under high-frequency airflow change conditions, providing a targeted calibration basis for subsequent system filtering constraints and preventing the system from overreacting to normal slight airflow fluctuations.

[0013] Preferably, the acquisition of pipeline turbulence dissipation indices includes: The arithmetic mean of the thermodynamic-pressure coupling characteristics within the continuous time window is used as the thermodynamic-pressure coupling index reference surface. The sum of the squared difference between the thermodynamic-pressure coupling characteristics at each moment and the thermodynamic-pressure coupling index reference surface, and the squared difference between the thermodynamic-pressure coupling characteristics at the previous moment, is used as the total transient fluctuation energy. The total transient fluctuation energy is multiplied by the corresponding spatiotemporal turbulence calibration factor and the preset exponential decay coefficient, and then accumulated within the continuous time window to obtain the total filtered energy. The square of the sum of the thermodynamic-pressure coupling index reference surface and the preset second small positive value is multiplied by the total number of sampling points in the continuous time window to obtain the dimensionless denominator. The total filtered energy is divided by the dimensionless denominator to obtain the pipeline turbulence dissipation index.

[0014] In multi-node networked high-pressure gas detection scenarios, this invention can filter out false feature discrepancies caused by spatial conduction, extract pure in-situ pipeline fluid true turbulent dissipation indicators, and improve the reliability of the physical state assessment of target nodes.

[0015] Preferably, obtaining the dynamic pressure compensation weight includes: Take the negative of the pipeline turbulence dissipation index as the exponent, calculate the exponential function value with the natural constant as the base, and subtract the exponential function value from 1 to obtain the exponential asymptotic boundary term; The square term of the pipeline turbulence dissipation index is calculated as the numerator, and the sum of the square term of the pipeline turbulence dissipation index and 1 is used as the denominator. The numerator is divided by the denominator to obtain the controlled term. The controlled logarithm is obtained by summing the natural constant with the controlled term and taking the natural logarithm; the dynamic pressure compensation weight is obtained by multiplying the exponential asymptotic boundary term with the controlled logarithm amplification term and adding 1.

[0016] Preferably, the step of aligning the relative pressure change rate of the upstream node according to the number of physical delay sampling steps to obtain the spatial delay gradient alignment index includes: Using the physical delay sampling steps as the backtracking offset, the historical pressure value at the corresponding moment is extracted from the real-time pressure sequence of the upstream node to obtain the relative pressure change rate of the upstream node, which is used as the spatial delay gradient alignment index of the corresponding target node at the current moment.

[0017] Preferably, the step of concatenating the weighted compensated pressure sequence and the real-time temperature sequence into a fused input feature matrix includes: The weighted compensated pressure sequence and the real-time temperature sequence are concatenated into a multi-dimensional matrix according to a unified timestamp alignment principle to obtain a fused input feature matrix.

[0018] Preferably, the deep neural network model is a multilayer perceptron network, which includes an input layer, multiple fully connected hidden layers, and an output layer.

[0019] Secondly, the present invention provides a parameter optimization system for high-pressure gas testing pressure, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned parameter optimization method for high-pressure gas testing pressure is implemented.

[0020] By adopting the above technical solution, a computer program is generated from the above-mentioned method for optimizing the parameters of a high-pressure gas test pressure, and stored in a memory so that it can be loaded and executed by a processor. A terminal device can then be made based on the memory and the processor for convenient use.

[0021] The beneficial effects of this invention are as follows: (1) This invention introduces multi-dimensional spatiotemporal feature extraction technology. By constructing a spatial delay alignment model, it can perform multi-dimensional linkage analysis on pressure signals at different spatial locations and time nodes in complex industrial pipelines, making up for the information blind spots of traditional single-point monitoring and improving the system's sensitivity and detection accuracy for capturing micro-leakage and high-frequency transient impacts. (2) The present invention constructs a thermodynamic pressure coupling feature model, which can deeply analyze the coupling relationship between the temperature field and the pressure field, identify and remove false pressure fluctuations caused by environmental disturbances or upstream routine operations, and reduce the false alarm rate and false alarm rate of the system by extracting pure turbulent dissipation index as the core criterion, thus ensuring the authenticity and reliability of the detection results. (3) This invention applies the nonlinear mapping capability of deep neural networks to pressure parameter optimization. Faced with the complex nonlinear fluid dynamics characteristics in industrial pipelines, deep neural networks can perform deep learning and adaptive weighted compensation on the extracted multidimensional features and pure turbulent dissipation index, enabling the system to dynamically optimize the pressure test parameters according to real-time operating conditions, thereby improving the intelligence level and automation control level of the high-pressure gas testing process. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a parameter optimization method for high-pressure gas testing pressure according to the present invention; Figure 2 This is a schematic diagram illustrating the timing of synchronous pressure acquisition at upstream and downstream nodes; Figure 3 This is a schematic diagram illustrating the thermodynamic-pressure coupling characteristics of the target node and the time series diagram of the spatiotemporal turbulence calibration factor; Figure 4 This is a schematic diagram showing the time series comparison between the original pressure and the weighted compensated pressure. Detailed Implementation

[0023] This invention discloses a parameter optimization method for high-pressure gas testing pressure, referring to... Figure 1 This includes steps S1-S4: S1: Obtain the preset sampling frequency; synchronously collect the real-time pressure of the upstream node, the real-time pressure of the target node, and the real-time temperature of the target node according to the preset sampling frequency; obtain the spatial pipeline length and theoretical speed of sound.

[0024] It should be noted that the high-pressure gas detection environment in industrial settings is often quite complex. The high-pressure gas state in a closed, long-distance pipeline network differs significantly from the static pressure model in an ideal laboratory environment. Not only is there pressure attenuation due to resistance along the route, but it is also susceptible to interference from equipment start-up and shutdown vibrations and high-frequency electrical noise. Relying solely on low-frequency detection data from a single node cannot capture the transient characteristics of airflow impact. Therefore, this invention pre-configures a high-frequency sampling rate to simultaneously acquire data at upstream and downstream nodes of the pipeline network, and incorporates the geometric length of the physical pipeline and the theoretical sound velocity of the medium, thereby capturing spatiotemporal multidimensional data consistent with the real industrial environment at the physical source.

[0025] Specifically, a preset sampling frequency is obtained; real-time pressure of upstream nodes, real-time pressure of target nodes, and real-time temperature of target nodes are synchronously collected according to the preset sampling frequency, including: A preset sampling frequency is used to control the temporal resolution of synchronous data acquisition across multiple nodes. For example, the preset sampling frequency is set to 200Hz. It should be noted that if the preset sampling frequency is too high, such as 1000Hz, it will overburden the underlying hardware data transmission and computation, and easily introduce invalid high-frequency electrical noise interference, masking the true airflow characteristics. If the preset sampling frequency is too low, such as 10Hz, it will result in insufficient system data resolution, making it impossible to accurately capture high-frequency transient airflow impact characteristics, leading to the loss of critical underlying data.

[0026] By installing pressure and temperature sensors at upstream and target nodes in the high-pressure gas detection pipeline network, the real-time pressure of the upstream node, the real-time pressure of the target node, and the real-time temperature of the target node are synchronously collected at the current sampling time and the previous sampling time according to the preset sampling frequency.

[0027] Preferably, obtaining the length of the spatial conduit and the theoretical velocity of sound includes: Read the 3D design drawing data of the pipeline in the pre-existing control system and extract the spatial pipeline length between the upstream node and the target node; based on the known formula for the speed of sound of an ideal gas, and according to the basic absolute temperature and gas composition settings of the current detection environment, obtain the theoretical speed of sound of the high-pressure gas in the current detection environment.

[0028] It should be noted that, as Figure 2 This is a time-series diagram showing the pressure synchronization between upstream and downstream nodes, illustrating the basic data acquisition process of this invention. The horizontal axis represents time in seconds; the vertical axis represents pressure in megapascals (MPa). The curves in the diagram represent the real-time pressure of the upstream node and the target node, respectively.

[0029] At this point, the collection and acquisition of the basic data required for the entire system has been completed.

[0030] S2: Based on the real-time pressure of the upstream node, the real-time pressure of the target node, and the real-time temperature of the target node, construct the thermodynamic pressure coupling characteristics of the target node; based on the spatial pipeline length, theoretical sound velocity, and preset sampling frequency, obtain the physical delay sampling steps and extract the spatial delay gradient index of the upstream node; based on the relative pressure change rate of the target node and the spatial delay gradient index of the upstream node, obtain the propagation wave decoupling ratio of the target node, and extract the spatiotemporal turbulence calibration factor of the target node within a continuous time window.

[0031] It should be noted that the true pressure state of high-pressure gases in closed pipelines is more significantly affected by the deep coupling of temperature abrupt changes and non-ideal gas effects compared to conventional fluids. For example, a sudden drop in external temperature or the heating caused by friction between the gas flow and the pipe wall can lead to the conversion of internal energy in gas molecules, thereby generating additional thermodynamic expansion stress. Relying solely on gauge pressure readings can mask the true pressure state of the pipeline, leading to dangerous deviations in subsequent pressure testing and control. Therefore, this invention first performs nonlinear deep fusion of the relative rate of change of pressure and the relative rate of change of temperature to construct the thermodynamic-pressure coupling characteristics of the target node that can reflect the true physical expansion stress. Simultaneously, since the normal operation of upstream valves in the pipeline network inevitably generates spatial pressure operation waves, these pressure waves are transmitted to the target node after a physical time delay. If not distinguished, the system is highly likely to misjudge this normal upstream transmission wave as a leak or abnormal oscillation within the target node. Therefore, this invention further utilizes the velocity of sound and pipeline length to calculate the physical delay sampling steps, and rigorously aligns and compares upstream and downstream data in the spatiotemporal dimension, thereby removing the false interference of the upstream transmission wave and extracting the spatiotemporal turbulence calibration factor of the target node that only reflects the in-situ fluid friction and high-frequency oscillation of the target node.

[0032] Specifically, based on the real-time pressure of the upstream node, the real-time pressure of the target node, and the real-time temperature of the target node, the thermodynamic-pressure coupling characteristics of the target node are constructed, including: It should be noted that, based on the well-known ideal gas law in chemical thermodynamics, the pressure of an ideal gas is positively correlated with its thermodynamic temperature. Under unsteady conditions, sudden temperature changes can cause changes in the internal energy of gas molecules, which in turn produces a thermodynamic coupling effect on pressure. Pressure differentials cannot reflect the transformation of internal energy of gas molecules caused by sudden temperature changes. This invention introduces the relative rate of change of temperature over time and combines it with the logarithmic decay term of the relative change of pressure to smooth out instantaneous step changes in data and truly reflect the comprehensive effect of pipeline internal energy transformation on pressure.

[0033] The thermodynamic-pressure coupling characteristics of the target node at any given time satisfy the expression: ; In the formula, It represents the thermodynamic-pressure coupling characteristics of the target node at time t, and its dimensions are consistent with those of pressure. , This represents the real-time pressure of the target node at time t and time t-1; , This represents the real-time temperature of the target node at time t and time t-1. Represents the natural constant; Indicates the absolute value symbol; Represent the natural logarithm function; This represents the first tiny positive value, with dimensions consistent with pressure, used to prevent the denominator from being zero when the pipeline is in an unpressurized, normal-pressure state. For example... .

[0034] In the formula, It represents the relative rate of temperature change at the target node, reflecting the degree of drastic change in the in-situ thermodynamic internal energy of the pipeline; This represents the relative rate of change of pressure at the target node, reflecting the amplitude of transient pressure fluctuations. The logarithmic decay term representing pressure change is used to achieve a smooth transition under pressure step abrupt changes by utilizing the logarithmic mapping of the natural constant; This is denoted as the adjustment coefficient; This comprehensively reflects the multiplication of the logarithmic terms of the relative rate of temperature change and the relative rate of pressure change, and superimposed on the base real-time pressure. This allows the thermodynamic-pressure coupling characteristics to more accurately capture the real thermodynamic expansion stress caused by the conversion of intramolecular energy in high-pressure gas under unsteady conditions.

[0035] Preferably, based on the spatial pipeline length, theoretical sound velocity, and preset sampling frequency, the physical delay sampling steps are obtained, and the spatial delay gradient index of the upstream node is extracted, including: It should be noted that the pressure fluctuations of high-pressure gas in pipelines are not transmitted instantaneously, but rather propagate using the speed of sound in the current environment as the physical limit. Due to the spatial physical distance between the upstream and target nodes, normal pressure operation waves originating upstream require a certain absolute propagation time to reach the target node. Directly using upstream and downstream data at the same timestamp for synchronous comparison would cause spatial misalignment of characteristic waveforms. Therefore, this invention accurately calculates the physical propagation time using the spatial pipeline length and theoretical speed of sound, and converts it into discrete sampling steps by combining the sampling frequency. This allows for accurate backtracking of the upstream historical state in a time series, extracting pressure change characteristics that are of the same origin and phase.

[0036] The absolute conduction time is obtained by dividing the length of the spatial pipeline by the theoretical speed of sound. The physical delay sampling steps are obtained by multiplying the absolute conduction time by the preset sampling frequency. The relative rate of change of pressure at the upstream node at the corresponding delay time is extracted to obtain the spatial delay gradient index of the upstream node.

[0037] Preferably, based on the relative pressure change rate of the target node and the spatial delay gradient index of the upstream node, the propagation wave decoupling ratio of the target node is obtained, and the spatiotemporal turbulence calibration factor of the target node is extracted within a continuous time window, including: A preset continuous time window is used to extract local continuous data segments for calculating spatiotemporal turbulence features. For example, the length of the continuous time window is set to include 150 sampling points. If the length of the continuous time window is too large, such as 500 sampling points, it will result in excessive time delay in feature extraction, making it impossible to respond promptly to transient airflow changes. If the length of the continuous time window is too small, such as 10 sampling points, it will result in the window not covering a complete turbulence fluctuation cycle, leading to incomplete dynamic features extracted.

[0038] It should be noted that, based on the well-known difference normalization method in the field of signal processing, this invention maps the difference between the upstream and downstream pressure change rates to the [0,1] interval through negative exponential mapping, thereby decoupling the upstream propagation wave from the in-situ turbulence of the target node; the improvement lies in accurately distinguishing between normal operating waves and abnormal turbulence signals through the pressure gradient difference aligned in time and space.

[0039] The decoupling ratio of the propagated wave at any time at the target node satisfies the expression: ; In the formula, This represents the number of consecutive time windows at time t for the target node. The decoupling ratio of the propagated wave at each moment; This represents the number of consecutive time windows at time t for the target node. The relative rate of change of pressure at the target node at each moment; This represents the number of consecutive time windows at time t for the target node. Spatial delay gradient alignment index at time step; Represents the natural exponential function; Represents the absolute value symbol.

[0040] In the formula, The absolute value of the difference between the relative rate of change of pressure at the target node at time t and the spatial delay gradient alignment index reflects the degree of physical mismatch between the upstream and downstream pressure waveforms. The suppression response function constructed using the negative exponential mapping comprehensively reflects the extraction mechanism of the decoupling ratio of the propagated wave. When the difference between the two approaches zero, it indicates that the current fluctuation is purely the spatial propagation of the upstream normal operating wave, and the decoupling ratio approaches zero. When the difference is large, the decoupling ratio increases, confirming the existence of strong in-situ fluid friction inside the target node.

[0041] It should be noted that, based on the well-known turbulence intensity characterization method in fluid mechanics, this invention combines the amplitude of the time-domain signal and the first-order differential rate of change, and comprehensively characterizes the absolute intensity and oscillation amplitude of turbulence through Euclidean distance; the improvement lies in extracting a calibration factor that only reflects the in-situ turbulence of the target node based on the proportion of the decoupled propagated wave, and removing the interference of the upstream propagated wave.

[0042] The spatiotemporal turbulence calibration factor of the target node at any time satisfies the expression: ; In the formula, This represents the number of consecutive time windows at time t for the target node. Spatiotemporal turbulence calibration factor at each moment; This represents the number of consecutive time windows at time t for the target node. The decoupling ratio of the propagated wave at each moment.

[0043] In the formula, The square of the absolute ground-state intensity represents the proportion of decoupling of the conducted wave; The square of the first-order differential rate of change, representing the proportion of decoupling of the conducted wave at adjacent moments, reflects the intensity of high-frequency abrupt changes in turbulence; the sum of the square of the absolute intensity of the ground state and the square of the first-order differential rate of change is defined as the superposition value of the ground state and the oscillation. This comprehensively reflects the absolute intensity and high-frequency oscillation amplitude of the real turbulence after decoupling within a continuous time window, thereby extracting a calibration factor that can be used for filtering constraints in subsequent spatial dimensions.

[0044] It should be noted that, as Figure 3The figure shows the time series plot of the thermodynamic-pressure coupling characteristics and spatiotemporal turbulence calibration factor of the target node, illustrating the nonlinear fusion results of temperature and pressure, as well as the turbulence feature extraction results after decoupling from the upstream conducted wave. The horizontal axis represents time in seconds; the left vertical axis represents the thermodynamic-pressure coupling characteristics in megapascals (MPA); and the right vertical axis represents the spatiotemporal turbulence calibration factor.

[0045] Thus, the construction of the basic characteristics of high-pressure gas and the extraction of spatiotemporal turbulence calibration factors have been completed.

[0046] S3: The spatiotemporal turbulence calibration factor of the target node is incorporated into the discreteness calculation of the thermodynamic pressure coupling characteristics of the target node, and filtering constraints are applied to obtain the pipeline turbulence dissipation index of the target node; a controlled logarithmic growth and exponential asymptotic boundary mechanism are introduced to calculate the dynamic pressure compensation weight of the target node with explosion-proof safety boundary based on the pipeline turbulence dissipation index of the target node.

[0047] It should be noted that, since the dissipation and fluctuation of airflow within the pipeline are not uniformly distributed, residual pressure waves and spurious dispersion caused by spatial propagation can interfere with the assessment of the true dissipation intensity of the current target detection node. If the dispersion of the original features is directly used, the system will include normal upstream operations, leading to cognitive bias. Therefore, this invention uses the extracted spatiotemporal turbulence calibration factor of the target node as a forced filtering constraint in the spatial dimension, combined with an exponential decay mechanism with a soft-time switching effect, to calculate the true turbulent dissipation index of the pipeline fluid after removing interference. Furthermore, under severe conditions such as extreme industrial shocks, extremely violent fluid oscillations can occur within the pipeline, resulting in abnormally large dissipation index values. If a linear ratio is used to map this value as a compensation parameter without restraint, the control system will output a dangerous pressure test command that exceeds the range, easily leading to pipeline overpressure rupture. Therefore, this invention further introduces a safety explosion-proof mechanism combining controlled logarithmic growth and exponential asymptotic soft boundary to reduce the divergence trend, ensuring that the dynamic pressure compensation weight of the target node always converges smoothly within the explosion-proof safety threshold range.

[0048] Specifically, the spatiotemporal turbulence calibration factor of the target node is incorporated into the discreteness calculation of the thermodynamic-pressure coupling characteristics of the target node, and filtering constraints are applied to obtain the pipeline turbulence dissipation index of the target node, including: Obtain the thermodynamic-pressure coupling index sequence within a continuous time window at any given time, and calculate the arithmetic mean of the thermodynamic-pressure coupling index sequence as the thermodynamic-pressure coupling index reference surface at that time.

[0049] It should be noted that, based on the statistically known method of calculating variance dispersion, variance is a general technical means to characterize the degree of signal fluctuation. This invention introduces a spatiotemporal turbulence calibration factor and a time exponential decay weight to impose weighted constraints on the dispersion calculation, thereby removing the false dispersion caused by upstream propagation waves and thus accurately extracting the true turbulence dissipation intensity of the pipeline.

[0050] The pipeline turbulence dissipation index at any given time at the target node satisfies the following expression: ; In the formula, The pipeline turbulence dissipation index at time t represents the target node, and is dimensionless. This represents the total number of sampling points within a continuous time window; , This represents the thermodynamic-pressure coupling characteristics at time i and time i-1 within a continuous time window at time t of the target node; The reference surface represents the thermodynamic-pressure coupling index of the target node at time t; This represents the number of consecutive time windows at time t for the target node. Spatiotemporal turbulence calibration factor at each moment; Represents the natural constant; This represents the second small positive value, consistent with the dimensions of the thermodynamic pressure coupling characteristic, used to avoid a denominator of 0. For example... .

[0051] In the formula, It represents the superposition of the global dispersion of the target node and the local oscillation intensity, reflecting the total transient fluctuation energy of the gas in the local pipeline; This represents the exponential decay coefficient based on a time-varying mechanism, which varies with the sequence number. Gradually approaching the end of the current time window The attenuation coefficient adaptively approaches Give greater computational weight to recent characteristic fluctuations; This represents the squared term in the denominator with a small positive value to prevent zero, ensuring dimensional consistency with the numerator's discreteness. This comprehensively reflects the process of multiplying the spatiotemporal turbulence calibration factor and time decay coefficient point by point in the discreteness calculation sequence, reducing spurious feature discrepancies caused by spatial wave propagation, and thus extracting pure pipeline turbulence dissipation indices. In this calculation, Defined as the total filtering energy, Defined as a dimensionless denominator.

[0052] Preferably, a controlled logarithmic growth and exponential asymptotic boundary mechanism is introduced to calculate the dynamic pressure compensation weight of the target node with explosion-proof safety boundary based on the pipeline turbulence dissipation index of the target node, including: It should be noted that, based on the well-known soft-limit saturation control mechanism in industrial process control, the adaptive adjustment of the compensation weight can be achieved through exponential asymptotic boundary and controlled logarithmic growth; this invention introduces a dual limiting mechanism to prevent the compensation weight from exceeding the limit under extreme operating conditions, thus ensuring the explosion-proof safety of high-pressure gas detection from the underlying algorithm.

[0053] The dynamic pressure compensation weight of the target node at any given time satisfies the following expression: ; In the formula, This represents the dynamic pressure compensation weight of the target node at time t. The pipeline turbulence dissipation index represents the target node at time t. Represents the natural constant; This represents the natural logarithm function.

[0054] In the formula, This represents the exponential asymptotic boundary term, ensuring that the additional dynamic compensation term can adaptively approach zero when turbulent dissipation is minimal and approaches a steady state. This indicates that by using a rational fraction structure with added first-order terms in the denominator, the explosive growth rate of the internal square terms is reduced, and this is defined as a controlled term; Defined as a controlled logarithmic amplification term; This comprehensively reflects that when the values ​​of the pipeline fluid turbulence dissipation index are abnormal, the growth rate of several internal parameters is subject to forced damping control, so that the output dynamic pressure compensation weight smoothly converges to the system's preset safe physical threshold boundary, and thoroughly prevents the output of dangerous parameters exceeding the limit from the underlying control algorithm.

[0055] Thus, dynamic pressure compensation weights with explosion-proof safety boundaries were obtained.

[0056] S4: Adaptively weight the real-time pressure sequence of the target node using the dynamic pressure compensation weight of the target node to obtain the weighted compensation pressure sequence of the target node; concatenate the weighted compensation pressure sequence of the target node with the real-time temperature sequence of the target node to generate the fusion input feature matrix of the target node; input the fusion input feature matrix of the target node into a pre-trained deep neural network model for nonlinear mapping to output the optimized parameters of the high-pressure gas detection pressure, thus completing the parameter optimization of the detection system.

[0057] It should be noted that due to the significant uncertainties in industrial environments, the raw sensor data often carries complex apparent measurement errors. The execution layer adjustment parameters of high-pressure gas testing systems exhibit a high-dimensional nonlinear relationship with multidimensional features, often resulting in lag in traditional PID linear control algorithms. Therefore, this invention concatenates weighted and enhanced pressure and temperature features, utilizing a neural network for nonlinear mapping in the high-dimensional implicit space. This intelligently infers the optimal testing parameters, avoiding the subjective lag of manual intervention and ensuring the safety of the automated high-pressure detection and control system.

[0058] Specifically, the real-time pressure sequence of the target node is adaptively weighted using the dynamic pressure compensation weights of the target node to obtain the weighted compensated pressure sequence of the target node; the weighted compensated pressure sequence of the target node is then concatenated with the real-time temperature sequence of the target node to generate the fusion input feature matrix of the target node, including: Obtain the real-time pressure sequence and real-time temperature sequence of the target node within the continuous time window corresponding to the current sampling time.

[0059] The dynamic pressure compensation weight of the target node is multiplied by the real-time pressure sequence of the target node to complete the adaptive data augmentation of the in-situ real dissipation state and obtain the weighted compensation pressure sequence of the target node. The weighted compensation pressure sequence of the target node and the real-time temperature sequence of the corresponding target node are concatenated into a multi-dimensional matrix according to a unified timestamp alignment principle to generate a fusion input feature matrix of the target node with unified dimensions and format.

[0060] It should be noted that, as Figure 4 This is a time-series graph comparing the original pressure and the weighted compensated pressure, demonstrating the optimization effect of this invention on the original pressure data and verifying the effectiveness of the dynamic pressure compensation weight. The horizontal axis represents time in seconds; the vertical axis represents pressure in megapascals (MPa). The curves in the graph represent the original pressure and the weighted compensated pressure of the target node, respectively.

[0061] Preferably, the fused input feature matrix of the target node is input into a pre-trained deep neural network model for nonlinear mapping, outputting optimized parameters for the high-pressure gas detection test pressure, thus completing the parameter optimization of the detection system, including: The learning rate parameter of the preset deep neural network model is set. For example, the learning rate parameter is set to 0.002. It should be noted that if the learning rate parameter is too large, such as 0.1, it will cause the model to oscillate violently or even cause gradient explosion, resulting in deviations from the optimized parameters in the output; if the learning rate parameter is too small, such as 0.00001, it will cause the model to fit slowly, easily get stuck in local optima, and fail to accurately adapt to the complex and ever-changing real industrial gas inspection conditions.

[0062] A deep neural network model is established using a multilayer perceptron architecture comprising one input layer, multiple fully connected hidden layers, and one output layer. The fused input feature matrix of the target node is input into the pre-trained deep neural network model for high-dimensional feature space data mapping processing. The output layer of the deep neural network model directly outputs the final optimized parameters for the high-pressure gas detection test pressure and sends these optimized parameters as electrical signals to the underlying control valve actuator of the detection pipeline, completing the automated final adjustment and parameter optimization of the high-pressure gas detection system. For example, the number of fully connected hidden layers is set to three, with 64, 32, and 16 neurons in each hidden layer, respectively.

[0063] This completes the parameter optimization and safety control of the high-pressure gas testing pressure.

[0064] This invention also discloses a parameter optimization system for high-pressure gas testing pressure, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a parameter optimization method for high-pressure gas testing pressure according to the present invention.

[0065] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0066] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for optimizing parameters of a high-pressure gas detector, characterized in that, include: The system collects real-time pressure and real-time temperature data of upstream and target nodes in the pipeline network to form real-time pressure and temperature sequences, calculates the relative rate of change of pressure and temperature at the corresponding nodes, and obtains the length of the spatial pipeline and the theoretical velocity of sound. The thermodynamic-pressure coupling characteristics are constructed based on the real-time relative change rates of pressure, temperature, and the logarithm of the relative change rate of pressure at the target node. The physical delay sampling steps are calculated based on the spatial pipeline length and the theoretical sound velocity. The relative pressure change rate of the upstream node is time-delay aligned according to the physical delay sampling steps to obtain a spatial delay gradient alignment index. The decoupling ratio of the conducted wave is obtained using the relative pressure change rate of the target node and the spatial delay gradient alignment index. A spatiotemporal turbulence calibration factor is extracted based on a preset continuous time window and the conducted wave decoupling ratio. The pipeline turbulence dissipation index is obtained by combining the dispersion of the spatiotemporal turbulence calibration factor and the thermodynamic pressure coupling characteristics. The pipeline turbulence dissipation index is sequentially subjected to exponential asymptotic transformation and controlled logarithmic amplification to obtain dynamic pressure compensation weights. The real-time pressure sequence of the target node is weighted using the dynamic pressure compensation weights to obtain a weighted compensated pressure sequence. The weighted compensated pressure sequence and the real-time temperature sequence are concatenated into a fusion input feature matrix and input into a deep neural network model to output optimized parameters for the test pressure.

2. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The construction of the thermodynamic pressure coupling feature includes: Calculate the absolute value of the real-time pressure difference between the target node at the current time and the previous time, divide it by the sum of the real-time pressure at the previous time and the preset first small positive value, and obtain the relative pressure change rate. The natural constant is summed with the relative rate of change of pressure, and the natural logarithm is taken to obtain the logarithmic decay term. Calculate the relative rate of temperature change of the target node, multiply it by the logarithmic decay term, and add 1 to obtain the adjustment coefficient; The thermodynamic pressure coupling feature is obtained by multiplying the real-time pressure at the current moment with the adjustment coefficient.

3. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The process of obtaining the decoupling ratio of the conducted wave includes: At each sampling moment of the continuous time window, the absolute value of the difference between the relative rate of change of pressure at the target node and the spatial delay gradient alignment index is calculated. Take the negative of the absolute value of the difference as the exponent, calculate the exponential function value with the natural constant as the base, and obtain the suppression response function; Subtract the suppression response function from 1 to obtain the decoupling ratio of the propagated wave at the target node at the corresponding sampling time.

4. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The extraction of spatiotemporal turbulence calibration factors includes: Calculate the difference between the decoupling ratio of the propagated wave at the current sampling time and the decoupling ratio of the propagated wave at the previous sampling time, and use it as the first-order differential rate of change. The square of the decoupling ratio of the conducted wave at the current sampling moment is summed with the square of the first-order differential rate of change to obtain the superposition value of the ground state and the oscillation. The spatiotemporal turbulence calibration factor of the target node at the current sampling time is extracted by taking the square root of the superposition value of the ground state and the oscillation.

5. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The acquisition of pipeline turbulence dissipation indices includes: The arithmetic mean of the thermodynamic-pressure coupling characteristics within the continuous time window is used as the thermodynamic-pressure coupling index reference surface. The sum of the squared difference between the thermodynamic-pressure coupling characteristics at each moment and the thermodynamic-pressure coupling index reference surface, and the squared difference between the thermodynamic-pressure coupling characteristics at the previous moment, is used as the total transient fluctuation energy. The total transient fluctuation energy is multiplied by the corresponding spatiotemporal turbulence calibration factor and the preset exponential decay coefficient, and then accumulated within the continuous time window to obtain the total filtered energy. The square of the sum of the thermodynamic-pressure coupling index reference surface and the preset second small positive value is multiplied by the total number of sampling points in the continuous time window to obtain the dimensionless denominator. The total filtered energy is divided by the dimensionless denominator to obtain the pipeline turbulence dissipation index.

6. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The process of obtaining dynamic stress compensation weights includes: Take the negative of the pipeline turbulence dissipation index as the exponent, calculate the exponential function value with the natural constant as the base, and subtract the exponential function value from 1 to obtain the exponential asymptotic boundary term; The square term of the pipeline turbulence dissipation index is calculated as the numerator, and the sum of the square term of the pipeline turbulence dissipation index and 1 is used as the denominator. The numerator is divided by the denominator to obtain the controlled term. The controlled logarithm is obtained by summing the natural constant with the controlled term and taking the natural logarithm; the dynamic pressure compensation weight is obtained by multiplying the exponential asymptotic boundary term with the controlled logarithm amplification term and adding 1.

7. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The step of aligning the relative pressure change rate of the upstream node according to the physical delay sampling steps to obtain the spatial delay gradient alignment index includes: Using the physical delay sampling steps as the backtracking offset, the historical pressure value at the corresponding moment is extracted from the real-time pressure sequence of the upstream node to obtain the relative pressure change rate of the upstream node, which is used as the spatial delay gradient alignment index of the corresponding target node at the current moment.

8. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The step of concatenating the weighted compensated pressure sequence and the real-time temperature sequence into a fused input feature matrix includes: The weighted compensated pressure sequence and the real-time temperature sequence are concatenated into a multi-dimensional matrix according to a unified timestamp alignment principle to obtain a fused input feature matrix.

9. The parameter optimization method for high-pressure gas detection test pressure according to claim 1, characterized in that, The deep neural network model is a multilayer perceptron network, which includes an input layer, multiple fully connected hidden layers, and an output layer.

10. A parameter optimization system for high-pressure gas detection pressure, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a parameter optimization method for high-pressure gas testing pressure according to any one of claims 1-9.

Images