Safety operation monitoring system for environmental protection gas cabinet based on leakage rate adaptive compensation

Abstract

The application discloses an environmental protection gas cabinet safety operation monitoring system based on leakage rate adaptive compensation and relates to the technical field of environmental protection gas cabinet safety monitoring. In view of the problem that the prior art cannot accurately measure and calculate the real micro leakage rate under the interference of multi-physical field coupling, the application acquires transient load current and pressure and other multi-source data through an acquisition module; estimates the real air pressure based on the Kalman filtering method of the augmented state space model by using a filtering module; unifies the time scale and calculates the minute-level dynamic volume by an aggregation module; calculates the dynamic compression factor by relying on the mixture property evolution network, combines the steady-state parameters and the dynamic volume to solve the total amount of substance of the mixed gas and to obtain the leakage rate by a solving module; and triggers the hierarchical early warning and the air supplement strategy according to the leakage rate and the compression factor deviation degree by an early warning module. The application effectively filters out the transient thermal shock error, adaptively compensates the volume and the property change, realizes the prospective early warning of the component imbalance and the insulation deterioration risk.

Description

Technical Field

[0001] This invention relates to the field of environmental gas cabinet safety monitoring technology, and in particular to an environmental gas cabinet safety operation monitoring system based on adaptive compensation for leakage rate. Background Technology

[0002] Ring main units (RNBs) are a fundamental component of power distribution networks. With the advancement of green development in the power grid, traditional SF6 insulating gas, which has a strong greenhouse effect, is gradually being replaced by environmentally friendly mixed gases such as perfluoroisobutyronitrile (C4F7N) and carbon dioxide (CO2). However, these environmentally friendly gas RNBs face potential problems such as flange aging and leakage, and decreased insulation performance during long-term operation. Without effective early detection and warning, even minor leaks can escalate over time, eventually leading to serious power failures. Existing technology discloses a method for detecting anomalies in RNBs based on gas state monitoring. This method primarily uses monitoring data such as the pressure, density, moisture content, and mixing ratio of the mixed gas. It calculates the reliability index of each data point using the coefficient of variation to measure the stability of the data. Subsequently, a nonlinear cumulative damage method is used to assess the abnormal state of the gas. This scheme combines multiple state parameters for comprehensive scoring and considers the working states of various sensors. In actual operation, the sudden change in the busbar load current can cause instantaneous heating and violent expansion of the local gas in the gas chamber, forming fluid convection interference similar to a low-frequency shock wave. This leads to transient measurement errors such as overshoot and high-frequency oscillation in the pressure sensor. The statistical evaluation and conventional filtering methods of existing technologies are difficult to isolate the dynamic errors caused by the coupling of electrothermal and multi-physics fields in the time domain. Existing monitoring methods usually treat the gas chamber volume as a constant, ignoring the thermal expansion and contraction effect of the metal cabinet due to changes in ambient temperature and operating temperature rise. When environmentally friendly mixed gases composed of different molecular sizes leak, selective leakage is very likely to occur, such as CO2 molecules preferentially escaping. At the same time, under extreme conditions, local liquefaction of C4F7N gas may also occur. Existing technologies have not been able to deeply analyze the dynamic evolution law of physical property parameters caused by the drift of mixed gas components. Summary of the Invention

[0003] The technical problem solved by this invention is that existing technologies are unable to accurately filter out the dynamic measurement errors of sensors caused by transient thermal shocks under complex multi-physics field coupling interference, and do not compensate for the dynamic changes in physical properties caused by thermal expansion and contraction of the cabinet and selective leakage of mixed gases, resulting in the inability to accurately calculate the true micro-leakage rate.

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a safety operation monitoring system for environmentally friendly gas cabinets based on adaptive leakage rate compensation, comprising: The data acquisition module is used to acquire multi-source data from the gas cabinet; The filtering module is used to estimate the real air pressure based on the transient load current and transient pressure in the multi-source data using the Kalman filtering method of the augmented state-space model; The aggregation module is used to unify the time scale of multi-source data to obtain minute-level steady-state parameters and calculate minute-level dynamic volume. The solution module is used to calculate the current dynamic compressibility factor based on the mixture evolution network, calculate the total amount of substance of the mixed gas by combining the minute-level steady-state parameters and minute-level dynamic volume, and calculate the leakage rate based on numerical differentiation. The early warning module is used to calculate the cumulative mass loss ratio based on the leakage rate and the compressibility factor deviation, and to trigger graded early warning and gas replenishment strategies based on the cumulative mass loss ratio and the compressibility factor deviation.

[0005] Preferably, the process of acquiring the multi-source data in the acquisition module specifically includes: Transient load current and transient pressure are collected at a preset high-frequency sampling rate; The rated cavity volume, calibrated temperature, and linear expansion coefficient of the cabinet material of the environmentally friendly gas cabinet are collected.

[0006] Preferably, in the filtering module, the Kalman filter based on the augmented state-space model has an augmented state vector that includes the actual air pressure, a control quantity that is the effective value current squared difference term, and a control matrix that includes coupling coefficients. The process of estimating the actual air pressure specifically includes: The root mean square of the transient load current within a preset sliding window is calculated to obtain the effective value current, and the square difference of the effective value current at adjacent times is taken as the effective value current square difference term. The transient pressure at each sampling moment and the squared difference term of the effective current are used as inputs to the Kalman filter based on the augmented state space model. Kalman filter prediction and update iteration operations are performed to output the optimal augmented state estimation vector. The component of the optimal augmented state estimation vector corresponding to the real air pressure is used as the real air pressure estimate at the current moment.

[0007] Preferably, the process of obtaining the effective value current squared difference term specifically includes: Calculate the root mean square of all transient load current data within the preset sliding window as the high-frequency effective value current at the current moment; Calculate the square of the high-frequency RMS current at the current moment, and subtract the square of the high-frequency RMS current at the previous moment from the square of the high-frequency RMS current at the current moment to obtain the RMS current square difference term.

[0008] Preferably, the process of obtaining the minute-level steady-state parameters specifically includes: Using a preset time interval as the dividing point, the arithmetic mean of each type of multi-source data and the estimated actual air pressure within the interval is calculated to obtain the minute load current, minute cabinet temperature, minute gas temperature and minute air pressure. The minute load current, minute cabinet temperature, minute gas temperature, and minute gas pressure constitute minute-level steady-state parameters.

[0009] Preferably, the process of calculating the minute-level dynamic volume specifically includes: Extract the coefficient of linear expansion and the calibration temperature, and use a preset multiple of the coefficient of linear expansion as the coefficient of volume expansion; Calculate the minute-level dynamic volume based on the rated cavity volume, volume expansion coefficient, and minute cabinet temperature.

[0010] Preferably, the hybrid evolution network is a fully connected feedforward neural network, comprising an input layer, a single hidden layer, and an output layer; The input layer contains four nodes, which correspond to the normalized minute pressure, minute gas temperature, preceding equivalent compressibility factor, and leakage rate of the previous minute, respectively. The preceding equivalent compressibility factor is calculated based on the ideal gas law using the minute-level gas pressure, minute-level gas temperature, minute-level dynamic volume, and the total amount of mixed gas in the previous minute. The single hidden layer contains 8 nodes and uses the tanh hyperbolic tangent activation function; The output layer contains one node and uses the purelin linear activation function to output a dynamic compression factor in segmented intervals.

[0011] Preferably, the process of obtaining the leakage rate specifically includes: The amount of substance is calculated based on the dynamic compressibility factor output by the mixture evolution network, combined with the minute-level gas pressure, the minute-level dynamic volume, the ideal gas constant, and the minute-level gas temperature. The calculated amount of substance is numerically differentiated to obtain the leakage rate.

[0012] Preferably, the logic for the tiered early warning system specifically includes: The absolute difference between the dynamic compressibility factor and the reference compressibility factor is taken as the compressibility factor deviation; the cumulative mass loss ratio is calculated based on the leakage rate and the initial total amount of material. When the cumulative quality loss ratio is less than a preset loss threshold and the compression factor deviation is less than a preset deviation threshold, it is determined to be in normal operating condition; When the cumulative mass loss ratio is greater than or equal to the preset loss threshold, and the compression factor deviation is less than the preset deviation threshold, an early warning of excessive cumulative leakage is triggered. When the deviation of the compressibility factor is greater than or equal to the preset deviation threshold, a gas component imbalance warning is triggered and the gas replenishment strategy is executed.

[0013] Preferably, the qi replenishment strategy includes: The pre-stored gas replenishment mapping table is invoked. The gas replenishment mapping table records the correspondence between different compressibility factor deviation nodes and the standard state volume of pure carbon dioxide gas to be replenished. Based on the current compressibility factor deviation, the corresponding gas replenishment volume is queried in the gas replenishment mapping table; if the current compressibility factor deviation falls between two adjacent nodes in the gas replenishment mapping table, the gas replenishment volume is determined by an interpolation algorithm.

[0014] The beneficial effects of this invention are as follows: This invention introduces a Kalman filtering method based on an augmented state-space model, using the effective value current squared difference term as the control input to establish a physical causal mapping relationship between electricity, heat, and force. This invention solves the problem of baseline drift in micro-leakage rate calculation by unifying the time scale of data and introducing a volume expansion coefficient to adaptively calculate the dynamic volume at the minute level. This invention constructs a mixture property evolution network with a fully connected feedforward neural network structure, and uses parameters such as the pre-order equivalent compressibility factor reflecting the drift of gas components to calculate the dynamic compressibility factor in the segmented interval, deeply analyzing the dynamic evolution law of the physical property parameters, and then combining numerical differentiation methods to obtain a truly accurate micro-leakage rate. Attached Figure Description

[0015] Figure 1 This is a basic flowchart of an environmental gas cabinet safety operation monitoring system based on adaptive leakage rate compensation, provided as an embodiment of the present invention. Detailed Implementation

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0017] Example, refer to Figure 1 It provides a monitoring system for the safe operation of environmentally friendly gas cabinets based on adaptive compensation for leakage rate, including: The data acquisition module is used to acquire multi-source data from the gas cabinet; The filtering module is used to estimate the real air pressure based on transient load current and transient pressure from multi-source data using the Kalman filtering method of an augmented state-space model. The aggregation module is used to unify the time scale of multi-source data to obtain minute-level steady-state parameters and calculate minute-level dynamic volume. The solution module is used to calculate the current dynamic compressibility factor based on the mixture evolution network, calculate the total amount of substance of the mixed gas by combining minute-level steady-state parameters and minute-level dynamic volume, and calculate the leakage rate based on numerical differentiation. The early warning module is used to calculate the cumulative mass loss ratio based on the leakage rate and the compressibility factor deviation, and to trigger graded early warning and gas replenishment strategies based on the cumulative mass loss ratio and the compressibility factor deviation.

[0018] This invention overcomes the technical problem of accurately measuring micro-leakage rate under multi-physics field coupling interference by integrating the acquisition, filtering, aggregation compensation and calculation network of multi-source data. It can adaptively eliminate the dynamic measurement error of thermal shock caused by transient large load current, and simultaneously compensate for the physical deformation of cabinet thermal expansion and contraction and the property variation caused by the imbalance of mixed gas components, thereby realizing the extraction of extremely weak real leakage characteristics from complex noise.

[0019] The data acquisition module's process of acquiring multi-source data specifically includes: Transient load current and transient pressure are collected at a preset high-frequency sampling rate; The rated cavity volume, calibrated temperature, and linear expansion coefficient of the cabinet material of the environmentally friendly gas cabinet are collected.

[0020] In one specific embodiment of the present invention, the transient load current of the busbar and the original transient pressure of the mixed gas are collected by a current transformer deployed on the busbar side of the gas cabinet and a piezoresistive gas pressure sensor at the gas chamber valve, with a preset high-frequency sampling rate of 1000Hz. Using a patch-type PT100 platinum resistance temperature sensor at a low sampling rate of 1Hz, the temperature of the metal cabinet shell surface is collected. By employing sleeve isolation or fiber optic temperature measurement technology, temperature sensors are deployed in the upper, middle and lower insulation safety areas inside the gas chamber to collect the gas temperature at three points. The arithmetic mean of these three temperature readings is then taken to obtain the average internal gas temperature as the gas temperature. For 1Hz temperature data, the zero-order hold method is used to interpolate upwards to the 1000Hz time axis. That is, within one temperature sampling period, the subsequent 1000Hz time points all use the most recent temperature sampling value, thereby aligning the high and low frequency data in time. The rated cavity volume of the environmentally friendly gas cabinet and the ambient temperature at which the volume was calibrated were obtained from the equipment's factory documentation; in this embodiment, it was 20°C. The linear expansion coefficient of the metal materials used in the manufacture of the cabinet is collected.

[0021] In the filtering module, the Kalman filter based on the augmented state-space model has an augmented state vector containing the actual air pressure, a control quantity that is the effective value current squared difference term, and a control matrix that contains coupling coefficients. The process of estimating the actual air pressure specifically includes: The root mean square of the transient load current within a preset sliding window is calculated to obtain the effective value current, and the square difference of the effective value current at adjacent times is taken as the effective value current square difference term. The transient pressure and the squared difference of the effective current at each sampling time are used as inputs to a Kalman filter based on an augmented state-space model. Kalman filter prediction and update iteration operations are performed to output the optimal augmented state estimation vector. The component of the optimal augmented state estimation vector corresponding to the real air pressure is used as the real air pressure estimate at the current time.

[0022] The process of obtaining the effective value of the current squared difference term specifically includes: Calculate the root mean square of all transient load current data within the preset sliding window as the high-frequency effective value current at the current moment; Calculate the square of the high-frequency RMS current at the current moment, and subtract the square of the high-frequency RMS current at the previous moment from the square of the high-frequency RMS current at the current moment to obtain the RMS current square difference term.

[0023] In a specific embodiment of the present invention, when the bus load current undergoes a step change, the heating power of the bus is proportional to the square of the effective value of the current. The sudden change in heating power will cause the local gas in the air chamber to be heated instantly and expand violently, forming fluid convection interference in the sealed cavity. The dynamic response of the piezoresistive pressure sensor itself is insufficient, and its sensitive element will generate forced vibration, resulting in the transient pressure containing overshoot spikes and high-frequency oscillations with a rate of change higher than the actual pressure. This is the dynamic measurement error of the sensor itself. Traditional frequency domain analysis methods will fail due to spectrum leakage and loss of time and frequency information. This embodiment solves the above problems by constructing a Kalman filter based on an augmented state-space model. The process of constructing a Kalman filter based on an augmented state-space model specifically includes: By combining laboratory calibration with subspace system identification algorithm, a fifth-order discrete-time state-space model of the same type of piezoresistive barometric pressure sensor as the field was established, and the 5×5 system matrix A, 5×1 input matrix B, 1×5 observation matrix C, and 1×1 direct transfer matrix D in the fifth-order discrete-time state-space model were obtained. To achieve online estimation of real air pressure, this embodiment uses real air pressure as one of the state variables to be estimated and constructs an augmented state vector. The mathematical expression for the augmented state-space model, which strips away sensor dynamic errors in real time within the time domain, is as follows: ; in, Let A be the predicted value of the augmented state vector at the next time step j+1, and let A be the system matrix, B be the input matrix, C be the observation matrix, and D be the direct transfer matrix. Let j be the augmented state vector at the current time. The coupling coefficient is... To observe the noise, The variance is 2.5 × 10⁻⁶. -7 , The effective value current squared difference term, For process noise, Transient pressure; In this embodiment, the effective value current squared difference term is used as the control input of the wide state space model, which eliminates AC high-frequency ripple, characterizes the transient step of Joule thermal power that leads to thermal expansion, and establishes a physical causal mapping relationship between electrothermal force. Coupling coefficient The linearized transfer gain characterizes the linearized transfer from the square step of the current to the resulting change in air pressure, in units of... In this embodiment The value is ; The process of calculating the effective value current squared difference term specifically includes: For the 1000Hz transient load current, a single-cycle (20 milliseconds) sliding window root mean square calculation is performed. That is, at a sampling rate of 1000Hz, a 20-millisecond window contains 20 transient load current data, including the current sampling point. The root mean square of all transient load current data is calculated as the high-frequency effective value current at the current moment. Calculate the square of the high-frequency RMS current at the current moment, and subtract the square of the high-frequency RMS current at the previous moment from the square of the high-frequency RMS current at the current moment to obtain the RMS current square difference term. Based on the above augmented state space model, the pressure and the squared difference of the effective current at each sampling time are used as inputs to the Kalman filter. Standard Kalman filter prediction and update iteration operations are performed to output the optimal augmented state estimation vector. The actual air pressure estimate at the current time is the component in the optimal augmented state estimation vector, and the unit is megapascal. In the Kalman filter prediction and update iteration calculation, the process noise covariance matrix is ​​a 6×6 diagonal matrix. Considering the slow drift characteristics of the actual air pressure in the augmented state, its corresponding process noise variance is set to a relatively small value, preset to 10. -4 The dynamic changes in the sensor's internal state are mainly driven by the model, and its process noise variance should be smaller, preset to 10. -6 The observed noise variance is set based on the static measurement noise variance calibration value of the piezoresistive sensor within its operating range; in this embodiment, it is preset to 2.5 × 10⁻⁶.-7 ; This method uses the actual air pressure as a state variable for optimal recursive estimation, effectively filtering out sensor overshoot and oscillation components caused by fluid impact, and is unaffected by the non-steady-state characteristics of the signal. The experimental calibration process of the fifth-order discrete-time state-space model of the piezoresistive barometric pressure sensor specifically includes: A piezoresistive pressure sensor of the same model as the field piezoresistive pressure sensor is used as the sensor under test. The sensor under test and the reference standard pressure sensor are installed in a sealed calibration chamber. The reference standard pressure sensor is a piezoelectric dynamic pressure sensor with the same range as the sensor under test, a natural resonant frequency of not less than 100kHz, a rise time of not more than 5 microseconds, and a nonlinearity error of not more than ±0.2% of the full scale output. The output signal of the reference standard pressure sensor is regarded as the pressure reference for this calibration. The calibration chamber is also equipped with a quick pressure relief valve, which is electromagnetically driven and has an opening time of no more than 0.5 milliseconds. After opening, it can release the pressure in the chamber from the target value to the normal pressure within no more than 2 milliseconds. The calibration chamber is equipped with a high-pressure gas cylinder, which is connected to the chamber through a shut-off valve and is used to fill the chamber with clean, dry nitrogen. Open the shut-off valve and fill the calibration chamber with nitrogen gas to stabilize the absolute pressure inside the chamber at 80% of the full scale of the sensor being tested. This pressure is recorded as the initial equilibrium pressure. After closing the shut-off valve and waiting for the pressure and temperature in the chamber to stabilize, start data acquisition and continuously record the pressure sequence output by the reference standard pressure sensor and the original voltage signal sequence output by the sensor under test. At the first second after the recording starts, trigger the quick pressure relief valve to open, and the gas in the chamber will be quickly released to normal pressure. Continue recording until the pressure relief is completed and stabilized, then stop data acquisition. The pressure sequence output by the reference standard pressure sensor is converted into a reference pressure sequence in MPa by its factory sensitivity coefficient, and the voltage sequence output by the sensor under test is converted into a measurement pressure sequence in MPa by its factory sensitivity coefficient, denoted as the measurement pressure sequence.

[0024] The reference pressure sequence is used as the input sequence of the system identification algorithm, the measured pressure sequence is used as the output sequence of the system identification algorithm, the moment when the pressure relief valve is triggered is used as the starting point of the input step, and data of a certain duration before and after the starting point are used as the identification dataset. The data duration should cover the entire transient response process of the sensor and extend to the steady state.

[0025] The above input and output data are processed by the numerical subspace state-space system identification algorithm N4SID. The model order of N4SID algorithm is specified as 5, and the direct transfer term is specified as existent, that is, the direct transfer matrix D is allowed to be non-zero. The algorithm output is the identified system matrix A, input matrix B, observation matrix C and direct transfer matrix D. The process of obtaining the coupling coefficients specifically includes: The coupling coefficient α is used to characterize the square step of the current ΔI. 2 The linearized transfer gain between the pressure change and the pressure change caused, expressed in MPa / A². The value of this coefficient is uniquely determined through the following reproducible experimental procedure: (1) Construction of the experimental platform; An environmentally friendly gas cabinet of the same model as the one used on site was used as the experimental object. The gas chamber was filled with C4F7N / CO2 mixed gas with the same ratio as that used on site to the rated filling pressure. A high-precision reference pressure sensor was installed at the valve of the gas chamber. Its static accuracy was not less than 0.05% of the full scale and the sampling frequency was not less than 1kHz. It was used to record the real pressure change in the gas chamber as the pressure reference for calibration. A programmable high-current source is connected to the busbar side inside the cabinet to generate step currents of different amplitudes. At the same time, the temperature of the housing is monitored by a PT100 temperature sensor attached to the surface of the cabinet to ensure that the temperature of the air chamber is consistent with the calibrated ambient temperature at the start of each experiment. After the air chamber and the environment reached full thermal equilibrium, the five sets of step current experiments shown in Table 1 were performed sequentially. The current in each set of experiments started from the steady-state initial value. Sudden increase to target value The current parameters for the five sets of step current experiments are shown in Table 1, and the pressure is maintained until it stabilizes again. Table 1 ; Sufficient cooling time was allowed between each set of experiments, allowing the air chamber temperature to return to the calibrated ambient temperature before proceeding to the next set. During each experiment, the bus current was recorded at a frequency of 1kHz (in A), and the actual air pressure output by the reference pressure sensor was recorded (in MPa). For the current sequence of each experiment, the root mean square calculation was performed using a 20-millisecond sliding window to obtain the effective value current sequence. In the effective current sequence, the moment when the effective current first exceeds 5% of the initial steady-state value is taken as the step start moment. The average effective current in the second before the step start moment is taken as the effective value before the step. The average effective current s in the period after the step start moment when the air pressure stabilizes again, that is, the period when the actual air pressure fluctuation does not exceed ±0.1% of the full scale, is taken as the effective value after the step. The difference between the square of the effective value after the step and the square of the effective value before the step is taken as the square step of the current. The average real air pressure within 1 second before the start of the step is taken as the air pressure before the step, and the average real air pressure within the air pressure re-stabilization period after the start of the step is taken as the air pressure after the step. The difference between the air pressure after the step and the air pressure before the step is taken as the air pressure change. For the above five sets of experiments, the corresponding changes in air pressure and the square step of current were calculated for each set. The five sets of data points were plotted on a rectangular coordinate system, with the horizontal axis representing the square step of current and the vertical axis representing the changes in air pressure. The least squares method was used to perform a linear fit on the five sets of data points through the origin. The slope of the fitted line is the coupling coefficient α.

[0026] By calculating the high-frequency effective current through a preset sliding window and obtaining the squared difference term between adjacent time points, high-frequency interference in the conventional AC load waveform is eliminated, and the core feature of the step change in bus heating power is extracted in real time. This not only provides driving control quantity for the augmented state space model, but also successfully avoids the problem of spectrum leakage and loss of time and frequency information that traditional frequency domain filtering methods are prone to when dealing with non-stationary change signals, ensuring the accuracy of feedforward compensation under complex load jump conditions. The Kalman filter of the augmented state-space model constructed in this invention introduces the effective value current squared difference term into the control parameters, establishes a causal mapping relationship between electrothermal and mechanical multi-physics fields, characterizes the transient step quantity of Joule thermal power that leads to thermal expansion, and recursively optimizes the estimation by using the real air pressure as the augmented state variable. It can separate the overshoot and high-frequency oscillation components caused by the violent expansion of the internal gas due to local instantaneous heating from the sensor signal under forced vibration, and is not affected by AC high-frequency ripple and signal non-stationary characteristics, thus improving the anti-dynamic disturbance capability of air pressure measurement.

[0027] The process of obtaining minute-level steady-state parameters specifically includes: Using a preset time interval as the dividing point, the arithmetic mean of each type of multi-source data and the estimated actual air pressure is calculated within the interval to obtain the minute load current, minute cabinet temperature, minute gas temperature and minute air pressure; The minute-level steady-state parameters are composed of minute-level load current, minute-level cabinet temperature, minute-level gas temperature, and minute-level gas pressure.

[0028] By arithmetic averaging, high-frequency multi-source transient signals and actual air pressure estimates are uniformly downsampled and aggregated to minute-level steady-state parameters, achieving cross-scale time alignment between high-frequency dynamic signals and low-frequency slow physical states, and further reducing the cumulative interference of random measurement white noise in the macroscopic time dimension. This data dimensionality reduction not only reduces the computational pressure of subsequent complex calculations such as neural network forward inference and differential solving, but also provides a smooth and high signal-to-noise ratio steady-state observation baseline for capturing the extremely slow evolution process of real trace leaks. The process of calculating minute-level dynamic volume specifically includes: Extract the coefficient of linear expansion and the calibration temperature, and use a preset multiple of the coefficient of linear expansion as the coefficient of volume expansion; Calculate the minute-level dynamic volume based on the rated cavity volume, volume expansion coefficient, and minute cabinet temperature.

[0029] In one specific embodiment of the present invention, a set of multi-source data is generated at each sampling point. The subsequent calculations of compression factor, leakage rate, and insulation warning are no longer performed per sampling point, but rather once every fixed-length time window (once per minute in this embodiment). Therefore, before proceeding to the next stage, the multi-source data needs to be converted to a different time scale. The specific process includes: Each complete minute is used as a segmented interval. The arithmetic mean of each type of multi-source data within the interval is calculated to obtain the minute load current, minute cabinet temperature, and minute gas temperature. The arithmetic mean of the actual gas pressure estimate output by the filter module within the interval is calculated to obtain the minute gas pressure. The purpose of this step is to address the issue of baseline drift in leakage rate calculation caused by changes in the gas chamber volume due to thermal expansion and contraction of the metal cabinet. The calculation of micro-leakage rates for environmentally friendly gases exhibits a strongly nonlinear coupling relationship with volume changes. This step employs a volume compensation method to resolve this problem. The specific process includes: Extract the cabinet temperature, coefficient of linear expansion, rated cavity volume and calibration temperature, and use three times the coefficient of linear expansion as the coefficient of volume expansion. Calculate the deformation volume of the air chamber at the current moment. The mathematical expression for the deformation volume is: ; in, The dynamic deformation volume of the air chamber at the current time t is expressed in cubic meters. The rated cavity volume, The coefficient of volume expansion is 1. For cabinet temperature, For calibration temperature; The mathematical expression for deformation volume is called, and the cabinet temperature is used to obtain the dynamic volume at the minute level.

[0030] By utilizing the rated cavity volume, volume expansion coefficient, and minute cabinet temperature, the dynamic deformation volume of the air chamber at the current moment is adaptively calculated, correcting the systematic defect in existing monitoring technologies that treat metal enclosed ring network cabinets as an absolutely rigid constant volume model. By compensating in real time for volume changes caused by thermal expansion and contraction due to the alternation of day and night in the external environment or the full load heating of internal equipment, false pressure fluctuations caused by weak volume deformation are effectively filtered out, and the calculation baseline drift phenomenon in the long-term solution process of micro-leakage rate is curbed.

[0031] The structure of the hybrid evolutionary network is a fully connected feedforward neural network, including an input layer, a single hidden layer, and an output layer; The input layer contains four nodes, which correspond to the normalized minute pressure, minute gas temperature, preceding equivalent compressibility factor, and leakage rate of the previous minute, respectively. The preceding equivalent compressibility factor is calculated based on the ideal gas law using the minute-level pressure, minute-level gas temperature, minute-level dynamic volume, and the total amount of substance of the mixed gas in the previous minute. A single hidden layer contains 8 nodes and uses the tanh hyperbolic tangent activation function; The output layer contains one node and uses the purelin linear activation function to output the dynamic compression factor in the segmented interval.

[0032] The purpose of this step is to solve the dynamic compressibility factor of the mixed gas in order to isolate the changes in physical properties caused by selective leakage and local liquefaction, and to calculate the true micro-leakage rate. The composition of the environmentally friendly mixed gas changes dynamically due to leakage and phase change. This step solves the above problems by constructing a mixture property evolution network. The structure of the hybrid evolutionary network is a fully connected feedforward neural network, including an input layer, a single hidden layer, and an output layer; The input layer contains four nodes, which correspond to the minute gas pressure, minute gas temperature, the preceding equivalent compressibility factor reflecting gas component drift, and the leakage rate of the previous minute after Min-Max normalization. The mathematical expression for the preorder equivalent compressibility factor is: ; in, The preceding equivalent compression factor of the current segment interval i, Pressure per minute For minute-level dynamic volume, This represents the previous interval, i.e., the total amount of substance of the mixed gas calculated in the previous minute. R is the ideal gas constant, with a value of [value missing]. , The Kelvin temperature value for the gas temperature per minute is obtained by adding 273.15 to the gas temperature per minute. When the preferential escape of CO2 leads to a decrease in the actual total number of moles n, the preceding equivalent compressibility factor increases, thus providing direct evidence for gas component drift in the mixture evolution network. This invention constructs a mixture evolution network with normalized steady-state environmental parameters, historical leakage rate, and preceding equivalent compressibility factor as characteristic inputs, and establishes a highly nonlinear evolution mapping between gas component drift and dynamic compressibility factor. By using the preceding equivalent compressibility factor recursively calculated from the previous physical quantity as network input, the network is directly provided with quantitative physical evidence of preferential escape of specific components in the mixed gas or local liquefaction under extreme conditions, ensuring the authenticity and accuracy of the core compressibility factor solution.

[0033] The process of obtaining the leakage rate specifically includes: The amount of substance is calculated based on the dynamic compressibility factor output by the mixture evolution network, combined with minute-level gas pressure, minute-level dynamic volume, ideal gas constant, and minute-level gas temperature. The calculated amount of substance is numerically differentiated to obtain the leakage rate.

[0034] In one specific embodiment of the present invention, the amount of matter in the current partition segment is calculated based on the dynamic compression factor output by the hybrid evolution network. The mathematical expression for the amount of matter is: ; in, Let be the amount of substance in the i-th partition segment. Pressure per minute For minute-level dynamic volume, Where R is the dynamic compressibility factor and R is the ideal gas constant. The Kelvin temperature value representing the gas temperature per minute; Based on the amount of substance, the leakage rate is obtained by numerical differentiation of the amount of substance. Leakage rate represents the rate of loss of a substance, measured in moles per minute. To align with the gas replenishment volume in actual engineering, the number of moles can be converted to volume in liters or cubic meters at standard atmospheric pressure of 0.101325 MPa and 20°C, based on the ideal gas law. The conversion factor is 24.055 L / mol. Initially, the number of accumulated partition segments is insufficient to support high-order difference formats. Starting from the first partition segment, the difference order is dynamically adjusted based on the number of accumulated partition segments: When the current partition segment number is 1, only the amount of substance in the current partition segment and the initial partition segment is available. The two-point first-order backward difference scheme is adopted, and the absolute leakage rate is obtained by dividing the difference in the amount of substance in the initial partition segment and the current partition segment by the partition segment time step. When the current partition segment number is 2, the amount of substance of three partition segments has been accumulated. The three-point second-order backward difference scheme is used to calculate the amount of substance of the first two partition segments and the current partition segment by weight. When the current partition segment number is 3, the amount of substance accumulated from four partition segments is used in the four-point third-order backward difference scheme. When the current partition segment number is 4, the amount of matter of five partition segments has been accumulated. The reduced-order five-point fourth-order backward difference scheme is adopted. Its specific expression is: the specific weighted combination of the amount of matter from the initial partition segment to the current partition segment is divided by the partition segment time step, and the weight coefficients correspond to the amount of matter of each partition segment. When the current partition segment number reaches 5 or above, it is uniformly switched to the standard five-point fourth-order backward difference format. The required data is the amount of substance of the current partition segment and the previous four consecutive partition segments. The weighting coefficient is fixed and the partition segment time step is fixed to the preset value. The pre-training process for hybrid evolutionary networks includes: In the same model of environmentally friendly gas cabinet, a standard C4F7N / CO2 mixed gas with the same ratio as on-site was filled in. The preferential escape of CO2 molecules was simulated through an integrated leak valve, and the leakage rate was controllable. During a continuous week of leakage, the gas pressure was synchronously collected at 1-minute intervals by an array of high-precision pressure and temperature sensors. and gas temperature And calculate the corresponding minute gas pressure and minute gas temperature; The density of the mixed gas is measured using a high-precision vibrating tube gas densitometer, and the real-time average molar mass of the gas is obtained by gas chromatography analysis. The actual dynamic compressibility factor at the current moment is calculated based on thermodynamic relationships and used as the output label. The mathematical expression for the dynamic compressibility factor is: ; in, The measured air pressure is given by [value], and M is the average molar mass. Let R be the density of the gas mixture, and R be the ideal gas constant. , for The Kelvin temperature, of which, The unit is MPa, multiplied by 10 6 Convert to Pa; The input to the hybrid evolution network is the minute pressure, minute gas temperature, preceding equivalent compressibility factor, and leakage rate of the previous minute after Min-Max normalization. The minute pressure and minute gas temperature can be obtained by filtering and averaging the original measurements, while the subsequent equivalent compressibility factor and leakage rate of the previous minute need to be calculated recursively. Recursive calculations require a starting point. Initially, the gas chamber contains a standard-proportion C4F7N / CO2 mixture. The initial amount of gas at this point is... Given directly from factory parameters, with an initial leakage rate of 0, the recursive process, starting from the first minute, specifically includes: For the m-th minute, use the minute pressure, minute volume, and minute gas temperature of the m-th minute, combined with the amount of gaseous substance in the previous batch of minutes, to calculate the preceding equivalent compressibility factor; The preceding equivalent compressibility factor, the gas pressure at minute m, the gas temperature at minute m, and the leakage rate of the previous minute are fed into an untrained hybrid evolution network to obtain the dynamic compressibility factor at minute m. The amount of gaseous substance in the m-th minute is calculated based on the dynamic compressibility factor, and the leakage rate in the m-th minute is calculated based on the amount of gaseous substance in the m-th minute. At this point, each minute segment has obtained a complete set of training samples; This embodiment uses a particle swarm optimization algorithm to optimize and solidify the parameters of a hybrid evolutionary network. The particle swarm optimization algorithm has a scale of 50, an inertia weight of 0.8, and both the individual learning factor and the swarm learning factor are set to 2.0. The velocity dimensions are limited to the range of [-0.5, 0.5], and the number of iterations is preset to 1000. The optimization objective is to minimize the mean square error between the network output dynamic compression factor and the label value.

[0035] This invention reconstructs the actual amount of material in the current segment based on a dynamically corrected compression factor, and continuously calculates the leakage rate using numerical differentiation techniques. It transforms and maps the pure physical state variables after multi-dimensional time-domain filtering and property compensation into an absolute mass loss rate that can intuitively reflect the sealing performance of the equipment cavity. Combined with a backward difference operation format that dynamically adjusts the order with the accumulated data volume, it can take into account both the data sparsity limitations in the early stage of operation and the high-order smoothing requirements of long-term operation, and output the leakage rate continuously and without lag. This avoids the technical problem that it is impossible to solve the problem in the initial stage using a single fixed-order difference.

[0036] The logic of tiered early warning specifically includes: The absolute difference between the dynamic compressibility factor and the reference compressibility factor is used as the compressibility factor deviation; the cumulative mass loss ratio is calculated based on the leakage rate and the initial total amount of material. When the cumulative mass loss ratio is less than the preset loss threshold and the compression factor deviation is less than the preset deviation threshold, it is determined to be in normal operating condition. When the cumulative mass loss ratio is greater than or equal to the preset loss threshold and the compression factor deviation is less than the preset deviation threshold, an early warning of excessive cumulative leakage is triggered. When the compressibility factor deviation is greater than or equal to the preset deviation threshold, regardless of whether the cumulative mass loss ratio is less than the preset loss threshold, a gas component imbalance warning is triggered and a gas replenishment strategy is executed.

[0037] This invention can diagnose and decouple two core fault risks: macroscopic loss of absolute gas inventory and microscopic imbalance of heterogeneous gas components. Relying on the preset threshold defense line obtained from the accelerated degradation inflection point experiment, it avoids frequent false alarms caused by normal slight temperature fluctuations and accurately triggers early warning before the partial discharge initiation voltage and breakdown voltage experience an irreversible cliff drop.

[0038] The absolute difference between the dynamic compressibility factor and the reference compressibility factor is taken as the compressibility factor deviation. The reference compressibility factor is the compressibility factor calculated according to the GERG-2008 equation of state for the mixed gas under the factory rated conditions, that is, the rated charging pressure and the calibrated temperature of 20°C. For the specific ratio of C4F7N / CO2 mixed gas targeted in this embodiment, the value of the reference compressibility factor is 0.98. The cumulative mass loss ratio is calculated using the following mathematical expression: ; in, This represents the cumulative quality loss ratio from the initial partition segment to the end of the Dth partition segment. Let i be the leakage rate of partition segment i. For 1 minute, The initial total amount of substance; The cumulative mass loss ratio characterizes the relative loss of the total amount of gas in the chamber. When this ratio exceeds a certain preset threshold, it indicates that a lot of mass loss has accumulated, even if the current leakage rate is very small. The average leakage rate of a number of recent partition segments is taken as the recent average leakage rate. In this embodiment, the average leakage rate of 60 partition segments is taken as the average leakage rate of the most recent hour. The recent average leakage rate represents the current leakage rate level and is used to capture the leakage acceleration trend in the short term. If the cumulative quality loss ratio is less than the preset loss threshold of 0.05 and the compression factor deviation is less than the preset deviation threshold of 0.05, it indicates normal operation. If the cumulative mass loss is greater than or equal to the preset loss threshold and the compression factor deviation is less than the preset deviation threshold, an alarm for excessive cumulative leakage will be triggered. If the compressibility factor deviation is greater than or equal to the preset deviation threshold, a gas component imbalance warning and gas replenishment strategy will be triggered. The loss threshold is used to determine whether the total amount of gas in the gas chamber has been lost to an unacceptable level due to long-term minor leakage. The process of determining the loss threshold specifically includes: In an environmentally friendly gas cabinet of the same model as the one on site, C4F7N / CO2 mixed gas with the factory standard ratio was filled to the rated filling pressure. The continuous micro-leakage was simulated by fine-tuning the leakage valve. The leakage rate was controlled at a slow level of 0.5% to 1% of the initial total gas volume per month. During the continuous six-month leakage process, the following data were recorded once a week. By weighing the mass difference of the gas cylinder before and after replenishing it with a precision balance, and combining the gas chamber volume and temperature, the actual amount of gas remaining in the gas chamber at that moment can be calculated, and thus the actual cumulative mass loss ratio can be obtained. A power frequency withstand voltage test device was used to measure the partial discharge initiation voltage of the gas chamber at that moment, which was used as a quantitative indicator of insulation performance. The experimental results are shown in Table 2; Table 2 ; As shown in the table above, when the cumulative mass loss ratio reaches 0.05, the partial discharge initiation voltage drop rate shows an accelerated inflection point, and the insulation performance begins to deteriorate at a rate higher than the linear trend. In order to trigger an early warning before this inflection point and leave sufficient maintenance response time, this embodiment presets the cumulative mass loss ratio threshold to 0.05. The deviation threshold is used to determine whether the gas composition imbalance caused by preferential CO2 escape has reached the inflection point of accelerated insulation performance degradation. The process of determining the deviation threshold includes: In an environmentally friendly gas cabinet of the same model as the one used on site, a C4F7N / CO2 mixed gas with the factory standard ratio was filled to the rated filling pressure. The selective permeation characteristics of the leak port were controlled by a precision mass flow controller, so that CO2 escaped at a rate that took precedence over C4F7N. At different leakage stages, the actual volume ratio of CO2 in the gas chamber was measured by a gas chromatograph, the compressibility factor deviation was recorded, and the breakdown voltage at that moment was measured by a power frequency withstand voltage test device. The experimental results are shown in Table 3. Table 3 ; As shown in the table above, when the compression factor deviation reaches 0.05, the breakdown voltage drop rate is close to 10%, and the insulation performance enters a clear inflection point of accelerated deterioration. Moreover, this value can effectively avoid the interference of small compression factor deviation caused by normal temperature fluctuations. Therefore, in this embodiment, the compression factor deviation trigger threshold is preset to 0.05.

[0039] Qi-replenishing strategies include: The pre-stored gas replenishment mapping table is invoked. The gas replenishment mapping table records the correspondence between different compressibility factor deviation nodes and the standard state volume of pure carbon dioxide gas to be replenished. Based on the current compressibility factor deviation, the corresponding gas replenishment volume is queried in the gas replenishment mapping table; if the current compressibility factor deviation falls between two adjacent nodes in the gas replenishment mapping table, the gas replenishment volume is determined by interpolation algorithm.

[0040] In one specific embodiment of the present invention, a pre-stored gas replenishment mapping table is invoked to automatically determine the specific volume of pure CO2 gas that needs to be replenished, in cubic meters. The process of generating the gas replenishment mapping table specifically includes: Fill the gas cabinet of the same model as the one on site with the factory standard ratio of C4F7N / CO2 mixed gas to the rated filling pressure, and let it stand until the temperature is balanced; Using a precision mass flow controller, the mixed gas is extracted from the gas chamber at a constant rate, causing the compressibility factor deviation to increase slowly. When the compressibility factor deviation reaches the set nodes of 0.05, 0.06, 0.07, 0.08, 0.09, and 0.10 respectively, the extraction is stopped, and the current compressibility factor deviation and the current gas chamber pressure are recorded. For each node, pure CO2 gas is injected into the gas chamber at a constant flow rate through another precision mass flow controller. During the injection process, the change in the compressibility factor deviation is monitored in real time. When the compressibility factor deviation just falls below 0.05 (i.e. 0.049±0.001), the injection is stopped, and the cumulative volume of CO2 injected is recorded and converted into the volume value under the factory rated condition. Record the one-to-one correspondence between the compression factor deviation value of each node and the corresponding standard state gas replenishment volume to form a univariate mapping table as the gas replenishment mapping table. In the table, the compression factor deviation is the index key and the gas replenishment volume is the mapping value.

[0041] To address the thermal shock and fluid convection interference caused by sudden changes in busbar load current during actual operation, this invention introduces a Kalman filter method based on an augmented state-space model. The effective value current squared difference term is used as the control input, establishing a physical causal mapping relationship between electricity, heat, and force. This filtering mechanism effectively eliminates transient dynamic measurement errors such as overshoot and high-frequency oscillation caused by instantaneous expansion shock in piezoresistive pressure sensors, enabling accurate estimation of the true pressure in the air chamber under complex operating conditions. Considering the thermal expansion and contraction of the metal cabinet caused by ambient temperature and operating temperature rise, this invention solves the problem of baseline drift in micro-leakage rate calculation caused by treating the gas chamber volume as a constant in traditional methods by unifying the time scale of data and introducing a volume expansion coefficient. On this basis, to overcome the deviation in property calculation caused by selective leakage or local liquefaction of environmentally friendly mixed gases, this invention constructs a mixture property evolution network with a fully connected feedforward neural network structure. By using parameters such as the pre-order equivalent compressibility factor reflecting the drift of gas components, the dynamic compressibility factor in segmented intervals is calculated, and the dynamic evolution law of property parameters is analyzed in depth. Then, the truly accurate micro-leakage rate is obtained by combining numerical differentiation methods.

[0042] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0043] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A safety operation monitoring system for environmentally friendly gas cabinets based on adaptive leakage rate compensation, characterized in that, include: The data acquisition module is used to acquire multi-source data from the gas cabinet; The filtering module is used to estimate the real air pressure based on the transient load current and transient pressure in the multi-source data using the Kalman filtering method of the augmented state-space model; The aggregation module is used to unify the time scale of multi-source data to obtain minute-level steady-state parameters and calculate minute-level dynamic volume. The solution module is used to calculate the current dynamic compressibility factor based on the mixture evolution network, calculate the total amount of substance of the mixed gas by combining the minute-level steady-state parameters and minute-level dynamic volume, and calculate the leakage rate based on numerical differentiation. The early warning module is used to calculate the cumulative mass loss ratio based on the leakage rate and the compressibility factor deviation, and to trigger graded early warning and gas replenishment strategies based on the cumulative mass loss ratio and the compressibility factor deviation.

2. The environmentally friendly gas cabinet safety operation monitoring system based on adaptive leakage rate compensation as described in claim 1, characterized in that, The process of acquiring the multi-source data in the acquisition module specifically includes: Transient load current and transient pressure are collected at a preset high-frequency sampling rate; The rated cavity volume, calibrated temperature, and linear expansion coefficient of the cabinet material of the environmentally friendly gas cabinet are collected.

3. The environmental protection gas cabinet safety operation monitoring system based on leakage rate adaptive compensation as described in claim 2, characterized in that, In the filtering module, the Kalman filter based on the augmented state-space model has an augmented state vector containing the actual air pressure, a control quantity that is the effective value current squared difference term, and a control matrix that contains coupling coefficients. The process of estimating the actual air pressure specifically includes: The root mean square of the transient load current within a preset sliding window is calculated to obtain the effective value current, and the square difference of the effective value current at adjacent times is taken as the effective value current square difference term. The transient pressure at each sampling moment and the squared difference term of the effective current are used as inputs to the Kalman filter based on the augmented state space model. Kalman filter prediction and update iteration operations are performed to output the optimal augmented state estimation vector. The component of the optimal augmented state estimation vector corresponding to the real air pressure is used as the real air pressure estimate at the current moment.

4. The environmentally friendly gas cabinet safety operation monitoring system based on adaptive leakage rate compensation as described in claim 3, characterized in that, The process of obtaining the effective value current squared difference term specifically includes: Calculate the root mean square of all transient load current data within the preset sliding window as the high-frequency effective value current at the current moment; Calculate the square of the high-frequency RMS current at the current moment, and subtract the square of the high-frequency RMS current at the previous moment from the square of the high-frequency RMS current at the current moment to obtain the RMS current square difference term.

5. The environmental protection gas cabinet safety operation monitoring system based on leakage rate adaptive compensation as described in claim 4, characterized in that, The process of obtaining the minute-level steady-state parameters specifically includes: Using a preset time interval as the dividing point, the arithmetic mean of each type of multi-source data and the estimated actual air pressure within the interval is calculated to obtain the minute load current, minute cabinet temperature, minute gas temperature and minute air pressure. The minute load current, minute cabinet temperature, minute gas temperature, and minute gas pressure constitute minute-level steady-state parameters.

6. The environmentally friendly gas cabinet safety operation monitoring system based on leakage rate adaptive compensation as described in claim 5, characterized in that, The process of calculating the minute-level dynamic volume specifically includes: Extract the coefficient of linear expansion and the calibration temperature, and use a preset multiple of the coefficient of linear expansion as the coefficient of volume expansion; Calculate the minute-level dynamic volume based on the rated cavity volume, volume expansion coefficient, and minute cabinet temperature.

7. The environmentally friendly gas cabinet safety operation monitoring system based on leakage rate adaptive compensation as described in claim 6, characterized in that, The hybrid evolution network is structured as a fully connected feedforward neural network, including an input layer, a single hidden layer, and an output layer. The input layer contains four nodes, which correspond to the normalized minute pressure, minute gas temperature, preceding equivalent compressibility factor, and leakage rate of the previous minute, respectively. The preceding equivalent compressibility factor is calculated based on the ideal gas law using the minute-level gas pressure, minute-level gas temperature, minute-level dynamic volume, and the total amount of mixed gas in the previous minute. The single hidden layer contains 8 nodes and uses the tanh hyperbolic tangent activation function; The output layer contains one node and uses the purelin linear activation function to output a dynamic compression factor in segmented intervals.

8. The environmental protection gas cabinet safety operation monitoring system based on leakage rate adaptive compensation as described in claim 7, characterized in that, The process of obtaining the leakage rate specifically includes: The amount of substance is calculated based on the dynamic compressibility factor output by the mixture evolution network, combined with the minute-level gas pressure, the minute-level dynamic volume, the ideal gas constant, and the minute-level gas temperature. The calculated amount of substance is numerically differentiated to obtain the leakage rate.

9. The environmentally friendly gas cabinet safety operation monitoring system based on adaptive leakage rate compensation as described in claim 8, characterized in that, The logic of the tiered early warning system specifically includes: The absolute difference between the dynamic compressibility factor and the reference compressibility factor is taken as the compressibility factor deviation; the cumulative mass loss ratio is calculated based on the leakage rate and the initial total amount of material. When the cumulative quality loss ratio is less than a preset loss threshold and the compression factor deviation is less than a preset deviation threshold, it is determined to be in normal operating condition; When the cumulative mass loss ratio is greater than or equal to the preset loss threshold, and the compression factor deviation is less than the preset deviation threshold, an early warning of excessive cumulative leakage is triggered. When the deviation of the compressibility factor is greater than or equal to the preset deviation threshold, a gas component imbalance warning is triggered and the gas replenishment strategy is executed.

10. The environmentally friendly gas cabinet safety operation monitoring system based on adaptive leakage rate compensation as described in claim 9, characterized in that, The qi replenishment strategy includes: The pre-stored gas replenishment mapping table is invoked. The gas replenishment mapping table records the correspondence between different compressibility factor deviation nodes and the standard state volume of pure carbon dioxide gas to be replenished. Based on the current compressibility factor deviation, the corresponding gas replenishment volume is queried in the gas replenishment mapping table; if the current compressibility factor deviation falls between two adjacent nodes in the gas replenishment mapping table, the gas replenishment volume is determined by an interpolation algorithm.

Images