Gas on-line detection system for transformer light gas on-line intelligent exhaust gas

Abstract

The invention provides a gas on-line detection system for on-line intelligent exhaust gas of transformer light gas, and relates to the technical field of gas on-line detection. Gas sample data of the transformer light gas is collected through a gas sample collection module; a separation detection window for detecting the light gas of the transformer is arranged through a sample deviation elimination module, and the concentration compensation deviation of the light gas of the transformer during emission detection is cooperatively eliminated according to pressure difference echelon information; the gas emission identification module is used for analyzing the gas flow derivation trend during online detection of the light gas of the transformer, and the emission risk identifier is used for dynamically identifying the gas dissipation level of the light gas of the transformer during emission detection; and carrying out on-line linkage detection on the light gas of the transformer through the on-line linkage detection module. According to the method and the device, the gas emission risk can be accurately identified in a complex environment of light gas emission of the transformer, so that the accuracy of light gas emission detection of the transformer is improved.

Description

Technical Field

[0001] The present application relates to the technical field of gas online detection, and more specifically, to a gas online detection system for online intelligent emission of light gas from a transformer. Background Art

[0002] The gas online detection system is a technical means to monitor gas emissions, concentration, composition and other related parameters in real time. In the monitoring of transformer light gas emissions, the gas online detection system can realize continuous, automatic and real-time monitoring of various indicators in the transformer light gas emission process. The system captures gas data at the transformer discharge port through a sensor array or gas analysis instrument, and transmits the collected values ​​to the monitoring system in real time through a data transmission module. The system can detect key data such as gas concentration, flow rate, pressure, temperature, and composition.

[0003] However, in the existing online gas detection system for transformer light gas online intelligent emission gas, there are certain errors in the collection of gas samples, the detection of gas concentration, and the derivative analysis of gas flow, especially in complex emission environments, which are easily affected by temperature, pressure changes and gas diffusion patterns, resulting in low gas detection accuracy, making it impossible to effectively eliminate concentration compensation deviations, inaccurate risk identification of gas emissions, and difficult to meet strict emission standards. The emission detection results may be inaccurate, thereby increasing the risk of environmental pollution and safety hazards. Therefore, how to accurately identify gas emission risks in the complex environment of transformer light gas emissions to improve the accuracy of transformer light gas emission detection is a difficult problem faced by the industry. Summary of the invention

[0004] The present application provides an online gas detection system for transformer light gas online intelligent emission gas, which can accurately identify gas emission risks in the complex environment of transformer light gas emission, so as to improve the accuracy of transformer light gas emission detection.

[0005] The present application provides a gas online detection system for online intelligent exhaust gas of transformer light gas, and the online detection system includes: A gas sample collection module is used to configure a gas component analysis array in the construction of a biomass energy project, and to collect gas sample data of transformer light gas through the gas component analysis array; A sample deviation elimination module, which is used to determine the trace gas characteristics of the transformer's light gas during emission based on the gas sample data, set a separation detection window for detecting the transformer's light gas, perform differential pressure resolution on the trace gas characteristics through the separation detection window, obtain the differential pressure gradient information when the transformer's light gas is emitted, and then cooperatively eliminate the concentration compensation deviation of the transformer's light gas during emission detection by the differential pressure gradient information; A gas emission identification module, which is used to analyze the airflow evolution trend during the on-line detection of the transformer's light gas, determine the emission risk identifier when the transformer's light gas is emitted through the airflow evolution trend, and then dynamically identify the gas escape level of the transformer's light gas during emission detection by the emission risk identifier; An on-line linkage detection module, which is used to perform on-line linkage detection of the transformer's light gas through the concentration compensation deviation after cooperative elimination and the gas escape level after dynamic identification.

[0006] In this embodiment, the specific process of collecting the gas sample data of the transformer's light gas by the gas component analysis array includes: Controlling the gas component analysis array to establish a communication channel with the emission port of the transformer's light gas, so that the transformer's light gas flows into the sampling area of the gas component analysis array; Collecting samples of the transformer's light gas flowing into the sampling area according to a preset sampling rule to obtain gas sample data.

[0007] In this embodiment, the specific process of determining the trace gas characteristics of the transformer's light gas during emission based on the gas sample data includes: Performing feature extraction on the gas sample data to obtain gas component information; Constructing a primary set of gas components through the gas component information; Determining the trace gas characteristics of the transformer's light gas during emission based on the primary set of gas components.

[0008] In this embodiment, the specific process of setting the separation detection window for detecting the transformer's light gas includes: Determining the associated distribution dimension of the transformer's light gas; Determining the window correction boundary for detecting the transformer's light gas according to the associated distribution dimension; Setting the separation detection window for detecting the transformer's light gas through the window correction boundary.

[0009] In this embodiment, the specific process of performing differential pressure resolution on the trace gas characteristics through the separation detection window to obtain the differential pressure gradient information when the transformer's light gas is emitted includes: Determine the differential pressure resolution feature according to the physical parameters corresponding to the trace gas characteristics in the separation detection window; Determine the differential pressure response amounts at different positions within the separation detection window through the differential pressure resolution feature; Determine the pressure evolution trend of the trace gas characteristics at different differential pressures according to the differential pressure response amounts; Determine the gas differential pressure limit during the light gas emission of the transformer; Determine the differential pressure gradient information during the light gas emission of the transformer according to the pressure evolution trend and the gas differential pressure limit;

[0010] In this embodiment, the collaborative elimination of the concentration compensation deviation during the emission detection of the light gas of the transformer by the differential pressure gradient information specifically includes: Match the differential pressure gradient information with a preset deviation correction model to determine the collaborative correction coefficient corresponding to the current differential pressure gradient; Use the collaborative correction coefficient to correct the original concentration data during the emission detection of the light gas of the transformer, thereby collaboratively eliminating the concentration compensation deviation.

[0011] In this embodiment, determining the emission risk identifier during the light gas emission of the transformer through the airflow evolution trend specifically includes: Real-time collect the dynamic parameters of the flow rate, component concentration, and pressure fluctuation during the light gas emission process of the transformer through a multi-modal sensor array to generate an original airflow evolution data set; Determine the evolution trend feature vector of the airflow during the light gas emission of the transformer according to the airflow evolution trend; Determine the critical risk probability during the light gas emission of the transformer through the evolution trend feature vector; Determine the emission risk identifier during the light gas emission of the transformer according to the original airflow evolution data set and the critical risk probability;

[0012] In this embodiment, the dynamic identification of the gas escape level during the emission detection of the light gas of the transformer by the emission risk identifier specifically includes: Construct an escape discrimination set during the light gas emission of the transformer based on the emission risk identifier; Perform multi-dimensional matching of the escape discrimination set with the light gas emission scenario of the transformer to obtain the critical diffusion coefficient during the light gas emission of the transformer; Dynamically match the gas escape level during the emission detection of the light gas of the transformer according to the critical diffusion coefficient.

[0013] In this embodiment, the online linkage detection of the light gas of the transformer by means of the concentration compensation deviation after collaborative elimination and the gas escape level after dynamic identification specifically includes: Fuse the concentration compensation deviation after collaborative elimination and the gas escape level after dynamic identification to generate a composite detection feature matrix including a dynamic calibration coefficient and an escape risk weight; Determine the gas emission constraint corresponding to the light gas of the transformer through the composite detection feature matrix; Perform online linkage detection on the light gas of the transformer according to the gas emission constraint.

[0014] In this embodiment, the airflow evolution trend refers to the change trend of the flow path and diffusion direction of gas over time under specific environmental conditions.

[0015] The technical solution provided by this application has the following beneficial effects: By configuring a gas component analysis array in the construction of a biomass energy project, collecting gas sample data of the light gas of the transformer through the gas component analysis array; determining the trace gas characteristics of the light gas of the transformer during emission according to the gas sample data, setting a separation detection window for detecting the light gas of the transformer, resolving the pressure difference of the trace gas characteristics through the separation detection window to obtain the pressure difference gradient information during the emission of the light gas of the transformer, and then collaboratively eliminating the concentration compensation deviation during the emission detection of the light gas of the transformer by the pressure difference gradient information; analyzing the airflow evolution trend during the online detection of the light gas of the transformer, determining the emission risk identifier during the emission of the light gas of the transformer through the airflow evolution trend, and then dynamically identifying the gas escape level during the emission detection of the light gas of the transformer by the emission risk identifier; performing online linkage detection on the light gas of the transformer by means of the concentration compensation deviation after collaborative elimination and the gas escape level after dynamic identification.

[0016] It can be seen that in this application, the concentration compensation deviation in the detection of the light gas of the transformer can be effectively eliminated; among them, by configuring the gas component analysis array in the biomass energy project construction to collect the sample data of the light gas of the transformer, the efficient and accurate gas component monitoring is realized, the accuracy and real-time performance of the gas sample collection are improved, and the reliability of the subsequent analysis is ensured; by setting the separation detection window and using the differential pressure resolution technology to analyze the characteristics of trace gases, the concentration compensation deviation is effectively eliminated, the error in the gas concentration measurement is reduced, and the emission detection of the light gas of the transformer is more accurate; by analyzing the airflow evolution trend and combining the emission risk identification for dynamic identification, the potential risk of gas emission can be evaluated in real time, more intelligent emission monitoring is provided, the safety in the process of the light gas emission of the transformer is ensured, and the risk caused by gas escape is effectively prevented; by jointly eliminating the concentration compensation deviation and dynamically identifying the gas escape level, the accurate monitoring and adjustment in the gas emission process can be realized, the overall performance of the gas detection system is optimized, and the comprehensive accuracy and reliability of the on-line detection are improved.

[0017] In summary, the technical solution adopted in this application can accurately identify the gas emission risk in the complex environment of the light gas emission of the transformer, so as to improve the accuracy of the light gas emission detection of the transformer. Brief Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 It is the module structure diagram of the on-line gas detection system for the on-line intelligent emission gas of the light gas of the transformer provided by the present application; Figure 2 It is the flow schematic diagram for determining the differential pressure gradient information provided by the present application; Figure 3 It is the flow schematic diagram for determining the emission risk identification provided by the present application. Detailed Embodiments

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0021] The embodiment of the present application provides a gas online detection system for online intelligent emission of transformer light gas. The core of the system is to configure a gas component analysis array in the construction of a biomass energy project, and collect gas sample data of transformer light gas through the gas component analysis array; determine the trace gas characteristics of the transformer light gas when it is emitted according to the gas sample data, set a separation detection window when detecting the transformer light gas, and perform pressure difference resolution on the trace gas characteristics through the separation detection window to obtain the pressure difference ladder information when the transformer light gas is emitted, and then use the pressure difference ladder information to coordinately eliminate the concentration compensation deviation of the transformer light gas during emission detection; analyze the airflow evolution trend during the online detection of the transformer light gas, determine the emission risk mark of the transformer light gas during emission through the airflow evolution trend, and then use the emission risk mark to dynamically identify the gas emission level of the transformer light gas during emission detection; and perform online linkage detection of the transformer light gas through the concentration compensation deviation after coordinated elimination and the gas emission level after dynamic identification. The above scheme can be used to accurately identify the gas emission risk in the complex environment of transformer light gas emission, so as to improve the accuracy of transformer light gas emission detection.

[0022] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods. Figure 1 As shown, the figure is a module structure diagram of a gas online detection system for online intelligent emission of light gas from a transformer according to the present embodiment of the present application. The online detection system includes: a gas sample collection module 100, a sample deviation elimination module 200, a gas emission identification module 300 and an online linkage detection module 400, which are described as follows: The gas sample collection module 100 is used to configure a gas component analysis array in the construction of a biomass energy project, and to collect gas sample data of transformer light gas through the gas component analysis array.

[0023] In specific implementation, the configuration of the gas composition analysis array in the construction of biomass energy projects can be achieved in the following ways: first, multiple types of gas sensors are deployed at key emission sources in the construction area (such as boiler exhaust ports, combustion furnaces, gas storage tanks, pipeline interfaces, etc.), including electrochemical sensors (detection of CO, O 2 ), non-dispersive infrared sensor (NDIR, detect CO 2 , CH 4),(Semiconductor sensors (for detecting VOCs), and ensure that the sensors have high selectivity and low detection limits (ppb level). Secondly, a wireless transmission module (LoRa, NB-IoT) is used to build a remote data transmission network to upload sensor data to the cloud monitoring platform in real time. The platform uses time series data analysis (ARIMA model) and anomaly detection algorithms (DBSCAN, PCA dimensionality reduction analysis) to identify the changing trend of gas concentration, and combines CFD (Computational Fluid Dynamics) simulation to predict the gas diffusion path. Finally, through the intelligent early warning system, a gas concentration threshold is set. When an over-standard situation is detected, the system triggers safety measures such as automatic ventilation and combustion optimization to ensure that the gas emissions during the biomass energy construction process are controllable. That is, the configuration of the gas component analysis array in the biomass energy project construction is completed.

[0024] It should be noted that in this application, the gas component analysis array refers to a monitoring system composed of a variety of high-precision gas sensors and a data processing unit, which is used to collect, analyze and evaluate the gas components and their changing trends in the environment or industrial processes in real time.

[0025] In this embodiment, the gas sample data of the transformer's light gas can be specifically collected by the gas component analysis array through the following steps, that is: Control the gas component analysis array to establish a communication channel with the transformer's light gas discharge port, so that the transformer's light gas flows into the sampling area of the gas component analysis array; Collect samples of the transformer's light gas flowing into the sampling area according to the preset sampling rules to obtain gas sample data.

[0026] In specific implementation, first, special pipes that are corrosion-resistant and oxidation-resistant are adopted. The special pipes can be PTFE pipes or stainless steel pipes to ensure that the gas is not secondarily contaminated. A flow regulating valve and a sampling interface are installed at the transformer discharge port. The flow regulating valve and the sampling interface can be a steady flow chamber or a buffer tank, which are used to regulate the gas flow rate and reduce the influence of transient fluctuations on the sampling accuracy. A vacuum pump or a micro air pump is used to control the gas flowing into the analysis array to ensure a stable sampling flow rate. The sampling flow rate is set to 50 - 200 mL / min. In other embodiments, the sampling flow rate can also be set according to expert experience, which is not limited here. A flow sensor, a temperature sensor, and a pressure sensor are integrated in the communication channel to real-time monitor the gas parameters and ensure stable sampling conditions. A programmable logic controller (PLC) or an embedded system is used to control the opening and closing of the solenoid valve, and automatically switch the gas flow direction according to the sampling requirements, so that the light gas of the transformer flows into the sampling area of the gas component analysis array. Then, an appropriate sampling interval is set according to the influence of the load change and temperature change of the light gas of the transformer, such as continuous sampling or periodic sampling. A dynamic adjustment strategy is adopted, such as combining the Bayesian optimization algorithm or the empirical regression model, and dynamically adjusting the sampling frequency according to the gas concentration change trend. A mass flow controller (MFC) or a precision needle valve is used to regulate the sampling flow rate to ensure that the flow rate is within the range of 50 - 200 mL / min, and to avoid the influence of being too fast or too slow on the detection accuracy. The minimum sampling volume can be set to 500 mL to ensure that the minimum detection limit requirements of the detection instrument are met. Inert materials are used to collect samples. The inert materials can be glass bottles or aluminum foil bags to prevent secondary reactions or adsorption of gas components. Combining the micro vacuum chamber technology, the sample is stored in short-term sealed storage, and a condensation dehumidification device is used to eliminate the water vapor interference. Through LoRa, NB-IoT, 4G / 5G communication, the real-time sampling data is uploaded to the remote monitoring platform, and the gas sample data is obtained by reading the remote monitoring platform.

[0027] It should be noted that in this application, the sampling area refers to the functional space in the gas component analysis array for receiving, stabilizing the flow, and preliminarily processing the gas to be measured; the gas sample data refers to the numerical information of the gas composition, concentration, and physical properties.

[0028] The sample deviation elimination module 200 is used to determine the trace gas characteristics of the light gas of the transformer during discharge according to the gas sample data, set a separation detection window for detecting the light gas of the transformer, perform differential pressure resolution on the trace gas characteristics through the separation detection window to obtain the differential pressure gradient information of the light gas of the transformer during discharge, and then cooperatively eliminate the concentration compensation deviation of the light gas of the transformer during discharge detection by the differential pressure gradient information.

[0029] In this embodiment, the trace gas characteristics of transformer light gas during emission can be determined according to the gas sample data in the following manner, namely: Extracting features from the gas sample data to obtain gas component information; Constructing a preliminary set of gas components through the gas component information; The trace gas characteristics of transformer light gas during emission are determined based on the preliminary set of gas components.

[0030] In the specific implementation, first, the wavelet transform algorithm is used to eliminate the environmental noise in the gas sensor signal. Then the data is normalized, where the data normalization can be Z-score normalization, which is not limited here, to ensure that the gas component data of different dimensions are comparable. Then, an abnormal detection algorithm is used, such as PCA principal component analysis to eliminate abnormal data points, and principal component analysis (PCA) is used for dimensionality reduction to screen out the most representative component information for transformer light gas. Mutual information is used to evaluate the impact of each gas component on the transformer state and extract the most critical gas component characteristics. Combined with dynamic time warping (DTW), the change pattern of gas components over time is analyzed, and time series characteristics such as emission rate and gas concentration gradient are extracted, and the time series characteristics are used as gas component information. Then, a method based on correlation analysis (Spearman / Kendall / Pearson correlation coefficient) is used to screen gas components that are highly correlated with transformer emission characteristics, and random forest feature selection (RF-FS) is used to evaluate the impact weight of each gas component on the transformer operating state and screen high-weight components. Combining expert knowledge with historical data ensures that the initial set contains the most important gas components, such as H 2 , CH 4 , C 2 H 2 , CO, etc., and K-means cluster analysis is used to classify similar gas components, and the classified similar gas components are used as the preliminary set of gas components. Finally, machine learning methods such as BP neural network are used to identify the change trend of trace gases. The Gaussian mixture model (GMM) is used to model the gas concentration distribution and identify abnormal emission patterns. The long short-term memory network (LSTM) is used to predict the trend of trace gas concentration changes and determine whether there is a potential fault risk. The Markov hidden state model (HMM) is combined to analyze the gas evolution law under different operating conditions, and the gas evolution law is used as the trace gas feature of the transformer light gas during emission.

[0031] It should be noted that in this application, gas component information refers to the gas components and their physical and chemical properties extracted from gas sample data; the initial gas component set refers to the key component set for identifying the characteristics of the light gas emission of the transformer; the trace gas characteristics refer to the characteristic gas components that can still reflect the operating state and fault trend of the transformer at extremely low concentrations when the light gas of the transformer is emitted.

[0032] In this embodiment, setting the separation detection window for detecting the light gas of the transformer can be specifically implemented by the following steps, that is: Determine the associated distribution dimension of the light gas of the transformer; Determine the window correction boundary for detecting the light gas of the transformer according to the associated distribution dimension; Set the separation detection window for detecting the light gas of the transformer through the window correction boundary.

[0033] Specifically, first, obtain the operating temperature, humidity, and pressure of the transformer, and use the operating temperature, humidity, and pressure of the transformer as the operating environment to collect gas component information, where the gas component information is H 2 , CH 4 , C 2 H 2The trend of concentration changing with time is recorded, along with the equipment load and operating status. Here, the load and operating status are reflected by high load, low load, no load, and abnormal emissions. Multivariate regression analysis is used to find the trend of gas concentration changing with environmental parameters. The Pearson correlation coefficient is used to calculate the correlation between different gas concentrations, and irrelevant or low-correlation data is excluded. Combining K-means clustering analysis, the gas distribution under different transformer states is classified, and the classified results are used as the associated distribution dimensions of the transformer's light gas. Then, using probability density estimation (Kernel Density Estimation, KDE), the probability distribution of gas components under different states is calculated to determine the concentration change range under normal conditions. Then, Box-Plot analysis is used to calculate the interquartile range (IQR) of gas concentration to identify the boundary between the normal range and outliers. Through dynamic threshold calculation based on time series analysis, it is ensured that the detection window can adapt to changes in external conditions such as environmental temperature and equipment load. Combining fuzzy logic reasoning (Fuzzy Logic), the gas concentration detection range is adjusted according to different operating states. For example, the boundary is appropriately widened during high load to reduce false alarms. According to the calculated normal range and abnormal identification threshold, where the normal range is the mean ± standard deviation of gas concentration detection, the upper and lower limits of the detection window are set, and the upper and lower limits of the detection window are used as the window correction boundaries when detecting the transformer's light gas. Finally, the separation detection window area is divided: the normal detection area is set as the green area, where the gas concentration is within the normal range, and the system only records basic data without triggering an alarm; the early warning detection area is set as the yellow area, where the gas concentration is close to the upper or lower limit, and the system triggers an early warning signal and requires further monitoring; the abnormal detection area is set as the red area, where the gas concentration exceeds the set threshold, and the system triggers an abnormal alarm and activates safety protection measures. A sliding window mechanism is used to calculate the trend change rate based on continuous data within a short period to determine whether the detection window needs to be adjusted. A time series prediction model is used, where the time series prediction model can be LSTM, RNN, which is not limited here, to predict the short-term gas concentration change and adjust the window range in advance to avoid false alarms or lagged responses, that is, the setting of the separation detection window for detecting the transformer's light gas is completed.

[0034] It should be noted that in this application, the associated distribution dimension is a key influencing factor describing the transformer's light gas under different environmental, operating, or time conditions; the window correction boundary represents the upper and lower limits of the detection range of the transformer's light gas; the separation detection window represents an area for independent, hierarchical, and targeted detection of the transformer's light gas.

[0035] Preferably, in this embodiment, the differential pressure of the trace gas characteristics is resolved through the separation detection window to obtain the differential pressure gradient information when the transformer's light gas is emitted, referring to Figure 2As shown, this figure is a schematic flowchart of determining differential pressure gradient information in some embodiments of the present application. In this embodiment, the determination of differential pressure gradient information can be achieved through the following steps: In step S21, determine the differential pressure resolution feature according to the physical parameters corresponding to the trace gas characteristics in the separation detection window; In step S22, determine the differential pressure response amounts at different positions within the separation detection window through the differential pressure resolution feature; In step S23, determine the pressure evolution trend of trace gas characteristics at different differential pressures according to the differential pressure response amounts; In step S24, determine the gas differential pressure limit during the light gas emission of the transformer; In step S25, determine the differential pressure gradient information during the light gas emission of the transformer according to the pressure evolution trend and the gas differential pressure limit.

[0036] In specific implementation, first, for the light gas, focus on analyzing its main components, and the main components include: H 2 、CH 4 、C 2 H 2Gas diffusivity, molecular motion characteristics, and rheological characteristics under different pressure differences. The ideal gas state equation is used to simulate the behavior of gas in different pressure difference environments. Combining with Bernoulli's equation in fluid mechanics, the flow characteristics of gas under different pressure differences are estimated, including velocity, flow rate changes, etc. Then, according to the diffusion coefficient and viscosity of gas components, gas characteristics sensitive to pressure difference changes are identified. Using gas sensor data, the relationship between concentration and pressure difference is determined through regression analysis to obtain pressure difference resolution characteristics. Next, based on the orifice flow formula, the pressure difference response amount at different positions is calculated, that is, the concentration change of gas under different pressure differences. Combining with the velocity gradient and flow distribution of gas at different positions, the difference equation or finite difference method is used to solve the pressure difference response amount. Multiple measurement points are set in the separation detection window to detect the gas concentration change and pressure change at different positions respectively, and the data at each position are integrated by the weighted average method, and the integrated result is used as the pressure difference response amount. Again, according to the pressure difference response amount, a graph of the relationship between pressure difference and gas concentration is drawn to describe the change trend of gas concentration under different pressure differences. A mathematical model of the change of gas concentration with pressure difference is obtained through linear regression, and computational fluid dynamics simulation is used to simulate the evolution trend of gas under different pressure differences, and the simulation result is used as the pressure evolution trend of trace gas characteristics under different pressure differences. Then, by analyzing the historical data of normal emissions and abnormal emissions, the gas pressure difference limit for gas emissions is set, and the range method is used to determine a reasonable pressure difference limit to ensure that emissions within the normal operating range do not exceed the set value. Finally, according to the set pressure difference limit, the emission process is divided into multiple echelon stages, which are normal, warning, and abnormal. Within each echelon stage, the changes in gas concentration and pressure difference are monitored, and the value with the largest slope in the changes of gas concentration and pressure difference is used as the key change point in the emission process, and this key change point is used as the pressure difference echelon information during the light gas emission of the transformer.

[0037] It should be noted that in this application, the pressure difference resolution characteristic refers to the influence mode of gas pressure difference on the detection result within the separation detection window; the pressure difference response amount refers to the reaction degree of gas at different positions within the separation detection window to the pressure difference change; the pressure evolution trend refers to the change law of gas concentration and distribution under different pressure differences; the gas pressure difference limit refers to the upper and lower pressure difference thresholds set according to the pressure change and gas characteristics during the light gas emission of the transformer; the pressure difference echelon information refers to the emission stage information of gas at different pressure differences when the gas pressure difference and concentration change.

[0038] In this embodiment, the collaborative elimination of the concentration compensation deviation of the light gas of the transformer during emission detection by the pressure difference echelon information can be specifically implemented by the following steps, that is: Match the differential pressure gradient information with a preset deviation correction model to determine a collaborative correction coefficient corresponding to the current differential pressure gradient; Use the collaborative correction coefficient to correct the original concentration data during the detection of the light gas emission of the transformer, thereby collaboratively eliminating the concentration compensation deviation.

[0039] When specifically implemented, first, based on historical data and experimental results, establish a correction model related to the differential pressure gradient. The model includes: the non-linear relationship between gas concentration and differential pressure; the sensor response curve, considering the error characteristics of the sensor; the influence of temperature and humidity on gas detection. Then, use statistical regression analysis to fit the deviation correction model, and match the real-time collected differential pressure gradient information with the preset deviation correction model. Use the K-nearest neighbor algorithm (KNN) to find the corresponding collaborative correction coefficient. Among them, the collaborative correction coefficient can be a constant or a dynamic variable that changes with the differential pressure gradient. Then, combine the real-time collected original concentration data with the matched collaborative correction coefficient, and use the following formula to correct the concentration data: C = A × B, where C represents the corrected concentration data, A represents the original concentration data, and B represents the collaborative correction coefficient determined by the differential pressure gradient information. Considering that there may be various factors affecting the concentration deviation, the collaborative correction coefficient can be multi-dimensional. Combine environmental variables such as temperature, humidity, and gas components to jointly correct the data, and use multiple linear regression or weighted average method to synthesize and correct various correction coefficients.

[0040] It should be noted that in this application, the collaborative correction coefficient refers to the quantity that adjusts the deviation between the current differential pressure gradient and the change in gas concentration; the preset deviation correction model refers to a mathematical model established in advance according to factors such as differential pressure gradient information, gas component characteristics, and sensor working principles, and is used to correct the errors in the detection data; the concentration compensation deviation refers to the concentration measurement deviation caused by changes during the gas emission process or sensor errors.

[0041] The gas emission identification module 300 is used to analyze the airflow evolution trend during the on-line detection of the light gas of the transformer, determine the emission risk identifier during the emission of the light gas of the transformer through the airflow evolution trend, and then dynamically identify the gas escape level during the emission detection of the light gas of the transformer by the emission risk identifier.

[0042] In this embodiment, the analysis of the airflow evolution trend during the on-line detection of the light gas in the transformer can be specifically implemented in the following manner, that is: First, establish an airflow simulation model around the transformer discharge port, and use computational fluid dynamics (CFD) technology to simulate the flow path after gas discharge. By considering parameters such as the velocity, temperature, and humidity of the airflow, predict the diffusion of the gas in the environment, analyze the interaction between the gas components and the airflow, and combine the diffusion coefficient and flow characteristics of the gas to analyze the propagation law of the gas under different temperature and pressure conditions. And through multi-point measurement data and wind speed information, evaluate how the airflow affects the distribution and evolution direction of the light gas. Then, for the discrimination of the evolution trend, use the real-time monitoring data and the simulation results to compare the real-time flow direction of the gas discharge and judge the specific trend of the airflow evolution. According to the speed and direction of the gas diffusion, predict whether the gas will deviate from the normal flow path and analyze whether it will cause potential leakage or danger, that is, complete the analysis of the airflow evolution trend during the on-line detection of the light gas in the transformer.

[0043] It should be noted that in this application, the airflow evolution trend refers to the change trend of the flow path and diffusion direction of the gas over time under specific environmental conditions.

[0044] Preferably, in this embodiment, determine the emission risk identifier when the transformer light gas is discharged through the airflow evolution trend, refer to Figure 3 As shown, this figure is a schematic flow chart for determining the emission risk identifier in some embodiments of this application. The determination of the emission risk identifier in this embodiment can be implemented by the following steps: In step S31, use the multi-modal sensor array to collect the dynamic parameters of the flow velocity, component concentration, and pressure fluctuation in real time during the discharge process of the transformer light gas, and generate the original data set of the airflow evolution. In step S32, determine the evolution trend feature vector of the airflow when the transformer light gas is discharged according to the airflow evolution trend. In step S33, determine the critical risk probability when the transformer light gas is discharged through the evolution trend feature vector. In step S34, determine the emission risk identifier when the transformer light gas is discharged according to the original data set of the airflow evolution and the critical risk probability.

[0045] In specific implementation, first, a multi-modal sensor array is adopted, combining devices such as gas concentration sensors, pressure sensors, and flow rate sensors. A sensor array is set at the transformer discharge port and its surroundings to collect parameters such as flow rate, gas component concentration, and pressure fluctuation in real time, ensuring that the data covers the entire process of gas emission. The data of each sensor is synchronously recorded through a data acquisition system, and an original dataset of airflow evolution is generated. This dataset contains dynamic change information generated during the emission process of the airflow. Then, according to information such as the speed of the airflow, gas concentration, and pressure fluctuation, key features in the airflow evolution process are extracted, including airflow direction, speed change, concentration attenuation, etc., and the principal component analysis (PCA) is used to extract the main change trend in the airflow evolution process. The main change trend is used as the evolution trend feature vector. Then, based on historical data or simulation results, a critical risk probability model is established. The model predicts potential emission risks by analyzing the airflow evolution pattern, and then according to the extracted evolution trend feature vector, the critical risk probability under the current airflow evolution pattern is calculated using this model. Among them, Bayesian classification can be used to calculate the critical risk probability when the transformer emits light gas. Finally, the original dataset of airflow evolution, where the original dataset of airflow evolution protects flow rate, gas concentration, pressure fluctuation, and critical risk probability, is combined to generate a comprehensive risk assessment dataset, and a rule engine or classification algorithm is used to generate an emission risk identifier for the emission process. The emission risk identifier includes different risk levels, and the risk levels include low, medium, and high risks), and then the emission process is monitored in real time. Based on the current airflow evolution pattern and risk probability, the emission risk identifier is automatically output.

[0046] It should be noted that in this application, the multi-modal sensor array refers to a sensor system that integrates multiple types of sensors to work together; the original dataset of airflow evolution refers to the original data set of airflow speed, gas concentration, and pressure fluctuation; the evolution trend feature vector refers to the feature data vector of the airflow diffusion trend; the critical risk probability refers to the possibility of a safety risk occurring during the gas emission process; the emission risk identifier refers to the safety risk level marked during the gas emission process.

[0047] In this embodiment, the dynamic identification of the gas escape level during the emission detection of the transformer's light gas by the emission risk identifier can be specifically implemented in the following way, that is: Construct a dispersion discrimination set for the emission of the transformer's light gas based on the emission risk identifier; Perform multi-dimensional matching between the dispersion discrimination set and the emission scenario of the transformer's light gas to obtain the critical diffusion coefficient during the emission of the transformer's light gas; Dynamically match the gas escape level during the emission detection of the transformer's light gas according to the critical diffusion coefficient.

[0048] In specific implementation, first, according to the gas emission risk identification, a dispersion discrimination set including different risk levels is constructed. This dispersion discrimination set includes various gas dispersion modes, such as slow gas diffusion, medium gas diffusion, and rapid gas diffusion that may form a dangerous gas cloud. Then, collect the diffusion characteristics during gas emission in different scenarios, including parameters such as wind speed, ambient temperature, air pressure, and gas concentration. Combine with the emission risk identification during gas emission to establish specific discrimination criteria for each gas dispersion mode, and obtain the dispersion discrimination set for the light gas emission of the transformer. Then, describe the current emission situation according to the emission environment and actual scenario of the transformer light gas, and compare the parameters of the emission scenario with each dispersion mode in the dispersion discrimination set to identify the gas diffusion mode that best matches the current emission conditions. Then, according to the matching result, use the Gaussian model to calculate the critical diffusion coefficient. Finally, according to the critical diffusion coefficient, dynamically evaluate the diffusion trend of the gas during the emission process. According to the magnitude of the diffusion coefficient, judge the level of gas dispersion, such as whether there is dispersion to a high-risk area. Then, according to the calculated critical diffusion coefficient, match the current gas emission situation with the defined dispersion levels. The levels can be divided into multiple grades. For example, the low grade is less gas dispersion, the medium grade is the gas starting to disperse, and the high grade is the gas rapidly dispersing and possibly affecting the surrounding environment.

[0049] It should be noted that in this application, the dispersion discrimination set refers to a discrimination data set including different gas dispersion modes and risk levels; the critical diffusion coefficient refers to a coefficient that measures the degree of gas diffusion during the emission process; the light gas emission scenario of the transformer represents the actual conditions when the gas is released from the transformer emission port to the environment, including gas flow rate, temperature, air pressure, wind speed, and emission position; the gas dispersion level refers to the non-dispersion levels divided according to the degree and risk of gas diffusion during the emission process.

[0050] The online linkage detection module 400 is used to perform online linkage detection on the transformer light gas through the concentration compensation deviation after collaborative elimination and the gas dispersion level after dynamic identification.

[0051] In this embodiment, performing online linkage detection on the transformer light gas through the concentration compensation deviation after collaborative elimination and the gas dispersion level after dynamic identification can specifically adopt the following method, that is: Fuse the concentration compensation deviation after collaborative elimination and the gas dispersion level after dynamic identification to generate a composite detection feature matrix including a dynamic calibration coefficient and a dispersion risk weight; Determine the gas emission constraint corresponding to the transformer light gas through the composite detection feature matrix; Perform online linkage detection on the transformer light gas according to the gas emission constraint.

[0052] In specific implementation, first, the concentration compensation deviation after collaborative elimination and the gas escape level after dynamic identification are combined into a matrix. Each row of the matrix represents a time period or an emission cycle to ensure that the emission changes of the gas can be reflected in real time. The columns of the matrix include the concentration compensation deviation after collaborative elimination and the gas escape level after dynamic identification. The columns can be divided into two groups. The first group is the concentration compensation deviation, and the second group is the gas escape level, forming a composite detection feature matrix, that is, the composite detection feature matrix. This composite detection feature matrix includes the corrected concentration compensation deviation and the gas escape level after dynamic identification. Then, according to the information in the composite detection feature matrix, the constraint conditions for the light gas emission of the transformer are defined. These constraints include but are not limited to that the concentration of the gas should be within a certain range to avoid excessive emissions, and at a specific escape level, the gas should follow a specific diffusion speed or diffusion range. Next, according to the composite detection feature matrix, the key parameters affecting gas emissions are determined, such as gas concentration, wind speed, emission port height, air flow direction, temperature, gas type, etc. The dynamic calibration coefficient and the escape risk weight in the composite detection feature matrix will be used as input parameters. According to environmental standards and safety regulations, boundary conditions such as the maximum allowable concentration of emissions, gas diffusion speed, and gas diffusion distance are set. The finite difference method is used to discretize the established mathematical model, and the discretized result is used as the gas emission constraint corresponding to the light gas of the transformer. Finally, an online monitoring system is established to collect data such as the concentration, flow rate, and pressure of the light gas of the transformer in real time, and combined with the real-time data of the gas and the gas emission constraints, linkage detection is implemented. When any parameter in the gas emission process deviates from the gas emission constraint range, the system will automatically issue an alarm or take corrective measures. The corrective measures can be adjusting the emission volume, shutting down the equipment, which are not limited here. During the online detection process, the system dynamically adjusts the detection strategy or emission control strategy according to the real-time data and the changes in gas diffusion to ensure that the emission process always complies with the gas emission constraints.

[0053] It should be noted that in this application, the dynamic calibration coefficient represents the coefficient for adjusting the deviation of gas detection data during real-time monitoring; the escape risk weight represents the risk importance of different escape levels during gas emission; the composite detection feature matrix refers to the multi-dimensional feature data set during gas emission; the gas emission constraint refers to the conditions for limiting the gas emission concentration, speed, and diffusion range.

[0054] It can be seen that in this application, the concentration compensation deviation in the detection of the light gas of the transformer can be effectively eliminated. Among them, by configuring the gas component analysis array in the biomass energy project construction to collect the sample data of the light gas of the transformer, the efficient and accurate gas component monitoring is realized, the accuracy and real-time performance of the gas sample collection are improved, and the reliability of the subsequent analysis is ensured. By setting the separation detection window and using the differential pressure resolution technology to analyze the trace gas characteristics, the concentration compensation deviation is effectively eliminated, the error in the gas concentration measurement is reduced, and the emission detection of the light gas of the transformer is made more accurate. By analyzing the airflow evolution trend and combining with the emission risk identification for dynamic identification, the potential risk of gas emission can be evaluated in real time, more intelligent emission monitoring is provided, the safety in the process of the light gas emission of the transformer is ensured, and the risk caused by gas leakage is effectively prevented. By synergistically eliminating the concentration compensation deviation and dynamically identifying the gas leakage level, the precise monitoring and adjustment in the gas emission process can be realized, the overall performance of the gas detection system is optimized, and the comprehensive accuracy and reliability of the on-line detection are improved.

[0055] In summary, the technical solution adopted in this application can accurately identify the gas emission risk in the complex environment of the light gas emission of the transformer, so as to improve the accuracy of the light gas emission detection of the transformer.

[0056] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general computer, a special computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one or more of the flows Figure 1 or a device for realizing the functions specified in one or more of the blocks.

[0057] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disc memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0058] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such process, method, commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, commodity or device including the element.

Claims

1. A gas online detection system for online intelligent exhaust gas of transformer light gas, characterized in that: The online detection system comprises: A gas sample collection module is used to configure a gas component analysis array in the construction of a biomass energy project, and to collect gas sample data of transformer light gas through the gas component analysis array; A sample deviation elimination module is used to determine the trace gas characteristics of the transformer light gas when it is discharged according to the gas sample data, set a separation detection window when detecting the transformer light gas, perform pressure difference resolution on the trace gas characteristics through the separation detection window, and obtain the pressure difference ladder information when the transformer light gas is discharged, and then use the pressure difference ladder information to collaboratively eliminate the concentration compensation deviation of the transformer light gas during emission detection; A gas emission identification module is used to analyze the gas flow evolution trend during the online detection of transformer light gas, determine the emission risk identification of transformer light gas during emission through the gas flow evolution trend, and then dynamically identify the gas emission level of transformer light gas during emission detection through the emission risk identification; The online linkage detection module is used to perform online linkage detection on transformer light gas through the concentration compensation deviation after collaborative elimination and the gas escape level after dynamic identification.

2. A gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1, characterized in that: The gas sample data of transformer light gas collected by the gas component analysis array specifically includes: Controlling the gas component analysis array to establish a communication channel with the transformer light gas discharge port, so that the transformer light gas flows into the sampling area of ​​the gas component analysis array; According to the preset sampling rules, samples of transformer light gas flowing into the sampling area are collected to obtain gas sample data.

3. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1 is characterized in that: Determining the trace gas characteristics of transformer light gas during emission based on the gas sample data specifically includes: Extracting features from the gas sample data to obtain gas component information; Constructing a preliminary set of gas components through the gas component information; The trace gas characteristics of transformer light gas during emission are determined based on the preliminary set of gas components.

4. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1 is characterized in that: The separation detection window for detecting transformer light gas specifically includes: Determine the associated distribution dimensions of transformer light methane gas; Determining a window correction boundary when detecting transformer light gas according to the associated distribution dimension; The window correction boundary is used to set a separate detection window for detecting transformer light gas.

5. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1 is characterized in that: The pressure difference information of the transformer light gas emission when the trace gas characteristics are subjected to pressure difference resolution through the separation detection window specifically includes: Determining a pressure difference resolution characteristic according to a physical parameter corresponding to the trace gas characteristic in the separation detection window; Determining the pressure difference response amount at different positions in the separation detection window by using the pressure difference resolution characteristic; Determining the pressure evolution trend of the trace gas characteristics under different pressure differences according to the pressure difference response; Determine the gas pressure difference limit when the transformer light gas is discharged; The pressure difference gradient information when the transformer light gas is discharged is determined according to the pressure evolution trend and the gas pressure difference limit.

6. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1, characterized in that: The coordinated elimination of the concentration compensation deviation of transformer light gas during emission detection by using the pressure difference ladder information specifically includes: Matching the pressure difference step information with a preset deviation correction model to determine a coordination correction coefficient corresponding to the current pressure difference step; The collaborative correction coefficient is used to correct the original concentration data during transformer light gas emission detection, thereby collaboratively eliminating concentration compensation deviation.

7. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1 is characterized in that: The emission risk identification when determining the light gas emission of the transformer through the airflow evolution trend specifically includes: The multimodal sensor array is used to collect the flow rate, component concentration and pressure fluctuation dynamic parameters of the transformer light gas emission process in real time to generate the original data set of gas flow derivative; Determining a derivative trend characteristic vector of the airflow when the transformer light gas is discharged according to the airflow derivative trend; Determine the critical risk probability of transformer light gas emission by using the derivative trend characteristic vector; An emission risk indicator for transformer light gas emission is determined based on the airflow-derived original data set and the critical risk probability.

8. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1 is characterized in that: Dynamically identifying the gas emission level of transformer light gas during emission detection by the emission risk identification specifically includes: Constructing a fugitive discrimination set for transformer light gas emission based on the emission risk identification; According to the fugitive discrimination set, multi-dimensional matching is performed with the transformer light gas emission scenario to obtain the critical diffusion coefficient of the transformer light gas emission; The gas escape level of the transformer light gas during emission detection is dynamically matched according to the critical diffusion coefficient.

9. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1, characterized in that: The online linkage detection of transformer light gas through the concentration compensation deviation after collaborative elimination and the gas escape level after dynamic identification specifically includes: The concentration compensation deviation after synergistic elimination and the gas emission level after dynamic identification are integrated to generate a composite detection feature matrix including dynamic calibration coefficients and emission risk weights; Determining the gas emission constraints corresponding to the transformer light gas through the composite detection feature matrix; The transformer light gas is subjected to online linkage detection according to the gas emission constraints.

10. The gas online detection system for online intelligent exhaust gas of transformer light gas as claimed in claim 1, characterized in that: The gas flow evolution trend refers to the change trend of the flow path and diffusion direction of the gas over time under specific environmental conditions.

Images