A gas emission monitoring system and method based on spectral analysis

Abstract

The present invention discloses a gas emission monitoring system and method based on spectral analysis, belonging to the field of artificial intelligence technology. The system starts a sampling pump to extract a gas sample from the environment to be tested, removes impurities, and transports the sample to a gas sample pool. A spectrum is generated and used to identify gas components and concentrations. When the spectrum identifies several gas components, a support vector machine algorithm and a random forest algorithm are used to establish a hazardous gas combination library. The system then detects the gas components and concentrations in the current environment in real time and compares them with a hazardous reaction library. If the detected gas combination meets the hazardous combination and the concentration reaches or exceeds a critical value, an alarm signal is generated. When the alarm is triggered, specific information is displayed on a control panel, an alarm is sounded, emergency operation instructions are displayed, the ventilation system is automatically started or increased, and the isolation device is automatically started. After the alarm is automatically released, data is recorded and the corresponding equipment failure is analyzed.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and particularly to a gas emission monitoring system and method based on spectral analysis. Background Art

[0002] With the continuous advancement of industrialization and urbanization, the problem of gas emissions generated during industrial production has become increasingly serious. These waste gases usually contain various harmful components, such as sulfur dioxide, carbon monoxide, nitrogen oxides, volatile organic compounds, etc. If not controlled, they will not only cause irreversible pollution to the environment around the factory, but also pose a serious threat to the health of personnel inside and near the factory. Therefore, the application of gas emission monitoring technology in industrial production has become crucial. Whether harmful gas emissions can be monitored and controlled in a timely and accurate manner is directly related to environmental protection and personnel safety.

[0003] Traditional gas monitoring methods often rely on chemical reagents or simple sensors. When detecting multiple gas components, these methods have low accuracy and sensitivity and are prone to errors. Many existing monitoring systems cannot achieve real-time monitoring, and there are lags in data transmission and processing, unable to promptly reflect changes in gas concentration, resulting in potential dangers not being discovered and handled in a timely manner. Existing systems usually rely on manual operation and analysis, with low automation levels, making it difficult to achieve continuous and stable gas monitoring and control. The process of manual operation is not only time-consuming and laborious, but also prone to errors in data collection and processing due to human factors. Many monitoring systems lack a perfect emergency response mechanism and cannot take effective emergency measures quickly when the concentration of dangerous gases exceeds the standard. This defect makes it difficult for the system to respond in a timely manner when facing sudden dangers and cannot effectively guarantee the safety of the environment and personnel. Summary of the Invention

[0004] The purpose of the present invention is to provide a gas emission monitoring system and method based on spectral analysis to solve the problems raised in the above background art.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] A gas emission monitoring method based on spectral analysis, the method comprising the following steps:

[0007] S100. Start the sampling pump, extract a gas sample from the environment to be measured, remove impurities through the sampling pipeline and filter, and transport it to the gas sample cell. The gas flow rate in the sample cell is adjusted by a flow controller; it flows through the inlet and outlet at both ends of the sample cell, and the gas in the sample cell is evenly distributed. The light beam emitted by the light source passes through the sample cell, and spectral measurement is performed by a spectrometer to generate a spectrogram, and the gas components and gas concentrations are identified using the spectrogram;

[0008] S200: When the spectrum identifies several gas components, a support vector machine algorithm and a random forest algorithm are used to establish a dangerous gas combination library, wherein the dangerous gas combination library lists gas combinations that may cause dangerous reactions and their critical concentrations; the gas components and concentrations in the current environment are detected in real time and compared with the dangerous reaction library. If the detected gas combination meets the dangerous combination and the concentration reaches or exceeds the critical value, an alarm signal is generated;

[0009] S300: When the alarm is triggered, the specific dangerous gas combination, corresponding concentration information and potential hazards are displayed on the control panel, an alarm is sounded through the sound and light alarm, emergency operation instructions are displayed, the ventilation system is automatically started or increased, and the isolation device is automatically started;

[0010] S400 monitors the gas concentration in the sample pool in real time. When the concentration of all hazardous gases has dropped below the safety threshold, it automatically lifts the alarm, linearly adjusts the operation of the ventilation system, closes the isolation device, records detailed data of the hazardous event, and uses the random forest algorithm to analyze the failure of the corresponding equipment.

[0011] According to step S100, a flow controller is installed on the air inlet pipe of the gas sample pool or on the pipe near the sampling pump. The flow controller is a mass flow controller. Before starting the system, the flow controller is calibrated to set a target flow range. The flow range is determined based on the characteristics of the gas to be tested and the sample pool.

[0012] When determining the flow range, it is necessary to obtain the physical properties of the gas to be measured, such as the type, density, viscosity, diffusion coefficient, etc. These parameters will affect the flow of the gas in the pipeline and the distribution in the sample pool. If there are highly chemically reactive components in the gas to be measured, the behavior of these components under high or low flow conditions must be considered to prevent possible chemical reactions. According to the volume of the sample pool, the flow range of the gas sample is determined so that the gas can be fully distributed in the sample pool, ensuring that the light source beam can accurately reflect the gas concentration when passing through the sample pool. The path length of the light source beam through the sample pool determines the detection sensitivity of the gas components. A longer optical path length requires a lower gas flow rate to ensure that the light beam has sufficient absorption and reflection in the gas. Before starting the system, calibrate the mass flow controller and set the target flow range of the flow controller, usually in the range of 0.1-10 L / min.

[0013] After starting the sampling pump, the flow controller begins to work, monitoring the flow rate of the gas sample in real time. The sensor inside the controller detects the flow signal and compares it with the set value. If the actual flow rate deviates from the set value, the controller automatically adjusts the valve opening or the pump speed to ensure that the flow rate is stable within the target range. The flow controller is equipped with a closed-loop control system to regulate the flow rate through real-time feedback. The monitoring interface displays the current flow rate value, allowing the operator to monitor in real time and make manual adjustments.

[0014] According to step S100, a xenon lamp is selected as the light source. The spectrum of the light source covers the absorption wavelength range of the target gas. The light source is introduced into the optical path system through a collimating lens and an optical fiber to form a parallel beam. The gas sample cell is designed with quartz glass as the material. The beam enters from one end of the sample cell, passes through the gas sample, and then exits from the other end. Gas molecules have an absorption effect on light of a specific wavelength. During the process of the beam passing through the sample cell, the intensity and wavelength of the beam change.

[0015] The beam entering the spectrometer is decomposed into spectra of different wavelengths by a grating or a prism. The decomposed spectra are projected onto a detector. The detector generates electrical signals according to the light intensity of different wavelengths and converts these electrical signals into digital signals. The light intensity data of different wavelengths are plotted into a spectrogram, which shows the absorption of the gas sample for light of each wavelength. The gas components and gas concentrations are identified by the position and intensity of the absorption peaks.

[0016] According to step S200, the spectral data of historical gas samples are collected. Each sample includes several gas components and their concentrations, and it is marked whether a dangerous reaction will occur. The marked sample labels are: dangerous is 1, safe is -1. Features are extracted from the spectral data, and the features include absorption peak position, absorption intensity, peak area, full width at half maximum, baseline offset, and peak shape.

[0017] Support vector machines are used to identify dangerous gas combinations. Its goal is to find a hyperplane to separate samples of different classes, and the classes include dangerous and safe. The spectral data feature vectors and the corresponding labels are composed into a first training data set, and the first training data set is used for training. The objective function is defined as follows:

[0018] ;

[0019] where w is the weight vector representing the direction of the hyperplane, and b is the bias term representing the displacement of the hyperplane.

[0020] For each training sample, the constraint conditions to be satisfied are defined as follows:

[0021] ;

[0022] where is the sample label of the i-th sample. is the feature vector extracted from the spectral data of the i-th sample.

[0023] Construct the Lagrangian function through Lagrange multipliers and solve the dual problem:

[0024] ;

[0025] where, are Lagrange multipliers for the dual solution of the optimization problem, representing the slack variables of the constraint conditions, and n is the number of training samples;

[0026] Solve the Lagrangian function, take the derivatives with respect to w and b, and set the derivatives to zero to obtain the dual problem:

[0027] ;

[0028] The constraint conditions of the dual problem are:

[0029] ;

[0030] Use the gradient descent algorithm to solve the dual problem to obtain the optimal ; Utilize the optimal to calculate the weight vector w and the bias term b, and the formulas are as follows:

[0031] ;

[0032] where, is the label of the support vector, is the feature vector of the support vector;

[0033] The classification decision function of the trained support vector machine model is:

[0034] ;

[0035] where, is the classification decision function for judging whether the gas combination is dangerous.

[0036] According to step S200, based on the dangerous gas combinations obtained through the support vector machine algorithm, collect the spectral data of the gas samples corresponding to the dangerous gas combinations; form the second training data set with the spectral data feature vectors of the gas samples of the dangerous gas combinations and the critical concentrations of the dangerous gas combinations, randomly sample several subsets from the second training data set, and each subset is used to train a decision tree; conduct decision tree training on each subset to establish a regression model; input the feature vector of the new sample into each decision tree for prediction, and take the average of the prediction results of all decision trees as the final critical concentration prediction value.

[0037] According to step S200, for each decision tree , use the training data subset for training, where is the spectral data feature vector of the gas sample of the hazardous gas combination, is the critical concentration of the hazardous gas combination; for the feature vector of the hazardous gas combination, the prediction result of each decision tree is , indicating the prediction result of the i-th decision tree for , and output the critical concentration of the hazardous gas combination;

[0038] The random forest gives the final prediction value by combining the prediction results of all decision trees :

[0039] ;

[0040] where is the number of decision trees;

[0041] Combine the critical concentrations of the hazardous gas combinations obtained by the support vector machine algorithm and the decision tree algorithm to establish a hazardous gas combination library.

[0042] According to step S300, the control panel will display the detected gas types, the concentrations of each gas, the danger level, and the description of potential hazards, and display the emergency operation guidelines on the control panel and the mobile devices of the staff; the emergency operation guidelines include how to wear personal protective equipment, the specific routes and steps for evacuating personnel, and the steps for starting the ventilation system; automatically start or increase the operation of the ventilation system to reduce the concentration of hazardous gases, and automatically adjust the wind speed and air volume of the ventilation system according to the gas concentration and combination; when the gas sensor detects that the concentration of hazardous gases exceeds the standard, automatically trigger the isolation device; when a fault in the isolation device is found, give an automatic alarm for the isolation device fault and provide troubleshooting guidelines.

[0043] On the control panel and the mobile devices of the staff, the detected gas types, the concentrations of each gas, the danger level, and the description of potential hazards are displayed in real time. Specifically, it shows how to wear personal protective equipment, including the types of protective clothing, masks, gloves, etc. to be used. According to the detected gas concentration and wind direction, a specific evacuation route map is displayed on the control panel, and the personnel are guided to evacuate safely through the navigation function. List the detailed steps, such as: shutting down the working equipment, quickly evacuating to a safe area, taking a headcount, etc. The steps for starting the ventilation system are displayed on the control panel, including the position of the manual start button, the automatic start conditions, etc.

[0044] When the gas sensor detects that the isolation device is not functioning properly, the system automatically generates an alarm and displays a fault message on the control panel. Detailed troubleshooting instructions are provided, including power supply checks, device restarts, and component replacement. The system's built-in fault diagnosis algorithm automatically analyzes the cause of the problem and provides appropriate repair recommendations.

[0045] According to step S400, the gas concentration in the sample pool is continuously monitored by sensors. When the concentration of all hazardous gases drops below a safety threshold, the alarm is automatically lifted. Based on the real-time monitored gas concentration data, the system automatically adjusts the wind speed and air volume of the ventilation system. The automatic control system shuts down previously activated isolation devices and restores the normal operation of the isolated area. The time of the hazardous event, the type of gas detected, the concentration change, and the emergency measures taken are automatically recorded.

[0046] Use the random forest algorithm for fault analysis. Input the preprocessed equipment operation data into the random forest model, randomly select feature subsets, and build several decision trees. Use the training data to build multiple decision trees. Each tree is trained independently and predicts the new data. The average prediction result of all decision trees is taken as the equipment fault analysis data and recorded in the hazardous gas combination library.

[0047] A gas emission monitoring system based on spectral analysis, comprising:

[0048] Sampling and filtration module: includes: gas sampling unit, gas filtration unit and flow control unit; the gas sampling unit is responsible for starting the sampling pump and extracting gas samples from the environment to be tested through the sampling probe; the gas filtration unit is responsible for controlling the gas sample to pass through the sampling pipeline and filter, remove impurities and then be transported to the gas sample pool; the flow control unit is responsible for adjusting the flow of the gas sample through the flow controller;

[0049] Spectral measurement module: includes: sample cell control unit, light source control unit and spectrometer control unit; among them, the sample cell control unit is responsible for controlling the flow of gas inlet and outlet at both ends of the sample cell when the gas sample enters the sample cell, so that the gas in the sample cell is evenly distributed; the light source control unit is responsible for controlling the light beam emitted by the light source to pass through the sample cell; the spectrometer control unit is responsible for controlling the spectrometer to perform spectral measurement, generate a spectrum graph, and use the spectrum graph to identify gas composition and gas concentration;

[0050] Data Analysis and Alarm Module: This module includes a hazardous gas combination library establishment unit and a real-time detection unit. The hazardous gas combination library establishment unit is responsible for establishing a hazardous gas combination library using a support vector machine algorithm and a random forest algorithm when the spectrum identifies several gas components. The hazardous gas combination library lists gas combinations that may cause hazardous reactions and their critical concentrations. The real-time detection unit is responsible for real-time detection of the current gas composition and concentration, comparing it with the hazardous reaction library. If a hazardous gas combination is detected and the concentration reaches a critical value, an alarm signal is generated.

[0051] Emergency response module: includes: display unit, notification unit and emergency control unit; among them, the display unit is responsible for displaying the specific dangerous gas combination, corresponding concentration information and potential hazards on the control panel when an alarm occurs. The notification unit is responsible for sounding the alarm through the sound and light alarm, and displaying emergency operation instructions on the control panel and the staff's mobile devices. The emergency control unit is responsible for automatically starting or increasing the ventilation system and automatically activating the isolation device;

[0052] Data recording and analysis module: includes: alarm cancellation unit, emergency control cancellation unit, data recording unit and random forest analysis unit; among them, the alarm cancellation unit is responsible for real-time monitoring of the gas concentration in the sample pool, and automatically cancels the alarm when the concentration of all hazardous gases has dropped below the safety threshold. The emergency control cancellation unit is responsible for linearly adjusting the ventilation system and closing the isolation device. The data recording unit is responsible for recording detailed data of dangerous events. The random forest analysis unit uses the random forest algorithm to analyze the failure of the corresponding equipment.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The present invention can collect and analyze gas samples in real time, quickly generate spectrograms, and detect gas composition and concentration in real time through machine learning algorithms, promptly discover potential dangerous gas combinations, and issue alarms.

[0055] The present invention adopts automated sampling, filtering, flow control, data processing and alarm mechanisms, which reduces manual intervention and improves the continuity and stability of monitoring.

[0056] The present invention utilizes support vector machine algorithm and random forest algorithm to establish a hazardous gas combination library, and combines it with big data technology to conduct in-depth analysis and mining of large amounts of complex data, providing accurate fault prediction and trend analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0058] Figure 1 It is a schematic diagram of the steps of a gas emission monitoring method based on spectral analysis according to the present invention;

[0059] Figure 2 It is a system structure diagram of a gas emission monitoring system based on spectral analysis according to the present invention. Detailed implementation manners

[0060] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0061] Please refer to Figure 1 and Figure 2 , the present invention provides the following technical solutions:

[0062] According to an embodiment of the present invention, as shown in the schematic diagram of the steps of a gas emission monitoring method based on spectral analysis, a gas emission monitoring method based on spectral analysis, the method includes the following steps: Figure 1 S100. Start the sampling pump, extract a gas sample from the environment to be measured, remove impurities through the sampling pipeline and filter, and transport it to the gas sample cell. The gas flow rate in the sample cell is adjusted by the flow controller; it flows through the inlet and outlet at both ends of the sample cell, and the gas in the sample cell is evenly distributed. The beam emitted by the light source passes through the sample cell, and spectral measurement is performed by the spectrometer to generate a spectrogram, and the gas components and gas concentrations are identified using the spectrogram;

[0063] S200. When several gas components are identified in the spectrogram, use the support vector machine algorithm and the random forest algorithm to establish a hazardous gas combination library, and the hazardous gas combination library lists the gas combinations that may undergo hazardous reactions and their critical concentrations; continuously detect the gas components and concentrations in the current environment, and compare them with the hazardous reaction library. If the detected gas combination meets the hazardous combination and the concentration reaches or exceeds the critical value, an alarm signal is generated;

[0064] S300. When the alarm is triggered, display the specific hazardous gas combination, the corresponding concentration information and potential hazards on the control panel, issue an alarm through the sound and light alarm, display the emergency operation guide, automatically start or increase the ventilation system, and automatically start the isolation device;

[0065] S300. When the alarm is triggered, display the specific hazardous gas combination, the corresponding concentration information and potential hazards on the control panel, issue an alarm through the sound and light alarm, display the emergency operation guide, automatically start or increase the ventilation system, and automatically start the isolation device;

[0066] S400 monitors the gas concentration in the sample pool in real time. When the concentration of all hazardous gases has dropped below the safety threshold, it automatically lifts the alarm, linearly adjusts the operation of the ventilation system, closes the isolation device, records detailed data of the hazardous event, and uses the random forest algorithm to analyze the failure of the corresponding equipment.

[0067] According to step S100, a flow controller is installed on the air inlet pipe of the gas sample pool or on the pipe near the sampling pump. The flow controller is a mass flow controller. Before starting the system, the flow controller is calibrated to set a target flow range. The flow range is determined based on the characteristics of the gas to be tested and the sample pool. In this embodiment, the flow range is 5 L / min.

[0068] After starting the sampling pump, the flow controller begins working and monitors the flow of the gas sample in real time. The sensor inside the controller detects the flow signal and compares it with the set value. If the actual flow deviates from the set value, the controller automatically adjusts the valve opening or the pump speed to ensure that the flow is stable within the target range. The flow controller is equipped with a closed-loop control system to adjust the flow through real-time feedback. The monitoring interface displays the current flow value, and the operator can monitor and adjust it manually in real time.

[0069] According to step S100, a xenon lamp is selected as the light source. The spectrum of the light source covers the absorption wavelength range of the target gas. The light source is introduced into the optical path system through a collimating lens and an optical fiber to form a parallel light beam. The gas sample cell is designed to be made of quartz glass. The light beam enters from one end of the sample cell, passes through the gas sample, and is emitted from the other end. Gas molecules have an absorption effect on light of specific wavelengths. As the light beam passes through the sample cell, its intensity and wavelength change.

[0070] The light beam entering the spectrometer is decomposed into spectra of different wavelengths by a grating or prism. The decomposed spectra are then projected onto a detector, which generates electrical signals based on the intensity of light at different wavelengths. The detector then converts these electrical signals into digital signals. The light intensity data at different wavelengths are plotted into a spectrum, which shows the absorption of light of each wavelength by the gas sample. The gas composition and concentration are identified by the position and intensity of the absorption peak.

[0071] According to step S200, spectral data of historical gas samples are collected, each sample including several gas components and their concentrations, and each sample is labeled as to whether a dangerous reaction will occur, with the labeled sample label being 1 for dangerous and -1 for safe; features are extracted from the spectral data, including absorption peak position, absorption intensity, peak area, half-peak width, baseline shift, and peak shape;

[0072] Part of the training data set is as follows: sample ID: 1, absorption peak position: 450, absorption intensity 0.8, peak area: 200, half-peak width: 50, baseline shift: 0.1, peak shape: 1.2, label: 1;

[0073] Sample ID: 2, Absorption peak position: 500, Absorption intensity: 0.6, Peak area: 180, Full width at half maximum: 45, Baseline offset: 0.15, Peak shape: 1.1, Label: -1;

[0074] Support vector machines are used to identify combinations of hazardous gases. The goal is to find a hyperplane that separates samples of different classes, which include hazardous and safe; the spectral data feature vectors and their corresponding labels are combined to form a first training dataset, and the first training dataset is used for training. The objective function is defined as follows:

[0075] ;

[0076] where w is the weight vector representing the direction of the hyperplane, and b is the bias term representing the displacement of the hyperplane;

[0077] For each training sample, the constraint conditions to be satisfied are defined as follows:

[0078] ;

[0079] where, is the sample label of the i-th sample, is the feature vector extracted from the spectral data of the i-th sample.

[0080] The Lagrangian function is constructed using Lagrange multipliers to solve the dual problem:

[0081] ;

[0082] where, are Lagrange multipliers used for the dual solution of the optimization problem, represents the slack variable of the constraint condition, and n is the number of training samples;

[0083] The Lagrangian function is solved, the derivatives of w and b are calculated and set to zero to obtain the dual problem:

[0084] ;

[0085] The constraint conditions of the dual problem are:

[0086] ;

[0087] The gradient descent algorithm is used to solve the dual problem to obtain the optimal ; Using the optimal the weight vector w and the bias term b are calculated as follows:

[0088] ;

[0089] Among them, is the label of the support vector, and is the feature vector of the support vector;

[0090] Finally, we get ;

[0091] The classification decision function of the trained support vector machine model is:

[0092] ;

[0093] Among them, to correspond to the absorption peak position, absorption intensity, peak area, full width at half maximum, baseline shift, and peak shape respectively, and is the classification decision function used to determine whether the gas combination is dangerous.

[0094] According to step S200, based on the hazardous gas combinations obtained through the support vector machine algorithm, collect the spectral data of the gas samples corresponding to the hazardous gas combinations; form the second training dataset with the spectral data feature vectors of the gas samples of the hazardous gas combinations and the critical concentrations of the hazardous gas combinations, randomly sample several subsets from the second training dataset, and each subset is used to train a decision tree; perform decision tree training on each subset to establish a regression model; input the feature vector of the new sample into each decision tree for prediction, and take the average of all the decision tree prediction results as the final critical concentration prediction value.

[0095] According to step S200, for each decision tree , use the training data subset for training, where is the spectral data feature vector of the gas sample of the hazardous gas combination, is the critical concentration of the hazardous gas combination; for the feature vector of the hazardous gas combination, the prediction result of each decision tree is , indicating the prediction result of the i-th decision tree for , and output the critical concentration of the hazardous gas combination;

[0096] The random forest gives the final prediction value by combining the prediction results of all decision trees :

[0097] ;

[0098] Among them, is the number of decision trees, and in this embodiment ;

[0099] Combine the critical concentrations of the hazardous gas combinations obtained by the support vector machine algorithm and the hazardous gas combinations obtained by the decision tree algorithm to establish a hazardous gas combination library.

[0100] The spectral feature vector of the new sample is , and substitute it into the SVM model for classification:

[0101] ;

[0102] Since , this sample is classified as dangerous. Then, use the random forest model to predict the critical concentration of this sample, and the prediction results of each decision tree are:

[0103] ;

[0104] The final predicted value is:

[0105] ;

[0106] The predicted value of the critical concentration of this sample is 10.95 units, which is recorded in the hazardous gas combination library.

[0107] According to step S300, the control panel will display the detected gas types, the concentrations of each gas, the danger level, and the description of potential hazards, and display the emergency operation guidelines on the control panel and the staff's mobile devices; the emergency operation guidelines include how to wear personal protective equipment, the specific routes and steps for evacuating personnel, and the steps for starting the ventilation system; automatically start or increase the operation of the ventilation system to reduce the concentration of hazardous gases, and automatically adjust the wind speed and air volume of the ventilation system according to the gas concentration and combination; when the gas sensor detects that the concentration of hazardous gases exceeds the standard, automatically trigger the isolation device; when a fault is found in the isolation device, an automatic isolation device fault alarm is given, and troubleshooting guidelines are provided.

[0108] On the control panel and the staff's mobile devices, the detected gas types, the concentrations of each gas, the danger level, and the description of potential hazards are displayed in real time. Specifically, it shows how to wear personal protective equipment, including what types of protective clothing, masks, gloves, etc. to use. According to the detected gas concentration and wind direction, a specific evacuation route map is displayed on the control panel, and the navigation function is used to guide personnel to evacuate safely. Detailed steps are listed, such as: shutting down the working equipment, quickly evacuating to a safe area, taking a headcount, etc. The steps for starting the ventilation system are displayed on the control panel, including the position of the manual start button, the automatic start conditions, etc.

[0109] When the gas sensor detects that the isolation device is not functioning properly, the system automatically generates an alarm and displays a fault message on the control panel. Detailed troubleshooting instructions are provided, including power supply checks, device restarts, and component replacement. The system's built-in fault diagnosis algorithm automatically analyzes the cause of the problem and provides appropriate repair recommendations.

[0110] According to step S400, the gas concentration in the sample pool is continuously monitored by sensors. When the concentration of all hazardous gases drops below a safety threshold, the alarm is automatically lifted. Based on the real-time monitored gas concentration data, the system automatically adjusts the wind speed and air volume of the ventilation system. The automatic control system shuts down previously activated isolation devices and restores the normal operation of the isolated area. The time of the hazardous event, the type of gas detected, the concentration change, and the emergency measures taken are automatically recorded.

[0111] Use the random forest algorithm for fault analysis. Input the preprocessed equipment operation data into the random forest model, randomly select feature subsets, and build several decision trees. Use the training data to build multiple decision trees. Each tree is trained independently and predicts the new data. The average prediction result of all decision trees is taken as the equipment fault analysis data and recorded in the hazardous gas combination library.

[0112] According to another embodiment of the present invention, Figure 2 As shown in the system structure diagram of a gas emission monitoring system based on spectral analysis, a gas emission monitoring system based on spectral analysis includes:

[0113] Sampling and filtration module: includes: gas sampling unit, gas filtration unit and flow control unit; the gas sampling unit is responsible for starting the sampling pump and extracting gas samples from the environment to be tested through the sampling probe; the gas filtration unit is responsible for controlling the gas sample to pass through the sampling pipeline and filter, remove impurities and then be transported to the gas sample pool; the flow control unit is responsible for adjusting the flow of the gas sample through the flow controller;

[0114] Spectral measurement module: includes: sample cell control unit, light source control unit and spectrometer control unit; among them, the sample cell control unit is responsible for controlling the flow of gas inlet and outlet at both ends of the sample cell when the gas sample enters the sample cell, so that the gas in the sample cell is evenly distributed; the light source control unit is responsible for controlling the light beam emitted by the light source to pass through the sample cell; the spectrometer control unit is responsible for controlling the spectrometer to perform spectral measurement, generate a spectrum graph, and use the spectrum graph to identify gas composition and gas concentration;

[0115] Data Analysis and Alarm Module: This module includes a hazardous gas combination library establishment unit and a real-time detection unit. The hazardous gas combination library establishment unit is responsible for establishing a hazardous gas combination library using a support vector machine algorithm and a random forest algorithm when the spectrum identifies several gas components. The hazardous gas combination library lists gas combinations that may cause hazardous reactions and their critical concentrations. The real-time detection unit is responsible for real-time detection of the current gas composition and concentration, comparing it with the hazardous reaction library. If a hazardous gas combination is detected and the concentration reaches a critical value, an alarm signal is generated.

[0116] Emergency response module: includes: display unit, notification unit and emergency control unit; among them, the display unit is responsible for displaying the specific dangerous gas combination, corresponding concentration information and potential hazards on the control panel when an alarm occurs. The notification unit is responsible for sounding the alarm through the sound and light alarm, and displaying emergency operation instructions on the control panel and the staff's mobile devices. The emergency control unit is responsible for automatically starting or increasing the ventilation system and automatically activating the isolation device;

[0117] Data recording and analysis module: includes: alarm cancellation unit, emergency control cancellation unit, data recording unit and random forest analysis unit; among them, the alarm cancellation unit is responsible for real-time monitoring of the gas concentration in the sample pool, and automatically cancels the alarm when the concentration of all hazardous gases has dropped below the safety threshold. The emergency control cancellation unit is responsible for linearly adjusting the ventilation system and closing the isolation device. The data recording unit is responsible for recording detailed data of dangerous events. The random forest analysis unit uses the random forest algorithm to analyze the failure of the corresponding equipment.

[0118] We conducted multiple experiments in a simulated factory environment. The environment included a variety of common industrial gases, simulating various hazardous gas combinations and concentrations exceeding specified limits. The experimental equipment included a spectrometer, gas sensors, sampling pumps, flow controllers, ventilation systems, isolation devices, control panels, and mobile equipment.

[0119] Gas samples were collected at different time points and conditions, and spectra were generated by a spectrometer. The spectral data characteristics of each sample were recorded, such as absorption peak position, absorption intensity, peak area, half-peak width, baseline offset and peak shape. Historical data were used to train support vector machines and random forest models to establish a library of hazardous gas combinations. SVM is used to identify hazardous gas combinations, and RF is used to predict the critical concentration of hazardous gas combinations. The gas concentration in the sample pool is monitored in real time, and the spectrometer continuously collects data and transmits it to the control system. The trained SVM model is used for real-time classification to determine whether a hazardous gas combination exists. When a hazardous gas combination is detected and the concentration reaches a critical value, the emergency response module is triggered.

[0120] The display unit shows detailed information on the control panel and the mobile device, including the hazardous gas combination, concentration information, hazard level, and potential hazards. The notification unit issues an alarm through an audible and visual alarm and displays emergency operation guidelines. The emergency control unit automatically starts or increases the ventilation system, adjusts the wind speed and air volume, and activates the isolation device. It monitors and adjusts in real time until the gas concentration drops below the safety threshold. It records the detailed data of the hazardous event and uses the random forest algorithm for fault analysis to identify potential fault sources.

[0121] In this embodiment, a hazardous combination of carbon monoxide and chlorine is detected, with concentrations of 35 ppm and 10 ppm respectively. The control panel and the mobile device display the detailed hazardous gas information and potential hazards. The audible and visual alarm immediately issues an alarm. The emergency operation guidelines recommend wearing a gas mask and isolation suit, and an evacuation route map is displayed on the control panel. The ventilation system is automatically started, the wind speed is increased to 20 m / s, and the air volume is adjusted to the maximum. The isolation device is automatically activated to seal off the contaminated area. The gas concentration is successfully reduced to the safe range, and the event lasts for 10 minutes.

[0122] A hazardous combination of ammonia and sulfur dioxide is detected, with concentrations of 25 ppm and 5 ppm respectively. The detailed information is displayed and an alarm is issued. The ventilation system is started normally, but the isolation device fails to start. The system automatically alarms and displays a fault message on the control panel. Troubleshooting guidelines are provided to guide the inspection of the power supply and the restart of the device. Through manual intervention and system guidelines, the fault is successfully eliminated and the function of the isolation device is restored. The event lasts for 15 minutes.

[0123] The preprocessed device operation data is input into the random forest model, including features such as wind speed, air volume, and gas concentration changes. Decision trees are trained using historical fault data to establish a fault prediction model. In the new experimental data, the random forest model successfully predicts an overload fault that may occur in the ventilation system, and preventive measures are taken in advance. Results: Prediction accuracy: 89%, Fault warning lead time: 5 minutes, Reduction in equipment failure rate: 30%.

[0124] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0125] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A gas emission monitoring method based on spectral analysis, characterized in that, The method includes the following steps: S100. Start the sampling pump to extract a gas sample from the environment to be measured. Remove impurities through the sampling pipeline and filter, and transport it to the gas sample cell. The gas flow rate in the sample cell is adjusted by a flow controller; it flows through the inlet and outlet at both ends of the sample cell, and the gas in the sample cell is evenly distributed. The light beam emitted by the light source passes through the sample cell, and spectroscopic measurement is performed by a spectrometer to generate a spectrogram, and the gas components and gas concentrations are identified using the spectrogram; According to step S100, a xenon lamp is selected as the light source. The spectrum of the light source covers the absorption wavelength range of the target gas. The light source is introduced into the optical path system through a collimating lens and an optical fiber to form a parallel light beam. The design material of the gas sample cell is quartz glass; the light beam enters from one end of the sample cell, passes through the gas sample, and then exits from the other end. Gas molecules have an absorption effect on light of a specific wavelength. During the process of the light beam passing through the sample cell, the intensity and wavelength of the light beam change; The light beam entering the spectrometer is decomposed into spectra of different wavelengths by a grating or a prism. The decomposed spectra are projected onto a detector. The detector generates electrical signals according to the light intensities of different wavelengths, and the detector converts these electrical signals into digital signals; the light intensity data of different wavelengths are plotted into a spectrogram. The absorption of the gas sample for light of each wavelength is shown on the spectrogram, and the gas components and gas concentrations are identified by the position and intensity of the absorption peak; S200. When several gas components are identified in the spectrogram, use the support vector machine algorithm and the random forest algorithm to establish a dangerous gas combination library, and the dangerous gas combination library lists the gas combinations that may undergo dangerous reactions and their critical concentrations; continuously detect the gas components and concentrations in the current environment, compare them with the dangerous reaction library. If the detected gas combination meets the dangerous combination and the concentration reaches or exceeds the critical value, an alarm signal is generated; According to step S200, according to the dangerous gas combinations obtained by the support vector machine algorithm, collect the spectral data of the gas samples corresponding to the dangerous gas combinations; form a second training data set with the feature vectors of the spectral data of the gas samples of the dangerous gas combinations and the critical concentrations of the dangerous gas combinations. Randomly sample from the second training data set to generate several subsets, and each subset is used to train a decision tree; perform decision tree training on each subset to establish a regression model; input the feature vectors of the new samples into each decision tree for prediction, and take the average of the prediction results of all decision trees as the final critical concentration prediction value; According to step S200, for each decision tree , it is trained using the training data subset , where is the spectral data feature vector of the gas sample of the hazardous gas combination, is the critical concentration of the hazardous gas combination; for the feature vector of the hazardous gas combination, the prediction result of each decision tree is , indicating the prediction result of the i-th decision tree for , and the critical concentration of the hazardous gas combination is output; The random forest gives the final prediction value by combining the prediction results of all decision trees : ； Among them, is the number of decision trees; Combine the dangerous gas combinations obtained by the support vector machine algorithm and the critical concentrations of the dangerous gas combinations obtained by the decision tree algorithm to establish a dangerous gas combination library; S300. When the alarm is triggered, display the specific dangerous gas combination, the corresponding concentration information and the potential hazards on the control panel, issue an alarm through an audible and visual alarm, display the emergency operation guide, automatically start or increase the ventilation system, and automatically start the isolation device; S400 monitors the gas concentration in the sample pool in real time. When the concentration of all hazardous gases has dropped below the safety threshold, it automatically lifts the alarm, linearly adjusts the operation of the ventilation system, closes the isolation device, records detailed data of the hazardous event, and uses the random forest algorithm to analyze the failure of the corresponding equipment.

2. The gas emission monitoring method based on spectral analysis according to claim 1, characterized in that: According to step S100, a flow controller is installed on the air inlet pipe of the gas sample pool or on the pipe near the sampling pump. The flow controller is a mass flow controller. Before starting the system, the flow controller is calibrated to set a target flow range. The flow range is determined based on the characteristics of the gas to be tested and the sample pool. After starting the sampling pump, the flow controller begins working and monitors the flow of the gas sample in real time. The sensor inside the controller detects the flow signal and compares it with the set value. If the actual flow deviates from the set value, the controller automatically adjusts the valve opening or the pump speed to ensure that the flow is stable within the target range. The flow controller is equipped with a closed-loop control system to adjust the flow through real-time feedback. The monitoring interface displays the current flow value, and the operator can monitor and adjust it manually in real time.

3. A gas emission monitoring method based on spectral analysis according to claim 1, characterized in that: According to step S200, spectral data of historical gas samples are collected, each sample including several gas components and their concentrations, and each sample is labeled as to whether a dangerous reaction will occur, with the labeled sample label being 1 for dangerous and -1 for safe; features are extracted from the spectral data, including absorption peak position, absorption intensity, peak area, half-peak width, baseline shift, and peak shape; The support vector machine is used to identify dangerous gas combinations. Its goal is to find a hyperplane that separates samples of different categories, including dangerous and safe. The spectral data feature vectors and corresponding labels form a first training dataset, which is used for training. The objective function is defined as follows: ； Among them, is the weight vector, representing the direction of the hyperplane, is the bias term, representing the displacement of the hyperplane; For each training sample, the constraints that need to be satisfied are defined as follows: ； Among them, is the sample label of the i-th sample, is the feature vector extracted from the spectral data of the i-th sample.

4. The gas emission monitoring method based on spectral analysis according to claim 3, characterized in that: Construct the Lagrange function through Lagrange multipliers and solve the dual problem: ； wherein, is a Lagrange multiplier for the dual solution of the optimization problem, representing the slack variable of the constraint condition, and n is the number of training samples; Solve the Lagrangian function and take the derivatives with respect to and and set the derivatives to zero to obtain the dual problem: ； The constraints of the dual problem are: ； Solve the dual problem using the gradient descent algorithm to obtain the optimal ; Use the optimal to calculate the weight vector and the bias term , and the formula is as follows: ， ； Among them, is the label of the support vector, is the feature vector of the support vector; The classification decision function of the trained support vector machine model is: ； Among them, is a classification decision function used to determine whether the gas combination is dangerous.

5. A gas emission monitoring method based on spectral analysis according to claim 1, characterized in that: According to step S300, the control panel will display the type of gas detected, the concentration of each gas, the hazard level and the potential hazard description, and display emergency operation instructions on the control panel and the staff's mobile device; the emergency operation instructions include how to wear personal protective equipment, the specific routes and steps for evacuating personnel and the steps for starting the ventilation system; automatically start or increase the operation of the ventilation system to reduce the concentration of dangerous gases, and automatically adjust the wind speed and air volume of the ventilation system according to the gas concentration and combination; when the gas sensor detects that the concentration of dangerous gases exceeds the standard, it automatically triggers the isolation device; when a fault is found in the isolation device, an automatic isolation device fault alarm is issued, and troubleshooting instructions are provided.

6. The gas emission monitoring method based on spectral analysis according to claim 1, characterized in that: According to step S400, the gas concentration in the sample pool is continuously monitored by sensors. When the concentration of all hazardous gases drops below a safety threshold, the alarm is automatically lifted. Based on the real-time monitored gas concentration data, the system automatically adjusts the wind speed and air volume of the ventilation system. The automatic control system shuts down previously activated isolation devices and restores the normal operation of the isolated area. The time of the hazardous event, the type of gas detected, the concentration change, and the emergency measures taken are automatically recorded. Use the random forest algorithm for fault analysis. Input the preprocessed equipment operation data into the random forest model, randomly select feature subsets, and build several decision trees. Use the training data to build multiple decision trees. Each tree is trained independently and predicts the new data. The average prediction result of all decision trees is taken as the equipment fault analysis data and recorded in the hazardous gas combination library.

7. A gas emission monitoring system based on spectral analysis, using a gas emission monitoring method based on spectral analysis according to any one of claims 1-6, characterized in that, include: Sampling and filtration module: includes: gas sampling unit, gas filtration unit and flow control unit; the gas sampling unit is responsible for starting the sampling pump and extracting gas samples from the environment to be tested through the sampling probe; the gas filtration unit is responsible for controlling the gas sample to pass through the sampling pipeline and filter, remove impurities and then be transported to the gas sample pool; the flow control unit is responsible for adjusting the flow of the gas sample through the flow controller; Spectral measurement module: includes: sample cell control unit, light source control unit and spectrometer control unit; among them, the sample cell control unit is responsible for controlling the flow of gas inlet and outlet at both ends of the sample cell when the gas sample enters the sample cell, so that the gas in the sample cell is evenly distributed; the light source control unit is responsible for controlling the light beam emitted by the light source to pass through the sample cell; the spectrometer control unit is responsible for controlling the spectrometer to perform spectral measurement, generate a spectrum graph, and use the spectrum graph to identify gas composition and gas concentration; Data Analysis and Alarm Module: This module includes a hazardous gas combination library establishment unit and a real-time detection unit. The hazardous gas combination library establishment unit is responsible for establishing a hazardous gas combination library using a support vector machine algorithm and a random forest algorithm when the spectrum identifies several gas components. The hazardous gas combination library lists gas combinations that may cause hazardous reactions and their critical concentrations. The real-time detection unit is responsible for real-time detection of the current gas composition and concentration, comparing it with the hazardous reaction library. If a hazardous gas combination is detected and the concentration reaches a critical value, an alarm signal is generated. Emergency response module: includes: display unit, notification unit and emergency control unit; among them, the display unit is responsible for displaying the specific dangerous gas combination, corresponding concentration information and potential hazards on the control panel when an alarm occurs. The notification unit is responsible for sounding the alarm through the sound and light alarm, and displaying emergency operation instructions on the control panel and the staff's mobile devices. The emergency control unit is responsible for automatically starting or increasing the ventilation system and automatically activating the isolation device; Data recording and analysis module: includes: alarm cancellation unit, emergency control cancellation unit, data recording unit and random forest analysis unit; among them, the alarm cancellation unit is responsible for real-time monitoring of the gas concentration in the sample pool, and automatically cancels the alarm when the concentration of all hazardous gases has dropped below the safety threshold. The emergency control cancellation unit is responsible for linearly adjusting the ventilation system and closing the isolation device. The data recording unit is responsible for recording detailed data of dangerous events. The random forest analysis unit uses the random forest algorithm to analyze the failure of the corresponding equipment.

Images