Dangerous source automatic identification and risk control method in chemical production process

Abstract

The invention provides a dangerous source automatic identification and risk control method in a chemical production process, and relates to the technical field of dangerous source automatic identification and risk control. Influence factors, such as harmful gas concentration, environment temperature and fluid flow, in dangerous equipment to be monitored in the chemical production process are recorded in a specific time period; secondly, extracting frequency domain features and statistical features from the preprocessed data, including a mean value and a standard deviation of each influence factor, and calculating the periodic change frequency of the factors and the influence of the factors on the internal pressure of the equipment; finally, according to the difference value between the calculated internal pressure data of the equipment and a set pressure threshold value, the risk state of the current dangerous equipment is judged, the risk level is output, a corresponding risk control scheme is formulated accordingly, and through the method, potential risks in the chemical production process can be effectively recognized and controlled.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic identification of hazardous sources and risk control, and in particular to a method for automatic identification of hazardous sources and risk control in a chemical production process. Background Art

[0002] In the chemical production process, internal pressure is one of the main sources of danger. Abnormal changes in the internal pressure of equipment may lead to serious consequences such as equipment failure, leakage or even explosion, which not only poses a threat to production safety, but also may cause immeasurable damage to personnel and the environment. Therefore, real-time monitoring and prediction of the internal pressure of dangerous equipment is crucial to ensure the safety and stability of chemical production.

[0003] However, the equipment operating environment is often complex and harsh, especially in chemical plants, where there may be high temperature, high pressure, corrosive gas and other conditions. This environment has a significant impact on the pressure sensor, which may cause damage or failure of the sensor, thereby affecting the accuracy and real-time performance of the pressure data. Therefore, there is a certain risk in relying on a single pressure sensor for monitoring.

[0004] Although there has been a certain amount of technical accumulation in the monitoring and prediction of internal pressure in hazardous equipment, there are still the following shortcomings: Many existing systems rely on a single type of pressure sensor for real-time monitoring. Once the sensor fails or the data is distorted, the effectiveness of the entire monitoring system will be seriously affected, resulting in missing information or wrong predictions, increasing safety risks. Existing technologies often only focus on pressure data, but do not fully consider the impact of other environmental factors such as temperature, flow, gas concentration, etc. on pressure. Ignoring the fusion of these multi-dimensional data may lead to misjudgment of the actual operating status of the equipment. Current systems usually use simple statistical analysis in data processing, and fail to fully utilize advanced data mining and machine learning technologies for deep feature extraction.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a method for automatically identifying and controlling hazards in a chemical production process, so as to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] The method for automatic identification and risk control of hazardous sources in the chemical production process includes the following specific steps:

[0009] Step 1: During the monitoring period of the chemical production process, record the influencing factors inside the dangerous equipment to be monitored, including harmful gas concentration, ambient temperature, and fluid flow data, and pre-process the recorded data;

[0010] Step 2: Extract characteristic data from the preprocessed data, and calculate the internal pressure of the dangerous equipment to be monitored based on the characteristic data, wherein the characteristic data includes frequency domain characteristics and statistical characteristics, wherein the statistical characteristics include mean and standard deviation data corresponding to various influencing factors, and the frequency domain characteristics include the main frequencies of various influencing factors;

[0011] Step 3: Based on the difference between the calculated internal pressure of the hazardous equipment to be monitored and the set equipment pressure threshold, determine the risk status inside the hazardous equipment to be monitored, output the risk level, and formulate a risk control plan for the hazardous equipment based on the risk level.

[0012] Furthermore, the monitoring time period is the time period before the current moment, and the duration of the monitoring time period is T. During the monitoring time period of the chemical production process, the harmful gas concentration, ambient temperature, and fluid flow data inside the monitored dangerous equipment at each time point are recorded in equal time intervals to obtain the following data sequence:

[0013] {C g (t i ),C t (t i ),C u (t i )|i=1,2,…,N}

[0014] Among them, C g (t i ) represents the time point t i The concentration of harmful gases recorded at t (t i ) represents the time point t i Ambient temperature recorded at the time, C u (t i ) represents the time point t i The flow rate data of the fluid recorded at t i represents the i-th time point in the monitoring period, i is the index of the time point, and i∈[1,N], N is the number of time points in the monitoring period;

[0015] The logic of the data preprocessing is: use the sliding average method to smooth the recorded data, then use the quartile method to eliminate outliers in the smoothed data, and finally use the mean filling method to supplement the missing values ​​of the data.

[0016] Furthermore, the statistical features are extracted to calculate the mean and standard deviation of harmful gas concentration, ambient temperature, and fluid flow data. Taking harmful gas concentration as an example, the formula is:

[0017]

[0018] Among them, μ g Represents the mean concentration of harmful gases after pretreatment, C' g (t i ) indicates the concentration of harmful gases C g (t i ) after preprocessing;

[0019]

[0020] Among them, σ g It represents the standard deviation of the concentration of harmful gases after pretreatment, that is, the degree of fluctuation of the concentration of harmful gases. The higher the standard deviation, the greater the degree of fluctuation of the concentration of harmful gases. Similarly, the ambient temperature and flow data are processed to obtain their mean values ​​μ t ,μ u and standard deviation σ t ,σ u ;

[0021] The logic for extracting the main frequencies of the harmful gas concentrations is as follows:

[0022] Fast Fourier transform is used to convert the time-series distribution of harmful gas concentration into frequency domain signal, and the formula is as follows:

[0023]

[0024] Among them, F g (k) is the kth frequency component in the frequency domain signal of the harmful gas concentration, expressed in the form of a complex number, k is the index of the frequency component in the frequency domain signal of the harmful gas concentration, and k∈[1,N], j is an imaginary unit;

[0025] The frequency resolution is calculated based on the following formula:

[0026]

[0027] Where Δf is the frequency resolution;

[0028] Calculate the amplitude corresponding to each frequency component, and determine the index of the frequency component corresponding to the maximum amplitude. This index is used as the main frequency index of the harmful gas concentration. The formula is as follows:

[0029] A g (k)=|F g (k)|

[0030]

[0031] Among them, A g (k) is the amplitude corresponding to the kth frequency component in the frequency domain signal of harmful gas concentration, k g is the main frequency index of harmful gas concentration, |F g (k)| indicates F g (k) taking the absolute value;

[0032] Based on the frequency resolution and the main frequency index of the harmful gas concentration, the main frequency of the harmful gas concentration is determined. The calculation formula is as follows:

[0033] f g =k g Δf

[0034] Among them, f g is the main frequency of the harmful gas concentration, and similarly the main frequency of the ambient temperature is obtained t , the main frequency f of the fluid flow u .

[0035] Furthermore, a pressure prediction model is established based on a deep neural network, and the LSTM network architecture is used to process time series features. One to two layers of LSTM units are set, and each layer is set to include 50 to 100 LSTM units; the hidden layer uses ReLU as the activation function, and the output layer uses a linear activation function;

[0036] Obtain the influencing factor data and the corresponding internal pressure of the dangerous equipment to be monitored in multiple historical monitoring time periods in a time series distribution, use the same method as step 1-2 to extract the features of the influencing factor data in the historical monitoring time period, obtain historical feature data, and use the historical feature data as a training set, the historical feature data as input, and the corresponding internal pressure of the dangerous equipment to be monitored as an output label to train the pressure prediction model. The model expression of the pressure prediction model is as follows:

[0037]

[0038] Among them, α, β, γ, δ, ζ are unknown parameters in the model expression, ε is an unknown constant to prevent the denominator from being zero, and ε>0;

[0039] Use the Batch training method to divide the entire training set into multiple small batches. In each training iteration, pass the input data into the LSTM model, calculate its output prediction value, use the MSE loss function to calculate the difference between the model's prediction value and the true value, and use the back-propagation algorithm to calculate the gradient of the loss function relative to the model parameters to update the model parameters.

[0040] During the training process, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of trainings, the number of batches, and the number of neurons in the hidden layer; wherein the number of network layers is set to a four-layer network structure, the number of iterations is set to 200, the learning rate is set to 0.001, the batch size is set to 64, the number of trainings is set to 500, the batch size is set to 256, and the number of neurons in the hidden layer is 32;

[0041] Gradually optimize the unknown parameters and unknown constants, and use RMSE to evaluate the model performance until the pressure prediction model training is completed. Input the feature data obtained in step 2 into the trained pressure prediction model to obtain the corresponding internal pressure of the hazardous equipment to be monitored.

[0042] Furthermore, the difference between the pressure data inside the dangerous equipment to be monitored and the set equipment pressure threshold is based on the formula:

[0043] ΔP=PP th

[0044] Where P represents the internal pressure of the dangerous equipment to be monitored, P th Indicates the set equipment pressure threshold, which is the maximum pressure limit for safe operation of the equipment. ΔP indicates the difference between the current pressure and the threshold, reflecting the degree to which the pressure deviates from the threshold.

[0045] According to the pressure difference ΔP: if ΔP>0, it means that the current pressure exceeds the safety threshold and the device is at risk; if ΔP≤0, it means that the device is within the safety range and there is no risk;

[0046] The risk levels can be divided into high risk, medium risk and low risk, and are classified by setting specific thresholds:

[0047]

[0048] Among them, ΔP high The pressure differential threshold, indicating a defined high-risk condition, is set at 5% or 10% of the device pressure threshold, ΔP medium The pressure difference threshold representing a defined medium risk condition is set at 1% or 3% of the equipment pressure threshold;

[0049] By analyzing the pressure data of the dangerous equipment, the risk status is automatically identified and corresponding control measures are formulated:

[0050] High Risk means that the equipment pressure exceeds the safety range, which may cause equipment damage or accidents. Take measures immediately: stop the equipment, conduct a comprehensive inspection, activate the equipment alarm system, notify relevant personnel, and open the safety valve to reduce the internal pressure;

[0051] Medium Risk means that the equipment pressure is close to the upper limit of the safety range. Close monitoring is required and preventive measures are taken: increase the frequency of pressure monitoring, observe pressure changes in real time, reduce the load on the equipment to reduce the risk of pressure increase, and arrange technicians to inspect and evaluate the equipment;

[0052] Low Risk means the equipment is operating normally within a safe range: maintain normal monitoring frequency to ensure that the equipment is within a safe range, continue to record pressure data, analyze pressure fluctuation trends to provide a basis for subsequent preventive measures, perform regular equipment maintenance to ensure normal equipment operation and prevent potential risks.

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

[0054] The present invention not only considers pressure data, but also comprehensively considers multiple influencing factors such as temperature, flow rate, and harmful gas concentration. Through multi-dimensional data fusion, it provides a more comprehensive and accurate equipment status assessment, improves the reliability of prediction, and adopts advanced data processing and feature extraction methods, such as the combination of statistical features and frequency domain features, so that the model can capture potential dynamic changes and key influencing factors, significantly improving the prediction accuracy.

[0055] The present invention integrates an intelligent risk assessment algorithm and establishes a hazard source identification model by performing deep learning analysis on the processed data. The model can not only identify typical hazard signals, but also has the ability to learn and adapt to new risk patterns. This provides a scientific basis for safety management in the chemical production process, realizes real-time monitoring and dynamic prediction of the internal pressure of the equipment, and can quickly respond to changes in the environment and operating conditions, ensuring that early warnings are issued in time when the pressure approaches the safety threshold, thereby reducing safety risks.

[0056] The present invention reduces dependence on human intervention and the risk of human operational errors through automated data collection and analysis, thereby improving the overall safety and reliability of the system, providing real-time data analysis and prediction results, and providing powerful decision-making support for equipment managers, which helps to formulate more scientific maintenance and operation strategies, thereby reducing downtime and maintenance costs. The early warning mechanism can issue an alarm before potential dangerous conditions occur, allowing the maintenance team to take quick measures to troubleshoot and handle faults, thereby reducing the risk of equipment damage and personal injury. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION

[0058] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0059] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0060] Example:

[0061] See also Figure 1 , the present invention provides a technical solution:

[0062] The method for automatic identification and risk control of hazardous sources in the chemical production process includes the following specific steps:

[0063] Step 1: During the monitoring period of the chemical production process, record the influencing factors inside the dangerous equipment to be monitored, including harmful gas concentration, ambient temperature, and fluid flow data, and pre-process the recorded data;

[0064] The monitoring time period is the time period before the current moment, and the duration of the monitoring time period is T. During the monitoring time period of the chemical production process, the harmful gas concentration, ambient temperature, and fluid flow data inside the monitored dangerous equipment at each time point are recorded in equal time intervals to obtain the following data sequence:

[0065] {C g (t i ),C t (t i ),C u (t i )|i=1,2,…,N}

[0066] Among them, C g (t i ) represents the time point t i The concentration of harmful gases recorded at t (t i ) represents the time point t iAmbient temperature recorded at time, °C u (t i ) represents the flow rate data of the fluid recorded at time point t i where t i represents the i-th time point within the monitoring time period, i is the index of the time point, and i ∈ [1, N], where N is the number of time points within the monitoring time period;

[0067] When performing short-term monitoring of the equipment: T ≤ 1 h, applicable to rapidly changing processes, or during equipment startup, shutdown, or failure;

[0068] When performing medium-term monitoring of the equipment: 1 h < T ≤ 12 h, applicable to general production processes, capturing fluctuations in daily operations;

[0069] When performing long-term monitoring of the equipment: T > 12 h, applicable to relatively stable production processes, or when observing long-term trends;

[0070] These data sequences are used to describe the changes in the internal environment of hazardous equipment during the chemical production process, providing the basis for real-time monitoring data. These data points can be used to analyze the trends of different variables (concentration of harmful gases, ambient temperature, fluid flow rate) over time and identify potential risk factors;

[0071] It should be noted that harmful gases include ammonia, formaldehyde, benzene, carbon monoxide, and hydrogen sulfide, etc. The concentration of harmful gases is the cumulative value of the concentrations of the above-mentioned harmful gases, comprehensively reflecting the concentration information of harmful gases inside the hazardous equipment to be monitored. The fluid flow rate is the maximum value between the average gas flow rate and the average liquid flow rate inside the hazardous equipment to be monitored. Since the maximum value can better reflect the pressure situation inside the hazardous equipment to be monitored, the maximum value between the average gas flow rate and the average liquid flow rate inside the hazardous equipment to be monitored is used as the fluid flow rate, which simplifies data input and reduces the calculation difficulty while ensuring subsequent pressure prediction;

[0072] The logic of the data preprocessing is as follows: use the moving average method to smooth the recorded data, then use the quartile method to剔除 outliers in the smoothed data, and finally use the mean filling method to supplement missing values in the data. The above-mentioned moving average method, quartile method, and mean filling method are all conventional technical means for data processing by those skilled in the art and will not be elaborated here;

[0073] It should be noted that if the window is too small, high-frequency noise may not be smoothed, and if the window is too large, important fluctuations may be masked. It is generally recommended that the size of the sliding window be 1 / 10 to 1 / 5 of the monitoring time period T as a preliminary reference; for example, if T = 2 h, the sliding window can be set to 12 to 24 minutes.

[0074] Step 2: Extract characteristic data from the preprocessed data, and calculate the internal pressure of the dangerous equipment to be monitored based on the characteristic data, wherein the characteristic data includes frequency domain characteristics and statistical characteristics, wherein the statistical characteristics include mean and standard deviation data corresponding to various influencing factors, and the frequency domain characteristics include the main frequencies of various influencing factors;

[0075] The statistical features are extracted to calculate the mean and standard deviation of harmful gas concentration, ambient temperature, and fluid flow data. Taking harmful gas concentration as an example, the formula is:

[0076]

[0077] Among them, μ g Represents the mean concentration of harmful gases after pretreatment, C' g (t i ) indicates the concentration of harmful gases C g (t i ) after preprocessing;

[0078] μ g The larger the value, the higher the gas concentration is, which may indicate a potential safety risk. g (t i ) is at time t i The recorded smoothed gas concentration value, the sample size N is the number of recorded points in the time period;

[0079]

[0080] Among them, σ g It represents the standard deviation of the concentration of harmful gases after pretreatment, that is, the degree of fluctuation of the concentration of harmful gases. The higher the standard deviation, the greater the degree of fluctuation of the concentration of harmful gases. Similarly, the ambient temperature and flow data are processed to obtain their mean values ​​μ t ,μ u and standard deviation σ t ,σ u ;

[0081] σ g The larger the value, the greater the range of gas concentration changes and the greater the volatility, which may indicate the instability of the equipment or environment. g (t i ) has a large dispersion, (C' g (t i )-μ g ) 2 The value of increases, resulting in σ g Increase, reflecting the instability of concentration;

[0082] The logic for extracting the main frequencies of the harmful gas concentrations is as follows:

[0083] Fast Fourier transform is used to convert the time-series distribution of harmful gas concentration into frequency domain signal, and the formula is as follows:

[0084]

[0085] Among them, F g (k) is the kth frequency component in the frequency domain signal of the harmful gas concentration, expressed in the form of a complex number, k is the index of the frequency component in the frequency domain signal of the harmful gas concentration, and k∈[1,N], j is an imaginary unit;

[0086] The frequency components in complex form can simultaneously reflect the intensity (amplitude) and phase (time delay) of the signal at a certain frequency. In this way, the Fourier transform converts the original time domain signal (the change of harmful gas concentration over time) into the frequency domain, revealing the frequency components of the signal;

[0087] The frequency resolution is calculated based on the following formula:

[0088]

[0089] Where Δf is the frequency resolution, which represents the minimum resolvable difference between different frequency components in the frequency domain;

[0090] The smaller the frequency resolution, the more accurately adjacent frequency components can be distinguished. If T is larger (the monitoring period is longer), the frequency resolution is smaller, which can better identify low-frequency signals. The larger the frequency resolution, that is, the smaller T, the larger the interval between frequency components, which may make it impossible to distinguish some closely spaced frequency components.

[0091] Calculate the amplitude corresponding to each frequency component, and determine the index of the frequency component corresponding to the maximum amplitude. This index is used as the main frequency index of the harmful gas concentration. The formula is as follows:

[0092] A g (k)=|F g (k)|

[0093]

[0094] Among them, A g (k) is the amplitude corresponding to the kth frequency component in the frequency domain signal of harmful gas concentration, k g is the main frequency index of harmful gas concentration, |F g (k)| indicates F g (k) taking the absolute value;

[0095] The amplitude is the intensity of the signal at this frequency component, indicating the amplitude of the change in the concentration of harmful gases at this frequency. g (k) is large, indicating that the frequency component contributes more to the signal, which means that the concentration of harmful gases changes significantly at this frequency; the amplitude is related to the intensity of the original signal, and usually a large amplitude means that the signal energy or intensity is high at this frequency point, reflecting the corresponding physical phenomenon, such as periodic changes;

[0096] k g is the index of the frequency component with the largest amplitude in the frequency domain signal of harmful gas concentration. If A g (k) is maximum at a certain k value, indicating that this frequency is the most important component in the signal and may correspond to a specific physical process or periodic change;

[0097] Based on the frequency resolution and the main frequency index of the harmful gas concentration, the main frequency of the harmful gas concentration is determined. The calculation formula is as follows:

[0098] f g =k g Δf

[0099] Among them, f g is the main frequency of the harmful gas concentration, indicating the frequency component k g The corresponding actual frequency value, similarly, the main frequency f of the ambient temperature is obtained t , the main frequency f of the fluid flow u ;

[0100] Main frequency f g Indicates the main cycle of harmful gas concentration changes, which can help identify the law or events of gas concentration changes (such as equipment failure, environmental impact, etc.). g It means that the concentration of harmful gases changes more frequently, reflecting a more active reaction process or influencing factor. The relationship between the main frequency and the amplitude can reflect the interaction between the intensity and frequency characteristics of the signal. Usually, the main frequency corresponding to the frequency component with greater intensity is easier to identify.

[0101] Step 3: According to the difference between the calculated internal pressure of the dangerous equipment to be monitored and the set equipment pressure threshold, the risk status inside the dangerous equipment to be monitored is judged, the risk level is output, and a risk control plan for the dangerous equipment is formulated according to the risk level;

[0102] A pressure prediction model is established based on a deep neural network. The LSTM network architecture is used to process time series features. One or two layers of LSTM units are set, and each layer is set to include 50 to 100 LSTM units. The hidden layer usually uses ReLU as the activation function, while the output layer can use a linear activation function to ensure the continuity of the output pressure value.

[0103] Obtain the influencing factor data and the corresponding internal pressure of the dangerous equipment to be monitored in multiple historical monitoring time periods in a time series distribution, use the same method as step 1-2 to extract the features of the influencing factor data in the historical monitoring time period, obtain historical feature data, and use the historical feature data as a training set, the historical feature data as input, and the corresponding internal pressure of the dangerous equipment to be monitored as an output label to train the pressure prediction model. The model expression of the pressure prediction model is as follows:

[0104]

[0105] Among them, α, β, γ, δ, ζ are unknown parameters in the model expression, ∈ is an unknown constant to prevent the denominator from being zero, and ∈>0;

[0106] Use the Batch training method to divide the entire training set into multiple small batches. In each training iteration, pass the input data into the LSTM model, calculate its output prediction value, use the MSE (mean square error) loss function to calculate the difference between the model's prediction value and the true value, and use the back-propagation algorithm to calculate the gradient of the loss function relative to the model parameters to update the model parameters.

[0107] During the training process, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of trainings, the number of batches, and the number of neurons in the hidden layer; wherein the number of network layers is set to a four-layer network structure, the number of iterations is set to 200, the learning rate is set to 0.001, the batch size is set to 64, the number of trainings is set to 500, the batch size is set to 256, and the number of neurons in the hidden layer is 32;

[0108] Gradually optimize the undetermined parameters and undetermined constants, and use RMSE (root mean square error) to evaluate the model performance until the pressure prediction model training is completed. Input the characteristic data obtained in step 2 into the trained pressure prediction model to obtain the corresponding internal pressure of the dangerous equipment to be monitored;

[0109] It means that as the concentration fluctuation of harmful gases increases, the effect of pressure decays nonlinearly. This is because excessive fluctuations may cause the safety mechanism of the equipment to start and reduce the pressure; represents the effect of temperature uncertainty on pressure, higher temperature fluctuations will increase the probability of pressure; It indicates the influence of the mean flow rate on the pressure, taking into account the influence of the concentration of harmful gases. The greater the flow rate, the higher the pressure. However, as the concentration of harmful gases increases, the pressure increase will decrease. Indicates the effect of periodic changes in temperature and flow on pressure. The higher the frequency, the faster the temperature and flow change, which may lead to rapid fluctuations in pressure. g Indicates the linear effect of gas concentration on pressure. The higher the gas concentration, the greater the pressure.

[0110] μ g is the mean value of the harmful gas concentration, reflecting the average harmful gas concentration level in a given period of time, σ g is the standard deviation of the concentration of harmful gases, reflecting the degree of concentration fluctuation; the larger the standard deviation, the greater the volatility, which may lead to changes in pressure, μ t is the mean value of the ambient temperature, which affects the physical characteristics of the device and its response, σ t is the standard deviation of the ambient temperature. Fluctuating temperature may cause changes in the internal pressure of the device. u is the mean value of the flow rate. The flow rate directly affects the flow characteristics of the fluid, thereby affecting the pressure. u is the standard deviation of the flow rate. Fluctuations in flow rate may lead to instantaneous changes in pressure. g ,f t and f u They represent the main frequencies of harmful gas concentration, ambient temperature and flow rate, respectively, and affect the periodic changes of these variables;

[0111] P represents the internal pressure of the equipment, which combines the effects of gas concentration, temperature, flow rate and their changing frequency on pressure. The larger the P value, the higher the internal pressure of the equipment, which may lead to increased equipment risks.

[0112] The difference between the pressure data inside the dangerous equipment to be monitored and the set equipment pressure threshold is based on the formula:

[0113] ΔP=PP th

[0114] Where P represents the internal pressure of the dangerous equipment currently monitored, P th Indicates the set equipment pressure threshold, which is the maximum pressure limit for safe operation of the equipment. ΔP indicates the difference between the current pressure and the threshold, which shows the difference between the currently monitored internal pressure P of the dangerous equipment and the set maximum pressure limit P for safe operation of the equipment. th The gap between them reflects the degree to which the pressure deviates from the threshold;

[0115] The greater the internal pressure P of the equipment, the greater the ΔP, indicating that the risk level faced by the equipment increases;

[0116] According to the pressure difference ΔP: if ΔP>0, it means that the current pressure exceeds the safety threshold and the device is at risk; if ΔP≤0, it means that the device is within the safety range and there is no risk;

[0117] The risk levels can be divided into high risk, medium risk and low risk, and are classified by setting specific thresholds:

[0118]

[0119] Among them, ΔP high The pressure differential threshold, indicating a defined high-risk condition, is set at 5% or 10% of the device pressure threshold, ΔP medium The pressure difference threshold representing a defined medium risk condition is set at 1% or 3% of the equipment pressure threshold;

[0120] The risk level (High Risk, Medium Risk, Low Risk) is the result of classification based on the pressure difference ΔP. The formula compares the pressure difference ΔP with the preset threshold to determine the risk status of the equipment. The risk level is directly determined by ΔP. An increase in ΔP will lead to an increase in the risk level. Specifically, as ΔP increases, the risk faced by the equipment increases from low risk to medium risk and then to high risk.

[0121] By analyzing the pressure data of the dangerous equipment, the risk status can be automatically identified and corresponding control measures can be formulated:

[0122] High Risk means that the equipment pressure exceeds the safety range, which may cause equipment damage or accidents. Immediate measures should be taken: stop the equipment, conduct a comprehensive inspection, activate the equipment alarm system, notify relevant personnel, and open the safety valve to reduce the internal pressure.

[0123] By identifying that ΔP is at high risk, managers can take quick action to prevent an incident;

[0124] Medium Risk means that the equipment pressure is close to the upper limit of the safety range and needs to be closely monitored. Preventive measures may also be needed: increase the frequency of pressure monitoring, observe pressure changes in real time, appropriately reduce the load of the equipment, reduce the risk of pressure increase, and arrange for technicians to inspect and evaluate the equipment;

[0125] Identifying that ΔP is at medium risk allows managers to take preventive measures to prevent equipment pressure from reaching dangerous levels;

[0126] Low Risk means the equipment is operating normally within a safe range: maintain normal monitoring frequency to ensure that the equipment is within a safe range, continue to record pressure data, analyze pressure fluctuation trends to provide a basis for subsequent preventive measures, perform regular equipment maintenance to ensure normal equipment operation and prevent potential risks.

[0127] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0128] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0129] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0130] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for automatic identification and risk control of hazardous sources in chemical production processes, characterized in that: The specific steps include: Step 1: During the monitoring period of the chemical production process, record the influencing factors inside the dangerous equipment to be monitored, including harmful gas concentration, ambient temperature, and fluid flow data, and pre-process the recorded data; Step 2: Extract characteristic data from the preprocessed data, and calculate the internal pressure of the dangerous equipment to be monitored based on the characteristic data, wherein the characteristic data includes frequency domain characteristics and statistical characteristics, wherein the statistical characteristics include mean and standard deviation data corresponding to various influencing factors, and the frequency domain characteristics include the main frequencies of various influencing factors; Step 3: Based on the difference between the calculated internal pressure of the hazardous equipment to be monitored and the set equipment pressure threshold, determine the risk status inside the hazardous equipment to be monitored, output the risk level, and formulate a risk control plan for the hazardous equipment based on the risk level.

2. The method for automatic identification and risk control of hazardous sources in a chemical production process according to claim 1 is characterized in that: The monitoring time period is the time period before the current moment, and the duration of the monitoring time period is T. During the monitoring time period of the chemical production process, the harmful gas concentration, ambient temperature, and fluid flow data inside the monitored dangerous equipment at each time point are recorded in equal time intervals to obtain the following data sequence: {C g (t i ),C t (t i ),C u (t i )|i=1,2,...,N} Among them, C g (t i ) represents the time point t i The concentration of harmful gases recorded at t (t i ) represents the time point t i Ambient temperature recorded at the time, C u (t i ) represents the time point t i The flow rate data of the fluid recorded at t i represents the i-th time point in the monitoring period, i is the index of the time point, and i∈[1,N], N is the number of time points in the monitoring period; The logic of the data preprocessing is: use the sliding average method to smooth the recorded data, then use the quartile method to eliminate outliers in the smoothed data, and finally use the mean filling method to supplement the missing values ​​of the data.

3. The method for automatic identification and risk control of hazardous sources in a chemical production process according to claim 2 is characterized by: The statistical features are extracted to calculate the mean and standard deviation of harmful gas concentration, ambient temperature, and fluid flow data. Taking harmful gas concentration as an example, the formula is: Among them, μ g Represents the mean value of the concentration of harmful gases after pretreatment, C′ g (t i ) indicates the concentration of harmful gases C g (t i ) after preprocessing; Among them, σ g It represents the standard deviation of the concentration of harmful gases after pretreatment, that is, the degree of fluctuation of the concentration of harmful gases. The higher the standard deviation, the greater the fluctuation of the concentration of harmful gases. Similarly, the ambient temperature and flow data are processed to obtain their mean values ​​μ t , μ u and standard deviation σ t , σ u ; The logic for extracting the main frequencies of the harmful gas concentrations is as follows: Fast Fourier transform is used to convert the time-series distribution of harmful gas concentration into frequency domain signal, and the formula is as follows: Among them, F g (k) is the kth frequency component in the frequency domain signal of the harmful gas concentration, expressed in the form of a complex number, k is the index of the frequency component in the frequency domain signal of the harmful gas concentration, and k∈[1,N], j is an imaginary unit; The frequency resolution is calculated based on the following formula: Where Δf is the frequency resolution; Calculate the amplitude corresponding to each frequency component, and determine the index of the frequency component corresponding to the maximum amplitude. This index is used as the main frequency index of the harmful gas concentration. The formula is as follows: A g (k)=|F g (k)| Among them, A g (k) is the amplitude corresponding to the kth frequency component in the frequency domain signal of harmful gas concentration, k g is the main frequency index of harmful gas concentration, |F g (k)| indicates F g (k) taking the absolute value; Based on the frequency resolution and the main frequency index of the harmful gas concentration, the main frequency of the harmful gas concentration is determined. The calculation formula is as follows: f g =k g ·Δf Among them, f g is the main frequency of the harmful gas concentration, and similarly the main frequency of the ambient temperature is obtained t , the main frequency f of the fluid flow u .

4. The method for automatic identification and risk control of hazardous sources in a chemical production process according to claim 3 is characterized by: A pressure prediction model is established based on a deep neural network. The LSTM network architecture is used to process time series features. One to two layers of LSTM units are set, and each layer is set to include 50 to 100 LSTM units. The hidden layer uses ReLU as the activation function, and the output layer uses a linear activation function. Obtain the influencing factor data and the corresponding internal pressure of the dangerous equipment to be monitored in multiple historical monitoring time periods in a time series distribution, use the same method as step 1-2 to extract the features of the influencing factor data in the historical monitoring time period, obtain historical feature data, and use the historical feature data as a training set, the historical feature data as input, and the corresponding internal pressure of the dangerous equipment to be monitored as an output label to train the pressure prediction model. The model expression of the pressure prediction model is as follows: Among them, α, β, γ, δ, ζ are unknown parameters in the model expression, ∈ is an unknown constant to prevent the denominator from being zero, and ∈>0; Use the Batch training method to divide the entire training set into multiple small batches. In each training iteration, pass the input data into the LSTM model, calculate its output prediction value, use the MSE loss function to calculate the difference between the model's prediction value and the true value, and use the back-propagation algorithm to calculate the gradient of the loss function relative to the model parameters to update the model parameters. During the training process, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of trainings, the number of batches, and the number of neurons in the hidden layer; wherein the number of network layers is set to a four-layer network structure, the number of iterations is set to 200, the learning rate is set to 0.001, the batch size is set to 64, the number of trainings is set to 500, the number of batches is set to 256, and the number of neurons in the hidden layer is 32; Gradually optimize the unknown parameters and unknown constants, and use RMSE to evaluate the model performance until the pressure prediction model training is completed. Input the feature data obtained in step 2 into the trained pressure prediction model to obtain the corresponding internal pressure of the hazardous equipment to be monitored.

5. The method for automatic identification and risk control of hazardous sources in a chemical production process according to claim 4 is characterized in that: The difference between the pressure data inside the dangerous equipment to be monitored and the set equipment pressure threshold is based on the formula: AP=PP th Where P represents the internal pressure of the dangerous equipment to be monitored, P th Indicates the set equipment pressure threshold, which is the maximum pressure limit for safe operation of the equipment. AP indicates the difference between the current pressure and the threshold, reflecting the degree to which the pressure deviates from the threshold. According to the pressure difference AP: if AP>0, it means that the current pressure exceeds the safety threshold and the device is at risk; if ΔP≤0, it means that the device is within the safety range and there is no risk; The risk levels can be divided into high risk, medium risk and low risk, and are classified by setting specific thresholds: Among them, ΔP high The pressure differential threshold, indicating a defined high-risk condition, is set at 5% or 10% of the device pressure threshold, ΔP medium The pressure difference threshold representing a defined medium risk condition is set at 1% or 3% of the equipment pressure threshold; By analyzing the pressure data of the dangerous equipment, the risk status is automatically identified and corresponding control measures are formulated: High Risk means that the equipment pressure exceeds the safety range, which may cause equipment damage or accidents. Take measures immediately: stop the equipment, conduct a comprehensive inspection, activate the equipment alarm system, notify relevant personnel, and open the safety valve to reduce the internal pressure; Medium Risk means that the equipment pressure is close to the upper limit of the safety range. Close monitoring is required and preventive measures are taken: increase the frequency of pressure monitoring, observe pressure changes in real time, reduce the load on the equipment to reduce the risk of pressure increase, and arrange technicians to inspect and evaluate the equipment; Low Risk means the equipment is operating normally within a safe range: maintain normal monitoring frequency to ensure that the equipment is within a safe range, continue to record pressure data, analyze pressure fluctuation trends to provide a basis for subsequent preventive measures, perform regular equipment maintenance to ensure normal equipment operation and prevent potential risks.

Images