Artificial intelligence-based industrial plant accident emergency response intelligent monitoring and judgment method

Abstract

This invention relates to the field of industrial emergency monitoring technology, and in particular to an intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence. The method accurately extracts thermal field-related features and smoke-related features based on infrared outer parameters. It calculates the light-invariant texture transmittance based on linear normalization, a noise compensation function, and normalized thermal field entropy flux density using visual side parameters. The method determines the scene state through a coupled state factor. For intense fire and dense smoke scenes, an iterative optimization algorithm is used to solve for the adaptive medium scattering compensation coefficient, while a default compensation coefficient is used for non-intense fire and dense smoke scenes. A signal-to-noise ratio gating function based on a baseline noise level, thermal field features, and smoke features is constructed. Combined with the adaptive medium scattering compensation coefficient, a smoke-compensated thermal turbulence singularity index is synthesized, enhancing the characterization ability of accident anomalies. This achieves accurate quantification of accident levels, providing precise level guidance for emergency response and effectively improving the timeliness and scientific nature of emergency handling.

Description

Technical Field

[0001] This invention relates to the field of industrial emergency monitoring technology, and in particular to an intelligent monitoring and analysis method for emergency response to industrial plant accidents based on artificial intelligence. Background Technology

[0002] Existing industrial plant accident emergency response monitoring and assessment technologies have many shortcomings, making it difficult to adapt to the needs of accurate early warning and efficient response in complex industrial scenarios. Traditional monitoring methods mostly adopt single infrared detection or visual detection modes, lacking the synchronous acquisition and fusion of infrared and visual parameters. This results in independent extraction of thermal field features and smoke features, making it impossible to accurately capture the coupling relationship between the two, and easily leading to misjudgments or omissions. In the feature processing process, existing technologies do not consider the interference of infrared detector thermal noise, changes in ambient light, and medium scattering in intense fire and dense smoke scenarios on the detection results. They only use fixed compensation coefficients or simple filtering, resulting in low accuracy in extracting thermal field features and smoke features. Especially in intense fire and dense smoke scenarios, the detection signal is severely distorted and cannot accurately reflect the true state of the accident. Meanwhile, existing technologies lack a scientifically sound signal-to-noise ratio gating mechanism and quantitative indicators for abnormal features. They rely heavily on single feature thresholds for early warning, making it difficult to consider adaptability to different scenarios. Furthermore, they fail to achieve accurate accident level classification through feature accumulation over time, resulting in a mismatch between early warning levels and the actual severity of accidents. This fails to provide precise level guidance for emergency response, thus affecting the timeliness and effectiveness of emergency handling. In addition, existing technologies have significant shortcomings in scenario adaptability. They do not design differentiated compensation strategies for scenarios with intense fires and dense smoke versus those without. They either oversimplify calculations, leading to insufficient accuracy, or use complex calculations that affect response efficiency, making it difficult to balance the requirements of assessment accuracy and real-time performance.

[0003] Therefore, there is an urgent need to propose a new intelligent monitoring and assessment method for emergency response to industrial plant accidents based on artificial intelligence. Summary of the Invention

[0004] The main objective of this invention is to provide an intelligent monitoring and assessment method for emergency response to industrial plant accidents based on artificial intelligence. It accurately extracts thermal and smoke-related features based on infrared side parameters. Through linear normalization, noise compensation functions, and calculation of normalized thermal entropy flux density, it offsets the interference of thermal noise from the infrared detector itself, improving the quantification accuracy of thermal features. Simultaneously, it calculates illumination-invariant texture transmittance based on visual side parameters, effectively eliminating the influence of ambient light variations on smoke feature extraction and ensuring the stability of feature extraction under different lighting conditions. To address the adaptability issue for different scenarios, it determines the scene state through a coupled state factor. For scenarios with intense fire and dense smoke, an iterative optimization algorithm is used to solve for the adaptive medium scattering compensation coefficient, while a default compensation coefficient is used for scenarios without intense fire and dense smoke. This solves the detection distortion problem caused by medium scattering in intense fire and dense smoke scenarios while also considering computational efficiency in non-intense fire and dense smoke scenarios, achieving a balance between assessment accuracy and real-time performance. Furthermore, a signal-to-noise ratio gating function based on baseline noise level, thermal field characteristics, and smoke characteristics was constructed. Combined with an adaptive medium scattering compensation coefficient, a smoke-compensated thermal turbulence singularity index was synthesized, which enhanced the characterization capability of accident anomalies. The accurate quantification of accident levels was achieved, making the warning level highly matched with the actual accident severity. This provides accurate level guidance for emergency response and effectively improves the timeliness and scientific nature of emergency handling.

[0005] The technical solution of the present invention is as follows:

[0006] A method for intelligent monitoring and assessment of emergency response to industrial plant accidents based on artificial intelligence is proposed. This method includes the following steps:

[0007] S1. Synchronously acquire infrared outer side parameters and visual side parameters and construct a time synchronization parameter vector;

[0008] S2. Extract thermal field-related features and smoke-related features based on the infrared outer parameters, and further obtain the coupling state factor based on the thermal field-related features and smoke-related features;

[0009] S3. Based on the numerical determination of the coupling state factor, when the state is determined to be a strong fire and dense smoke state, the adaptive medium scattering compensation coefficient is solved by an iterative optimization algorithm. When the state is determined to be a non-strong fire and dense smoke state, the adaptive medium scattering compensation coefficient adopts the default value.

[0010] S4. Based on the infrared outer parameters, the baseline noise level is obtained. The signal-to-noise ratio gating function is constructed by combining the baseline noise level, thermal field correlation characteristics, and smoke correlation characteristics. The smoke-compensated thermal turbulence singularity index is then synthesized by combining the adaptive medium scattering compensation coefficient.

[0011] S5. Extract the smoke-compensated thermal turbulence singularity index sequence within the historical time window, perform numerical integration using the trapezoidal method, generate the accident response level index, and guide early warning based on the magnitude of the accident response level index.

[0012] A further improvement of the present invention is that the specific content of S1 is: synchronously acquiring infrared-outer parameters and visual-side parameters, wherein the infrared-outer parameters include the raw radiation values ​​acquired by the infrared detector. Infrared detector target surface temperature and the integration time of the infrared detector The visual side parameters include the energy value of the image texture portion of the visual detection region. With the ambient light intensity of the detection area Construct time synchronization parameter vector , t is the index of the time node.

[0013] A further improvement of the present invention is that step S2 includes the following specific steps:

[0014] S21. Extract the raw radiation values ​​collected by the infrared detector. It is mapped to a gray level k of 0-255 through linear normalization. The mapping formula is: Where k is the gray level index of the normalized grayscale histogram, with a value ranging from 0 to 255. This is the maximum detectable radiation value of the infrared detector. This is the minimum detectable radiation value for the infrared detector. To perform floor operations, construct a normalized grayscale histogram probability distribution. , ; This represents the probability value of the k-th gray level appearing in the normalized gray-level histogram at time point t. The frequency of occurrence of gray level k. To prevent division by zero of constants;

[0015] S22, Extract the infrared detector target surface temperature Construct a noise compensation function, the expression of which is: ;in, This is the noise compensation coefficient. This represents the thermal noise compensation value of the equipment at time node t.

[0016] S23, Integration time based on infrared detector Normalized grayscale histogram probability distribution and noise compensation function Calculate the normalized thermal field entropy flux density, the formula for which is:

[0017] ;

[0018] in, This represents the normalized thermal entropy flux density at time point t. This is the rated integration time of the infrared detector.

[0019] A further improvement of the present invention is that S2 further includes:

[0020] S24. Based on visual side parameters, including the energy value of the image texture portion of the visual detection region. With the ambient light intensity of the detection area The transmittance of the illumination-invariant texture is calculated using the following formula:

[0021] ;

[0022] in, The transmittance of the illumination-invariant texture at time point t. As a correction factor, This is the ambient light compensation coefficient;

[0023] S25. Based on the normalized thermal field entropy flux density and the illumination-invariant texture transmittance, the coupling state factor is further obtained. , ; Let be the coupling state factor at time node t. The reference time constant for the change in entropy flow in the thermal field is denoted as . This is the reference time constant for the change in texture transmittance.

[0024] A further improvement of the present invention is that step S3 includes the following specific steps:

[0025] S31, when the coupling state factor Not greater than the preset threshold At that time, it was determined to be a non-strong fire and dense smoke state, and the adaptive medium scattering compensation coefficient was used. The default value is set to 1; otherwise, it is determined to be a state of intense fire and dense smoke.

[0026] S32. When the condition is determined to be a strong fire and dense smoke state, the initial scattering rate is calculated based on the combustion medium type of the detection area and the corresponding prediction formula. The calculated initial scattering rate Substituting the equations into the combustion medium mass diffusion-thermal entropy flow coupling control equations for verification, the normalized thermal field entropy flow density is obtained. The expression for the combustion medium mass diffusion-thermal entropy flow coupling control equation is as follows: ;in, For a preset empirical density based on the type of combustion medium, The preset empirical diffusion rate of the combustion medium, For divergence calculation, The scattering rate to be tested is... The thermal entropy flow-mass diffusion coupling coefficient;

[0027] S33, Calculated value of normalized thermal field entropy flux density With normalized thermal field entropy flux density When the absolute value of the difference between them is greater than 0.1, the initial scattering rate... Make corrections as follows: If The corrected initial scattering rate is ,like The corrected initial scattering rate is The corrected initial scattering rate is then substituted back into the combustion medium mass diffusion-thermal entropy flow coupled control equation for verification. This process is repeated iteratively until the absolute value of the difference between the initial scattering rates of two adjacent iterations is less than [a certain value]. The iteration stops when the initial scattering rate is reached, and the final value of the initial scattering rate is output as the adaptive medium scattering compensation coefficient. .

[0028] A further improvement of this invention is that, in step S32, the initial scattering rate is calculated based on the combustion medium type of the detection area using a corresponding estimation formula, including: when the combustion medium type of the detection area is ethylene black smoke, the estimation formula is: When the combustion medium in the detection area is cable white smoke, the prediction formula is: .

[0029] A further improvement of the present invention is that step S4 includes the following specific steps:

[0030] S41, Based on infrared detector target surface temperature By using the linear fitting formula: ; Obtain the baseline noise level of the infrared detection module , where a is the noise floor fitting coefficient and b is the noise floor fitting constant;

[0031] S42, Based on the reference noise floor level of the infrared detection module Normalized thermal field entropy flux density and illumination-invariant texture transmittance Construct a signal-to-noise ratio (SNR) gating function, the expression of which is:

[0032] ;

[0033] in, The signal-to-noise ratio gating coefficient represents the signal-to-noise ratio at time point t. The first judgment threshold is... 2 is the second judgment threshold. is the gain coefficient of the signal-to-noise ratio gating function. This is the rated reference noise floor of the infrared detector;

[0034] S43, Based on normalized thermal field entropy flux density Adaptive medium scattering compensation coefficient Light-invariant texture transmittance and signal-to-noise ratio gating coefficient A smoke-compensated thermal turbulence singularity index is generated, and the expression for the smoke-compensated thermal turbulence singularity index is:

[0035] ;

[0036] in, The singularity index of smoke-compensated thermal turbulence at time node t is denoted by t.

[0037] A further improvement of the present invention is that the specific content of S5 is: introducing a historical time window length w to obtain the historical time node interval. The accident response level index is obtained by numerically integrating the squared terms of the smoke-compensated thermal turbulence singularity index sequence using the trapezoidal method. The formula for calculating the accident response level index is as follows: ;in, This represents the accident response level index at time point t. The sampling time interval for the smoke-compensated thermal turbulence singularity index sequence. , n represents the historical time interval. The number of sampling points within.

[0038] The technical effects of this invention are as follows:

[0039] An AI-based intelligent monitoring and assessment method for emergency response to industrial plant accidents was constructed. Through feature extraction, thermal and smoke-related features were accurately extracted based on infrared outer parameters. Linear normalization, noise compensation functions, and normalized thermal entropy flux density calculations were used to offset the interference of infrared detector thermal noise, improving the quantification accuracy of thermal features. Simultaneously, illumination-invariant texture transmittance was calculated based on visual side parameters, effectively eliminating the influence of ambient light variations on smoke feature extraction and ensuring the stability of feature extraction under different lighting conditions. To address the adaptability issue for different scenarios, a coupled state factor was used to determine the scene state. For high-fire, dense smoke scenarios, an iterative optimization algorithm was used to solve for the adaptive medium scattering compensation coefficient, while a default compensation coefficient was used for non-high-fire, dense smoke scenarios. This approach not only solved the detection distortion problem caused by medium scattering in high-fire, dense smoke scenarios but also considered computational efficiency in non-high-fire, dense smoke scenarios, achieving a balance between assessment accuracy and real-time performance. Furthermore, a signal-to-noise ratio gating function based on baseline noise level, thermal field characteristics, and smoke characteristics was constructed. Combined with an adaptive medium scattering compensation coefficient, a smoke-compensated thermal turbulence singularity index was synthesized, which enhanced the characterization capability of accident anomalies. The accurate quantification of accident levels was achieved, making the warning level highly matched with the actual accident severity. This provides accurate level guidance for emergency response and effectively improves the timeliness and scientific nature of emergency handling. Attached Figure Description

[0040] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0041] Figure 1 This is a flowchart illustrating the intelligent monitoring and analysis method for emergency response to industrial plant accidents based on artificial intelligence, as described in Embodiment 1 of the present invention. Detailed Implementation

[0042] Example 1

[0043] This embodiment proposes an AI-based intelligent monitoring and assessment method for emergency response to industrial plant accidents. Through feature extraction, it accurately extracts thermal and smoke-related features based on infrared outer parameters. Linear normalization, noise compensation functions, and normalized thermal entropy flux density calculations counteract the interference of thermal noise from the infrared detector itself, improving the quantification accuracy of thermal features. Simultaneously, it calculates illumination-invariant texture transmittance based on visual side parameters, effectively eliminating the impact of ambient light variations on smoke feature extraction and ensuring the stability of feature extraction under different lighting conditions. Addressing the adaptability issue for different scenarios, it determines the scene state through a coupled state factor. For scenarios with intense fire and dense smoke, an iterative optimization algorithm is used to solve for the adaptive medium scattering compensation coefficient, while a default compensation coefficient is used for scenarios without intense fire and dense smoke. This solves the detection distortion problem caused by medium scattering in intense fire and dense smoke scenarios while also considering computational efficiency in non-intense fire and dense smoke scenarios, achieving a balance between assessment accuracy and real-time performance. Furthermore, a signal-to-noise ratio (SNR) gating function based on baseline noise floor level, thermal field characteristics, and smoke characteristics was constructed. Combined with an adaptive medium scattering compensation coefficient, a smoke-compensated thermal turbulence singularity index was synthesized, enhancing the characterization capability of accident anomalies. This achieved precise quantification of accident levels, ensuring a high degree of match between warning levels and actual accident severity, providing precise level guidance for emergency response, and effectively improving the timeliness and scientific rigor of emergency handling. Specifically, such as... Figure 1 As shown in this embodiment, the intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence includes the following specific steps:

[0044] S1. Synchronously acquire infrared outer side parameters and visual side parameters and construct a time synchronization parameter vector;

[0045] S2. Extract thermal field-related features and smoke-related features based on the infrared outer parameters, and further obtain the coupling state factor based on the thermal field-related features and smoke-related features;

[0046] S3. Based on the numerical determination of the coupling state factor, when the state is determined to be a strong fire and dense smoke state, the adaptive medium scattering compensation coefficient is solved by an iterative optimization algorithm. When the state is determined to be a non-strong fire and dense smoke state, the adaptive medium scattering compensation coefficient adopts the default value.

[0047] S4. Based on the infrared outer parameters, the baseline noise level is obtained. The signal-to-noise ratio gating function is constructed by combining the baseline noise level, thermal field correlation characteristics, and smoke correlation characteristics. The smoke-compensated thermal turbulence singularity index is then synthesized by combining the adaptive medium scattering compensation coefficient.

[0048] S5. Extract the smoke-compensated thermal turbulence singularity index sequence within the historical time window, perform numerical integration using the trapezoidal method, generate the accident response level index, and guide early warning based on the magnitude of the accident response level index.

[0049] In this embodiment, S1 specifically involves: simultaneously acquiring infrared-outer parameters and visual-side parameters, wherein the infrared-outer parameters include the raw radiation values ​​acquired by the infrared detector. Infrared detector target surface temperature and the integration time of the infrared detector The visual side parameters include the energy value of the image texture portion of the visual detection region. With the ambient light intensity of the detection area Construct time synchronization parameter vector , t is the index of the time node.

[0050] In this embodiment, the dimension of the raw radiation value collected by the infrared detector is W / m², the dimension of the infrared detector target surface temperature is K, and the value range is the normal operating temperature range of the infrared detector. The dimension of the infrared detector integration time is μs, and the value range is the integration time adjustment range supported by the infrared detector. The dimension of the energy value of the image texture part of the visual detection area is the dimensionless image energy statistical value, and the value range is 0 to the product of the total number of image pixels and the maximum gray value of a single pixel. The dimension of the ambient light intensity of the detection area is lux, and the value range is 0 to the maximum value of normal ambient light in industrial plants.

[0051] In this embodiment, step S2 includes the following specific steps:

[0052] S21. Extract the raw radiation values ​​collected by the infrared detector. It is mapped to a gray level k of 0-255 through linear normalization. The mapping formula is: Where k is the gray level index of the normalized grayscale histogram, with a value ranging from 0 to 255. This is the maximum detectable radiation value of the infrared detector. This is the minimum detectable radiation value for the infrared detector. To perform floor operations, construct a normalized grayscale histogram probability distribution. , ; This represents the probability value of the k-th gray level appearing in the normalized gray-level histogram at time point t. The frequency of occurrence of gray level k. To prevent division by zero, the preferred value is [value missing]. This value is derived from the principle of minimizing interference with the statistical characteristics of the original data, avoiding a denominator of 0 while not changing the probability distribution characteristics of the gray-level histogram.

[0053] S22, Extract the infrared detector target surface temperature Construct a noise compensation function, the expression of which is: ;in, The noise compensation coefficient has the following dimensions: The preferred value is This value is derived from the fitting results of the target surface temperature and thermal noise of the industrial-grade infrared detector within its normal operating temperature range, achieving optimal thermal noise compensation. This represents the thermal noise compensation value of the equipment at time node t.

[0054] S23, Integration time based on infrared detector Normalized grayscale histogram probability distribution and noise compensation function Calculate the normalized thermal field entropy flux density, the formula for which is:

[0055] ;

[0056] in, This represents the normalized thermal entropy flux density at time point t. The rated integration time of the infrared detector is in μs, preferably 100 μs, and this value comes from the rated operating parameters calibrated by the infrared detector at the factory.

[0057] The design concept of this step is that traditional infrared monitoring judges temperature changes only by radiation value, which is easily affected by detector integration time and target surface thermal noise. However, thermal field entropy flux density can characterize the disorder of the thermal field in the monitoring area. During a fire, the disorder of the thermal field will change rapidly and continuously, which has higher sensitivity than a single radiation value or temperature value. At the same time, by compensating for integration time and target surface temperature, the calculation deviation caused by changes in the detector's own working state can be eliminated, ensuring the stability of thermal field related parameters.

[0058] In this embodiment, S2 further includes:

[0059] S24. Based on visual side parameters, including the energy value of the image texture portion of the visual detection region. With the ambient light intensity of the detection area The transmittance of the illumination-invariant texture is calculated using the following formula:

[0060] ;

[0061] in, The transmittance of the illumination-invariant texture at time point t. The correction factor is preferably set to 0.8. This value is derived from the correlation fitting results between ambient light intensity and image texture energy within the normal lighting range of an industrial plant area, which can achieve the best illumination invariance correction effect. The ambient light compensation coefficient has the following dimensions: The preferred value is 1.2. This value comes from the calibration results of the photoelectric response characteristics of the visual image acquisition device.

[0062] The design idea behind this step is that the smoke produced by a fire will cause the texture details of the visual image to be obscured. Texture transmittance can directly represent the concentration and degree of obscuration of smoke. At the same time, by correcting the ambient light intensity, the interference of daytime and nighttime light changes on texture parameters can be eliminated, ensuring the stability of smoke-related parameters under different lighting conditions.

[0063] S25. Based on the normalized thermal field entropy flux density and the illumination-invariant texture transmittance, the coupling state factor is further obtained. , ; Let be the coupling state factor at time node t. The reference time constant for the change in entropy flow in the thermal field is preferably 1 s. The reference time constant for the change in texture transmittance is preferably 1 s.

[0064] The design idea behind this step is that, under normal conditions, the rate of change of thermal field disorder and texture transmittance are both at a low level and there is no strong correlation between the two. However, when a fire occurs, the thermal field disorder will rise rapidly, and at the same time, smoke will cause the texture transmittance to drop rapidly. The product of the normalized rate of change of the two will increase by an order of magnitude. By coupling the state factor, the changes in thermal field and smoke can be correlated simultaneously, so as to achieve a preliminary judgment of the fire scene.

[0065] In this embodiment, step S3 includes the following specific steps:

[0066] S31, when the coupling state factor Not greater than the preset threshold At that time, it was determined to be a non-strong fire and dense smoke state, and the adaptive medium scattering compensation coefficient was used. The default value is set to 1; otherwise, it is determined to be a state of intense fire and dense smoke.

[0067] S32. When the condition is determined to be a strong fire and dense smoke state, the initial scattering rate is calculated based on the combustion medium type of the detection area and the corresponding prediction formula. The calculated initial scattering rate Substituting the equations into the combustion medium mass diffusion-thermal entropy flow coupling control equations for verification, the normalized thermal field entropy flow density is obtained. The expression for the combustion medium mass diffusion-thermal entropy flow coupling control equation is as follows: ;in, The preset empirical density is based on the matching of combustion medium type, with dimensions of , The preset empirical diffusion rate of the combustion medium, with dimensions of , For divergence calculation, The scattering rate to be tested is... The thermal entropy flow-mass diffusion coupling coefficient has the following dimensions: .

[0068] S33, Calculated value of normalized thermal field entropy flux density With normalized thermal field entropy flux density When the absolute value of the difference between them is greater than 0.1, the initial scattering rate... Make corrections as follows: If The corrected initial scattering rate is ,like The corrected initial scattering rate is The corrected initial scattering rate is then substituted back into the combustion medium mass diffusion-thermal entropy flow coupled control equation for verification. This process is repeated iteratively until the absolute value of the difference between the initial scattering rates of two adjacent iterations is less than [a certain value]. The iteration stops when the initial scattering rate is reached, and the final value of the initial scattering rate is output as the adaptive medium scattering compensation coefficient. .

[0069] In this embodiment, the calculation of the initial scattering rate in step S32, based on the combustion medium type of the detection area and the corresponding estimation formula, includes: when the combustion medium type of the detection area is ethylene black smoke, the estimation formula is: When the combustion medium in the detection area is cable white smoke, the prediction formula is: .

[0070] The design concept of this step is that in scenarios with intense fire and dense smoke, the smoke particles generated by the fire will strongly scatter infrared radiation, causing distortion of the original radiation values ​​collected by the infrared detector, which in turn affects the accuracy of the thermal field parameters. The smoke scattering characteristics produced by different combustion media are significantly different. By matching the initial prediction formula of the combustion media type and combining iterative optimization of the coupled control equation, a compensation coefficient that perfectly matches the smoke scattering characteristics of the current scenario can be obtained, eliminating the interference of smoke scattering on infrared detection. In scenarios without intense fire and dense smoke, the influence of smoke scattering is negligible, so the default value of 1 can be used without additional iterative calculations, which can reduce the computational overhead of the system.

[0071] In this embodiment, step S4 includes the following specific steps:

[0072] S41, Based on infrared detector target surface temperature By using the linear fitting formula: ; Obtain the baseline noise level of the infrared detection module , where the dimension is the digital quantity DN output by the infrared detector, a is the noise floor fitting coefficient with the dimension DN / K, preferably 0.05 DN / K, and b is the noise floor fitting constant with the dimension DN, preferably 10DN. These two values ​​are derived from the linear calibration results of the noise floor level of the infrared detector at different target surface temperatures.

[0073] S42, Based on the reference noise floor level of the infrared detection module Normalized thermal field entropy flux density and illumination-invariant texture transmittance Construct a signal-to-noise ratio (SNR) gating function, the expression of which is:

[0074] ;

[0075] in, The signal-to-noise ratio gating coefficient represents the signal-to-noise ratio at time point t. The first threshold value is 10. 2 is the second judgment threshold, and the preferred value is 0.5. The gain coefficient of the signal-to-noise ratio (SNR) gating function is preferably set to 100. These three values ​​are derived from the statistical distribution results of normalized thermal field entropy flux density and illumination-invariant texture transmittance in fire-free scenarios, which can achieve smooth suppression of the signal when the SNR is insufficient. The rated reference noise floor of the infrared detector is measured in DN, with a preferred value of 10DN. This value is derived from the reference noise floor value under the rated operating conditions specified by the infrared detector at the factory.

[0076] The design concept of this step is based on the fact that the noise floor level of the infrared detector fluctuates with changes in target surface temperature. When the thermal field signal in the monitored area is weak, the noise floor dominates the signal, leading to a significant decrease in the reliability of thermal field-related parameters. By constructing a sigmoid-type gating function based on the signal-to-noise ratio (SNR), a gating coefficient approaching 0 can be output when the SNR is low, suppressing false judgments caused by noise. When the SNR is high, a gating coefficient approaching 1 can be output, preserving the effective signal. At the same time, the smooth gating curve can avoid signal jumps caused by hard thresholds, ensuring the continuity of subsequent exponential synthesis.

[0077] S43, Based on normalized thermal field entropy flux density Adaptive medium scattering compensation coefficient Light-invariant texture transmittance and signal-to-noise ratio gating coefficient A smoke-compensated thermal turbulence singularity index is generated, and the expression for the smoke-compensated thermal turbulence singularity index is:

[0078] ;

[0079] in, This represents the singularity index of smoke-compensated thermal turbulence at time point t. Since intense thermal turbulence occurs during fire development, singular changes in thermal turbulence are a key indicator of a fire's progression from its initial stage to its out-of-control phase.

[0080] In this embodiment, the specific content of S5 is: introducing the historical time window length w, and obtaining the historical time node interval. The accident response level index is obtained by numerically integrating the squared terms of the smoke-compensated thermal turbulence singularity index sequence using the trapezoidal method. The formula for calculating the accident response level index is as follows: ;in, This represents the accident response level index at time point t. The sampling time interval for the smoke-compensated thermal turbulence singularity index sequence. , n represents the historical time interval. The number of sampling points within.

[0081] The design concept of this step is that instantaneous singular index jumps may come from environmental interference or equipment noise. By numerically integrating the squared terms of the singular index within the historical time window, the weight of high-risk signals can be amplified, while the cumulative risk level over a period of time can be represented, effectively suppressing false alarms caused by instantaneous interference. The trapezoidal method, as a numerical integration method, has the characteristics of low computational cost and high accuracy, and can be adapted to the real-time computing needs of embedded devices in industrial monitoring scenarios. The final accident response level index can directly correspond to different warning levels, guiding the emergency response operations of industrial plants.

[0082] Example 2

[0083] This embodiment provides an electronic device, including a processor and a memory, wherein the memory stores a computer program that can be called by the processor; the processor executes the above-mentioned intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence by calling the computer program stored in the memory.

[0084] The electronic device can vary considerably depending on its configuration or performance. It may include one or more Central Processing Units (CPUs) and one or more memories, wherein the memory stores at least one computer program, which is loaded and executed by the processor to implement the AI-based intelligent monitoring and analysis method for emergency response to industrial plant accidents provided in the above-described embodiment. The electronic device may also include other components for implementing its functions; for example, it may have wired or wireless network interfaces and input / output interfaces for data input and output. Further details are omitted in this embodiment.

[0085] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this disclosure can be embodied in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, the invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0086] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0087] This invention is described with reference to flowchart illustrations and block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and block diagrams, as well as combinations of blocks in the flowchart illustrations and block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0088] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and boxes Figure 1 The steps of the function specified in one or more boxes.

[0089] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. An artificial intelligence-based industrial plant accident emergency response intelligent monitoring and judgment method, characterized in that: The specific steps include the following: S1. Synchronously acquire infrared outer side parameters and visual side parameters and construct a time synchronization parameter vector; S2. Extract thermal field-related features and smoke-related features based on the infrared outer parameters, and further obtain the coupling state factor based on the thermal field-related features and smoke-related features; S3. Based on the numerical determination of the coupling state factor, when the state is determined to be a strong fire and dense smoke state, the adaptive medium scattering compensation coefficient is solved by an iterative optimization algorithm. When the state is determined to be a non-strong fire and dense smoke state, the adaptive medium scattering compensation coefficient adopts the default value. S4. Based on the infrared outer parameters, the baseline noise level is obtained. The signal-to-noise ratio gating function is constructed by combining the baseline noise level, thermal field correlation characteristics, and smoke correlation characteristics. The smoke-compensated thermal turbulence singularity index is then synthesized by combining the adaptive medium scattering compensation coefficient. S5. Extract the smoke-compensated thermal turbulence singularity index sequence within the historical time window, perform numerical integration using the trapezoidal method, generate the accident response level index, and guide early warning based on the magnitude of the accident response level index.

2. The intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence as described in claim 1, characterized in that: The specific content of S1 is: synchronously acquiring infrared-outer parameters and visual-side parameters, wherein the infrared-outer parameters include the raw radiation values ​​acquired by the infrared detector. Infrared detector target surface temperature and the integration time of the infrared detector The visual side parameters include the energy value of the image texture portion of the visual detection region. With the ambient light intensity of the detection area Construct time synchronization parameter vector , t is the index of the time node.

3. The intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence as described in claim 2, characterized in that: S2 includes the following specific steps: S21. Extract the raw radiation values ​​collected by the infrared detector. It is mapped to a gray level k of 0-255 through linear normalization. The mapping formula is: Where k is the gray level index of the normalized gray-level histogram, with a value ranging from 0 to 255. This is the maximum detectable radiation value of the infrared detector. This is the minimum detectable radiation value for the infrared detector. To perform floor operations, construct a normalized grayscale histogram probability distribution. , ; This represents the probability value of the k-th gray level appearing in the normalized gray-level histogram at time point t. The frequency of occurrence of gray level k. To prevent division by zero of constants; S22, Extract the infrared detector target surface temperature Construct a noise compensation function, the expression of which is: ;in, This is the noise compensation coefficient. This represents the thermal noise compensation value of the equipment at time node t. S23, Integration time based on infrared detector Normalized grayscale histogram probability distribution and noise compensation function Calculate the normalized thermal field entropy flux density, the formula for which is: ； in, This represents the normalized thermal entropy flux density at time point t. This is the rated integration time of the infrared detector.

4. The intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence as described in claim 3, characterized in that: S2 further includes: S24. Based on visual side parameters, including the energy value of the image texture portion of the visual detection region. With the ambient light intensity of the detection area The transmittance of the illumination-invariant texture is calculated using the following formula: ； in, The transmittance of the illumination-invariant texture at time point t. As a correction factor, This is the ambient light compensation coefficient; S25. Based on the normalized thermal field entropy flux density and the illumination-invariant texture transmittance, the coupling state factor is further obtained. , ; Let be the coupling state factor at time node t. The reference time constant for the change in entropy flow in the thermal field is denoted as . This is the reference time constant for the change in texture transmittance.

5. The intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence as described in claim 4, characterized in that: S3 includes the following specific steps: S31, when the coupling state factor Not greater than the preset threshold At that time, it was determined to be a non-strong fire and dense smoke state, and the adaptive medium scattering compensation coefficient was used. The default value is set to 1; otherwise, it is determined to be a state of intense fire and dense smoke. S32. When the situation is determined to be a strong fire and dense smoke, the initial scattering rate is calculated based on the type of combustion medium in the detection area and the corresponding prediction formula. The calculated initial scattering rate Substituting the equations into the combustion medium mass diffusion-thermal entropy flow coupling control equations for verification, the normalized thermal field entropy flow density is obtained. The expression for the combustion medium mass diffusion-thermal entropy flow coupling control equation is as follows: ;in, For a preset empirical density based on the type of combustion medium, The preset empirical diffusion rate of the combustion medium, For divergence calculation, The scattering rate to be tested is... The thermal entropy flow-mass diffusion coupling coefficient; S33, Calculated value of normalized thermal field entropy flux density With normalized thermal field entropy flux density When the absolute value of the difference between them is greater than 0.1, the initial scattering rate... Make corrections as follows: If The corrected initial scattering rate is ,like The corrected initial scattering rate is The corrected initial scattering rate is then substituted back into the combustion medium mass diffusion-thermal entropy flow coupled control equation for verification. This process is repeated iteratively until the absolute value of the difference between the initial scattering rates of two adjacent iterations is less than [a certain value]. The iteration stops when the initial scattering rate is reached, and the final value of the initial scattering rate is output as the adaptive medium scattering compensation coefficient. .

6. The intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence as described in claim 5, characterized in that: The initial scattering rate calculation in S32, based on the combustion medium type of the detection area and matching the corresponding estimation formula, includes: when the combustion medium type of the detection area is ethylene black smoke, the estimation formula is: When the combustion medium in the detection area is cable white smoke, the prediction formula is: .

7. The intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence as described in claim 6, characterized in that: S4 includes the following specific steps: S41, Based on infrared detector target surface temperature By using the linear fitting formula: ; Obtain the baseline noise level of the infrared detection module. , where a is the noise floor fitting coefficient and b is the noise floor fitting constant; S42, Based on the reference noise floor level of the infrared detection module Normalized thermal field entropy flux density and illumination-invariant texture transmittance Construct a signal-to-noise ratio (SNR) gating function, the expression of which is: ； in, The signal-to-noise ratio gating coefficient represents the signal-to-noise ratio at time point t. The first judgment threshold is... 2 is the second judgment threshold. is the gain coefficient of the signal-to-noise ratio gating function. This is the rated reference noise floor of the infrared detector; S43, Based on normalized thermal field entropy flux density Adaptive medium scattering compensation coefficient Light-invariant texture transmittance and signal-to-noise ratio gating coefficient A smoke-compensated thermal turbulence singularity index is generated, and the expression for the smoke-compensated thermal turbulence singularity index is: ； in, The singularity index of smoke-compensated thermal turbulence at time node t is denoted by t.

8. The intelligent monitoring and judgment method for emergency response to industrial plant accidents based on artificial intelligence as described in claim 7, characterized in that: The specific content of S5 is as follows: Introduce the historical time window length w, and obtain the historical time node interval. The accident response level index is obtained by numerically integrating the squared terms of the smoke-compensated thermal turbulence singularity index sequence using the trapezoidal method. The formula for calculating the accident response level index is as follows: ;in, This represents the accident response level index at time point t. The sampling time interval for the smoke-compensated thermal turbulence singularity index sequence. , n represents the historical time interval. The number of sampling points within.

Images