Fire outbreak prediction frequency conversion inspection method, medium and fire outbreak prediction frequency conversion inspection system

Abstract

The invention provides a fire outbreak prediction frequency conversion inspection method, medium and system, and belongs to the technical field of fire fighting, and the method comprises the steps: collecting a thermal imaging image through an integrated intelligent fire extinguisher, detecting a temperature change point and a temperature abnormal point through a frame difference method, constructing and decomposing a time sequence matrix, and calculating a fire outbreak risk index matrix. The inspection strategy of the holder is dynamically adjusted according to the fire risk index, the pause time is prolonged for a high-risk area, and the pause time is shortened for a low-risk area. And when the fire risk index of a certain area exceeds a preset threshold value, determining an initial fire and triggering an alarm. And the inspection strategy of the integrated intelligent fire extinguisher can be dynamically adjusted. The inspection time is increased for high-risk areas, and the inspection frequency is reduced for low-risk areas, so that monitoring resources are effectively centralized at key parts. Once the risk index of a certain area exceeds a preset threshold value, it can be judged that initial fire occurs and an alarm is given immediately. The problem that the inspection strategy cannot be dynamically adjusted for fire risks of different areas in the prior art is solved.

Description

Technical Field

[0001] The present invention belongs to the field of fire protection technology, and in particular, relates to a fire prediction variable frequency inspection method, medium and system. Background Art

[0002] Fire accidents have always been one of the key issues of social concern. How to effectively prevent and control fire accidents is an important issue that all sectors must face. Existing fire prevention technologies mainly include firefighting and monitoring. Firefighting technology mainly responds to fire accidents that have already occurred by equipping fire-fighting equipment and establishing fire brigades; while monitoring technology focuses on real-time detection and early warning of potential fire hazards. Currently, the most widely used monitoring technologies include smoke monitoring, temperature monitoring, video monitoring, etc.

[0003] Smoke monitoring technology can quickly detect early fires by real-time detection of smoke concentration in the environment. However, this technology has certain limitations and it is difficult to accurately locate the specific location of the fire. Temperature monitoring technology analyzes the changes in ambient temperature in real time and issues an early warning when abnormal temperatures are found. This method can better determine the location of the fire, but it is not effective in detecting potential fires with slow temperature changes. Video surveillance technology uses a camera to capture the monitored area in real time and analyzes video image information to detect abnormal conditions that may cause fires. However, this technology is limited by the field of view and imaging quality, and it is difficult to fully cover the monitored area.

[0004] In addition, most existing monitoring technologies use fixed deployments and are unable to dynamically adjust inspection strategies based on fire risks in different areas. Summary of the invention

[0005] In view of this, the present invention provides a fire prediction variable frequency inspection method, medium and system, which can solve the technical problem that fire inspections in the prior art often use a fixed speed or frequency and cannot dynamically adjust the inspection strategy according to the fire risks in different areas.

[0006] The present invention is achieved in that:

[0007] A first aspect of the present invention provides a fire prediction variable frequency inspection method, which comprises the following steps:

[0008] S10, acquiring in real time a thermal imaging image collected during the rotation of the integrated intelligent fire extinguisher; the integrated intelligent fire extinguisher has a pan / tilt, and a thermal imager is arranged on the pan / tilt for collecting thermal imaging images;

[0009] S20, using a frame difference method to perform moving target detection on adjacent thermal imaging images, and identifying temperature change points and temperature abnormality points in the thermal imaging images; the temperature change point refers to a point where the temperature change between adjacent frames exceeds a preset first threshold; the temperature abnormality point refers to a point where the temperature is higher than the sum of the ambient average temperature and a preset second threshold;

[0010] S30, combining the temperature change points and temperature abnormality points corresponding to all adjacent thermal imaging images into a temperature change point time series matrix and a temperature abnormality point time series matrix respectively;

[0011] S40, decomposing the temperature change point time series matrix by singular value decomposition to obtain a temperature change point uniform speed change matrix and a temperature change point variable speed change matrix;

[0012] S50, decomposing the temperature anomaly point time series matrix by principal component analysis to obtain a temperature anomaly point position stability matrix and a temperature anomaly point position variation matrix;

[0013] S60, calculating a fire risk index matrix using a preset fire risk matrix calculation equation according to the temperature change point speed change matrix and the temperature abnormal point position change matrix; each element in the fire risk index matrix is ​​a vector, including coordinates and a fire risk index, and the higher the fire risk index, the greater the possibility of an initial fire occurring in the area corresponding to the coordinates;

[0014] S70, dynamically adjusting the inspection strategy of the PTZ according to the calculated fire risk index matrix: for areas with high fire risk index, lengthening the corresponding pause time during the rotation process of the integrated intelligent fire extinguishing system; for areas with low fire risk index, reducing the corresponding pause time during the rotation process of the integrated intelligent fire extinguishing system;

[0015] S80: When the fire risk index of a certain area exceeds the preset initial fire determination threshold, it is determined that an initial fire has occurred in the area, the initial fire position coordinates are recorded, and an alarm is triggered.

[0016] Specifically, step S10 includes: collecting thermal imaging images of the monitored area in real time from a thermal imager set on the pan / tilt of the integrated intelligent fire extinguisher. The thermal imager can capture the surface temperature distribution information of objects in the monitored area and convert it into a thermal imaging image. During the variable frequency inspection process, the thermal imager will continuously collect real-time thermal imaging images to provide a data basis for subsequent temperature change analysis.

[0017] The specific steps of step S20 include: using the frame difference method to detect moving targets on adjacent thermal imaging images, and identifying temperature change points and temperature abnormality points in the thermal imaging images. Specifically, two temperature thresholds are first set. For the judgment of temperature change points, if the temperature difference of a certain coordinate point at adjacent moments exceeds the first threshold, the point is considered to be a temperature change point; for the judgment of temperature abnormality points, if the temperature of a certain coordinate point is higher than the current environmental average temperature plus the second threshold, the point is considered to be a temperature abnormality point. Through this step, areas with large temperature changes and areas with high temperatures can be extracted from the thermal imaging images, laying the foundation for subsequent fire risk analysis.

[0018] The specific steps of step S30 include: combining the temperature change points and temperature anomaly points corresponding to all adjacent thermal imaging images into a temperature change point time series matrix and a temperature anomaly point time series matrix. The temperature change point time series matrix records the temperature change information of each coordinate point in the time series, and the temperature anomaly point time series matrix records the absolute temperature value information of each temperature anomaly point in the time series. The purpose of this step is to organize the temperature change point and temperature anomaly point information extracted in the previous step into two two-dimensional matrices to prepare for the subsequent matrix decomposition analysis.

[0019] The specific steps of step S40 include: decomposing the temperature change point time series matrix by singular value decomposition to obtain the temperature change point uniform change matrix and the temperature change point variable speed change matrix. Singular value decomposition is a commonly used matrix decomposition method, which can decompose the temperature change point time series matrix into two components: one represents the uniform change characteristics of the temperature change points, and the other represents the variable speed change characteristics of the temperature change points. The purpose of this step is to extract two modes of temperature change from the time series to provide a basis for subsequent fire risk analysis.

[0020] The specific steps of step S50 include: decomposing the temperature anomaly point time series matrix by principal component analysis to obtain the temperature anomaly point position stability matrix and the temperature anomaly point position change matrix. Principal component analysis is a commonly used data dimension reduction and feature extraction method, which can decompose the temperature anomaly point time series matrix into two components: one represents the stability characteristics of the temperature anomaly point position, and the other represents the change characteristics of the temperature anomaly point position. The purpose of this step is to extract two patterns of position change from the time series of the temperature anomaly point to provide a basis for subsequent fire risk analysis.

[0021] The specific steps of step S60 include: according to the temperature change point speed change matrix and the temperature abnormal point position change matrix, a preset fire risk matrix calculation equation is used to calculate the fire risk index matrix. The fire risk matrix calculation equation comprehensively considers multiple factors such as the temperature change rate, the degree of temperature deviation, the temperature change trend and the second-order derivative of the temperature field, and can more comprehensively reflect the possibility of the initial fire in a certain area. Through this step, the fire risk index of each area can be calculated based on the temperature change characteristics extracted previously, providing a basis for subsequent intelligent inspection decisions.

[0022] The specific steps of step S70 include: dynamically adjusting the inspection strategy of the integrated intelligent fire extinguisher according to the calculated fire risk index matrix. Specifically, for areas with high fire risk index, increase the pause time of the integrated intelligent fire extinguisher in the area; for areas with low fire risk index, reduce the pause time of the integrated intelligent fire extinguisher in the area. Through this step, the inspection strategy of the integrated intelligent fire extinguisher can be dynamically adjusted according to the fire risk calculated in real time, so as to better focus on high-risk areas and detect early fires in time.

[0023] The specific steps of step S80 include: when the fire risk index of a certain area exceeds the preset initial fire judgment threshold, it is determined that the initial fire occurs in the area, the coordinates of the initial fire location are recorded, and an alarm is triggered. By analyzing the historical fire risk index data, a reasonable initial fire judgment threshold can be determined. Once the fire risk index of a certain area exceeds the threshold, it means that the initial fire is likely to have occurred in the area. The system will record the coordinate location information of the area, and promptly trigger an alarm and take further fire extinguishing measures. The purpose of this step is to make a final initial fire judgment on the detected high-risk areas, and to promptly alarm and respond to initial fire accidents.

[0024] The fire risk matrix calculation equation is specifically expressed as follows:

[0025] The fire risk matrix calculation equation is specifically expressed as follows:

[0026]

[0027] Where R(x,y,t) is the fire risk index of the coordinate (x,y) at time t; T(x,y,t) is the temperature value of the coordinate (x,y) at time t; T avg (t) is the average ambient temperature at time t; θ(x, y, t) is the phase angle of temperature change; α, β, γ, δ are weight coefficients; i is an imaginary unit; is the Laplace operator.

[0028] Parameter acquisition method:

[0029] T(x,y,t) is directly measured by a thermal imager.

[0030] T avg (t) is obtained by calculating the average temperature of the entire thermal imaging image.

[0031] It is calculated by the temperature difference between adjacent time frames.

[0032] θ(x,y,t) is obtained by calculating the phase of temperature change through Fourier transform.

[0033] It is obtained by calculating the second-order derivative of the temperature field using the finite difference method.

[0034] α, β, γ, and δ are obtained through machine learning methods (such as support vector machine regression) based on historical data training.

[0035] The determination formula of temperature change point:

[0036] ΔT(x,y,t)=|T(x,y,t)-T(x,y,t-1)|>λ 1 ;

[0037] The determination formula of temperature abnormal point:

[0038] T(x,y,t)-(T avg (t)+λ 2 )>0;

[0039] Where ΔT(x,y,t) is the temperature change; λ 1 is the first threshold; 2 is the second threshold.

[0040] Temperature change point timing matrix:

[0041]

[0042] Temperature anomaly point timing matrix:

[0043]

[0044] Singular value decomposition in step S40:

[0045]

[0046] In the formula, is the uniform change matrix of the temperature change point; is the speed change matrix of the temperature change point.

[0047] Principal component analysis in step S50:

[0048]

[0049] In the formula, is the temperature anomaly point position stability matrix; is the temperature anomaly point position change matrix.

[0050] The step of obtaining the first threshold value specifically includes:

[0051] Collect historical temperature change data and calculate the mean value μ of the temperature change ΔT and standard deviation σ ΔT ;

[0052] Choose a suitable adjustment factor k 1 , usually between 1.5 and 3;

[0053] Apply the formula λ 1 =μ ΔT +k 1 σ ΔT A first threshold is calculated.

[0054] The default value of the first threshold is 5°C;

[0055] The step of obtaining the second threshold value specifically includes:

[0056] Analyze historical temperature data to determine the maximum temperature T max and the average temperature T avg ;

[0057] Choose a suitable adjustment factor k 2 , usually between 0.1 and 0.3;

[0058] Apply the formula λ 2 =k 2 (T max -T avg ) calculates the second threshold.

[0059] The default value of the second threshold is 10°C;

[0060] The step of determining the initial fire threshold specifically includes:

[0061] Collect historical fire risk index data and calculate its mean μ R and standard deviation σ R ;

[0062] Choose a suitable adjustment factor k 3 , usually between 2.5 and 3.5;

[0063] Apply the formula λ 3 =μ R +k 3 σ RCalculate the initial fire determination threshold.

[0064] The default value of the initial fire determination threshold is 0.8;

[0065] For the area with high fire risk index, the corresponding pause time of the integrated intelligent fire extinguishing rotation process is lengthened; for the area with low fire risk index, the corresponding pause time of the integrated intelligent fire extinguishing rotation process is reduced, specifically:

[0066] Set the basic pause time t base , usually 2 seconds;

[0067] Calculate the average fire risk index R in the current scanning cycle avg , Maximum fire risk index R max and minimum fire risk index R min ;

[0068] For areas with high fire risk index (R(x,y,t)>R avg ), the pause duration is calculated as:

[0069]

[0070] For areas with low fire risk index (R(x,y,t)≤R avg ), the pause duration is calculated as:

[0071]

[0072] The upper limit of the pause duration is set to 5 seconds and the lower limit is set to 0.5 seconds. If the calculated result exceeds this range, the corresponding upper or lower limit value is taken.

[0073] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the above-mentioned fire prediction variable frequency inspection method.

[0074] A third aspect of the present invention provides a fire prediction variable frequency inspection system, which includes the above-mentioned computer-readable storage medium.

[0075] Compared with the prior art, the fire prediction variable frequency inspection method, medium and system provided by the present invention have the following beneficial effects:

[0076] 1. Real-time and efficient fire hazard detection. This method uses a thermal imager to collect temperature distribution information of the monitored area in real time, and uses the frame difference method and threshold judgment to accurately identify areas with abnormal temperature changes and high temperatures, providing a reliable basis for subsequent fire warnings. Compared with single temperature monitoring or video monitoring, this method can detect potential fire hazards more comprehensively and accurately.

[0077] 2. Intelligent inspection strategy optimization. This method calculates the fire risk index of each area through matrix analysis of temperature change characteristics, and dynamically adjusts the inspection strategy of the integrated intelligent fire extinguisher based on risk assessment. The inspection time is increased for high-risk areas, and the inspection frequency is reduced for low-risk areas, so as to effectively focus monitoring resources on key areas. This intelligent inspection strategy based on fire risk can greatly improve the early warning efficiency of the monitoring system.

[0078] 3. Timely and rapid initial fire detection and alarm. Once the fire risk index of a certain area exceeds the preset threshold, this method can quickly determine that the initial fire has occurred in the area, record the fire location and immediately trigger an alarm. This initial fire determination mechanism based on quantitative risk assessment can timely detect fire hazards and respond quickly, providing valuable emergency support for the fire department.

[0079] 4. Flexible and adaptable integrated design. This method uses an integrated intelligent fire extinguisher as the core execution unit, integrating thermal imaging monitoring, risk analysis, inspection decision-making and other functions. This integrated design is not only convenient for system deployment and maintenance, but also can flexibly respond to monitoring needs in different environments, improving overall applicability and reliability.

[0080] In summary, the thermal imaging fire prediction variable frequency inspection method proposed in the present invention is significantly superior to existing firefighting and monitoring technologies in terms of real-time monitoring, intelligent analysis, and rapid response. It solves the technical problem that fire inspections in the prior art often use fixed speeds or frequencies and cannot dynamically adjust inspection strategies for fire risks in different areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 A flow chart of the method provided by the present invention;

[0082] Figure 2 This is a schematic diagram of the specific structure of the integrated intelligent fire extinguisher in Example 2;

[0083] Figure 3 This is a schematic diagram of the inspection area of ​​the integrated intelligent fire extinguisher in Example 2;

[0084] Figure 4 is a thermal imaging schematic diagram of the temperature distribution inside the warehouse in Example 2;

[0085] Figure 5 is a schematic diagram of the distribution of temperature change points and temperature anomaly points in Example 2;

[0086] Figure 6 is a schematic diagram of the distribution of fire risk indexes in various areas in Example 2;

[0087] Figure 7 This is a schematic diagram of the changes in the inspection path and pause duration of the integrated intelligent fire extinguisher in Example 2. DETAILED DESCRIPTION

[0088] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0089] like Figure 1 FIG. 1 is a flow chart of a fire prediction variable frequency inspection method provided by the first aspect of the present invention. The method comprises the following steps:

[0090] S10, real-time acquisition of thermal imaging images collected during the rotation of the integrated intelligent fire extinguisher; the integrated intelligent fire extinguisher has a pan / tilt, and a thermal imager is arranged on the pan / tilt for collecting thermal imaging images;

[0091] S20, using a frame difference method to perform moving target detection on adjacent thermal imaging images, and identifying temperature change points and temperature abnormality points in the thermal imaging images; a temperature change point refers to a point where the temperature change between adjacent frames exceeds a preset first threshold; a temperature abnormality point refers to a point where the temperature is higher than the sum of the ambient average temperature and a preset second threshold;

[0092] S30, combining the temperature change points and temperature abnormality points corresponding to all adjacent thermal imaging images into a temperature change point time series matrix and a temperature abnormality point time series matrix respectively;

[0093] S40, decomposing the temperature change point time series matrix by singular value decomposition to obtain a temperature change point uniform speed change matrix and a temperature change point variable speed change matrix;

[0094] S50, decomposing the temperature anomaly point time series matrix by principal component analysis to obtain a temperature anomaly point position stability matrix and a temperature anomaly point position change matrix;

[0095] S60, calculating a fire risk index matrix using a preset fire risk matrix calculation equation according to the temperature change point speed change matrix and the temperature abnormal point position change matrix; each element in the fire risk index matrix is ​​a vector, including coordinates and a fire risk index, and the higher the fire risk index, the greater the possibility of an initial fire occurring in the area corresponding to the coordinates;

[0096] S70, dynamically adjusting the inspection strategy of the PTZ according to the calculated fire risk index matrix: for areas with high fire risk index, lengthening the corresponding pause time during the rotation of the integrated intelligent fire extinguishing system; for areas with low fire risk index, reducing the corresponding pause time during the rotation of the integrated intelligent fire extinguishing system;

[0097] S80: When the fire risk index of a certain area exceeds the preset initial fire determination threshold, it is determined that an initial fire has occurred in the area, the initial fire position coordinates are recorded, and an alarm is triggered.

[0098] The specific implementation methods of the above steps are described in detail below:

[0099] The specific implementation method of step S10 is: real-time acquisition of thermal imaging images collected during the rotation of the integrated intelligent fire extinguisher. The integrated intelligent fire extinguisher is designed to be installed with a pan-tilt head, which is equipped with a thermal imager. The thermal imager can capture the surface temperature distribution information of objects in the monitored area and convert it into a thermal image. During the variable frequency inspection process, the thermal imager will continuously collect real-time thermal imaging images to provide a data basis for subsequent temperature change analysis.

[0100] The specific implementation of step S20 is: using the frame difference method to detect moving targets on adjacent thermal imaging images, and identifying temperature change points and temperature abnormal points in the thermal imaging images. The so-called frame difference method refers to detecting the temperature change area in the image by calculating the temperature difference between two adjacent frames of thermal imaging images. Specifically, first set two temperature thresholds λ 1 and λ 2 For the determination of temperature change points, if the temperature difference between a certain coordinate (x, y) at time t and the previous time t-1 |ΔT(x, y, t)|=|T(x, y, t)-T(x, y, t-1)| is greater than the first threshold λ 1 , then the point is considered to be a temperature change point. For the judgment of temperature abnormality points, if the temperature T(x,y,t) of a certain coordinate (x,y) at time t is higher than the average ambient temperature T at the current time avg (t) plus the second threshold λ 2 , then the point is considered to be a temperature abnormality point. The purpose of this step is to extract the areas with large temperature changes and high temperature areas from the thermal imaging image, laying the foundation for the subsequent fire risk analysis. 1 and λ 2 The following method can be used to set the temperature: First, collect historical temperature change data and calculate the mean value μ of the temperature change. ΔT and standard deviation σ ΔT Then choose the appropriate adjustment coefficient k 1 (usually between 1.5 and 3), apply the formula λ 1 =μ ΔT +k 1 σ ΔT Calculate the first threshold λ 1 ; Analyze historical temperature data to determine the maximum temperature T max and the average temperature T avg , select the adjustment coefficient k 2(usually between 0.1 and 0.3), apply the formula λ 2 =k 2 (T max -T avg ) Calculate the second threshold λ 2 . 1 The default value is 5 degrees Celsius, 2 The default value is 10 degrees Celsius.

[0101] The specific implementation of step S30 is: combining the temperature change points and temperature abnormal points corresponding to all adjacent thermal imaging images into a temperature change point timing matrix and a temperature abnormal point timing matrix. change The construction process of is as follows: Assume that the resolution of the thermal imaging image is m×n pixels and the length of the captured time series is n frames, then M change is an m×n matrix, where the (i,j)th element ΔT(x i ,y i ,t j ) represents the coordinate (x i ,y i ) is the temperature change at the jth moment compared to the previous moment. The temperature anomaly point time series matrix M abnormal The construction process of is as follows: Assume that the number of temperature anomaly points is k, then M abnormal is a k×n matrix, where the (p,q)th element T(x p ,y p ,t q ) represents the absolute temperature value of the pth temperature anomaly point at the qth moment. The purpose of this step is to organize the temperature change points and temperature anomaly point information extracted in the previous step into two two-dimensional matrices to prepare for the subsequent matrix decomposition analysis.

[0102] The specific implementation of step S40 is: change Singular value decomposition is used to decompose the uniform change matrix of temperature change points and the variable change matrix of temperature change points. Singular value decomposition is a commonly used matrix decomposition method, which can decompose any m×n matrix M into M=U∑V T In the form of, U is an m×n orthogonal matrix, Σ is an m×n diagonal matrix, and V is an n×n orthogonal matrix. For the temperature change point timing matrix M change , which can be decomposed into in Represents the uniform change component of the temperature change point, The variable speed change component represents the temperature change point. The purpose of this step is to extract two modes of temperature change from the time series: one is relatively stable temperature change, and the other is drastically fluctuating temperature change, which provides a basis for subsequent fire risk analysis.

[0103] The specific implementation of step S50 is: abnormal The principal component analysis is used to decompose the temperature anomaly point position stability matrix and the temperature anomaly point position change matrix. Principal component analysis is a commonly used data dimension reduction and feature extraction method, which can represent an n-dimensional data set as several mutually orthogonal principal components. For the temperature anomaly point time series matrix M abnormal , which can be decomposed into in represents the stable component of the temperature anomaly point location, The variable component representing the location of the temperature anomaly point. The purpose of this step is to extract two modes of location change from the time series of the temperature anomaly point: one is a relatively stable location distribution, and the other is a constantly changing location distribution, which provides a basis for the subsequent fire risk analysis.

[0104] The specific implementation of step S60 is: according to the temperature change point speed change matrix and the temperature abnormal point position change matrix, a preset fire risk matrix calculation equation is used to calculate the fire risk index matrix. The specific form of the fire risk matrix calculation equation is:

[0105]

[0106] Where R(x,y,t) is the fire risk index of the coordinate (x,y) at time t; T(x,y,t) is the temperature value of the coordinate point at that moment; T avg (t) is the average ambient temperature at that moment; θ(x,y,t) is the phase angle of temperature change; α, β, γ, δ are weight coefficients. This fire risk matrix calculation formula comprehensively considers factors such as temperature change rate, temperature deviation degree, temperature change trend and second-order derivative of temperature field, and can more comprehensively reflect the possibility of initial fire in a certain area. Among them, T(x,y,t) is directly measured by thermal imager, T avg (t) is obtained by calculating the average temperature of the entire thermal imaging image, It is obtained by calculating the temperature difference of adjacent time frames, and θ(x,y,t) is obtained by calculating the phase of temperature change through Fourier transform. The second-order derivative of the temperature field is calculated by the finite difference method. The four weight coefficients α, β, γ, and δ need to be trained based on historical data through machine learning methods (such as support vector machine regression). The purpose of this step is to comprehensively calculate the fire risk index of each area based on the temperature change characteristics extracted previously, providing a basis for subsequent intelligent inspection decisions.

[0107] The specific implementation of step S70 is: dynamically adjust the inspection strategy of the PTZ according to the calculated fire risk index matrix. Specifically, for areas with high fire risk index, increase the pause time of the integrated intelligent fire extinguisher in the area; for areas with low fire risk index, reduce the pause time of the integrated intelligent fire extinguisher in the area. Set the basic pause time t base is 2 seconds, calculate the average fire risk index R in the current scanning cycle avg , Maximum fire risk index R max and minimum fire risk index R min For areas with high fire risk index (R(x,y,t)>R avg ), the pause duration is calculated as For areas with low fire risk index (R(x,y,t)≤R avg ), the pause duration is calculated as The upper limit of the pause time is set to 5 seconds and the lower limit is set to 0.5 seconds. If the calculated result exceeds this range, the corresponding upper or lower limit value is taken. The purpose of this step is to dynamically adjust the inspection strategy of the integrated intelligent fire extinguisher based on the real-time calculated fire risk situation, so as to better focus on high-risk areas and detect early fires in time.

[0108] The specific implementation of step S80 is: when the fire risk index of a certain area exceeds the preset initial fire determination threshold, the initial fire is determined to have occurred in the area, the initial fire position coordinates are recorded, and an alarm is triggered. The steps for obtaining the initial fire determination threshold are as follows: first, collect historical fire risk index data, calculate its mean μ R and standard deviation σ R Then choose the appropriate adjustment coefficient k 3 (usually between 2.5 and 3.5), apply the formula λ 3 =μ R +k 3 σ R Calculate the initial fire judgment threshold λ 3 . 3 The default value is 0.8. Once the fire risk index of a certain area exceeds λ 3, it means that the area is likely to have a fire. The system will record the coordinates of the area, trigger an alarm, and take further fire-fighting measures. The purpose of this step is to make a final judgment on the fire in the detected high-risk area, and to promptly alarm and respond to the fire accident.

[0109] In order to better understand and implement the present invention, a specific embodiment 1 of the method of the first aspect of the present invention is provided below. The specific implementation of each step of this embodiment 1 is described in detail as follows: The specific implementation of step S10 is: real-time acquisition of thermal imaging images collected during the rotation of the integrated intelligent fire extinguisher. The integrated intelligent fire extinguisher is designed to be installed with a pan-tilt head, which is equipped with a thermal imager. The thermal imager can capture the surface temperature distribution information of objects in the monitored area and convert it into a thermal imaging image. During the variable frequency inspection process, the thermal imager will continuously collect real-time thermal imaging images to provide a data basis for subsequent temperature change analysis.

[0110] The specific implementation of step S20 is: using the frame difference method to detect moving targets on adjacent thermal imaging images, and identifying temperature change points and temperature abnormal points in the thermal imaging images. The so-called frame difference method refers to detecting the temperature change area in the image by calculating the temperature difference between two adjacent frames of thermal imaging images. Specifically, first set two temperature thresholds λ 1 and λ 2 For the determination of temperature change points, if the temperature difference between a certain coordinate (x, y) at time t and the previous time t-1 |ΔT(x, y, t)|=|T(x, y, t)-T(x, y, t-1)| is greater than the first threshold λ 1 , then this point is considered to be a temperature change point, which can be expressed by the following formula:

[0111] ΔT(x,y,t)=|T(x,y,t)-T(x,y,t-1)|>λ 1 ;

[0112] For the judgment of temperature anomaly points, if the temperature T(x,y,t) of a certain coordinate (x,y) at time t is higher than the average ambient temperature T at the current time avg (t) plus the second threshold λ 2 , then the point is considered to be a temperature anomaly point, which can be expressed by the following formula:

[0113] T(x,y,t)-(T avg (t)+λ 2 )>0;

[0114] The purpose of this step is to extract areas with large temperature changes and areas with high temperature from the thermal imaging image, laying the foundation for subsequent fire risk analysis.1 and λ 2 The following method can be used to set the temperature: First, collect historical temperature change data and calculate the mean value μ of the temperature change. ΔT and standard deviation σ ΔT Then choose the appropriate adjustment coefficient k 1 (usually between 1.5 and 3), apply the formula λ 1 =μ ΔT +k 1 σ ΔT Calculate the first threshold λ 1 ; Analyze historical temperature data to determine the maximum temperature T max and the average temperature T avg , select the adjustment coefficient k 2 (usually between 0.1 and 0.3), apply the formula λ 2 =k 2 (T max -T avg ) Calculate the second threshold λ 2 . 1 The default value is 5 degrees Celsius, 2 The default value is 10 degrees Celsius.

[0115] The specific implementation of step S30 is: combining the temperature change points and temperature abnormal points corresponding to all adjacent thermal imaging images into a temperature change point timing matrix and a temperature abnormal point timing matrix. change The construction process of is as follows: Assume that the resolution of the thermal imaging image is m×n pixels and the length of the captured time series is n frames, then M change is an m×n matrix, where the (i,j)th element ΔT(x i ,y i ,t j ) represents the coordinate (x i ,y i ) The temperature change at the jth moment compared to the previous moment can be expressed by the following matrix:

[0116]

[0117] Temperature abnormal point time series matrix M abnormal The construction process of is as follows: Assume that the number of temperature anomaly points is k, then M abnormal is a k×n matrix, where the (p,q)th element T(x p ,y p ,t q ) represents the absolute temperature value of the pth temperature anomaly point at the qth moment, which can be represented by the following matrix:

[0118]

[0119] The purpose of this step is to organize the temperature change points and temperature anomaly points information extracted in the previous step into two two-dimensional matrices to prepare for the subsequent matrix decomposition analysis.

[0120] The specific implementation of step S40 is: change Singular value decomposition is used to decompose the uniform change matrix of temperature change points and the variable change matrix of temperature change points. Singular value decomposition is a commonly used matrix decomposition method, which can decompose any m×n matrix M into M=U∑V T In the form of, U is an m×m orthogonal matrix, ∑ is an m×n diagonal matrix, and V is an n×n orthogonal matrix. For the temperature change point timing matrix M change , which can be decomposed into in Represents the uniform change component of the temperature change point, The speed change component representing the temperature change point can be expressed by the mathematical formula as follows:

[0121]

[0122] The purpose of this step is to extract two patterns of temperature change from the time series: one is relatively stable temperature change, and the other is drastically fluctuating temperature change, providing a basis for subsequent fire risk analysis.

[0123] The specific implementation of step S50 is: abnormal The principal component analysis is used to decompose the temperature anomaly point position stability matrix and the temperature anomaly point position change matrix. Principal component analysis is a commonly used data dimension reduction and feature extraction method, which can represent an n-dimensional data set as several mutually orthogonal principal components. For the temperature anomaly point time series matrix M abnormal , which can be decomposed into in represents the stable component of the temperature anomaly point location, The variation component representing the location of the temperature anomaly point can be expressed by the mathematical formula as follows:

[0124]

[0125] The purpose of this step is to extract two patterns of location changes from the time series of temperature anomaly points: one is a relatively stable location distribution, and the other is a constantly changing location distribution, which provides a basis for subsequent fire risk analysis.

[0126] The specific implementation of step S60 is: according to the temperature change point speed change matrix and the temperature abnormal point position change matrix, a preset fire risk matrix calculation equation is used to calculate the fire risk index matrix. The specific form of the fire risk matrix calculation equation is:

[0127]

[0128] Where R(x,y,t) is the fire risk index of the coordinate (x,y) at time t; T(x,y,t) is the temperature value of the coordinate point at that moment; T avg (t) is the average ambient temperature at that moment; θ(x,y,t) is the phase angle of temperature change; α, β, γ, δ are weight coefficients. This fire risk matrix calculation formula comprehensively considers factors such as temperature change rate, temperature deviation degree, temperature change trend and second-order derivative of temperature field, and can more comprehensively reflect the possibility of initial fire in a certain area. Among them, T(x,y,t) is directly measured by thermal imager, T avg (t) is obtained by calculating the average temperature of the entire thermal imaging image, It is obtained by calculating the temperature difference of adjacent time frames, and θ(x,y,t) is obtained by calculating the phase of temperature change through Fourier transform. The second-order derivative of the temperature field is calculated by the finite difference method. The four weight coefficients α, β, γ, and δ need to be trained based on historical data through machine learning methods (such as support vector machine regression). The purpose of this step is to comprehensively calculate the fire risk index of each area based on the temperature change characteristics extracted previously, providing a basis for subsequent intelligent inspection decisions.

[0129] The specific implementation of step S70 is: dynamically adjust the inspection strategy of the PTZ according to the calculated fire risk index matrix. Specifically, for areas with high fire risk index, increase the pause time of the integrated intelligent fire extinguisher in the area; for areas with low fire risk index, reduce the pause time of the integrated intelligent fire extinguisher in the area. Set the basic pause time t base is 2 seconds, calculate the average fire risk index R in the current scanning cycle avg , Maximum fire risk index R max and minimum fire risk index R min For areas with high fire risk index (R(x,y,t)>R avg ), the pause duration is calculated as:

[0130]

[0131] For areas with low fire risk index (R(x,y,t)≤R avg ), the pause duration is calculated as:

[0132]

[0133] The upper limit of the pause time is set to 5 seconds and the lower limit is set to 0.5 seconds. If the calculated result exceeds this range, the corresponding upper or lower limit value is taken. The purpose of this step is to dynamically adjust the inspection strategy of the integrated intelligent fire extinguisher based on the real-time calculated fire risk situation, so as to better focus on high-risk areas and detect early fires in time.

[0134] The specific implementation of step S80 is: when the fire risk index of a certain area exceeds the preset initial fire determination threshold, the initial fire is determined to have occurred in the area, the initial fire position coordinates are recorded, and an alarm is triggered. The steps for obtaining the initial fire determination threshold are as follows: first, collect historical fire risk index data, calculate its mean μ R and standard deviation σ R Then choose the appropriate adjustment coefficient k 3 (usually between 2.5 and 3.5), apply the formula λ 3 =μ R +k 3 σ R Calculate the initial fire judgment threshold λ 3 . 3 The default value is 0.8. Once the fire risk index of a certain area exceeds λ 3 , it means that the area is likely to have a fire. The system will record the coordinates of the area, trigger an alarm, and take further fire-fighting measures. The purpose of this step is to make a final judgment on the fire in the detected high-risk area, and to promptly alarm and respond to the fire accident.

[0135] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the above-mentioned fire prediction variable frequency inspection method.

[0136] A third aspect of the present invention provides a fire prediction variable frequency inspection system, which includes the above-mentioned computer-readable storage medium.

[0137] Specifically, the principle of the present invention is to make full use of thermal imaging technology to obtain the temperature distribution information of the monitored area, and to accurately model and conduct risk assessment on these temperature change characteristics through complex mathematical analysis methods. Specifically, the method mainly includes the following key steps:

[0138] First, the thermal imager equipped on the integrated intelligent fire extinguisher will collect thermal images of the monitored area in real time. These thermal images contain rich temperature change information, which is an important data basis for subsequent identification of fire hazards.

[0139] Secondly, the frame difference method is used to detect moving targets in adjacent thermal imaging images, which can accurately locate the points with large temperature changes and high temperatures in the monitored area. These temperature change points and temperature abnormalities reflect the potential fire hazards in the local environment.

[0140] Then, these temperature change points and temperature anomaly points are organized in a matrix on the time series to obtain the temperature change point time series matrix and the temperature anomaly point time series matrix. These two matrices record the trend characteristics of temperature change and the distribution characteristics of high temperature respectively.

[0141] Next, singular value decomposition is used on the temperature change point time series matrix to decompose it into uniform change components and variable change components of the temperature change points; principal component analysis is used on the temperature anomaly point time series matrix to decompose it into stable components and variable components of the temperature anomaly point locations. These decomposition results reveal the potential laws of temperature changes and anomaly point locations.

[0142] Finally, according to the temperature change characteristics extracted above, the fire risk index of each area can be calculated through the designed fire risk matrix calculation formula. This formula comprehensively considers multiple factors such as temperature change rate, temperature deviation degree, temperature change trend and second-order derivative of temperature field, and can more comprehensively reflect the possibility of initial fire in a certain area.

[0143] Based on the calculated fire risk index matrix, this method can dynamically adjust the inspection strategy of the integrated intelligent fire extinguisher. For high-risk areas, the inspection time is increased, and for low-risk areas, the inspection frequency is reduced, so as to effectively concentrate monitoring resources on key areas. Once the risk index of a certain area exceeds the preset threshold, it can be determined that the initial fire has occurred and an alarm can be immediately issued.

[0144] In order to further better understand and implement the present invention, the following provides a specific application scenario of the present invention in combination with a specific integrated intelligent fire extinguisher. The integrated intelligent fire extinguisher used in the integrated intelligent fire extinguisher system in this embodiment 2 is as follows: Figure 2 As shown, it includes: ceiling-mounted fire extinguishing agent tank; if there is a sprinkler pipe / fire fighting pipe; this fire extinguishing agent tank is not required; patrol robot base; 90° patrol camera; built-in battery / pump / valve; sound and light alarm; 360° patrol camera; smoke sensor; various detection sensors / AI cameras; fire extinguishing nozzle; the initial position is set to 0° manually, and 0 can also be deployed as a sensor. After working, according to the actual measurement of the stepper motor and structure, the starting positions of 360° and 90° are recorded. After power-on, 360° patrol, according to the set parameters, rotate within the set patrol range, up to 360° (can be set to 0-360°), and then reciprocate. 90° patrol, the initial position is vertical to the ground, can rotate up to 90°, according to the set parameters, up to 90° (can be set to 0-90°).

[0145] Among them, the action logic of 360° inspection, 90° inspection, and thermal imaging detection is as follows: taking 360° inspection as the benchmark, each rotation changes a set angle by 90°, and this angle is projected to the inspection area to form a complete coverage, such as ceiling + wall, wall + ground, ground + deep ground. After setting the "jump degree", for each 360° inspection cycle, the 90° jump degree + 1 jump degree, for example, 3 jump degrees are set. After all the jump degrees are completed, reverse and add 1 jump degree, and so on. After the sensor structure is locked, the sensor is stationary at the set angle in each 360° inspection cycle, so it will not affect the thermal imaging response speed. When the fire point is captured, it can automatically track and control the rotation of the stepper motor, but the program should record it to find the 0 point, such as Figure 3 As shown. When the fire point is captured, the 360° stepper motor stops and enters the automatic tracking state of left and right rotation. Of course, the number of steps should also be recorded to return to 0. During the inspection process, the integrated intelligent fire extinguisher considers adopting the solution of the present invention to achieve more efficient variable frequency inspection; the specific description is as follows.

[0146] Before the intelligent monitoring system was officially put into use, the operation and maintenance personnel calibrated and tested the key parameters of the system. First, by collecting the historical temperature change data of the warehouse, the mean and standard deviation of the temperature change were calculated, and the first threshold λ was determined. 1 Specifically, the mean value of the temperature change μ ΔT is 3.2 degrees Celsius, standard deviation σ ΔT is 1.0 degrees Celsius, select the adjustment coefficient k 1 is 2, then the first threshold λ 1 The calculated result is 5.2 degrees Celsius.

[0147] Similarly, after analyzing the historical temperature data of the warehouse, the highest temperature T max The average temperature is 35 degrees Celsius. avg For 20 degrees Celsius, select the adjustment coefficient k 2 is 0.2, then the second threshold λ 2 The calculated result is 3 degrees Celsius.

[0148] In addition, the operation and maintenance personnel also collected the warehouse's past fire accident data, calculated the mean and standard deviation of the historical fire risk index, and thus determined the initial fire judgment threshold λ 3 According to statistics, the mean value of the historical fire risk index is μ R is 0.6, standard deviation σ R is 0.1, select the adjustment coefficient k 3 is 3, then the initial fire determination threshold λ 3 The calculated result is 0.9.

[0149] The above three key thresholds λ 1 , 2 and λ 3 It is set as the default parameter of the system and does not need to be adjusted frequently unless there is a special need during actual operation.

[0150] After the installation and commissioning work was completed, the integrated intelligent fire extinguishing system was officially put into use. The daily routine inspection process is as follows:

[0151] 1. The thermal imager collects thermal images inside the warehouse in real time to obtain real-time temperature distribution information.

[0152] 2. Use the frame difference method to analyze adjacent thermal imaging images and identify temperature change points and temperature abnormal points. According to the above judgment formula:

[0153] ΔT(x,y,t)=|T(x,y,t)-T(x,y,t-1)|>λ 1 ;

[0154] T(x,y,t)-(T avg (t)+λ 2 )>0;

[0155] Pixels that meet the above conditions are determined as temperature change points and temperature anomaly points.

[0156] 3. Organize all temperature change points and temperature anomaly points into two matrices in the time series: temperature change point time series matrix M change and the temperature abnormal point time series matrix M abnormal .

[0157] Temperature change point timing matrix:

[0158]

[0159] Temperature anomaly point timing matrix:

[0160]

[0161] 4. Temperature change point timing matrix M change Perform singular value decomposition to obtain the uniform change component and variable speed change component of the temperature change point:

[0162]

[0163] 5. For the temperature anomaly point timing matrix M abnormal By performing principal component analysis, we can obtain the stable component and variable component of the temperature anomaly point location:

[0164]

[0165] 6. According to the temperature change characteristics extracted above, the fire risk index of each area is calculated by the fire risk matrix calculation formula:

[0166]

[0167] Among them, the weight coefficients trained by machine learning methods are: α=0.4, β=0.3, γ=0.2, and δ=0.1.

[0168] 7. According to the calculated fire risk index matrix, dynamically adjust the inspection strategy of the integrated intelligent fire extinguisher. For areas where the fire risk index is higher than the average, increase the inspection time; for areas where the fire risk index is lower than the average, reduce the inspection time. The specific calculation method is as follows:

[0169] Assume basic pause time t base = 2 seconds. During a certain inspection, the calculated parameters are: R avg =0.5, R max =0.8, R min =0.2. Then the actual pause time of each area is calculated as follows:

[0170] For high risk areas (R(x,y,t)>0.5): t pause (x,y)=2(1+)=2(1+2);

[0171] For low risk areas (R(x,y,t)≤0.5):

[0172]

[0173] The upper limit of the pause duration is set to 5 seconds and the lower limit is set to 0.5 seconds.

[0174] 8. When the fire risk index R(x,y,t) of a certain area exceeds the initial fire judgment threshold λ 3 =0.9, the area is immediately judged to have a primary fire, the coordinates of the primary fire are recorded, and an alarm is triggered.

[0175] Below, some charts are used to visually demonstrate how the system operates in a warehouse environment.

[0176] First, the changes in temperature distribution inside the warehouse during a certain inspection were plotted. Figure 4 The thermal imaging images at three preset time points (t1, t2, and t3) are displayed, and it can be clearly seen that there are large differences in the temperature distribution inside the warehouse.

[0177] Figure 5The distribution of temperature change points and temperature anomaly points during a certain inspection process is shown. As can be seen from the figure, the temperature change points are mainly concentrated in some local areas of the warehouse, while the temperature anomaly points are relatively scattered.

[0178] Figure 6 The distribution of fire risk index in each area during a certain inspection process is shown. From the figure, it can be seen that the northwest corner and southeast corner of the warehouse have a higher fire risk, while the risk in the middle area is relatively low.

[0179] at last, Figure 7 The changes in the inspection path and pause time of the integrated intelligent fire extinguisher after the dynamic inspection strategy is adopted are shown. As can be seen from the figure, the system focuses more inspection time on high-risk areas, while reducing the inspection time for low-risk areas accordingly.

[0180] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A fire prediction variable frequency inspection method, It is characterized in that The following steps are involved: S10, acquiring in real time a thermal imaging image collected during the rotation of the integrated intelligent fire extinguisher; the integrated intelligent fire extinguisher has a pan / tilt, and a thermal imager is arranged on the pan / tilt for collecting thermal imaging images; S20, using a frame difference method to perform moving target detection on adjacent thermal imaging images, and identifying temperature change points and temperature abnormality points in the thermal imaging images; the temperature change point refers to a point where the temperature change between adjacent frames exceeds a preset first threshold; the temperature abnormality point refers to a point where the temperature is higher than the sum of the ambient average temperature and a preset second threshold; S30, combining the temperature change points and temperature abnormality points corresponding to all adjacent thermal imaging images into a temperature change point time series matrix and a temperature abnormality point time series matrix respectively; S40, decomposing the temperature change point time series matrix by singular value decomposition to obtain a temperature change point uniform speed change matrix and a temperature change point variable speed change matrix; S50, decomposing the temperature anomaly point time series matrix by principal component analysis to obtain a temperature anomaly point position stability matrix and a temperature anomaly point position variation matrix; S60, calculating a fire risk index matrix using a preset fire risk matrix calculation equation according to the temperature change point speed change matrix and the temperature abnormal point position change matrix; Each element in the fire risk index matrix is ​​a vector, including coordinates and a fire risk index. The higher the fire risk index, the greater the possibility of an initial fire occurring in the area corresponding to the coordinates. S70, dynamically adjusting the inspection strategy of the PTZ according to the calculated fire risk index matrix: for areas with high fire risk index, lengthening the corresponding pause time during the rotation process of the integrated intelligent fire extinguishing system; for areas with low fire risk index, reducing the corresponding pause time during the rotation process of the integrated intelligent fire extinguishing system; S80: When the fire risk index of a certain area exceeds the preset initial fire determination threshold, it is determined that an initial fire has occurred in the area, the initial fire position coordinates are recorded, and an alarm is triggered.

2. A fire prediction variable frequency inspection method according to claim 1, It is characterized in that The fire risk matrix calculation equation is specifically expressed as follows: Where R(x, y, t) is the fire risk index of the coordinate (x, y) at time t; T(x, y, t) is the temperature value of the coordinate (x, y) at time t; T avg (t) is the average ambient temperature at time t; θ(x, y, t) is the phase angle of temperature change; α, β, γ, δ are weight coefficients; i is an imaginary unit; is the Laplace operator.

3. A fire prediction variable frequency inspection method according to claim 2, It is characterized in that The determination formula of the temperature change point is as follows: ΔT(x,y,t)=|T(x,y,t)-T(x,y,t-1)|>λ 1 ; The determination formula of temperature abnormal point: T(x,y,t)-(T avg (t)+λ 2 )>0; Where ΔT(x, y, t) is the temperature change; λ 1 is the first threshold; 2 is the second threshold.

4. A fire prediction variable frequency inspection method according to claim 3, It is characterized in that The temperature change point timing matrix is ​​expressed as follows: The temperature anomaly point timing matrix is ​​expressed as follows:

5. A fire prediction variable frequency inspection method according to claim 4, It is characterized in that The specific expression of singular value decomposition is as follows: In the formula, is the uniform change matrix of the temperature change point; is the speed change matrix of the temperature change point.

6. A fire prediction variable frequency inspection method according to claim 5, It is characterized in that The step of obtaining the first threshold value specifically includes: Collect historical temperature change data and calculate the mean value μ of the temperature change ΔT and standard deviation σ ΔT ; Choose a suitable adjustment factor k 1 ; Apply the formula λ 1 =μ ΔT +k 1 σ ΔT A first threshold is calculated.

7. A fire prediction variable frequency inspection method according to claim 6, It is characterized in that The step of obtaining the second threshold value specifically includes: Analyze historical temperature data to determine the maximum temperature T max and the average temperature T avg ; Choose a suitable adjustment factor k 2 ; Apply the formula λ 2 =k 2 (T max -T avg ) calculates the second threshold.

8. A fire prediction variable frequency inspection method according to claim 7, It is characterized in that The step of determining the initial fire threshold specifically includes: Collect historical fire risk index data and calculate its mean μ R and standard deviation σ R ; Choose a suitable adjustment factor k 3 ; Apply the formula λ 3 =μ R +k 3 σ R Calculate the initial fire determination threshold.

9. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores program instructions, and when the program instructions are executed in a computer, they are used to execute a fire prediction variable frequency inspection method according to any one of claims 1 to 8.

10. A fire prediction variable frequency inspection system, It is characterized in that A computer-readable storage medium comprising the computer-readable storage medium of claim 9.

Images