Fire cooperative detection method and system based on multi-modal fusion

Abstract

The invention relates to the field of fire cooperative detection, in particular to a fire cooperative detection method and system based on multi-modal fusion, and the method comprises the steps: simulating the smoke diffusion and temperature change process of a fire in a machine room through a fire simulation method, determining the distribution region of smoke concentration and temperature rise, and constructing a smoke detector network and a temperature detector network; smog concentration data and temperature data of the detection points are collected and transmitted to a data processing center for data preprocessing, a comprehensive fire feature matrix is generated through a data fusion method, and the fire area of the fire is positioned through a cross analysis method according to the comprehensive fire feature matrix; the problem that after a machine room has a fire, due to smoke diffusion, the visualization degree is low, and the fire area cannot be accurately judged is solved.

Description

Technical Field

[0001] The present invention relates to the field of collaborative fire detection, and in particular to a collaborative fire detection method and system based on multimodal fusion. Background Art

[0002] With the rapid development of science and technology, modern computer rooms are beginning to develop in the direction of unmanned operation and remote monitoring. Since a large number of electrical equipment are arranged in the computer room, such as UPS systems, air conditioners, power distribution equipment and other power supply and power consumption equipment, most of these equipment are in 24-hour working state, long-term working state is prone to cause fatigue or aging of wires, cables or electronic components, as well as insulation materials in the equipment, resulting in short circuits or poor contact and other faults, which are easy to cause fires in the computer room.

[0003] Therefore, the demand for fire prevention and control in computer rooms is increasing, and traditional fire detection methods can no longer meet the intelligent fire detection needs of modern computer rooms and other complex places. In a complex computer room environment, smoke often fills the air after a fire occurs, resulting in a low degree of visualization in the computer room. The monitoring cannot effectively observe the fire area at this time, making it inaccurate to judge the fire area after the fire occurs, and unable to provide effective information to the monitoring center and rescue personnel, resulting in the fire not being effectively and accurately isolated and extinguished.

[0004] Therefore, in a complex computer room environment, there are still significant deficiencies in the precise location of the fire source, and a more intelligent and efficient fire collaborative detection method is urgently needed. Summary of the invention

[0005] (1) Technical issues to be resolved The purpose of the present invention is to provide a collaborative fire detection method and system based on multimodal fusion to solve the problem that after a fire occurs in a computer room, smoke fills the computer room, resulting in low visualization in the computer room and inability to accurately determine the fire area.

[0006] (2) Technical solution To achieve the above object, on the one hand, the present invention provides a fire collaborative detection method based on multimodal fusion, the method comprising: S1. Acquire the spatial structure and layout information of the computer room, and simulate the smoke diffusion and temperature change process of the fire in the computer room through a fire simulation method and record it as a simulation result, determine the distribution area of ​​the smoke concentration and temperature increase according to the simulation result, obtain the layout density and position of the smoke detector and the temperature detector according to the distribution area, and construct a smoke detector network and a temperature detector network.

[0007] S2. The smoke detector network is used to collect smoke concentration data of the detection points. The temperature detector network and the smoke detector network are collaboratively arranged to collect temperature data of the detection points. After the smoke concentration data and the temperature data are transmitted to a data processing center, data preprocessing is performed using a signal evolution analysis method based on a time window.

[0008] S3. In the data processing center, the smoke density data after data preprocessing is used to generate a comprehensive fire feature matrix through a data fusion method; and the fire ignition area of ​​the fire is located by a cross analysis method using the comprehensive fire feature matrix.

[0009] Furthermore, the method of generating a comprehensive fire characteristic matrix by using a data fusion method for the smoke density data after data preprocessing in the data processing center includes: Get test points In time Smoke density at ; The smoke concentration The smoke concentration gradient is calculated using the numerical discretization method. ; The smoke concentration gradient The calculation formula is: .

[0010] in, Indicates the detection point In time The smoke concentration is rate of change of direction; Indicates the detection point In time The smoke concentration is The rate of change of direction.

[0011] The calculation formula for the temperature rise rate is: .

[0012] in, Indicates the detection point In time Temperature at Indicates the detection point Temperature over time Variation of the temperature rise rate.

[0013] The smoke concentration , Smoke density gradient and temperature rise rate After data standardization, it is recorded as standard smoke concentration , Standard smoke concentration gradient and standard temperature rise rate , and perform data fusion to generate a comprehensive fire feature matrix , the comprehensive fire characteristic matrix for: .

[0014] in, is the absolute value of the smoke concentration gradient.

[0015] Furthermore, the method of locating the fire origin area of ​​the fire by using the comprehensive fire feature matrix through a cross analysis method includes: Calculate the comprehensive characteristic value for each detection point in the computer room according to the comprehensive fire characteristic matrix : .

[0016] in, Indicates the weight value of smoke density, The weight value representing the absolute value of the smoke density gradient, Represents the weight value of the temperature rise rate, Obtained through data analysis of historical data.

[0017] A set of detection points whose comprehensive characteristic values ​​are greater than a preset characteristic threshold is obtained; if the detection point with the highest comprehensive characteristic value in the detection point set is a single detection point, the single detection point is determined as the fire area; if the detection point with the highest comprehensive characteristic value in the detection point set is multiple detection points, they are recorded as candidate points, and the fire area is obtained by analyzing the candidate points through dynamic change trends.

[0018] Furthermore, the method of obtaining the fire area by analyzing the dynamic change trend of the candidate points includes: Obtain comprehensive feature values ​​of candidate points and construct time series data ; In the preset local time window The comprehensive characteristic value slope is calculated using the linear regression method ; The comprehensive characteristic value slope The calculation formula is: .

[0019] in, is the preset local time window length; It is the starting time point index of the preset local time window, which is used to dynamically adjust the position of the preset local time window.

[0020] The comprehensive eigenvalue slopes of the candidate points are sorted and the detection point with the largest comprehensive eigenvalue slope is selected as the fire area.

[0021] Furthermore, the method of transmitting the smoke concentration data and temperature data to the data processing center and then performing data preprocessing by a signal evolution analysis method based on a time window includes: The fire simulation experiment data of the computer room is obtained, and the average delay of the signal change of the smoke detector and the temperature detector is obtained according to the fire simulation experiment data. The size of the time window is set according to the average delay to obtain the signal analysis time window.

[0022] The smoke concentration data and temperature data of the detection point collected within the signal analysis time window are obtained, and the abnormal data in the smoke concentration data and temperature data are analyzed and eliminated through data statistical analysis methods.

[0023] Based on the same inventive concept, the present invention also provides a fire collaborative detection system based on multimodal fusion, the system comprising: a detection network construction module, a data acquisition management module, and a fire area analysis module, and the modules are sequentially connected in communication; The detection network construction module is used to obtain the spatial structure and layout information of the computer room, and simulate the smoke diffusion and temperature change process of the fire in the computer room through a fire simulation method and record it as the simulation result, determine the distribution area of ​​the smoke concentration and temperature increase according to the simulation result, obtain the layout density and position of the smoke detector and temperature detector according to the distribution area, and construct the smoke detector network and the temperature detector network.

[0024] The data acquisition management module is used to collect smoke concentration data of the detection points through the smoke detector network, and the temperature detector network and the smoke detector network are collaboratively arranged to collect temperature data of the detection points; after the smoke concentration data and temperature data are transmitted to the data processing center, the data are preprocessed through a signal evolution analysis method based on a time window.

[0025] The fire area analysis module is used to generate a comprehensive fire feature matrix by using a data fusion method for the smoke density data after data preprocessing in the data processing center; and locate the fire area of ​​the fire by using the comprehensive fire feature matrix through a cross analysis method.

[0026] Furthermore, the fire area analysis module includes the following steps: Get test points In time Smoke density at ; The smoke concentration The smoke concentration gradient is calculated using the numerical discretization method. ; The smoke concentration gradient The calculation formula is: .

[0027] in, Indicates the detection point In time The smoke concentration is rate of change of direction; Indicates the detection point In time The smoke concentration is The rate of change of direction.

[0028] The calculation formula for the temperature rise rate is: .

[0029] in, Indicates the detection point In time Temperature at Indicates the detection point Temperature over time Variation of the temperature rise rate.

[0030] The smoke concentration , Smoke density gradient and temperature rise rate After data standardization, it is recorded as standard smoke concentration , Standard smoke concentration gradient and standard temperature rise rate , and perform data fusion to generate a comprehensive fire feature matrix , the comprehensive fire characteristic matrix for: .

[0031] in, is the absolute value of the smoke concentration gradient.

[0032] Furthermore, the fire area analysis module further comprises the following steps: The fire area analysis submodule is used to calculate the comprehensive characteristic value of each detection point in the computer room according to the comprehensive fire characteristic matrix : .

[0033] in, Indicates the weight value of smoke density, The weight value representing the absolute value of the smoke density gradient, Represents the weight value of the temperature rise rate, Obtained through data analysis of historical data; A set of detection points whose comprehensive characteristic values ​​are greater than a preset characteristic threshold is obtained; if the detection point with the highest comprehensive characteristic value in the detection point set is a single detection point, the single detection point is determined as the fire area; if the detection point with the highest comprehensive characteristic value in the detection point set is multiple detection points, they are recorded as candidate points, and the fire area is obtained by analyzing the candidate points through dynamic change trends.

[0034] Furthermore, the fire area analysis submodule includes the following steps: Obtain comprehensive feature values ​​of candidate points and construct time series data .

[0035] In the preset local time window The comprehensive characteristic value slope is calculated using the linear regression method ; The comprehensive characteristic value slope The calculation formula is: .

[0036] in, is the preset local time window length; It is the starting time point index of the preset local time window, which is used to dynamically adjust the position of the preset local time window.

[0037] The comprehensive eigenvalue slopes of the candidate points are sorted and the detection point with the largest comprehensive eigenvalue slope is selected as the fire area.

[0038] Furthermore, the data acquisition management module includes: The time window analysis module is used to obtain the fire simulation experiment data of the computer room, analyze the fire simulation experiment data to obtain the average delay of the signal changes of the smoke detector and the temperature detector, and set the size of the time window according to the average delay to obtain the signal analysis time window.

[0039] The data processing module is used to obtain the smoke concentration data and temperature data of the detection point collected within the signal analysis time window, and analyze and eliminate the abnormal data in the smoke concentration data and temperature data through data statistical analysis methods.

[0040] (3) Beneficial effects Compared with the prior art, the present invention has the following beneficial effects: 1. By integrating the data from smoke concentration and temperature change detectors and comprehensively considering the multi-dimensional characteristics of fire in space and time, it is possible to capture fire signs more comprehensively and accurately locate the fire area, thereby significantly improving the accuracy of fire detection.

[0041] 2. Through dynamic trend analysis, candidate fire points are further screened to identify areas where fires actually exist, thereby greatly improving the reliability of fire detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flowchart of a fire collaborative detection method based on multimodal fusion according to Example 1 of the present invention.

[0043] Figure 2 This is a schematic diagram of the module composition of a fire collaborative detection system based on multimodal fusion according to Example 2 of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Before giving examples, it is necessary to explain the application scenarios conceived by the present invention. The present invention is a collaborative fire detection method and system based on multimodal fusion. The data center of some large cloud computing companies is located in a high-rise building in the urban area and has multiple computer rooms and server rooms. The data center is equipped with a large number of servers, network equipment, and storage devices, and data storage and processing tasks are crucial to the continued operation of the business. Due to the dense equipment and complex system, once a fire occurs, it may cause equipment damage, data loss, and even endanger personnel safety. Therefore, the data center has extremely strict requirements for the fire warning system, which requires real-time and accurate monitoring and identification of the location and development trend of the fire.

[0046] Example 1: Figure 1 As shown, this embodiment provides a fire collaborative detection method based on multimodal fusion, the method comprising: S1. Obtain the spatial structure and layout information of the computer room, and simulate the smoke diffusion and temperature change process of the fire in the computer room through a fire simulation method and record it as the simulation result. According to the simulation result, the distribution area of ​​the smoke concentration and the temperature rise is determined, and the layout density and position of the smoke detector and the temperature detector are obtained according to the distribution area, and the smoke detector network and the temperature detector network are constructed; obtain the physical structure of the computer room, including the size of the computer room, equipment layout, ventilation system and other information. This information is the basis for fire simulation and determines the air flow, smoke diffusion path and heat conduction method. For example, the area of ​​a computer room is 100 square meters, with multiple server cabinets, air conditioning ventilation systems and an emergency exit. The walls of the computer room are made of metal, the ground is paved with fireproof materials, the ceiling has air conditioning outlets, and the doors and windows are fixed. Based on the spatial structure of the computer room, fire simulation software (such as FDS, PyroSim, etc.) can be used to simulate the smoke diffusion and temperature change process after the fire occurs. The fire simulation software predicts the distribution of smoke and temperature in the computer room during the fire through fluid mechanics models and thermodynamics models. The simulation results can help identify the development trend of the fire in space and then determine the areas where the smoke concentration and temperature increase. According to the results of the fire simulation, the system will determine the areas where the smoke concentration and temperature change, especially the areas with high smoke concentration and high temperature, which are usually the "core" areas of the fire. By analyzing the simulation results, the path of the fire spread can be accurately found and which areas may reach the critical temperature first, forming high-risk areas for fire. The layout and density of smoke detectors and temperature detectors are determined according to the distribution areas of smoke concentration and temperature increase, ensuring that there are enough detectors in key areas to detect abnormal changes in smoke or temperature in time. For example, more smoke detectors are laid in areas with high smoke concentration, and more temperature detectors are laid in areas with drastic temperature changes. The layout of detectors should cover every corner of the computer room as much as possible, especially areas with poor ventilation. For example, according to the simulation results, the area near the fire source in the upper left corner of the computer room has high temperature and smoke concentration, so multiple smoke detectors and temperature detectors are laid in these areas. The layout of the detectors is determined, and the smoke detector network and temperature detector network are constructed through the wireless transmission modules of the smoke detectors and temperature detectors, for example, the network is constructed based on NB-IoT cellular mobile Internet of Things connection technology.

[0047] S2. The smoke detector network collects smoke concentration data at the detection point, and the temperature detector network and the smoke detector network are arranged in a coordinated manner to collect temperature data at the detection point; after transmitting the smoke concentration data and temperature data to the data processing center, data preprocessing is performed using a signal evolution analysis method based on a time window; coordinated arrangements are performed based on the type, sensitivity, installation location of the temperature detector and the smoke detector, as well as the layout and characteristics of the computer room to ensure that data can be synchronized in real time. The data processing center is the core part for data analysis and decision-making, where data from smoke and temperature detectors will be summarized, processed and analyzed.

[0048] S3. In the data processing center, the smoke density data after data preprocessing is used to generate a comprehensive fire feature matrix through a data fusion method; the fire starting area of ​​the fire is located by the comprehensive fire feature matrix through a cross analysis method. The data fusion method is to integrate the data obtained from the smoke detector and the temperature detector, extract valuable features and generate a comprehensive fire feature matrix. The installation position of each detector is defined as a detection point, and the effective detection area corresponding to each detection is defined as the fire starting area.

[0049] Furthermore, the method of generating a comprehensive fire characteristic matrix by using a data fusion method for the smoke density data after data preprocessing in the data processing center includes: Get test points In time Smoke density at ; The smoke concentration The smoke concentration gradient is calculated using the numerical discretization method. ; The smoke concentration gradient The calculation formula is: .

[0050] in, Indicates the detection point In time The smoke concentration is rate of change of direction; Indicates the detection point In time The smoke concentration is The rate of change of direction.

[0051] The calculation formula for the temperature rise rate is: .

[0052] in, Indicates the detection point In time Temperature at Indicates the detection point Temperature over time Variation of the temperature rise rate.

[0053] The smoke concentration , Smoke density gradient and temperature rise rate After data standardization, it is recorded as standard smoke concentration , Standard smoke concentration gradient and standard temperature rise rate , and perform data fusion to generate a comprehensive fire feature matrix , the comprehensive fire characteristic matrix for: .

[0054] in, is the absolute value of the smoke concentration gradient.

[0055] Calculated by the gradient operator, . Among them, smoke concentration is usually expressed in ppm (parts per million), and the smoke concentration gradient represents the rate of change of smoke concentration in space, reflecting the speed and direction of smoke diffusion. The temperature rise rate represents the rate of change of temperature over time, which helps to identify the fire source area, because the fire source is usually accompanied by a rapid increase in temperature. Data normalization is to ensure that the dimensions of smoke concentration, smoke concentration gradient and temperature rise rate are consistent. Minimum-maximum normalization and zero-mean normalization (Z-score normalization) can be used. Assume that the following values ​​are obtained after normalization: (1,1,t1) = 0.8 (normalized smoke concentration); (absolute value of normalized smoke concentration gradient), (normalized temperature rise rate), the comprehensive fire characteristic matrix is: .

[0056] Furthermore, the method of locating the fire origin area of ​​the fire by using the comprehensive fire feature matrix through a cross analysis method includes: Calculate the comprehensive characteristic value for each detection point in the computer room according to the comprehensive fire characteristic matrix : .

[0057] in, Indicates the weight value of smoke density, The weight value representing the absolute value of the smoke density gradient, Represents the weight value of the temperature rise rate, Obtained through data analysis of historical data.

[0058] A set of detection points whose comprehensive characteristic values ​​are greater than a preset characteristic threshold is obtained; if the detection point with the highest comprehensive characteristic value in the detection point set is a single detection point, the single detection point is determined as the fire area; if the detection point with the highest comprehensive characteristic value in the detection point set is multiple detection points, they are recorded as candidate points, and the fire area is obtained by analyzing the candidate points through dynamic change trends.

[0059] For example, multiple smoke detectors and temperature detectors are arranged in the computer room. Through fire simulation and real-time monitoring, the following data is collected:

[0060] Assume the weights are: = 0.4, = 0.3, = 0.3, calculate the comprehensive eigenvalue: P(1,1, )=0.574; P(2,1, ) =0.786; P(1,2, ) = 0.929; P(2,2, ) = 0.85; Assuming the preset feature threshold is 0.8, the detection points (1,2) and (2,2) greater than the threshold are selected as candidate points. Assuming the comprehensive feature value calculated from the data collected from the detection point (2,2) is 0.75, there is only one detection point (1,2) greater than the preset feature threshold, and the detection point (1,2) is taken as the fire area.

[0061] Furthermore, the method of obtaining the fire area by analyzing the dynamic change trend of the candidate points includes: Obtain comprehensive feature values ​​of candidate points and construct time series data .

[0062] In the preset local time window The comprehensive characteristic value slope is calculated using the linear regression method ; The comprehensive characteristic value slope The calculation formula is: .

[0063] in, is the preset local time window length, which indicates the number of time points contained in the time window, for example , it means that the time window covers 5 consecutive time points; is the starting time point index of the preset local time window, which is used to dynamically adjust the position of the preset local time window. For example, when When incrementing, the time window slides to a new starting position. Indicates the starting time point, Indicates the end time point. , .

[0064] The comprehensive eigenvalue slopes of the candidate points are sorted and the detection point with the largest comprehensive eigenvalue slope is selected as the fire area.

[0065] Assume that the detection point (1,1) is at time The comprehensive eigenvalue of The comprehensive eigenvalue of The comprehensive characteristic value is 0.86; the detection point (2,1) at time The comprehensive eigenvalue of The comprehensive eigenvalue of The comprehensive characteristic value is 0.9; the detection point (1,2) at time The comprehensive eigenvalue of The comprehensive eigenvalue of The comprehensive characteristic value is 0.9; assuming that the time window is selected ,Right now , calculate the comprehensive eigenvalue slope of each candidate point. The comprehensive eigenvalue slope of the detection point (1,1) , ; The comprehensive characteristic value slope of the detection point (2,1) ; Comprehensive characteristic value slope of detection point (1,2) . According to the sorting of the comprehensive eigenvalue slope, the detection point with the largest comprehensive eigenvalue slope is (1,2), which is taken as the fire area.

[0066] Furthermore, the method of transmitting the smoke concentration data and temperature data to the data processing center and then performing data preprocessing by a signal evolution analysis method based on a time window includes: Obtain the fire simulation experiment data of the computer room, analyze the fire simulation experiment data to obtain the average delay of the signal change of the smoke detector and the temperature detector, and set the size of the time window according to the average delay to obtain the signal analysis time window; in the fire detection system, the response of the smoke detector and the temperature detector to the fire is not instantaneous, but there is a certain signal delay (delay), which is determined by factors such as the fire development process, the smoke diffusion speed, the temperature change rate, and the response time of the detector itself. Through the fire simulation experiment, the data of the detector signal changing over time can be obtained, and the average delay of the smoke detector and the temperature detector to the fire signal can be obtained by analysis. A time window is set according to this average delay to analyze the change trend of the signal within a certain period of time. This time window is called the signal analysis time window.

[0067] Obtain the smoke density data and temperature data of the detection point collected within the signal analysis time window, and analyze and remove the abnormal data in the smoke density data and temperature data through data statistical analysis methods. The smoke density data and temperature data of the detection point collected within the signal analysis time window may contain abnormal values ​​caused by equipment failure, environmental interference or data transmission errors. Through data statistical analysis methods (such as threshold judgment, box plot analysis, Z score detection, etc.), these abnormal data can be identified and removed to ensure the accuracy of subsequent data fusion and analysis.

[0068] Embodiment 2: Based on the same inventive concept, Figure 2 As shown, this embodiment also provides a fire collaborative detection system based on multimodal fusion, the system comprising: a detection network construction module, a data acquisition management module, and a fire area analysis module, and each module is sequentially connected in communication; The detection network construction module is used to obtain the spatial structure and layout information of the computer room, and simulate the smoke diffusion and temperature change process of the fire in the computer room through a fire simulation method and record it as the simulation result, determine the distribution area of ​​the smoke concentration and temperature increase according to the simulation result, obtain the layout density and position of the smoke detector and temperature detector according to the distribution area, and construct the smoke detector network and the temperature detector network.

[0069] The data acquisition management module is used to collect smoke concentration data of the detection points through the smoke detector network, and the temperature detector network and the smoke detector network are collaboratively arranged to collect temperature data of the detection points; after the smoke concentration data and temperature data are transmitted to the data processing center, the data are preprocessed through a signal evolution analysis method based on a time window.

[0070] The fire area analysis module is used to generate a comprehensive fire feature matrix by using a data fusion method for the smoke density data after data preprocessing in the data processing center; and locate the fire area of ​​the fire by using the comprehensive fire feature matrix through a cross analysis method.

[0071] Furthermore, the fire area analysis module includes the following steps: Get test points In time Smoke density at ; The smoke concentration The smoke concentration gradient is calculated using the numerical discretization method. ; The smoke concentration gradient The calculation formula is: .

[0072] in, Indicates the detection point In time The smoke concentration is rate of change of direction; Indicates the detection point In time The smoke concentration is rate of change of direction; The calculation formula for the temperature rise rate is: .

[0073] in, Indicates the detection point In time Temperature at Indicates the detection point Temperature over time Variation of the temperature rise rate.

[0074] The smoke concentration , Smoke density gradient and temperature rise rate After data standardization, it is recorded as standard smoke concentration , Standard smoke concentration gradient and standard temperature rise rate , and perform data fusion to generate a comprehensive fire feature matrix , the comprehensive fire characteristic matrix for: .

[0075] in, is the absolute value of the smoke concentration gradient.

[0076] Furthermore, the fire area analysis module further comprises the following steps: The fire area analysis submodule is used to calculate the comprehensive characteristic value of each detection point in the computer room according to the comprehensive fire characteristic matrix : .

[0077] in, Indicates the weight value of smoke density, The weight value representing the absolute value of the smoke density gradient, Represents the weight value of the temperature rise rate, Obtained through data analysis of historical data.

[0078] A set of detection points whose comprehensive characteristic values ​​are greater than a preset characteristic threshold is obtained; if the detection point with the highest comprehensive characteristic value in the detection point set is a single detection point, the single detection point is determined as the fire area; if the detection point with the highest comprehensive characteristic value in the detection point set is multiple detection points, they are recorded as candidate points, and the fire area is obtained by analyzing the candidate points through dynamic change trends.

[0079] Furthermore, the fire area analysis submodule includes the following steps: Obtain comprehensive feature values ​​of candidate points and construct time series data .

[0080] In the preset local time window The comprehensive characteristic value slope is calculated using the linear regression method ; The comprehensive characteristic value slope The calculation formula is: .

[0081] in, is the preset local time window length; It is the starting time point index of the preset local time window, which is used to dynamically adjust the position of the preset local time window.

[0082] The comprehensive eigenvalue slopes of the candidate points are sorted and the detection point with the largest comprehensive eigenvalue slope is selected as the fire area.

[0083] Furthermore, the data acquisition management module includes: The time window analysis module is used to obtain the fire simulation experiment data of the computer room, analyze the fire simulation experiment data to obtain the average delay of the signal changes of the smoke detector and the temperature detector, and set the size of the time window according to the average delay to obtain the signal analysis time window.

[0084] The data processing module is used to obtain the smoke concentration data and temperature data of the detection point collected within the signal analysis time window, and analyze and eliminate the abnormal data in the smoke concentration data and temperature data through data statistical analysis methods.

[0085] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0086] Finally, it should be noted that: Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A fire collaborative detection method based on multimodal fusion, characterized in that: The method comprises: Acquire the spatial structure and layout information of the computer room, and simulate the smoke diffusion and temperature change process of the fire in the computer room by a fire simulation method and record it as a simulation result, determine the distribution area of ​​the smoke concentration and temperature increase according to the simulation result, obtain the layout density and position of the smoke detector and the temperature detector according to the distribution area, and construct a smoke detector network and a temperature detector network; The smoke detector network is used to collect smoke concentration data of the detection points, and the temperature detector network and the smoke detector network are collaboratively arranged to collect temperature data of the detection points; after the smoke concentration data and the temperature data are transmitted to the data processing center, data preprocessing is performed using a signal evolution analysis method based on a time window; In the data processing center, the smoke density data after data preprocessing is used to generate a comprehensive fire feature matrix through a data fusion method; the fire starting area of ​​the fire is located by the comprehensive fire feature matrix through a cross analysis method.

2. The fire collaborative detection method based on multimodal fusion according to claim 1 is characterized in that: The method of generating a comprehensive fire characteristic matrix by using a data fusion method for the smoke density data after data preprocessing in the data processing center includes: Get test points In time Smoke density at ; The smoke concentration The smoke concentration gradient is calculated using the numerical discretization method. ; The smoke concentration gradient The calculation formula is: ； in, Indicates the detection point In time The smoke concentration is rate of change of direction; Indicates the detection point In time The smoke concentration is rate of change of direction; The calculation formula for the temperature rise rate is: ； in, Indicates the detection point In time Temperature at Indicates the detection point Temperature over time Variation of the rate of temperature rise; The smoke concentration , Smoke density gradient and temperature rise rate After data standardization, it is recorded as standard smoke concentration , Standard smoke concentration gradient and standard temperature rise rate , and perform data fusion to generate a comprehensive fire feature matrix , the comprehensive fire characteristic matrix for: ； in, is the absolute value of the smoke concentration gradient.

3. The fire collaborative detection method based on multimodal fusion according to claim 2 is characterized in that: The method of locating the fire origin area of ​​the fire by using the comprehensive fire feature matrix through a cross analysis method comprises: Calculate the comprehensive characteristic value of each detection point in the computer room according to the comprehensive fire characteristic matrix : ； in, Indicates the weight value of smoke density, The weight value representing the absolute value of the smoke density gradient, represents the weight value of the temperature rise rate, Obtained through data analysis of historical data; A set of detection points whose comprehensive characteristic values ​​are greater than a preset characteristic threshold is obtained; if the detection point with the highest comprehensive characteristic value in the detection point set is a single detection point, the single detection point is determined as the fire area; if the detection point with the highest comprehensive characteristic value in the detection point set is multiple detection points, they are recorded as candidate points, and the fire area is obtained by analyzing the candidate points through dynamic change trends.

4. The fire collaborative detection method based on multimodal fusion according to claim 3 is characterized in that: The method of obtaining the fire area by analyzing the candidate points through dynamic change trends includes: Obtain comprehensive feature values ​​of candidate points and construct time series data ; In the preset local time window The comprehensive characteristic value slope is calculated using the linear regression method ; The comprehensive characteristic value slope The calculation formula is: ； in, is the preset local time window length; is the starting time point index of the preset local time window, used to dynamically adjust the position of the preset local time window; The comprehensive eigenvalue slopes of the candidate points are sorted and the detection point with the largest comprehensive eigenvalue slope is selected as the fire area.

5. The fire collaborative detection method based on multimodal fusion according to claim 1 is characterized in that: The method of transmitting the smoke concentration data and temperature data to the data processing center and then performing data preprocessing by using a signal evolution analysis method based on a time window comprises: Obtain fire simulation experiment data of the computer room, analyze the fire simulation experiment data to obtain the average delay of signal changes of the smoke detector and the temperature detector, and set the size of the time window according to the average delay to obtain the signal analysis time window; The smoke concentration data and temperature data of the detection point collected within the signal analysis time window are obtained, and the abnormal data in the smoke concentration data and temperature data are analyzed and eliminated through data statistical analysis methods.

6. The fire collaborative detection system based on multimodal fusion is characterized by: The system comprises: a detection network construction module, a data acquisition management module, and a fire area analysis module, and each module is sequentially connected in communication; A detection network construction module is used to obtain the spatial structure and layout information of the computer room, and simulate the smoke diffusion and temperature change process of the fire in the computer room by a fire simulation method and record it as a simulation result, determine the distribution area of ​​the smoke concentration and temperature increase according to the simulation result, obtain the layout density and position of the smoke detector and the temperature detector according to the distribution area, and construct a smoke detector network and a temperature detector network; A data acquisition management module is used to collect smoke concentration data of the detection points through the smoke detector network, and the temperature detector network and the smoke detector network are arranged in a coordinated manner to collect temperature data of the detection points; after transmitting the smoke concentration data and temperature data to the data processing center, the data is preprocessed by a signal evolution analysis method based on a time window; The fire area analysis module is used to generate a comprehensive fire feature matrix by using a data fusion method for the smoke density data after data preprocessing in the data processing center; and locate the fire area of ​​the fire by using the comprehensive fire feature matrix through a cross analysis method.

7. The fire collaborative detection system based on multimodal fusion according to claim 6 is characterized in that: The fire area analysis module comprises the following steps: Get test points In time Smoke density at ; The smoke concentration The smoke concentration gradient is calculated using the numerical discretization method. ; The smoke concentration gradient The calculation formula is: ； in, Indicates the detection point In time The smoke concentration is rate of change of direction; Indicates the detection point In time The smoke concentration is rate of change of direction; The calculation formula for the temperature rise rate is: ； in, Indicates the detection point In time Temperature at Indicates the detection point Temperature over time Variation of the rate of temperature rise; The smoke concentration , Smoke density gradient and temperature rise rate After data standardization, it is recorded as standard smoke concentration , Standard smoke concentration gradient and standard temperature rise rate , and perform data fusion to generate a comprehensive fire feature matrix , the comprehensive fire characteristic matrix for: ； in, is the absolute value of the smoke concentration gradient.

8. The fire collaborative detection system based on multimodal fusion according to claim 7 is characterized in that: The fire area analysis module includes a fire area analysis submodule: The fire area analysis submodule is used to calculate the comprehensive characteristic value of each detection point in the computer room according to the comprehensive fire characteristic matrix : ； in, Indicates the weight value of smoke density, The weight value representing the absolute value of the smoke density gradient, represents the weight value of the temperature rise rate, Obtained through data analysis of historical data; A set of detection points whose comprehensive characteristic values ​​are greater than a preset characteristic threshold is obtained; if the detection point with the highest comprehensive characteristic value in the detection point set is a single detection point, the single detection point is determined as the fire area; if the detection point with the highest comprehensive characteristic value in the detection point set is multiple detection points, they are recorded as candidate points, and the fire area is obtained by analyzing the candidate points through dynamic change trends.

9. The fire collaborative detection system based on multimodal fusion according to claim 8, characterized in that: The fire area analysis submodule comprises the following steps: Obtain comprehensive feature values ​​of candidate points and construct time series data ; In the preset local time window The comprehensive characteristic value slope is calculated using the linear regression method ; The comprehensive characteristic value slope The calculation formula is: ； in, is the preset local time window length; The starting time point index of the preset local time window is used to dynamically adjust the position of the preset local time window; The comprehensive eigenvalue slopes of the candidate points are sorted and the detection point with the largest comprehensive eigenvalue slope is selected as the fire area.

10. The fire collaborative detection system based on multimodal fusion according to claim 6, characterized in that: The data acquisition management module includes: A time window analysis module is used to obtain the fire simulation experiment data of the computer room, analyze the fire simulation experiment data to obtain the average delay of the signal changes of the smoke detector and the temperature detector, and set the size of the time window according to the average delay to obtain the signal analysis time window; The data processing module is used to obtain the smoke concentration data and temperature data of the detection point collected within the signal analysis time window, and analyze and eliminate the abnormal data in the smoke concentration data and temperature data through data statistical analysis methods.

Images