An edge computing building component fire resistance prediction system of multi-core heterogeneous architecture

Abstract

The present application relates to a fire-resistant prediction system, in particular to a multi-core heterogeneous architecture edge computing building component fire-resistant prediction system.The multi-core heterogeneous architecture edge computing building component fire-resistant prediction system comprises a data acquisition module capable of acquiring combustion data; an edge computing node, the combustion data acquired by the data acquisition module can be transmitted to the edge computing node through an intelligent gateway, and the edge computing node can generate a combustion characteristic curve according to the combustion data.The system adopts edge computing technology, and the advantage lies in faster response efficiency.Because the data does not need to be uploaded to the cloud or the remote server, the time for data uploading and waiting for instruction feedback is saved, and the Internet of Things system based on edge computing can respond to emergency situations more quickly and independently intervene without human intervention.Therefore, it is also very suitable for fire alarm, earthquake warning, security alarm and other applications with high time efficiency requirements.

Description

Technical Field

[0001] This invention relates to a fire resistance prediction system for building components, and more particularly to a multi-core heterogeneous architecture edge computing fire resistance prediction system for building components. Background Technology

[0002] The fire protection industry plays a vital role in national security and safety. With rapid social and economic development, the fire protection industry has become one of the key sectors determining national security and prosperity. In 2023, China's fire protection industry will accelerate the research and development and continuous optimization of intelligent fire protection technologies, promoting their widespread application in safety protection. In the future, intelligent alarms, intelligent monitoring, and intelligent flood alarms will become key areas of research and development in the fire protection industry. Fire is one of the most frequent and widespread major disasters threatening public safety and social development in recent years. Fires are easy to occur and difficult to control. To better protect people's lives and health and ensure the safety of various buildings nationwide, the reproduction, identification, and prediction of fires are crucial for fire prevention and saving lives and property.

[0003] Existing fire monitoring and early warning systems primarily employ a centralized data processing model, transmitting large amounts of data from the fire scene to a remote data center for processing and analysis. This model not only suffers from high data transmission latency and bandwidth limitations but also struggles to meet the real-time requirements of fire emergency response. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a fire resistance prediction system for edge computing building components with a multi-core heterogeneous architecture.

[0005] This invention is achieved through the following technical solution:

[0006] A fire resistance prediction system for edge computing building components with a multi-core heterogeneous architecture, the fire resistance prediction system comprising:

[0007] The data acquisition module is capable of acquiring combustion data;

[0008] The combustion data acquired by the data acquisition module can be transmitted to the edge computing node through the smart gateway, and the edge computing node can generate combustion characteristic curves based on the combustion data.

[0009] Preferably, the data acquisition module includes at least one of a temperature detection device, a smoke detection device, a gas detection device, an image detection device, an audio detection device, a light detection device, a pressure detection device, and an acceleration detection device.

[0010] Preferably, the edge computing node includes:

[0011] The data processing module is capable of processing the combustion data and extracting features from the processed combustion data.

[0012] The prediction module is able to obtain the combustion characteristic curve of the burning object based on the features.

[0013] Preferably, the data processing module includes:

[0014] The preprocessing submodule is capable of preprocessing the combustion data acquired by the data acquisition module;

[0015] The fusion processing submodule is capable of fusing the preprocessed combustion data and extracting the features from the fused combustion data.

[0016] Preferably, the data processing module further includes:

[0017] The storage and transmission submodule is capable of storing the combustion data and transmitting the combustion data to the preprocessing submodule and the fusion processing submodule.

[0018] Preferably, the preprocessing submodule and the storage and transmission submodule are deployed in an FPGA processing unit within a multi-core heterogeneous architecture; and / or

[0019] The fusion processing submodule is deployed in the DSP processing unit of a multi-core heterogeneous architecture.

[0020] Preferably, the prediction module includes a prediction model, which comprises a fuzzy control system and a neural network model. The fuzzy control system can control combustion parameters, and the neural network model can obtain the combustion characteristic curve of the combustion object based on the features. The neural network model can also control the parameters of the fuzzy control system; and / or

[0021] The prediction model is deployed on an ARM processing unit in a multi-core heterogeneous architecture.

[0022] Preferably, the fire resistance prediction system further includes a visualization module, which can display the combustion data acquired by the data acquisition module.

[0023] Preferably, the control method of the fuzzy control system includes the following steps:

[0024] Collect combustion data, which includes multiple input variables;

[0025] The combustion data is preprocessed;

[0026] Based on the preprocessed combustion data, multiple membership functions of the input variable are set, and the membership degree of the input variable in the membership functions is calculated.

[0027] Based on the membership degree, obtain the corresponding rule membership degree;

[0028] Aggregate all the membership degrees of the aforementioned rules into an output fuzzy set;

[0029] The control signal is obtained based on the output fuzzy set, and the control signal is used to adjust the control parameters of combustion.

[0030] The beneficial effects of this invention are: the data acquisition module can acquire combustion data from the combustion site and transmit it to the edge computing node through a smart gateway. The edge computing node can then generate a combustion characteristic curve based on the combustion data. This system employs edge computing technology, which offers the advantage of faster response efficiency. Since there is no need to upload data to the cloud or remote server, the time spent on data upload and waiting for command feedback is eliminated. This allows the edge computing-based IoT system to respond more quickly to emergencies and intervene autonomously without human intervention. Therefore, it is also highly suitable for applications with high timeliness requirements, such as fire alarms, earthquake early warning, and security alarms.

[0031] Furthermore, the fire prediction system is deployed on a multi-core heterogeneous architecture. A multi-core heterogeneous architecture integrates multiple types of processor cores within a single computing system. This architecture combines the advantages of ARM processors with those of DSP and FPGA processors, enabling the system to efficiently handle complex computational tasks and real-time data analysis. ARM processors are known for their low power consumption and high performance, making them suitable for handling complex control tasks and running operating systems. DSP processors can quickly execute tasks such as matrix operations and excel at real-time signal processing and data analysis. FPGAs can implement highly parallel computing tasks, a parallelism particularly important for applications requiring high-speed processing, such as deep learning, image processing, and signal processing. Fire prediction and judgment are based on a large amount of different types of combustion data, and the processing units designed with a multi-core heterogeneous architecture significantly improve the system's real-time performance, flexibility, and reliability. By combining the advantages of different processors, the system can efficiently handle complex computational tasks and real-time data analysis, providing strong technical support for fire early warning and emergency response.

[0032] Furthermore, the system's data acquisition module includes various data acquisition devices such as temperature detectors, smoke detectors, gas detectors, image detectors, audio detectors, light detectors, pressure detectors, and acceleration detectors, enabling the acquisition of multiple data points from the combustion site, i.e., multimodal data. Even further, the data processing module can fuse multimodal data to achieve the reproduction of combustion in different combustion furnace components and improve judgment accuracy.

[0033] Furthermore, the prediction module includes a prediction model comprising a neural network model and a fuzzy control system. The neural network model can acquire the combustion characteristic curve of the burning object based on its features, enabling the prediction of component combustion characteristics and the quantification and determination of combustion results, thereby improving fire response efficiency and accuracy and reducing fire losses. The fuzzy control system can control combustion parameters during combustion experiments as needed, ensuring that the data from the combustion experiments meet the temperature rise requirements of different control curves and adapting to different types of combined furnaces, thus improving the control accuracy and robustness of the temperature combustion curve. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the fire resistance prediction system for edge computing building components with a multi-core heterogeneous architecture according to the present invention.

[0035] Figure 2 This is a schematic diagram of the multimodal data acquisition module of the present invention.

[0036] Figure 3 This is a schematic diagram of the structure of the multimodal data fusion and processing module of the present invention.

[0037] Figure 4 This is a schematic diagram of the multi-core heterogeneous architecture fire resistance prediction generation module of the present invention. Detailed Implementation

[0038] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and preferred embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0039] In the description of the invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention.

[0040] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "setting," and "connection" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection, an indirect connection through an intermediate medium, or a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0041] This invention provides a fire resistance prediction system using edge computing, comprising a data acquisition module and edge computing nodes. The data acquisition module acquires combustion data from the combustion site, which is then transmitted to the edge computing nodes via a smart gateway. The edge computing nodes generate combustion characteristic curves based on the combustion data. This fire resistance prediction system utilizes edge computing technology to achieve real-time processing and rapid transmission of combustion data, enabling efficient, low-latency, and high-precision fire resistance prediction. It avoids data transmission delays and network bandwidth limitations, thereby improving the safety and reliability of building components.

[0042] The data acquisition module can collect multimodal data from the combustion site. The module includes temperature, smoke, gas, image, audio, light, pressure, and acceleration sensors. Specifically, the temperature sensors monitor the temperature and its trends in real time to obtain temperature curve data. The smoke sensors detect the concentration and distribution of smoke in the air at the combustion site. The gas sensors detect the concentration of harmful and combustible gases to determine the combustion status. The image sensors capture visual and thermal images of the combustion site. The audio sensors collect ambient sound and vibration information and detect combustion-related audio signals. The light sensors detect the ambient light intensity and the degree of light obstruction by smoke to analyze the smoke. The pressure sensors measure gas pressure changes to identify pressure fluctuations caused by combustion. The acceleration detection component is specifically an acceleration sensor, used to detect the structural vibration trend of building components caused by combustion.

[0043] The edge computing node is equipped with a data processing module and a prediction module.

[0044] The data processing module can process the combustion data and extract features, i.e., multimodal feature data, from the processed combustion data. The data processing module includes a preprocessing submodule, a fusion processing submodule, and a storage and transmission submodule.

[0045] The storage and transmission submodule can store the combustion data collected by the data acquisition module and transmit the combustion data to the preprocessing submodule and the fusion processing submodule.

[0046] The preprocessing submodule performs preprocessing on the combustion data acquired by the data acquisition module. Preprocessing includes data cleaning, data calibration, and data format conversion. Data cleaning primarily removes noise and invalid values, such as missing and outliers. Data calibration involves calibrating data from different sensors to ensure consistency and accuracy. Data format conversion transforms data from different formats into a unified format for easier processing and analysis. The preprocessing submodule is deployed within an FPGA processing unit in a multi-core heterogeneous architecture.

[0047] The fusion processing submodule can fuse pre-processed combustion data and extract features from it. Specifically, the fusion processing submodule uses data fusion algorithms to combine multiple combustion data sets to improve data accuracy and reliability. Data fusion algorithms include, but are not limited to, Kalman filtering and particle filtering. Features can be extracted from the fused data for analysis. The fusion processing submodule is deployed in a DSP processing unit within a multi-core heterogeneous architecture.

[0048] Furthermore, the preprocessing submodule and the storage and transmission submodule are deployed in the FPGA processing unit of the multi-core heterogeneous architecture, while the fusion processing submodule is deployed in the DSP processing unit of the multi-core heterogeneous architecture.

[0049] The prediction module contains a prediction model deployed on an ARM processing unit within a multi-core heterogeneous architecture. This prediction model comprises a fuzzy control system and a neural network model. The neural network model is trained using deep learning based on a large amount of historical combustion data. This neural network model can obtain the combustion characteristic curve of the combustion object based on features extracted by the data processing module. Furthermore, after each combustion prediction, the combustion data is saved as training data to adjust and modify the neural network model.

[0050] The fuzzy control system can control combustion parameters to improve the control accuracy of the combustion characteristic curve and the robustness of the fire resistance prediction system. The fire resistance prediction generation module introduces a combination of fuzzy control and neural network models. The fuzzy control system can handle uncertainties and nonlinear problems, improving the system's adaptability and robustness. Furthermore, this fire resistance prediction system can set various temperature rise control curves, such as linear rise, stepped rise, and exponential rise. The fuzzy control system can control combustion parameters to ensure that the combustion characteristic curve of the combustion experiment conforms to the expected curve.

[0051] The specific control method of the fuzzy control system includes the following steps:

[0052] S1: Collect combustion data. In this embodiment, combustion data is collected through a data acquisition module.

[0053] S2: Preprocess the combustion data. In this embodiment, the combustion data is preprocessed by the data processing module.

[0054] S3: Based on the preprocessed combustion data, set the membership functions for the input variables and the output variables, and calculate the membership degree of the input variables in the membership functions. In this embodiment, the input variables include temperature, smoke concentration, and gas composition, and the output variables include fuel supply and damper opening. The membership functions for the input variables are:

[0055] Temperature membership function:

[0056]

[0057]

[0058]

[0059] In the above formula, T is the temperature value of the input variable, μ low For temperature to belong to a lower membership degree, μ medium μ represents the membership degree of temperature in the membership set. high Temperature represents a high degree of membership.

[0060] Smoke concentration membership function:

[0061]

[0062]

[0063]

[0064] In the above formula, S is the smoke concentration value of the input variable, μ low For concentrations to belong to a lower membership degree, μ medium μ represents the membership degree of the concentration in the set. high The concentration belongs to a higher degree of membership.

[0065] Gas composition membership function:

[0066]

[0067]

[0068]

[0069] In the above formula, G is the gas composition value of the input variable, μ low For gas content to belong to a low membership degree, μ medium The degree of membership of the gas content to which it belongs, μ high The gas content belongs to a high degree of membership.

[0070] The membership function of the output variable is:

[0071] Fuel supply membership function:

[0072]

[0073]

[0074]

[0075] In the above formula, F represents the output variable, specifically the fuel supply amount, μ. low For fuel supply to belong to a low membership degree, μ medium μ represents the membership degree of the fuel supply in the equation. high The fuel supply belongs to a high degree of membership.

[0076] Damper opening membership function:

[0077]

[0078]

[0079]

[0080] In the above formula, A represents the output variable, the fuel supply amount, and μ low For the damper opening to belong to a low membership degree, μ medium μ represents the membership degree of the damper opening in the set. high The damper opening degree belongs to a higher membership degree.

[0081] In a specific implementation, if the input variable temperature T is 40 degrees Celsius, then this temperature belongs to the lowest membership level.

[0082] The membership degree of this temperature to the intermediate temperature range is:

[0083] The membership degree of this temperature to higher temperatures is:

[0084] S4: Obtain the corresponding rule membership degree based on the membership degree. In this embodiment, an example of a rule for the fuzzy control system is as follows:

[0085] a) If the temperature is low, the smoke concentration is low, and the gas composition is low, then the fuel supply is low and the damper opening is small.

[0086] b) If the temperature is low, the smoke concentration is low, and the gas composition is medium, then the fuel supply is medium and the damper opening is medium.

[0087] c) If the temperature is low, the smoke concentration is low, and the gas composition is high, then the fuel supply is high and the damper opening is large.

[0088] d) If the temperature is medium, the smoke concentration is medium, and the gas composition is medium, then the fuel supply is medium, and the damper opening is medium.

[0089] e) If the temperature is high, the smoke concentration is high, and the gas composition is high, then the fuel supply is high and the damper opening is large.

[0090] Other rules can be derived based on the above examples and actual situations.

[0091] After calculating the membership degree of each input variable in S3, the rule membership degree is obtained according to the above fuzzy rules. In a specific embodiment, if the input variables are temperature 40℃, smoke concentration 20ppm, and gas composition 10ppm, then in S3, the membership degree of temperature (belonging to a low level) is calculated to be 0.86, the membership degree of smoke concentration (belonging to a low level) is 0.8, and the membership degree of gas composition (belonging to a low level) is 0.6. In S4, the rule membership degree for fuzzy rule a) is calculated to be min(0.86, 0.80, 0.60) = 0.6, and the membership degree of the output fuel supply (belonging to a low level) is 0.6, and the membership degree of the damper opening (belonging to a small level) is 0.6.

[0092] S5: Aggregate all rule membership degrees into an output fuzzy set. Specifically, in S4, the rule membership degrees of all fuzzy rules are calculated, and then in S5, a weighted average method is applied to all rule membership degrees to obtain the output fuzzy set.

[0093] S6: Obtain the control signal based on the output fuzzy set. The control signal is used to adjust the combustion control parameters. Specifically, the output fuzzy set obtained in S5 is converted into a control signal using the centroid method. The centroid method calculation formula is:

[0094]

[0095] Among them, F crisp It is the explicit output value after defuzzification, μ(F) i ) is the membership function of the output variable.

[0096] Furthermore, the neural network model can learn historical data from the fuzzy control system to adjust the parameters of the fuzzy control system based on actual combustion parameters, such as adjusting the membership function in S3 and the fuzzy rules in S4.

[0097] The visualization module displays combustion data acquired by the data acquisition module. Based on the pyQT platform, it enables real-time distributed furnace monitoring, detection signal curve monitoring, control signal input, alarm recording, and automated experimental operation. It features multi-condition switching display and detection capabilities, and the display interface and operational functions can be expanded according to different experimental needs. Specifically, leveraging the powerful multi-node synchronous detection capabilities of the pyQT platform, the visualization module utilizes a multi-modal data acquisition module to monitor the furnace status distributed across various locations in real time and transmit the data to the central control system. The visualization module provides curve monitoring for various detection signals, supports historical data playback, and can simultaneously display curves of multiple detection signals. It supports signal comparison analysis, and the intuitive graphical display helps operators better understand the changing trends during the experiment. Multi-signal comparison assists operators in performing multi-dimensional data analysis and decision-making. Through the pyQT platform, the visualization module provides flexible control signal input functionality, ensuring automated and precise control of the experimental process. Operators can manually adjust control signals through the user interface, enabling real-time intervention and adjustment of the experimental process, ensuring the flexibility and safety of the experiment.

[0098] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A fire resistance prediction system for building components based on a multi-core heterogeneous architecture edge computing, characterized in that, include: The data acquisition module is capable of acquiring combustion data; The combustion data acquired by the data acquisition module can be transmitted to the edge computing node through the smart gateway, and the edge computing node can generate a combustion characteristic curve based on the combustion data. The edge computing node includes: The data processing module is capable of processing the combustion data and extracting features from the processed combustion data. The prediction module is able to obtain the combustion characteristic curve of the burning object based on the features; The prediction module is equipped with a prediction model, which includes a fuzzy control system and a neural network model. The fuzzy control system can control combustion parameters, and the neural network model can obtain the combustion characteristic curve of the combustion object based on the features. The neural network model can also control the parameters of the fuzzy control system. The prediction model is deployed in an ARM processing unit in a multi-core heterogeneous architecture; The control method of the fuzzy control system includes the following steps: Collect combustion data, which includes multiple input variables; The combustion data is preprocessed; Based on the preprocessed combustion data, multiple membership functions of the input variable are set, and the membership degree of the input variable in the membership functions is calculated. Based on the membership degree, obtain the corresponding rule membership degree; Aggregate all the membership degrees of the aforementioned rules into an output fuzzy set; The control signal is obtained based on the output fuzzy set, and the control signal is used to adjust the control parameters of combustion. The input variables include temperature, smoke concentration, and gas composition; the output variables include fuel supply and damper opening; and the membership function of the input variables is: Temperature membership function: In the above formula, T is the temperature value of the input variable. Temperature belongs to a lower membership degree. The degree of membership of temperature in the membership list. Temperature belongs to a higher degree of membership; Smoke concentration membership function: In the above formula, S is the smoke concentration value of the input variable. For concentrations belonging to a lower degree of membership, The degree of membership of the concentration is the degree of membership in the equation. The concentration belongs to a higher degree of membership; Gas composition membership function: In the above formula, G represents the gas composition value of the input variable. The gas content belongs to a low degree of membership. The degree of membership of the gas content to which it belongs. The gas content belongs to a high degree of membership; Fuel supply membership function: In the above formula, F represents the fuel supply as an output variable. The fuel supply belongs to a low degree of membership. The degree of membership of the fuel supply is represented by its degree of membership. The fuel supply belongs to a high degree of membership; Damper opening membership function: In the above formula, A represents the fuel supply quantity as an output variable. The damper opening degree belongs to a low membership degree. The membership degree of the damper opening is defined in the equation. The damper opening degree belongs to a higher membership degree. If the temperature T of the input variable is 40 degrees Celsius, then the temperature belongs to the lowest membership degree. ; The degree of membership of the temperature to the intermediate temperature is: ; The membership degree of the temperature belonging to higher temperatures is: .

2. The fire resistance prediction system according to claim 1, characterized in that, The data acquisition module includes at least one of the following: temperature detection device, smoke detection device, gas detection device, image detection device, audio detection device, light detection device, pressure detection device, and acceleration detection device.

3. The fire resistance prediction system according to claim 1, characterized in that, The data processing module includes: The preprocessing submodule is capable of preprocessing the combustion data acquired by the data acquisition module; The fusion processing submodule is capable of fusing the preprocessed combustion data and extracting the features from the fused combustion data.

4. The fire resistance prediction system according to claim 3, characterized in that, The data processing module further includes: The storage and transmission submodule is capable of storing the combustion data and transmitting the combustion data to the preprocessing submodule and the fusion processing submodule.

5. The fire resistance prediction system according to claim 4, characterized in that, The preprocessing submodule and the storage and transmission submodule are deployed in an FPGA processing unit within a multi-core heterogeneous architecture; and / or The fusion processing submodule is deployed in the DSP processing unit of a multi-core heterogeneous architecture.

6. The fire resistance prediction system according to claim 1, characterized in that, The fire resistance prediction system also includes a visualization module, which can display the combustion data acquired by the data acquisition module.

Images