Low-sensitivity optimization method for fire safety situation based on multimodal perception
By preprocessing the historical safety situation data of fire protection facilities and multi-objective optimization algorithm analysis, feature parameter sets are generated to improve fire protection safety situation awareness, solving the problem of insufficient low-sensitivity situation awareness in the existing technology, and achieving the correct Effective detection of potential weak signals and improved stability of fire protection facilities.
Patent Information
- Application Number
- CN202410273486.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-11
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-03-11
AI Technical Summary
The existing technology ignores low-sensitivity safety situation awareness in urban fire safety management, making it difficult to effectively detect potential weak signals, affecting the efficient operation of fire protection facilities.
By collecting and preprocessing the historical safety situation data of fire protection facilities, a multi-objective optimization algorithm is used to generate a feature parameter set, and import it into the optimization model for learning, output the optimal solution to adjust parameters to improve perception ability, and increase the ability to low-sensitivity situation Detection capability.
It improves the detection capability of potential weak signals, enhances the operating stability and safety situation awareness of fire protection facilities.
Smart Images

Figure CN118095643B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of smart security, and in particular to a low-sensitivity optimization method for fire safety situation based on multimodal perception. Background Art
[0002] Urban fires are characterized by high frequency, destructive consequences, and complex mechanisms. In addition to causing casualties and property losses, they also affect the overall function of the city, damage the social and economic environment, and disrupt the order of work and life. Moreover, under different meteorological conditions, different population concentration conditions, or after developments and changes such as dynamic adjustments to fire protection plans, they will cause the spatiotemporal evolution and reconstruction of urban fire risks and fire safety situations.
[0003] How to identify and analyze the fire safety situation of a city and make a more reasonable, accurate and objective system evaluation is a key issue that needs to be urgently addressed under the current situation. It is an inevitable trend to improve the fire safety level of "resilient cities". Its importance and urgency are self-evident.
[0004] At present, with the establishment of smart venue security systems and the application of artificial intelligence technology, the effective operation of urban security infrastructure has been ensured. Although the current smart venues based on multimodal perception have diverse perception capabilities, they often focus on discovering early warning information of strong detection signals and ignore potential weak signal trends. This low-sensitivity security situation perception capability is inappropriate in efficient security management work. How to achieve low-sensitivity optimization of multimodal perception of fire safety situation and improve the detection capability of potential weak signals is a difficult problem that needs to be solved urgently. Summary of the invention
[0005] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.
[0006] In view of the above existing problems, the present invention is proposed.
[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: collecting historical safety situation data of the operation of fire protection facilities in smart places and preprocessing the data to obtain a sample data set;
[0008] The sample data set is divided and subjected to feature target mixed encoding processing to generate a feature parameter set;
[0009] Importing the characteristic parameter set into an optimization model constructed based on a multi-objective optimization algorithm to perform optimization learning of the objective function, and outputting the optimal solution corresponding to each characteristic parameter in the characteristic parameter set;
[0010] The optimal solutions are regarded as a solution set and sent to the smart security system in the form of instructions for operation, so as to adaptively adjust the parameters of the safety situation awareness state of the fire protection facilities during operation:
[0011] When the security situation awareness state is lower than the standard state, the optimal solution increases the awareness index of each intelligent control terminal in the fire protection facility in the form of instructions;
[0012] When the security situation awareness state is higher than the standard state, it is ignored.
[0013] As a preferred solution of the low-sensitivity optimization method for fire safety situation based on multimodal perception described in the present invention, the preprocessing includes:
[0014] Screen out repeated, useless, and irrelevant parameters to the equipment operation performance in the historical safety situation data of the fire protection facilities in the smart venue;
[0015] The coding strategy is used to transform the information parameters after screening into uniform dummy variables;
[0016] Standardizing the dummy variable by subtracting its mean value and dividing it by the standard deviation, so that different configuration values of the parameter are replaced by mathematical values to form a numerical parameter;
[0017] The set of numerical parameters is the sample data set.
[0018] As a preferred solution of the low-sensitivity optimization method for fire safety situation based on multimodal perception described in the present invention, the sample data set is divided, including:
[0019] Use deep learning strategies and four-element information description principles to write target recognition and extraction programs;
[0020] Running the target extraction program, reading the sample data set using the protocol interface, and annotating and classifying it;
[0021] After the division is completed, the protocol interface is used to output to the structure encoder.
[0022] As a preferred solution of the low-sensitivity optimization method for fire safety situation based on multimodal perception described in the present invention, the divided sample data set is converted into a two-dimensional table structure in the structure encoder and exists in the form of a string. When each column of feature target data in each row of feature target data record is a character or a number, it is directly stored in the corresponding position in the two-dimensional table in sequence;
[0023] If the feature target data is missing, it will be supplemented with 0;
[0024] The characteristic target data stored in the two-dimensional table is taken as a set, namely the characteristic parameter set.
[0025] As a preferred solution of the low-sensitivity optimization method for fire safety situation based on multimodal perception described in the present invention, the optimization model selects the primitive function as the objective function of the learning machine, and its mathematical expression is as follows:
[0026]
[0027] Where a={a1,a2,…,a n}: is the characteristic matrix composed of the amplitude-frequency characteristic vectors of the historical security situation data that affect the optimization in the sample data set, b: is the amplitude-frequency characteristic vector that affects the optimization in the sample data set, σ: is the target vector, that is, the distribution or range characteristics of the sample data set, μ: is the weight of the feature parameter set.
[0028] As a preferred solution of the low-sensitivity optimization method for fire safety situation based on multimodal perception described in the present invention, the optimization model needs to be trained during construction, including:
[0029] Initializing the penalty parameter and the target vector, and using the sample data set to train and learn the target function;
[0030] If the objective function does not meet the accuracy threshold requirement, the penalty parameter and the target vector are assigned and optimized according to the error;
[0031] Until the accuracy threshold requirement is met, the optimization model is output.
[0032] As a preferred solution of the low-sensitivity optimization method for fire safety situation based on multimodal perception described in the present invention, obtaining the optimal solution includes:
[0033] Solving the objective function, obtaining multiple groups of feasible solutions that meet the optimization conditions and comparing them to obtain a frontier solution;
[0034] The solution that satisfies the minimum perceived value of the safety situation of fire-fighting facilities in the frontier solution is extracted as the optimal solution.
[0035] Beneficial effects of the present invention: The present invention analyzes the historical safety situation data of fire-fighting facilities, takes characteristic targets as observation points, obtains the optimal solution of key features in the optimization model, and uses it as the standard for adjusting parameters to improve the ability to perceive low-sensitivity fire safety situations, improve the detection capability under signal lines, and increase the operational stability of smart fire-fighting facilities. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:
[0037] Figure 1 It is a flow chart of the low-sensitivity optimization method for fire safety situation based on multimodal perception shown in the present invention. DETAILED DESCRIPTION
[0038] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.
[0039] Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without making any creative work should fall within the scope of protection of the present invention.
[0040] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0041] Example 1
[0042] According to an embodiment of the present invention, Figure 1 The flowchart shown is a low-sensitivity optimization method for fire safety situation based on multimodal perception, which specifically includes the following steps:
[0043] S1: Collect historical safety situation data of fire protection facilities in smart places and preprocess them to obtain a sample data set. It should be noted that the preprocessing includes:
[0044] Eliminate duplicate, useless, and irrelevant parameters to equipment performance in the historical safety situation data of fire protection facilities in smart places;
[0045] The coding strategy is used to transform the information parameters after screening into uniform dummy variables;
[0046] Standardize the dummy variables by subtracting their mean and dividing by the standard deviation, so that the different configuration values of the parameters are replaced by mathematical values to form numerical parameters;
[0047] Among them, the set of numerical parameters is the sample data set.
[0048] S2: Divide the sample data set and perform feature target mixed encoding processing to generate a feature parameter set.
[0049] Among them, it should be noted that this step is to divide the sample data set, including:
[0050] Use deep learning strategies and four-element information description principles to write target recognition and extraction programs;
[0051] Run the target extraction program, use the protocol interface to read the sample data set, and annotate and classify it;
[0052] After the division is completed, it is output to the structure encoder using the protocol interface.
[0053] Furthermore, the divided sample data set is converted into a two-dimensional table structure in the structure encoder and exists in the form of a string. When each column of feature target data in each row of feature target data record is a character or a number, it is directly stored in the corresponding position in the two-dimensional table in sequence.
[0054] If the feature target data is missing, it will be supplemented with 0;
[0055] The characteristic target data stored in the two-dimensional table is taken as a set, namely, a characteristic parameter set.
[0056] S3: Import the feature parameter set into the optimization model built based on the multi-objective optimization algorithm to optimize the objective function and output the optimal solution corresponding to each feature parameter in the feature parameter set. It should be noted that in this step, the optimization model selects the primitive function as the objective function of the learning machine, and its mathematical expression is as follows:
[0057]
[0058] Where a={a1,a2,…,a n}: is the characteristic matrix composed of the amplitude-frequency characteristic vectors of the historical security situation data that affect the optimization in the sample data set, b: is the amplitude-frequency characteristic vector that affects the optimization in the sample data set, σ: is the target vector, that is, the distribution or range characteristics of the sample data set, μ: is the weight of the feature parameter set.
[0059] Furthermore, the optimization model needs to be trained during construction, including:
[0060] Initialize the penalty parameters and target vector, and use the sample data set to train and learn the target function;
[0061] If the objective function does not meet the accuracy threshold requirement, the penalty parameter and the target vector are assigned and optimized according to the error;
[0062] Until the accuracy threshold requirement is met, the optimized model is output.
[0063] Specifically, obtaining the optimal solution includes:
[0064] Solve the objective function, obtain multiple feasible solutions that meet the optimization conditions and compare them to obtain the frontier solution;
[0065] The solution that meets the minimum perception value of the safety situation of fire-fighting facilities in the frontier solution is extracted as the optimal solution.
[0066] S4: Take each optimal solution as a solution set and send it to the smart security system in the form of instructions to run it, and adaptively adjust the parameters of the safety situation awareness state of the fire protection facilities during operation:
[0067] a: When the security situation awareness status is lower than the standard status, the optimal solution is to increase the perception indicators of each intelligent control terminal in the fire protection facility in the form of instructions;
[0068] b: When the security situation awareness status is higher than the standard status, it is ignored.
[0069] Preferably, the present invention analyzes the historical safety situation data of fire-fighting facilities, takes characteristic targets as observation points, obtains the optimal solution of key features in the optimization model, and uses it as the standard for adjusting parameters to improve the awareness of low-sensitivity fire safety situations, improve the detection capability under signal lines, and increase the operational stability of smart fire-fighting facilities.
[0070] The target recognition and extraction program written by the aforementioned deep learning strategy and four-element information description principle can be implemented using existing technologies and will not be described in detail in this embodiment.
[0071] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A low-sensitivity optimization method for fire safety situation based on multimodal perception, characterized in that: include: Collect historical safety situation data of fire protection facilities in smart places and pre-process it to obtain a sample data set; The sample data set is divided and subjected to feature target mixed encoding processing to generate a feature parameter set; The sample data set is divided, including: Use deep learning strategies and four-element information description principles to write target recognition and extraction programs; Running the target recognition and extraction program, using the protocol interface to read the sample data set, and annotating and classifying it; After the division is completed, the protocol interface is used to output to the structure encoder; The divided sample data set is converted into a two-dimensional table structure in the structure encoder and exists in the form of a character string. When each column of feature target data in each row of feature target data record is a character or a number, it is directly stored in the corresponding position in the two-dimensional table in sequence; If the feature target data is missing, it will be supplemented with 0; Among them, the characteristic target data stored in the two-dimensional table is taken as a set, namely the characteristic parameter set; the characteristic parameter set is imported into the optimization model constructed based on the multi-objective optimization algorithm to perform optimization learning of the objective function, and the optimal solution corresponding to each characteristic parameter in the characteristic parameter set is output; the optimization model selects the primitive function as the objective function of the learning machine, and its mathematical expression is as follows: in, : is the characteristic matrix composed of the amplitude-frequency characteristic vectors of the historical security situation data that affect the optimization in the sample data set, : is the amplitude-frequency characteristic vector that affects the optimization in the sample data set, : is the target vector, i.e. the distribution or range characteristics of the sample data set, : is the weight of the feature parameter set; Obtaining the optimal solution includes: Solving the objective function, obtaining multiple groups of feasible solutions that meet the optimization conditions and comparing them to obtain a frontier solution; Extracting the solution that satisfies the minimum perception value of the safety situation of fire-fighting facilities from the frontier solution as the optimal solution; The optimal solutions are regarded as a solution set and sent to the smart security system in the form of instructions for operation, so as to adaptively adjust the parameters of the safety situation awareness state of the fire protection facilities during operation: When the security situation awareness state is lower than the standard state, the optimal solution increases the awareness index of each intelligent control terminal in the fire protection facility in the form of instructions; When the security situation awareness state is higher than the standard state, it is ignored.
2. The low-sensitivity optimization method for fire safety situation based on multimodal perception according to claim 1 is characterized in that: The pre-processing comprises: Screen out repeated, useless, and irrelevant parameters to the equipment operation performance in the historical safety situation data of the fire protection facilities in the smart venue; The coding strategy is used to transform the information parameters after screening into uniform dummy variables; Standardizing the dummy variable by subtracting its mean value and dividing it by the standard deviation, so that different configuration values of the parameter are replaced by mathematical values to form a numerical parameter; The set of numerical parameters is the sample data set.
3. The low-sensitivity optimization method for fire safety situation based on multimodal perception according to claim 1 is characterized in that: The optimization model needs to be trained during construction, including: Initializing the penalty parameter and the target vector, and using the sample data set to train and learn the target function; If the objective function does not meet the accuracy threshold requirement, the penalty parameter and the target vector are assigned and optimized according to the error; Until the accuracy threshold requirement is met, the optimization model is output.
