Cable tunnel fire early warning method and device, storage medium and program product

Abstract

The invention relates to a cable tunnel fire early warning method and device, a storage medium and a program product. The method comprises the following steps: acquiring current monitoring data of a cable tunnel, wherein the current monitoring data comprises current monitoring values of a plurality of preset influence factors; a target fire early warning model is determined, the target fire early warning model is obtained through training of multiple pieces of historical monitoring data, each piece of historical monitoring data comprises historical monitoring values of multiple preset influence factors, and initial model parameters of the target fire early warning model are determined through a genetic algorithm; and inputting the current monitoring data into the target fire early warning model to obtain a fire prediction result corresponding to the current monitoring data output by the target fire early warning model. By adopting the method, a fire prediction result can be accurately obtained.

Description

Technical Field

[0001] The present application relates to the field of intelligent detection technology, and in particular to a cable tunnel fire early warning method, device, storage medium and program product. Background Art

[0002] As an important part of urban power transmission lines, the safety of cable tunnels is crucial to ensuring the stable operation of urban power grids. The closed space characteristics of tunnels and potential combustibles, such as cable insulation materials, make fire risk management particularly critical. Therefore, accurate early warning of cable tunnel fire risks can greatly reduce the probability of cable tunnel fires and reduce fire losses. Summary of the invention

[0003] Based on this, it is necessary to provide a cable tunnel fire warning method, device, storage medium and program product to address the above technical problems, which can accurately obtain fire prediction results.

[0004] In a first aspect, the present application provides a cable tunnel fire early warning method, the method comprising:

[0005] Acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of multiple preset influencing factors;

[0006] Determine a target fire warning model, the target fire warning model is trained using multiple historical monitoring data, each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm;

[0007] The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0008] In one embodiment, the step of training the target fire warning model includes:

[0009] Acquire multiple historical monitoring data of the cable tunnel, and obtain labeled fire results corresponding to the multiple historical monitoring data respectively; input the multiple historical monitoring data into the fire warning model to be trained respectively, and obtain the predicted fire results corresponding to the multiple historical monitoring data respectively; for each historical monitoring data in the multiple historical monitoring data, determine the difference between the labeled fire result corresponding to the historical monitoring data and the corresponding predicted fire result, and determine the result difference corresponding to the historical monitoring data; comprehensively consider the result differences corresponding to the multiple historical monitoring data respectively, and determine the model loss value of the fire warning model; use the model loss value to adjust the model parameters of the fire warning model to obtain the target fire warning model.

[0010] In one embodiment, the step of determining the fire warning model to be trained includes:

[0011] Generate multiple initial models for the first round, and the model parameters of each initial model are randomly generated; use multiple sample monitoring data to train each initial model respectively, and obtain trained models corresponding to the multiple initial models; determine the fitness characterization value of each trained model, and the fitness characterization value is used to reflect the fitness; when the iteration stop condition is not met, select at least part of the initial models from the multiple initial models according to the fitness characterization value, and update the model parameters of at least part of the initial models to obtain multiple initial models for the next round, and return to the step of training each initial model using multiple sample monitoring data to obtain trained models corresponding to the multiple initial models; when the iteration stop condition is met, select the target model from the multiple initial models of the current round according to the fitness characterization value; determine the target model as the fire warning model to be trained.

[0012] In one embodiment, selecting at least part of the initial models from a plurality of initial models according to the fitness representation value comprises:

[0013] Comprehensively calculate the fitness representation values ​​of each of the multiple initial models to obtain a total fitness representation value; for each initial model, calculate the ratio of the initial model fitness representation value to the total fitness representation value to obtain the selection probability of the initial model; select at least some of the initial models from the multiple initial models based on the selection probability.

[0014] In one embodiment, the early warning model includes an input layer, a hidden layer, and an output layer, and the step of determining the number of nodes in the hidden layer includes:

[0015] Determine the number of nodes in the input layer to obtain the number of input nodes; determine the number of nodes in the output layer to obtain the number of output nodes; determine the number of nodes in the hidden layer based on the number of input nodes and the number of output nodes.

[0016] In one embodiment, the current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model, including:

[0017] Obtain the fire intensity correlations corresponding to a plurality of preset influencing factors; for each preset influencing factor, multiply the current monitoring value of the preset influencing factor by the corresponding fire intensity correlation to obtain an updated monitoring value of the preset influencing factor; obtain updated monitoring data based on each updated monitoring value; input the updated monitoring data into the target fire warning model to obtain a predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0018] In a second aspect, the present application also provides a cable tunnel fire warning device, the device comprising:

[0019] A data acquisition module, used to acquire current monitoring data of the cable tunnel, the current monitoring data including current monitoring values ​​of multiple preset influencing factors;

[0020] A model determination module is used to determine a target fire warning model. The target fire warning model is obtained by training with multiple historical monitoring data. Each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors. The initial model parameters of the target fire warning model are determined by using a genetic algorithm.

[0021] The result determination module is used to input the current monitoring data into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0022] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0023] Acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of multiple preset influencing factors;

[0024] Determine a target fire warning model, the target fire warning model is trained using multiple historical monitoring data, each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm;

[0025] The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0026] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0027] Acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of multiple preset influencing factors;

[0028] Determine a target fire warning model, the target fire warning model is trained using multiple historical monitoring data, each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm;

[0029] The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0030] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0031] Acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of multiple preset influencing factors;

[0032] Determine a target fire warning model, the target fire warning model is trained using multiple historical monitoring data, each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm;

[0033] The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0034] The cable tunnel fire warning method, device, storage medium and program product described above determine the target fire warning model through multiple historical monitoring data and genetic algorithms; further, the current monitoring data of the cable tunnel (i.e., the current monitoring values ​​of multiple preset influencing factors) obtained are input into the target fire warning model, and the target fire warning model processes and analyzes the current monitoring data of the cable tunnel, and determines the current status of the cable tunnel based on the final analysis results, and obtains the predicted fire results corresponding to the current monitoring data. This solution improves the calculation speed by introducing preset influencing factors while eliminating the influence of irrelevant interference factors; furthermore, the model parameters of the target fire warning model are determined based on the genetic algorithm, which lays the foundation for quickly and accurately determining the predicted fire results. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] FIG1 is an application environment diagram of a cable tunnel fire early warning method provided in an embodiment of the present application;

[0037] FIG2 is a schematic diagram of a process flow of cable tunnel fire warning provided in an embodiment of the present application;

[0038] FIG3 is a schematic diagram of a flow chart of a training target fire warning model provided in an embodiment of the present application;

[0039] FIG4 is a schematic diagram of a flow chart of obtaining a predicted fire result provided in an embodiment of the present application;

[0040] FIG5 is a structural block diagram of a cable tunnel fire warning device provided in an embodiment of the present application;

[0041] FIG. 6 is an internal structure diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] The cable tunnel fire warning method provided in the embodiment of the present application can be applied in the application environment shown in FIG. 1 . Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The terminal 102 collects the monitoring data of the cable tunnel and sends the collected monitoring data to the server 104; further, the server 104 can directly obtain the monitoring data of the cable tunnel (including current monitoring data and historical monitoring data), and determine the target fire warning model based on the acquired historical monitoring data; then, the current monitoring data is input into the target fire warning model, and finally the predicted fire result corresponding to the current monitoring data is determined. Among them, the terminal 102 can be, but is not limited to, various acquisition devices and monitoring devices. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0044] In an exemplary embodiment, as shown in FIG2 , a cable tunnel fire early warning method is provided, which is described by taking the method applied to the server 104 in FIG1 as an example, and may include the following steps:

[0045] S201, obtaining current monitoring data of the cable tunnel.

[0046] Among them, the current monitoring data may include current monitoring values ​​of multiple preset influencing factors, such as the preset influencing factors may include temperature, humidity, carbon monoxide, oxygen, hydrogen sulfide and smoke concentration.

[0047] Optionally, a corresponding monitoring device, namely, terminal 102, may be installed in the cable tunnel to monitor the current status of the cable tunnel.

[0048] Exemplarily, the terminal 102 (ie, the monitoring device installed in the cable tunnel) monitors the current status of the cable tunnel, and sends the monitored current monitoring data to the server 104 through the communication module, so that the server 104 analyzes the collected current monitoring data.

[0049] S202, determining a target fire warning model.

[0050] The target fire warning model can be trained using multiple historical monitoring data, each of which can include historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm.

[0051] Exemplarily, the model parameters in the target fire warning model can be initially determined through a genetic algorithm based on historical monitoring data; further, the target fire warning model after the model parameters are updated is trained again based on the historical monitoring data, so that the trained target fire warning model can make accurate predictions about the current conditions of the cable tunnel to reduce the probability of fire.

[0052] S203, inputting the current monitoring data into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0053] Among them, the predicted fire results may include fire intensity; or the predicted fire results may include fire intensity, fire occurrence time, the development relationship between preset influencing factors and time, abnormal preset influencing factors, and response strategies.

[0054] Exemplarily, the current monitoring data is input into the target fire warning model, which extracts and analyzes features of the current monitoring data, determines the current condition of the cable tunnel based on the final analysis results, and generates predicted fire results corresponding to the current monitoring data, thereby enabling early detection of abnormal conditions in the cable tunnel and taking countermeasures in advance to reduce the risk of fire.

[0055] The above cable tunnel fire warning method determines the target fire warning model through multiple historical monitoring data and genetic algorithms; further, the current monitoring data of the cable tunnel (i.e., the current monitoring values ​​of multiple preset influencing factors) obtained are input into the target fire warning model, and the target fire warning model processes and analyzes the current monitoring data of the cable tunnel, and determines the current status of the cable tunnel based on the final analysis results, and obtains the predicted fire results corresponding to the current monitoring data. This solution improves the calculation speed by introducing preset influencing factors, eliminating the influence of irrelevant interference factors; furthermore, the model parameters of the target fire warning model are determined based on the genetic algorithm, which lays the foundation for quickly and accurately determining the predicted fire results.

[0056] Based on the above embodiment, the present application embodiment explains the above embodiment S202 in detail. Specifically, the process of training the target fire warning model in the present application embodiment, as shown in FIG3, specifically includes the following steps:

[0057] S301, obtaining a plurality of historical monitoring data of the cable tunnel, and obtaining labeled fire results corresponding to the plurality of historical monitoring data respectively.

[0058] Among them, historical monitoring data may include but are not limited to historical monitoring values ​​of multiple preset influencing factors; labeled fire results may include but are not limited to historical fire intensity, historical fire occurrence time, historical development relationship between preset influencing factors and time, historical abnormal preset influencing factors, and historical response strategies.

[0059] Exemplarily, based on the structural data of the cable tunnel, multiple historical monitoring data of cable tunnels with similar structures to the cable tunnel and the labeled fire results corresponding to the multiple historical monitoring data can be directly screened from the database or data processing platform, so that the fire warning model can be trained based on the historical monitoring data to accurately learn the characteristics of the fire.

[0060] S302, inputting a plurality of historical monitoring data into the fire warning model to be trained respectively, and obtaining the predicted fire results corresponding to the plurality of historical monitoring data respectively.

[0061] Among them, the fire warning model can represent the target fire warning model before training.

[0062] It should be noted that the step of determining the fire warning model to be trained may include: generating multiple initial models for the first round, the model parameters of each initial model are randomly generated; using multiple sample monitoring data to train each initial model respectively, and obtaining trained models corresponding to the multiple initial models; determining the fitness characterization value of each trained model, and the fitness characterization value is used to reflect the fitness; when the iteration stop condition is not met, selecting at least part of the initial models from the multiple initial models according to the fitness characterization value, and updating the model parameters of at least part of the initial models to obtain multiple initial models for the next round, returning to the step of training each initial model respectively using multiple sample monitoring data to obtain trained models corresponding to the multiple initial models; when the iteration stop condition is met, selecting the target model from the multiple initial models of the current round according to the fitness characterization value; and determining the target model as the fire warning model to be trained.

[0063] The sample monitoring data may be part of the historical monitoring data; the iteration stopping conditions may include but are not limited to reaching the maximum number of iterations and the fitness characterization value reaching a preset standard.

[0064] Exemplarily, for each trained model, the fitness representation value F of the trained model can be calculated by the following formula:

[0065] (1)

[0066] Where n can represent the number of output layer nodes of the initial model; k can represent the preset coefficient, O i It can represent the predicted output of the i-th node, y i It can represent the expected output of the i-th node of the initial model.

[0067] Furthermore, the fitness characterization values ​​of each of the multiple initial models can be comprehensively calculated to obtain a total fitness characterization value; for each initial model, the ratio of the initial model fitness characterization value to the total fitness characterization value is calculated to obtain the selection probability of the initial model; and at least some of the initial models are selected from the multiple initial models based on the selection probability.

[0068] For example, based on the roulette wheel selection strategy, a plurality of initial models may be selected according to the fitness representation value according to the probability p. i , determine at least part of the initial model; where the probability of selection p i The calculation process is as follows:

[0069] (2)

[0070] Among them, F iIt can represent the fitness representation value of the initial model i; N can represent the number of initial models; k can represent a preset coefficient.

[0071] Furthermore, the model parameters of at least some of the initial models may be updated, which may be specifically expressed as:

[0072] (3)

[0073] Among them, b can represent a random number between [0,1].

[0074] Furthermore, the jth model parameter of the i-th initial model can be selected and mutated:

[0075] (4)

[0076] in, ; r2 can represent a random number; g can represent the current number of iterations; g max It can represent the maximum number of iterations; r can represent a random number between [0,1].

[0077] S303, for each historical monitoring data in the plurality of historical monitoring data, determining the difference between the labeled fire result corresponding to the historical monitoring data and the corresponding predicted fire result, and determining the result difference corresponding to the historical monitoring data.

[0078] Exemplarily, based on each labeled fire result of historical monitoring data and the corresponding predicted fire result, a mean square error between the two can be calculated, and the mean square error can be determined as the corresponding result difference of the historical monitoring data.

[0079] S304, determining a model loss value of the fire warning model by integrating the result differences corresponding to the plurality of historical monitoring data.

[0080] Exemplarily, the result differences corresponding to the multiple historical monitoring data may be summed, or the result differences corresponding to the multiple historical monitoring data may be averaged, and the final processing result may be determined as the model loss value of the fire warning model.

[0081] S305: adjusting the model parameters of the fire warning model using the model loss value to obtain a target fire warning model.

[0082] Exemplarily, the adjustment of model parameters can be achieved through the LM (Levenberg-Marquardt) algorithm; the adjustment process of model parameters may include but is not limited to: initializing algorithm parameters in the LM algorithm, calculating the Hessian matrix based on the target fire warning model, adjusting the Hessian matrix based on the algorithm parameters and adjusting the model parameters, etc.; further, after completing the adjustment of the model parameters of the fire warning model, the target fire warning model is obtained.

[0083] In the embodiment of the present application, by calculating the model loss value, a data basis is provided for adjusting the model parameters of the fire warning model, and a foundation is laid for the precise measurement accuracy of the target fire warning model.

[0084] Based on the above embodiment, the embodiment of the present application explains the above embodiment S202 in detail. Specifically, the target fire warning model may include an input layer, a hidden layer and an output layer. The process of determining the number of nodes in the hidden layer in the embodiment of the present application may specifically include:

[0085] Determine the number of nodes in the input layer to obtain the number of input nodes; determine the number of nodes in the output layer to obtain the number of output nodes; determine the number of nodes in the hidden layer based on the number of input nodes and the number of output nodes.

[0086] Exemplarily, the acquired current monitoring data is input into the target fire warning model, that is, the number of nodes m in the input layer is the number of preset influencing factors. If the preset influencing factors are temperature, humidity, carbon monoxide, oxygen, hydrogen sulfide and smoke concentration, the number of nodes in the input layer is m=6; further, if the output of the target fire warning model is fire intensity, the number of nodes in the output layer is q=1; further, based on the number of input nodes m and the number of output nodes q, the number of nodes p in the hidden layer can be determined, which can be specifically expressed as:

[0087] (5)

[0088] in, is an integer between 1 and 10; based on m=6 and q=1, it can be obtained that the number of nodes in the hidden layer of the target fire warning model is between 3 and 12; furthermore, the target fire warning model with the number of nodes in the hidden layer ranging from 3 to 12 can be tested in turn, and the mean square error of each test can be recorded, and the number of nodes in the hidden layer corresponding to the minimum mean square error can be determined, and the number of nodes in the hidden layer can be determined as the number of nodes in the hidden layer in the final target fire warning model.

[0089] It should be noted that the nodes in the hidden layer tend to use sigmoid transfer functions, such as the tansig function, to facilitate the processing of nonlinear features; the nodes in the output layer usually use linear transfer functions, such as the purelin function.

[0090] In the embodiment of the present application, by accurately calculating the number of hidden layer nodes in the target fire warning model, a foundation is laid for obtaining accurate fire prediction results.

[0091] Based on the above embodiment, the present application embodiment explains the above embodiment S203 in detail. Specifically, the process of obtaining the predicted fire result in the present application embodiment, as shown in FIG4, specifically includes the following steps:

[0092] S401, obtaining fire intensity correlations corresponding to a plurality of preset influencing factors.

[0093] The fire intensity correlation degree may represent the correlation between each preset influencing factor and the fire intensity.

[0094] Optionally, based on the grey correlation analysis formula, each historical monitoring data X={x1, x2, …, x N} and the labeled fire results y corresponding to multiple historical monitoring data are calculated, and the fire intensity correlation p corresponding to each preset influencing factor X is obtained, thereby providing an important scientific basis for fire early warning. Among them, N can represent the number of each historical monitoring data or the number of labeled fire results.

[0095] S402: For each preset influencing factor, multiply the current monitoring value of the preset influencing factor by the corresponding fire intensity correlation degree to obtain an updated monitoring value of the preset influencing factor.

[0096] Exemplarily, for each preset influencing factor, the current monitoring value x of the preset influencing factor is i By multiplying the corresponding fire intensity correlation degree p, the updated monitoring value of the preset influencing factor can be obtained, that is, x i ×p.

[0097] S403: Obtain updated monitoring data based on each updated monitoring value.

[0098] S404, inputting the updated monitoring data into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0099] Exemplarily, after the updated monitoring data is input into the target fire warning model, the updated monitoring data can be formatted, such as removing outliers and normalizing, etc.; further, the processed updated monitoring data is input into the target fire warning model, which processes and analyzes it and obtains the predicted fire result corresponding to the current monitoring data. Optionally, the target fire warning model can be composed of a BP (Back Propagation) neural network.

[0100] In the embodiment of the present application, by introducing the fire intensity correlation, the degree of correlation between each preset influencing factor and the fire intensity is taken into account; further, through the fire intensity correlation, the current measurement value of each preset influencing factor is updated, so that the data input into the target fire warning model has more reference value, and ultimately ensures the accuracy of predicting fire results.

[0101] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0102] Based on the same inventive concept, the embodiment of the present application also provides a cable tunnel fire warning device for implementing the cable tunnel fire warning method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more cable tunnel fire warning device embodiments provided below can refer to the limitations of the cable tunnel fire warning method above, and will not be repeated here.

[0103] In an exemplary embodiment, as shown in FIG5 , a cable tunnel fire warning device 1 is provided, comprising: a data acquisition module 10, a model determination module 20 and a result determination module 30, wherein:

[0104] The data acquisition module 10 is used to acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of a plurality of preset influencing factors.

[0105] The model determination module 20 is used to determine the target fire warning model. The target fire warning model is trained using multiple historical monitoring data. Each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors. The initial model parameters of the target fire warning model are determined using a genetic algorithm.

[0106] The result determination module 30 is used to input the current monitoring data into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0107] In one embodiment, the model determination module 20 is specifically used to:

[0108] Acquire multiple historical monitoring data of the cable tunnel, and obtain labeled fire results corresponding to the multiple historical monitoring data respectively; input the multiple historical monitoring data into the fire warning model to be trained respectively, and obtain the predicted fire results corresponding to the multiple historical monitoring data respectively; for each historical monitoring data in the multiple historical monitoring data, determine the difference between the labeled fire result corresponding to the historical monitoring data and the corresponding predicted fire result, and determine the result difference corresponding to the historical monitoring data; comprehensively consider the result differences corresponding to the multiple historical monitoring data, and determine the model loss value of the fire warning model; use the model loss value to adjust the model parameters of the fire warning model to obtain the target fire warning model.

[0109] In one embodiment, the model determination module 20 is further configured to:

[0110] Generate multiple initial models for the first round, and the model parameters of each initial model are randomly generated; use multiple sample monitoring data to train each initial model respectively, and obtain trained models corresponding to the multiple initial models; determine the fitness characterization value of each trained model, and the fitness characterization value is used to reflect the fitness; when the iteration stop condition is not met, select at least part of the initial models from the multiple initial models according to the fitness characterization value, and update the model parameters of at least part of the initial models to obtain multiple initial models for the next round, and return to the step of training each initial model using multiple sample monitoring data to obtain trained models corresponding to the multiple initial models; when the iteration stop condition is met, select the target model from the multiple initial models of the current round according to the fitness characterization value; determine the target model as the fire warning model to be trained.

[0111] In one embodiment, the model determination module 20 is further configured to:

[0112] Comprehensively calculate the fitness representation values ​​of each of the multiple initial models to obtain a total fitness representation value; for each initial model, calculate the ratio of the initial model fitness representation value to the total fitness representation value to obtain the selection probability of the initial model; select at least some of the initial models from the multiple initial models based on the selection probability.

[0113] In one embodiment, the model determination module 20 is further configured to:

[0114] Determine the number of nodes in the input layer to obtain the number of input nodes; determine the number of nodes in the output layer to obtain the number of output nodes; determine the number of nodes in the hidden layer based on the number of input nodes and the number of output nodes.

[0115] In one embodiment, the result determination module 30 is specifically used to:

[0116] Obtain the fire intensity correlations corresponding to a plurality of preset influencing factors; for each preset influencing factor, multiply the current monitoring value of the preset influencing factor by the corresponding fire intensity correlation to obtain an updated monitoring value of the preset influencing factor; obtain updated monitoring data based on each updated monitoring value; input the updated monitoring data into the target fire warning model to obtain a predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0117] Each module in the cable tunnel fire warning device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0118] In an exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be shown in FIG6. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store monitoring data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a cable tunnel fire early warning method is implemented.

[0119] Those skilled in the art will understand that the structure shown in FIG. 6 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0120] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0121] Acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of multiple preset influencing factors;

[0122] Determine a target fire warning model, the target fire warning model is trained using multiple historical monitoring data, each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm;

[0123] The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0124] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0125] Acquire multiple historical monitoring data of the cable tunnel, and obtain labeled fire results corresponding to the multiple historical monitoring data respectively; input the multiple historical monitoring data into the fire warning model to be trained respectively, and obtain the predicted fire results corresponding to the multiple historical monitoring data respectively; for each historical monitoring data in the multiple historical monitoring data, determine the difference between the labeled fire result corresponding to the historical monitoring data and the corresponding predicted fire result, and determine the result difference corresponding to the historical monitoring data; comprehensively consider the result differences corresponding to the multiple historical monitoring data respectively, and determine the model loss value of the fire warning model; use the model loss value to adjust the model parameters of the fire warning model to obtain the target fire warning model.

[0126] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0127] Generate multiple initial models for the first round, and the model parameters of each initial model are randomly generated; use multiple sample monitoring data to train each initial model respectively, and obtain trained models corresponding to the multiple initial models; determine the fitness characterization value of each trained model, and the fitness characterization value is used to reflect the fitness; when the iteration stop condition is not met, select at least part of the initial models from the multiple initial models according to the fitness characterization value, and update the model parameters of at least part of the initial models to obtain multiple initial models for the next round, and return to the step of training each initial model using multiple sample monitoring data to obtain trained models corresponding to the multiple initial models; when the iteration stop condition is met, select the target model from the multiple initial models of the current round according to the fitness characterization value; determine the target model as the fire warning model to be trained.

[0128] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0129] Comprehensively calculate the fitness representation values ​​of each of the multiple initial models to obtain a total fitness representation value; for each initial model, calculate the ratio of the initial model fitness representation value to the total fitness representation value to obtain the selection probability of the initial model; select at least some of the initial models from the multiple initial models based on the selection probability.

[0130] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0131] Determine the number of nodes in the input layer to obtain the number of input nodes; determine the number of nodes in the output layer to obtain the number of output nodes; determine the number of nodes in the hidden layer based on the number of input nodes and the number of output nodes.

[0132] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0133] Obtain the fire intensity correlations corresponding to a plurality of preset influencing factors; for each preset influencing factor, multiply the current monitoring value of the preset influencing factor by the corresponding fire intensity correlation to obtain an updated monitoring value of the preset influencing factor; obtain updated monitoring data based on each updated monitoring value; input the updated monitoring data into the target fire warning model to obtain a predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0134] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0135] Acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of multiple preset influencing factors;

[0136] Determine a target fire warning model, the target fire warning model is trained using multiple historical monitoring data, each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm;

[0137] The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0138] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0139] Acquire multiple historical monitoring data of the cable tunnel, and obtain labeled fire results corresponding to the multiple historical monitoring data respectively; input the multiple historical monitoring data into the fire warning model to be trained respectively, and obtain the predicted fire results corresponding to the multiple historical monitoring data respectively; for each historical monitoring data in the multiple historical monitoring data, determine the difference between the labeled fire result corresponding to the historical monitoring data and the corresponding predicted fire result, and determine the result difference corresponding to the historical monitoring data; comprehensively consider the result differences corresponding to the multiple historical monitoring data respectively, and determine the model loss value of the fire warning model; use the model loss value to adjust the model parameters of the fire warning model to obtain the target fire warning model.

[0140] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0141] Generate multiple initial models for the first round, and the model parameters of each initial model are randomly generated; use multiple sample monitoring data to train each initial model respectively, and obtain trained models corresponding to the multiple initial models; determine the fitness characterization value of each trained model, and the fitness characterization value is used to reflect the fitness; when the iteration stop condition is not met, select at least part of the initial models from the multiple initial models according to the fitness characterization value, and update the model parameters of at least part of the initial models to obtain multiple initial models for the next round, and return to the step of training each initial model using multiple sample monitoring data to obtain trained models corresponding to the multiple initial models; when the iteration stop condition is met, select the target model from the multiple initial models of the current round according to the fitness characterization value; determine the target model as the fire warning model to be trained.

[0142] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0143] Comprehensively calculate the fitness representation values ​​of each of the multiple initial models to obtain a total fitness representation value; for each initial model, calculate the ratio of the initial model fitness representation value to the total fitness representation value to obtain the selection probability of the initial model; select at least some of the initial models from the multiple initial models based on the selection probability.

[0144] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0145] Determine the number of nodes in the input layer to obtain the number of input nodes; determine the number of nodes in the output layer to obtain the number of output nodes; determine the number of nodes in the hidden layer based on the number of input nodes and the number of output nodes.

[0146] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0147] Obtain the fire intensity correlations corresponding to a plurality of preset influencing factors; for each preset influencing factor, multiply the current monitoring value of the preset influencing factor by the corresponding fire intensity correlation to obtain an updated monitoring value of the preset influencing factor; obtain updated monitoring data based on each updated monitoring value; input the updated monitoring data into the target fire warning model to obtain a predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0148] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0149] Acquire current monitoring data of the cable tunnel, where the current monitoring data includes current monitoring values ​​of multiple preset influencing factors;

[0150] Determine a target fire warning model, the target fire warning model is trained using multiple historical monitoring data, each historical monitoring data includes historical monitoring values ​​of multiple preset influencing factors, and the initial model parameters of the target fire warning model are determined using a genetic algorithm;

[0151] The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0152] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0153] Acquire multiple historical monitoring data of the cable tunnel, and obtain labeled fire results corresponding to the multiple historical monitoring data respectively; input the multiple historical monitoring data into the fire warning model to be trained respectively, and obtain the predicted fire results corresponding to the multiple historical monitoring data respectively; for each historical monitoring data in the multiple historical monitoring data, determine the difference between the labeled fire result corresponding to the historical monitoring data and the corresponding predicted fire result, and determine the result difference corresponding to the historical monitoring data; comprehensively consider the result differences corresponding to the multiple historical monitoring data, and determine the model loss value of the fire warning model; use the model loss value to adjust the model parameters of the fire warning model to obtain the target fire warning model.

[0154] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0155] Generate multiple initial models for the first round, and the model parameters of each initial model are randomly generated; use multiple sample monitoring data to train each initial model respectively, and obtain trained models corresponding to the multiple initial models; determine the fitness characterization value of each trained model, and the fitness characterization value is used to reflect the fitness; when the iteration stop condition is not met, select at least part of the initial models from the multiple initial models according to the fitness characterization value, and update the model parameters of at least part of the initial models to obtain multiple initial models for the next round, and return to the step of training each initial model using multiple sample monitoring data to obtain trained models corresponding to the multiple initial models; when the iteration stop condition is met, select the target model from the multiple initial models of the current round according to the fitness characterization value; determine the target model as the fire warning model to be trained.

[0156] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0157] Comprehensively calculate the fitness representation values ​​of each of the multiple initial models to obtain a total fitness representation value; for each initial model, calculate the ratio of the initial model fitness representation value to the total fitness representation value to obtain the selection probability of the initial model; select at least some of the initial models from the multiple initial models based on the selection probability.

[0158] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0159] Determine the number of nodes in the input layer to obtain the number of input nodes; determine the number of nodes in the output layer to obtain the number of output nodes; determine the number of nodes in the hidden layer based on the number of input nodes and the number of output nodes.

[0160] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0161] Obtain the fire intensity correlations corresponding to a plurality of preset influencing factors; for each preset influencing factor, multiply the current monitoring value of the preset influencing factor by the corresponding fire intensity correlation to obtain an updated monitoring value of the preset influencing factor; obtain updated monitoring data based on each updated monitoring value; input the updated monitoring data into the target fire warning model to obtain a predicted fire result corresponding to the current monitoring data output by the target fire warning model.

[0162] It should be noted that the information (including but not limited to cable tunnel information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0163] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0164] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0165] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A cable tunnel fire early warning method, characterized in that: The method comprises: Acquire current monitoring data of the cable tunnel, wherein the current monitoring data includes current monitoring values ​​of a plurality of preset influencing factors; Determine a target fire warning model, wherein the target fire warning model is obtained by training with a plurality of historical monitoring data, each of the historical monitoring data includes historical monitoring values ​​of the plurality of preset influencing factors, and initial model parameters of the target fire warning model are determined by using a genetic algorithm; The current monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

2. The method according to claim 1, characterized in that The steps of training the target fire warning model include: Acquire multiple historical monitoring data of the cable tunnel, and acquire labeled fire results corresponding to the multiple historical monitoring data respectively; Inputting the plurality of historical monitoring data into the fire warning model to be trained respectively, and obtaining the predicted fire results corresponding to the plurality of historical monitoring data respectively; For each historical monitoring data among the plurality of historical monitoring data, determining a difference between a labeled fire result corresponding to the historical monitoring data and a corresponding predicted fire result, and determining a result difference corresponding to the historical monitoring data; Determine the model loss value of the fire warning model by combining the result differences corresponding to the plurality of historical monitoring data; The model loss value is used to adjust the model parameters of the fire warning model to obtain a target fire warning model.

3. The method according to claim 2, characterized in that The step of determining the fire warning model to be trained comprises: Generate a plurality of initial models of the first round, wherein the model parameters of each initial model are randomly generated; Using a plurality of sample monitoring data to train each of the initial models respectively, to obtain trained models corresponding to the plurality of initial models respectively; Determine a fitness representation value of each of the trained models, wherein the fitness representation value is used to reflect fitness; In the case where the iteration stop condition is not met, at least some initial models are selected from the multiple initial models according to the fitness characterization value, and the model parameters of the at least some initial models are updated to obtain multiple initial models for the next round, and the step of training each of the initial models using the multiple sample monitoring data to obtain trained models corresponding to the multiple initial models respectively is returned; When the iteration stop condition is met, a target model is selected from multiple initial models of the current round according to the fitness representation value; The target model is determined as the fire warning model to be trained.

4. The method according to claim 3, characterized in that The selecting at least part of the initial models from the multiple initial models according to the fitness representation value comprises: Comprehensively calculating the fitness representation values ​​of the respective multiple initial models to obtain a total fitness representation value; For each of the initial models, calculating the ratio of the fitness representation value of the initial model to the total fitness representation value, and obtaining the selection probability of the initial model; At least part of the initial models are selected from the plurality of initial models based on the selection probability.

5. The method according to claim 1, characterized in that: The target fire warning model includes an input layer, a hidden layer and an output layer, and the step of determining the number of nodes in the hidden layer includes: Determine the number of nodes in the input layer to obtain the number of input nodes; Determine the number of nodes in the output layer to obtain the number of output nodes; The number of nodes in the hidden layer is determined according to the number of input nodes and the number of output nodes.

6. The method according to claim 1, characterized in that The step of inputting the current monitoring data into the target fire warning model to obtain a predicted fire result corresponding to the current monitoring data output by the target fire warning model includes: Obtaining fire intensity correlations corresponding to the plurality of preset influencing factors respectively; For each of the preset influencing factors, multiplying the current monitoring value of the preset influencing factor by the corresponding fire intensity correlation degree to obtain an updated monitoring value of the preset influencing factor; Obtaining updated monitoring data based on each of the updated monitoring values; The updated monitoring data is input into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

7. A cable tunnel fire warning device, characterized in that: The device comprises: A data acquisition module, used to acquire current monitoring data of the cable tunnel, wherein the current monitoring data includes current monitoring values ​​of a plurality of preset influencing factors; A model determination module, used to determine a target fire warning model, wherein the target fire warning model is obtained by training with a plurality of historical monitoring data, each of the historical monitoring data includes historical monitoring values ​​of the plurality of preset influencing factors, and the initial model parameters of the target fire warning model are determined by using a genetic algorithm; The result determination module is used to input the current monitoring data into the target fire warning model to obtain the predicted fire result corresponding to the current monitoring data output by the target fire warning model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images