Wireless networking fire-fighting early warning method and system based on multi-source data

Abstract

The invention discloses a wireless networking fire-fighting early warning method and system based on multi-source data, and relates to the technical field of wireless networking. The method comprises the following steps: deploying an array integrated with multiple types of sensors in a target space in a wireless networking mode, and collecting and uploading electrical, gas and temperature multi-source parameters to a data medium station in real time; when monitoring parameter triggering is abnormal, obtaining a time verification credibility array based on multi-source data evolution time verification, and performing spatial evolution verification on gas parameters based on an air flow field to obtain spatial verification credibility; basic early warning parameters are calculated according to the multi-source parameters, correction is carried out on the basic early warning parameters by fusing the credibility of space-time dual verification, final early warning parameters are generated, and fire-fighting early warning judgment is executed. According to the invention, through a time sequence and space dual verification mechanism, false alarms caused by sensor faults or environmental interference are effectively discriminated, the early warning accuracy is improved, and early and reliable fire risk identification and hierarchical linkage response are realized.

Description

Technical Field

[0001] This invention relates to the field of wireless networking technology, and specifically to a wireless networking fire early warning method and system based on multi-source data. Background Technology

[0002] In modern industrial and built environments, continuous and effective fire status monitoring of critical spaces is crucial. Traditional monitoring methods mainly rely on deploying several independent fire detectors, judging whether a single physical quantity exceeds a fixed threshold. The core of this approach lies in point-based monitoring and threshold comparison.

[0003] However, relying on the independent judgment of a single or a few types of sensors is susceptible to interference from non-fire factors in the environment, leading to frequent false alarms. Secondly, existing technologies generally lack collaborative verification mechanisms for abnormal signals. When a single sensor triggers a threshold, it is usually directly judged as a fire alarm without analyzing the collaborative evolution of multi-source data from a temporal perspective to verify fire characteristics, or verifying the rationality of abnormal signal propagation based on physical principles such as airflow diffusion from a spatial perspective. This results in a high false alarm rate and low reliability, potentially leading to resource waste and weakening the timeliness and accuracy of early warnings. Summary of the Invention

[0004] This invention addresses the technical problems of existing fire early warning systems that rely on a single sensor and lack a data collaborative verification mechanism, resulting in high false alarm rates and insufficient early warning accuracy. It provides a wireless networked fire early warning method and system based on multi-source data.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a wireless networking fire early warning method based on multi-source data, comprising: By deploying a sensor array of multiple types of sensors in the target space, multi-source data monitoring is performed to obtain electrical parameter arrays, gas parameter arrays, and temperature parameter arrays, which are then transmitted to the data center via wireless networking. When the trigger parameters are abnormal, multi-source data evolution time verification is performed based on the electrical parameter array, gas parameter array, and temperature parameter array to obtain a time verification credibility array. The airflow field within the target space is acquired, and the spatial evolution of the gas parameter array is verified to obtain the spatial verification credibility. Based on the electrical parameter array, gas parameter array, and temperature parameter array, basic early warning parameters are obtained by classification. These parameters are then combined with the time verification credibility array and the spatial verification credibility array for credibility correction to obtain early warning parameters, which are then used for fire early warning judgment.

[0006] Secondly, the present invention provides a wireless networked fire early warning system based on multi-source data, comprising: The data acquisition module is used to monitor multi-source data through a sensor array of multiple types of sensors deployed in the target space, obtain electrical parameter arrays, gas parameter arrays and temperature parameter arrays, and transmit them to the data platform through wireless networking; The evolution time verification module is used to perform multi-source data evolution time verification based on the electrical parameter array, gas parameter array, and temperature parameter array when the trigger parameters are abnormal, and obtain a time verification credibility array. The spatial evolution verification module is used to acquire the air flow field in the target space, perform spatial evolution verification on the gas parameter array, and obtain the spatial verification credibility. The early warning discrimination module is used to classify and obtain basic early warning parameters based on the electrical parameter array, gas parameter array, and temperature parameter array, and perform credibility correction by combining the time verification credibility array and spatial verification credibility to obtain early warning parameters for fire early warning discrimination.

[0007] The beneficial effects of this invention are: Compared to existing technologies, this invention firstly utilizes a flexible wireless networking approach to deploy a multi-dimensional sensor array, enabling simultaneous acquisition and real-time transmission of multiple source parameters, including electrical, gas, and temperature parameters. This solves the problems of inflexible wired deployment and limited monitoring dimensions. Secondly, it innovatively introduces multi-source data evolution time verification when parameters are abnormal. By analyzing the coordinated changes of multiple parameters over time, it effectively identifies anomalies caused by transient interference, improving the accuracy of time-series judgments. Thirdly, by acquiring airflow fields and performing spatial evolution verification on gas parameters, the rationality of abnormal signals is validated from the perspective of physical diffusion principles, enhancing the reliability of the spatial dimension. Finally, the credibility obtained from both time and spatial verification is used to correct the basic warning parameters, ultimately generating more accurate warning parameters. This reduces the system's false alarm and false negative rates, achieving earlier, more accurate, and more reliable intelligent warnings for fire hazards. Attached Figure Description

[0008] Figure 1 A flowchart illustrating the wireless networking fire early warning method based on multi-source data provided by the present invention; Figure 2 This is a schematic diagram of the structure of the wireless network fire early warning system based on multi-source data provided by the present invention.

[0009] In the attached diagram, the components represented by each number are as follows: Data acquisition module 11, evolution time verification module 12, spatial evolution verification module 13, early warning and discrimination module 14. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0012] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0013] Example 1, as Figure 1 As shown, this embodiment of the invention provides a wireless networking fire early warning method based on multi-source data, including: S10: Through a sensor array of multiple types of sensors deployed in the target space, multi-source data monitoring is performed to obtain electrical parameter arrays, gas parameter arrays and temperature parameter arrays, and the data is transmitted to the data center via wireless networking; First, multi-source data monitoring is achieved through a sensor array of various types of sensors deployed within the target space. The target space refers to a specific area or location requiring fire early warning monitoring, such as a battery compartment in a power storage station, a data center server room, or an industrial warehouse. The sensor array, comprising various types of sensors, is a collection of monitoring nodes integrated with multiple sensing units, deployed according to a specific spatial distribution pattern. Each sensor is a device used to sense specific physical or chemical parameters, such as current / voltage sensors, smoke or specific gas sensors, and temperature sensors. By deploying the sensor array within the target space, real-time status data from multiple points and dimensions can be acquired, including electrical parameter arrays reflecting electrical status, gas parameter arrays reflecting gas composition and concentration, and temperature parameter arrays reflecting heat distribution.

[0014] Specifically, multi-source data monitoring is performed using a sensor array of various types of sensors deployed within the target space to obtain electrical parameter arrays, gas parameter arrays, and temperature parameter arrays. This data is then transmitted to a data platform via a wireless network, including: Multi-source data is monitored by a sensor array of multiple types of sensors deployed in the target space. The sensor array includes a combination of sensors at multiple location coordinates in the target space. Each sensor combination includes an electrical sensor, a gas sensor, and a temperature sensor. According to the multi-source data type, the multi-source data array is divided to obtain an electrical parameter array, a gas parameter array, and a temperature parameter array, and then sent to the data platform via wireless networking.

[0015] First, multi-source data is monitored using a sensor array of various types of sensors deployed within the target space, thereby obtaining the original multi-source data array. Preferably, the sensor array is constructed by pre-defining multiple monitoring points with clearly defined location coordinates within the target space. At each monitoring point, a sensor combination integrating at least three types of sensors is deployed. This sensor combination typically includes an electrical sensor for monitoring current, voltage, or leakage current; a gas sensor for monitoring smoke concentration or specific gases such as carbon monoxide or volatile organic compounds; and a temperature sensor for monitoring ambient temperature. All sensor combinations at the monitoring points work collaboratively to form a monitoring network covering the target space area, enabling real-time acquisition of multi-dimensional raw data reflecting the overall operating status of the target space, thus constituting the multi-source data array.

[0016] The multi-source data array obtained above is divided and reorganized according to the sensor types from which it originates. Since each sensor combination at each location coordinate outputs three types of parameters—electrical, gas, and temperature—it is necessary to classify and extract the data mixed within the multi-source data array. Specifically, data collected by electrical sensors at all location coordinates is extracted from the multi-source data array and reorganized according to the location coordinate relationships to form a data set specifically reflecting the electrical state of each point in the target space, i.e., the electrical parameter array. Similarly, data from all gas sensors is extracted to form a gas parameter array, and data from all temperature sensors is extracted to form a temperature parameter array. This division allows subsequent processing to perform specialized analysis based on the characteristics of different physical quantities.

[0017] Finally, the electrical parameter array, gas parameter array, and temperature parameter array obtained after classification are sent to the remote data center via a wireless network deployed at the monitoring site.

[0018] Specifically, this wireless network utilizes communication links built with wireless self-organizing networks or low-power wide-area network (LPWAN) technologies. This wireless network enables remote data transmission between monitoring nodes and the data platform, avoiding the complex cabling required by traditional wired deployments and enhancing the deployment and expansion flexibility of the monitoring system in complex or difficult-to-wire environments. The data platform, a centralized data processing platform located in the cloud or on a local server, serves as the data aggregation and core processing center for the entire early warning architecture. It reliably receives and stores multi-dimensional parameter arrays uploaded in real-time from various monitoring nodes, providing a unified, complete, and time-consistent data foundation for subsequent multi-source data evolution time verification, gas parameter spatial evolution verification, and final fire early warning determination.

[0019] S20: When the trigger parameter is abnormal, perform multi-source data evolution time verification based on the electrical parameter array, gas parameter array and temperature parameter array to obtain a time verification credibility array; Specifically, when an abnormality is triggered, multi-source data evolution time verification is performed based on the electrical parameter array, gas parameter array, and temperature parameter array to obtain a time verification reliability array, including: When any electrical parameter, gas parameter, or temperature parameter exceeds the threshold value for that parameter, gas parameter, or temperature parameter, a parameter anomaly is triggered. The electrical parameter array, gas parameter array, and temperature parameter array are divided and combined according to multiple location coordinates to obtain multiple multi-source data combinations. Based on multiple multi-source data combinations, multi-source data evolution time verification is performed to obtain a time verification credibility array.

[0020] First, the real-time acquired electrical, gas, and temperature parameters are continuously compared with their respective preset thresholds. When any electrical parameter at any monitoring location exceeds its threshold, or a gas parameter exceeds its threshold, or a temperature parameter exceeds its threshold, an abnormal parameter is identified. This threshold triggering mechanism serves as the initial condition for initiating subsequent advanced verification processes.

[0021] The electrical parameter threshold, gas parameter threshold, and temperature parameter threshold are pre-determined critical values ​​through experiments, standard specifications, or historical data analysis. They represent the criteria for judging abnormal operating conditions of electrical circuits or equipment, excessive concentrations of specific harmful gases or smoke in the air, and abnormally high ambient temperatures, respectively. These thresholds are comprehensively set based on the specific nature of the monitored space, equipment type, safety standards, and environmental background values. For example, in the application scenario of a lithium-ion battery energy storage compartment, the electrical parameter threshold can be set when the circuit current continuously exceeds 20% of the rated value; the gas parameter threshold can be set when the hydrogen concentration reaches 0.5% by volume; and the temperature parameter threshold can be set when the temperature at the monitoring point exceeds 60°C. The setting of these electrical, gas, and temperature parameter thresholds aims to capture early abnormal signals while balancing sensitivity and anti-interference capabilities.

[0022] When an abnormality is triggered, a collaborative analysis of the overall data will be immediately initiated. Specifically, based on multiple predefined location coordinates, the electrical parameter array, gas parameter array, and temperature parameter array at the current moment will be aligned and combined. Then, the electrical parameter value, gas parameter value, and temperature parameter value corresponding to the same location coordinate will be extracted, forming a multi-source data combination representing the overall state of that location coordinate. Performing this operation on all location coordinates will yield multiple multi-source data combinations corresponding to the number of location coordinates.

[0023] Furthermore, after obtaining multiple multi-source data combinations, multi-source data evolution time verification is performed. Specifically, the core of this multi-source data evolution time verification lies in analyzing whether the sequence and synergistic patterns of changes in the three types of parameters—electrical, gas, and temperature—over time conform to the precursor evolution model of typical fires, especially electrical fires. For example, before an electrical fire occurs, abnormal electrical parameters usually appear first due to abnormal current or insulation degradation; secondly, abnormal heating or electric arcs may ignite surrounding materials or produce characteristic gases, leading to abnormal gas parameters; finally, the continuous accumulation of heat causes a significant increase in ambient temperature.

[0024] Specifically, multi-source data evolution time verification analyzes current and historical parameter combination sequences to determine whether observed electrical, gas, and temperature parameter anomalies exhibit a causal and temporal relationship consistent with the physical processes of fire precursors. This effectively distinguishes the characteristic evolution patterns of genuine fire hazards from isolated anomalous signals caused by momentary sensor malfunctions or transient environmental interference. Simultaneously, based on the analysis, a quantitative score is calculated for each monitoring point. This score characterizes the reliability of the current anomalous pattern at that point in the temporal dimension; it is the time verification reliability score.

[0025] Specifically, based on multiple combinations of multi-source data, multi-source data evolution time verification is performed to obtain a time verification reliability array, including: Multiple data evolution predictors corresponding to multiple location coordinates are obtained, wherein each data evolution predictor includes a gas evolution prediction branch and a temperature evolution prediction branch; The electrical and gas parameters in the multiple multi-source data combinations are input into multiple data evolution predictors, and multiple evolution prediction gas parameters and multiple evolution prediction temperature parameters are output. These are then combined to obtain multiple evolution prediction multi-source data combinations. Calculate the similarity between multiple evolution prediction multi-source data combinations and multiple multi-source data combinations to obtain a time-verified confidence array.

[0026] First, multiple data evolution predictors are obtained, each corresponding to a specific location coordinate. Each predictor is a pre-trained computational model for a particular monitoring point, containing two functional branches: a gas evolution prediction branch and a temperature evolution prediction branch. Specifically, the gas evolution prediction branch predicts the gas parameter values ​​that should appear at that point in the next moment or stage, based on the input electrical parameters and historical and current gas parameters. The temperature evolution prediction branch predicts the corresponding temperature parameter values ​​based on the input gas parameters and historical and current temperature parameters.

[0027] Specifically, multiple data evolution predictors corresponding to multiple location coordinates are obtained, including: Obtain facility characteristics at multiple location coordinates, and index multiple similar location coordinates with similar facility characteristics within historical fire monitoring data; Based on historical monitoring data of multiple similar location coordinates, multiple sets of sample electrical parameters were collected when a fire alarm was issued, and gas parameters were collected when different sample electrical parameters were issued, to obtain multiple sets of sample evolution prediction gas parameters. Based on historical monitoring data of multiple similar location coordinates, multiple sets of sample gas parameters were collected when a fire alarm was issued, and temperature parameters were collected when different sample gas parameters appeared, to obtain multiple sets of sample evolution prediction temperature parameters. Based on multiple sets of sample electrical parameters and multiple sets of sample evolution prediction gas parameters, multiple gas evolution prediction branches are constructed. Based on multiple sets of sample gas parameters and multiple sets of sample evolution prediction temperature parameters, multiple temperature evolution prediction branches are constructed. By combining multiple gas evolution prediction branches and multiple temperature evolution prediction branches, multiple data evolution predictors are obtained.

[0028] First, the facility characteristics corresponding to the location coordinates of each monitoring point within the current target space are obtained. Specifically, facility characteristics may include information describing the environmental and functional attributes of the point, such as the type of equipment, rated power, surrounding material properties, and ventilation conditions. Second, based on the aforementioned facility characteristics, a matching search is performed on historical fire monitoring data to index multiple similar location coordinates with highly similar facility characteristics. The historical monitoring data accumulated from these similar location coordinates provides effective training samples with consistent physical backgrounds and referable evolutionary patterns for constructing a data evolution predictor for the current monitoring point.

[0029] The historical time refers to a continuous or discontinuous monitoring period in the past that is used for model training and feature retrieval. This time range is set comprehensively based on the system deployment time, data accumulation cycle, and typical development time scale of the fire hazards of concern. For example, it can be set to cover historical data covering at least one full year of operation.

[0030] Secondly, based on historical monitoring data from multiple similar location coordinates, a sample set is constructed to train the gas evolution prediction branch. Specifically, electrical parameter data recorded before and after a fire alarm event are collected from historical monitoring data, forming multiple sample electrical parameter sets. Simultaneously, for each sample electrical parameter set, corresponding gas parameter data is collected as the expected gas evolution result for that electrical state, thus obtaining multiple sample evolution prediction gas parameter sets. A temporal mapping relationship from electrical parameters to gas parameters is established between the sample electrical parameter sets and the sample evolution prediction gas parameter sets.

[0031] Similarly, a sample set is constructed for training the temperature evolution prediction branch. Gas parameter data for all times before and after fire warning events are collected from the same historical monitoring data, forming multiple sample gas parameter sets. For each sample gas parameter set, its corresponding temperature parameter data is collected, which serves as the expected temperature evolution result for that gas state, thus obtaining multiple corresponding sample evolution prediction temperature parameter sets. The sample gas parameter sets and the sample evolution prediction temperature parameter sets constitute a time-series mapping relationship from gas parameters to temperature parameters.

[0032] Furthermore, based on the constructed sets of multiple sample electrical parameters and multiple sets of sample evolution prediction gas parameters, multiple gas evolution prediction branches are constructed. Specifically, there are two implementation paths for constructing these gas evolution prediction branches, which can be selected based on actual data conditions. One is to establish a mapping relationship based on physical experience or statistical laws, for example, by establishing a lookup table or linear regression model corresponding to specific electrical parameter ranges and characteristic gas concentration growth rates. The other is to use data-driven machine learning methods for training. When there is sufficient historical sample data, this method can more effectively capture complex nonlinear relationships; for example, using the sample electrical parameter sets as input features and the corresponding sample evolution prediction gas parameter sets as target labels, a long short-term memory network model can be trained. After learning, this long short-term memory network model can predict future gas parameter trends based on the input electrical parameter sequence. In practical deployment, the machine learning training method that allows for better data conditions can be preferred to obtain more accurate prediction capabilities.

[0033] Similarly, based on the multiple sets of sample gas parameters and multiple sets of sample evolution prediction temperature parameters constructed above, multiple temperature evolution prediction branches are built. Specifically, the construction of temperature evolution prediction branches also follows the two optional paths mentioned above: an analytical model based on empirical formulas for gas concentration and temperature rise can be built, or machine learning methods can be used for training. For example, in scenarios with abundant data, the sample gas parameter set can be used as input, and the sample evolution prediction temperature parameter set can be used as output to train a gradient boosting decision tree model, enabling it to learn the prediction mapping from a specific gas concentration change pattern to subsequent temperature changes. Preferably, based on data completeness, machine learning training is preferred to construct high-performance temperature evolution prediction branches.

[0034] Finally, the gas evolution prediction branch and temperature evolution prediction branch corresponding to a given location coordinate are functionally combined and encapsulated to form a complete data evolution predictor. By repeating the above process from feature matching and sample collection to model training and combination for all location coordinates where data evolution predictors need to be built, multiple data evolution predictors corresponding to each monitoring point can be obtained, providing a model foundation for subsequent time-based validation.

[0035] Furthermore, the multiple multi-source data combinations obtained at the current moment are input into the data evolution predictors at their corresponding location coordinates. Specifically, the input data consists of the electrical and gas parameters within each multi-source data combination. After receiving the input data, each data evolution predictor's internal gas evolution prediction branch outputs a predicted gas parameter for that location, while its temperature evolution prediction branch outputs a predicted temperature parameter. Combining the obtained predicted gas and temperature parameters with the actual monitored electrical parameters at that location at the current moment constitutes the multi-source data combination for evolution prediction.

[0036] This evolutionary prediction multi-source data ensemble is a data structure composed of predicted values ​​that characterizes the expected state of the model. It includes electrical parameters derived from current actual monitoring, evolutionary predicted gas parameters generated from the gas evolution prediction branch, and evolutionary predicted temperature parameters generated from the temperature evolution prediction branch. This evolutionary prediction multi-source data ensemble represents the mathematical prediction by the data evolution predictor, based on the learned temporal evolution patterns of fire precursors, of the complete multi-parameter state that the monitoring point should exhibit in the near future under the current electrical and gas parameter inputs.

[0037] By performing the above operations on all location coordinates, multiple evolution prediction multi-source data combinations that correspond one-to-one with the original multi-source data combinations can be obtained.

[0038] Finally, by calculating the similarity between each original multi-source data combination and its corresponding evolutionary prediction multi-source data combination, the degree of agreement between the actual observed data and the predictions made by the data evolution predictor based on the time-series evolution law is evaluated. Specifically, the similarity calculation can be based on the difference or correlation between the predicted gas parameters, predicted temperature parameters, and actual observed values. There are various calculation methods to choose from. For example, Euclidean distance can be used to calculate the overall closeness between the predicted and actual values, cosine similarity can be used to measure the directional consistency between the two in multi-dimensional space, or the prediction accuracy can be quantified by calculating the reciprocal of the mean square error. Preferably, the absolute difference between the predicted and actual gas parameters, and the absolute difference between the predicted and actual temperature parameters can be calculated separately. After normalizing the two differences, the average value is calculated, and then one is subtracted from the average value to obtain a similarity value between 0 and 1. The closer the similarity value is to 1, the better the agreement between the prediction and the actual observation.

[0039] The calculated similarity value represents the credibility score of the current anomalous signal at that location in the time evolution dimension. Organizing and arranging the credibility scores of all locations according to their location coordinates forms a time-verified credibility array. This time-verified credibility array quantifies the degree to which the anomalous signals at each monitoring point within the target space conform to historical evolution patterns.

[0040] S30: Obtain the air flow field in the target space, perform spatial evolution verification on the gas parameter array, and obtain the spatial verification credibility; Because time-verified reliability arrays primarily assess the reliability of anomalous signals based on the inherent patterns of parameter changes over time, they struggle to identify false anomalies caused by localized sensor malfunctions or independent interference sources at monitoring points. These anomalies may coincidentally follow a certain pattern in their time series but violate physical diffusion principles in their spatial distribution. Therefore, it is necessary to introduce a spatial verification mechanism. Based on the physical diffusion characteristics of gas parameters in the air, this mechanism verifies whether the distribution and propagation of anomalous signals within the target space conforms to known airflow patterns. By constructing a complete reliability assessment system with dual temporal and spatial verification, false alarms can be more effectively eliminated, and the accuracy of determining the true location of fire sources can be improved.

[0041] Specifically, the airflow field within the target space is acquired, and the spatial evolution of the gas parameter array is verified to obtain the spatial verification credibility, including: The test obtains the airflow direction within the target space and thus the airflow field. Based on the airflow field, several combinations of position coordinates within multiple position coordinates are selected. Each combination of position coordinates includes upstream position coordinates and downstream position coordinates. Each set of upstream and downstream position coordinates is distributed in the airflow field at the upstream and downstream positions of the airflow direction. Gas parameters are extracted from several multi-source data sets corresponding to several combinations of location coordinates to obtain several gas parameter combinations, where each gas parameter combination includes upstream gas parameters and downstream gas parameters; The system obtains the positional distances of several combinations of location coordinates and extracts several airflow features within the airflow field. These features, combined with the upstream gas parameters within each gas parameter combination, are input into a space gas evolution predictor, which outputs several predicted downstream gas parameters. Calculate the similarity between downstream gas parameters within several gas parameter combinations and several evolution-predicted downstream gas parameters, and calculate the mean to obtain the spatial verification credibility.

[0042] First, the airflow direction within the target space is obtained through on-site measurements or computational fluid dynamics simulations, thereby constructing an airflow field describing the macroscopic path of gas diffusion in the target space. Specifically, this airflow field is a physical field model containing the airflow velocity vector and direction information at various points within the space. It characterizes the overall pattern and intensity distribution of airflow in the target space under normal or specific operating conditions, and can be used to analyze and predict the migration paths and diffusion range of substances such as gases and smoke.

[0043] Secondly, based on the obtained airflow field, several coordinate combinations that conform to the upstream-downstream relationship are selected from the coordinates of all monitoring points. Specifically, each coordinate combination contains two coordinate points, one defined as the upstream coordinate and the other as the downstream coordinate. The criterion for this is that the two points are located exactly on the same streamline or upstream and downstream of the prevailing wind direction within the airflow field. For example, in a scenario with a clear prevailing wind direction, the point located upwind can be considered upstream, and the point located downwind can be considered downstream. The resulting coordinate combinations form a set of monitoring point pairs paired according to the airflow direction, representing potential gas anomaly propagation paths within the target space, and can be used for spatial consistency verification analysis based on diffusion models.

[0044] Then, from the stored multi-source data set, data corresponding to the selected location coordinate combinations is extracted. Specifically, the gas parameter values, such as gas concentration, recorded at the upstream and downstream location coordinates in each location coordinate combination are extracted. Each location coordinate combination will correspond to a gas parameter combination containing both upstream and downstream gas parameter values.

[0045] Then, the actual physical distance between the upstream and downstream points in each location coordinate combination is obtained, i.e., the location distance. Simultaneously, airflow characteristics related to the path of that combination are extracted from the airflow field, such as the average airflow velocity along that path. The upstream gas parameter values, the corresponding location distances between the two points, and the relevant airflow characteristics in each gas parameter combination are used as input data and fed into a pre-trained space gas evolution predictor. This space gas evolution predictor can output a predicted downstream gas parameter value that should theoretically be monitored at that downstream point, based on the input upstream gas state, location distance, and airflow conditions. Performing the above operation on all selected location coordinate combinations yields several predicted downstream gas parameters, the same number as the number of location coordinate combinations.

[0046] Specifically, the steps for obtaining the space gas evolution predictor include: Based on the monitoring data of spatial gas parameter evolution over a historical period, the upstream gas parameter set, the sample air flow characteristic set, and the sample location distance set are collected, and the downstream gas parameters at the downstream location are collected as the downstream gas parameter set for sample evolution prediction. A space gas evolution predictor is constructed based on machine learning; The space gas evolution predictor is trained and tested using the upstream gas parameter set, the sample airflow feature set, the sample location distance set, and the sample evolution prediction downstream gas parameter set as training and testing data until convergence.

[0047] First, a training sample set needs to be constructed from historical monitoring data on the evolution of space gas parameters, which includes spatial correlation information. The historical period refers to a continuous monitoring period used for training the space gas evolution predictor. This timeframe is determined by a combination of factors, including the system's data archiving cycle, the typical timescale of gas diffusion phenomena, and the required number of samples for training. For example, it can be set to cover historical monitoring data spanning at least three complete seasonal operating cycles.

[0048] The data acquisition process focuses on paired monitoring points with clear upstream and downstream relationships. For each pair of historical upstream and downstream points, long-term gas parameter sequences for the upstream point are extracted from the spatial gas parameter evolution monitoring data, forming the sample's upstream gas parameter set. Simultaneously, airflow characteristics recorded concurrently with historical monitoring of the pair, or obtained through simulation, such as average airflow velocity, are extracted, forming the sample's airflow characteristic set. Furthermore, the actual physical distance between the pair of points is recorded, forming the sample's location distance set. Finally, actual gas parameter data for the downstream points corresponding to the aforementioned upstream data are collected. This actual gas parameter data will serve as the prediction target, forming the sample's evolution prediction downstream gas parameter set.

[0049] Furthermore, a space gas evolution predictor is constructed based on machine learning. This space gas evolution predictor is a regression prediction model that takes multivariate data as input and outputs a single gas parameter value. Its input feature dimensions include at least upstream gas parameters, airflow characteristics, and location distance. For example, the structure of this space gas evolution predictor can be a neural network, gradient boosting tree, or other suitable machine learning model.

[0050] Specifically, the upstream gas parameter set, sample airflow feature set, and sample location distance set constructed using the above steps are used as input features for the space gas evolution predictor, while the downstream gas parameter set predicted by the sample evolution is used as the target output that the space gas evolution predictor needs to fit. The space gas evolution predictor is trained and tested until convergence, thus obtaining a trained and usable space gas evolution predictor. This space gas evolution predictor can learn the variation law of downstream gas concentration when the upstream gas state, spatial distance, and airflow conditions are known. The convergence condition is set according to the optimization objective of the model training and the overfitting control requirements. For example, it can be set as the root mean square error of the space gas evolution predictor on the independent validation set decreasing by less than 0.1% for 10 consecutive training cycles, or the total number of model training iterations reaching 1000.

[0051] For example, the space gas evolution predictor can be constructed using a gradient boosting decision tree model, which exhibits good performance in handling multi-source feature fusion and nonlinear regression prediction tasks and has a strong ability to capture the interaction relationships between features.

[0052] Specifically, this space gas evolution predictor mainly consists of a feature input layer, a multi-tree ensemble learning layer, and a concentration prediction output layer. The input layer receives a normalized input feature vector containing data in three dimensions: upstream gas parameters, airflow characteristics, and location distance. The multi-tree ensemble learning layer is composed of multiple decision tree-based learners arranged sequentially. The maximum depth of each decision tree is set between 5 and 8 to balance model complexity and generalization ability. The learning objective of each tree is to fit the residuals of the predictions from all previous trees. By introducing feature subsampling and early stopping mechanisms during training, the risk of overfitting is effectively controlled. The output layer performs a weighted sum of the predictions from all decision trees, ultimately outputting a continuous numerical value as the downstream gas parameter value for evolution prediction.

[0053] During training, key hyperparameters included a learning rate of 0.05, a number of decision trees of 200, and a subsampling ratio of 0.8. The learning rate controlled the fitting strength of each tree to the residuals, the number of decision trees ensured sufficient capacity for the model to learn complex gas diffusion patterns, and the subsampling ratio enhanced the model's robustness. Specifically, a supervised learning training method was adopted. The constructed upstream gas parameter set, sample airflow feature set, and sample location distance set were merged to form the input feature sample set, while the sample evolution prediction of downstream gas parameters served as the target output sample set. The input feature sample set and the corresponding target output sample set were divided into training, validation, and test sets in an 8:1:1 ratio.

[0054] Furthermore, the feature vectors of the samples in the training set are used as input, with the corresponding downstream gas parameters as supervision signals, and a decision tree sequence is iteratively constructed using a gradient boosting algorithm. The mean squared error loss function is used to measure the deviation between the predicted gas concentration values ​​and the actual observed values. The training process is monitored using a validation set. When the value of the validation set loss function decreases by less than 0.1% for 10 consecutive training epochs, the model is considered to have reached convergence, training is terminated, and the final space gas evolution predictor is obtained. This space gas evolution predictor can effectively learn the variation law of downstream gas concentration under the combined effects of upstream gas state, spatial distance, and airflow conditions, achieving accurate predictions consistent with the principles of physical diffusion.

[0055] Furthermore, the upstream gas parameter values ​​recorded in each gas parameter combination, the calculated location distance, and the extracted airflow features are combined to form a multi-dimensional feature vector, which is then input into the trained space gas evolution predictor. This space gas evolution predictor calculates and outputs a corresponding evolution prediction downstream gas parameter value, which represents the theoretical expected gas concentration at the downstream location under given upstream conditions and environmental airflow.

[0056] Performing this process on all selected location coordinate combinations will yield several evolution prediction downstream gas parameters, the same number as the number of location coordinate combinations.

[0057] Finally, for each combination of location coordinates, the similarity between the actual monitored downstream gas parameter values ​​and the evolution-predicted downstream gas parameter values ​​output by the space gas evolution predictor is calculated. Specifically, the similarity calculation can be based on methods such as absolute error, relative error, or correlation coefficient. Furthermore, the arithmetic mean of the similarities for all combinations of location coordinates is calculated, and this mean serves as the spatial verification credibility of the entire target space in this verification. This spatial verification credibility reflects the degree to which the currently monitored anomalous gas distribution conforms to physical diffusion laws in the spatial dimension. The higher the spatial verification credibility value, the more consistent the actual observed gas spatial distribution pattern is with the prediction results based on physical laws, and the higher the probability that the anomalous signal originates from a unified source conforming to diffusion laws; the lower the value, the more significant the deviation between the actual observed values ​​and the theoretical prediction values, and the more likely the current anomalous gas distribution is caused by local interference or sensor errors, resulting in lower spatial credibility.

[0058] S40: Based on the electrical parameter array, gas parameter array, and temperature parameter array, basic early warning parameters are obtained by classification. The parameters are then combined with the time verification credibility array and the spatial verification credibility array for credibility correction to obtain early warning parameters and to perform fire early warning judgment.

[0059] Specifically, based on the electrical parameter array, gas parameter array, and temperature parameter array, basic early warning parameters are obtained by classification. These are then combined with the time-verification reliability array and spatial verification reliability array for reliability correction to obtain early warning parameters. Fire early warning discrimination is then performed, including: Based on the electrical parameter array, gas parameter array, and temperature parameter array, and combined with the electrical parameter threshold, gas parameter threshold, and temperature parameter threshold, the basic early warning parameters are calculated. Based on the time-verification credibility array and spatial verification credibility, the basic early warning parameters are reliably corrected to obtain early warning parameters for fire early warning judgment.

[0060] First, based on real-time monitoring of electrical, gas, and temperature parameter arrays, and combined with pre-set thresholds for these parameters, a preliminary basic warning parameter reflecting the current breadth and intensity of the anomaly is calculated. Preferably, the proportion of monitoring points in each parameter array whose values ​​exceed their corresponding thresholds can be statistically analyzed out of the total number of monitoring points. For example, the proportions of points where electrical parameters exceed their thresholds, gas parameters exceed their thresholds, and temperature parameters exceed their thresholds can be calculated. These three proportions are then weighted and summed to obtain a comprehensive basic warning parameter value.

[0061] Specifically, the basic early warning parameter value is calculated as α*PE + γ*PG + δ*PT. Here, PE, PG, and PT represent the proportions of monitoring points exceeding their corresponding thresholds for electrical, gas, and temperature parameters, respectively. α, γ, and δ are the corresponding weighting coefficients, which are comprehensively set based on the importance of each parameter in fire early warning, historical false alarm statistics, and the specificity of the monitoring environment. For example, in scenarios with a high risk of electrical fires, α can be set to 0.5, γ to 0.3, and δ to 0.2. The higher the calculated basic early warning parameter value, the wider or deeper the range of anomalies within the target space, and the greater the initial likelihood of a fire hazard.

[0062] Furthermore, using the aforementioned temporal verification credibility array and spatial verification credibility, the basic early warning parameters are corrected to obtain more reliable final early warning parameters. Specifically, only when the abnormal signal exhibits high credibility in both temporal evolution and spatial diffusion dimensions should the corresponding early warning be strengthened; conversely, if the credibility is low, the early warning level should be suppressed.

[0063] Specifically, based on the time-verification credibility array and spatial verification credibility, the basic early warning parameters are reliably corrected to obtain early warning parameters, and fire early warning discrimination is performed, including: The mean of the temporal verification confidence array is calculated, and combined with the spatial verification confidence, the fusion confidence is obtained. Obtain benchmark credibility; Based on the fusion credibility and the baseline credibility, the basic early warning parameters are calculated with credibility correction to obtain early warning parameters. It is then determined whether the parameters are greater than or equal to the early warning parameter threshold, and a fire warning is issued.

[0064] First, the arithmetic mean of all elements in the temporal verification credibility array is calculated. This mean represents the overall credibility level of the abnormal signal in the temporal evolution dimension. Simultaneously, this mean is fused with the spatial verification credibility to calculate a fused credibility that comprehensively reflects the results of both spatiotemporal verification. Preferably, the fused credibility is calculated using a weighted average: Fusion Credibility = ω1 × Temporal Comprehensive Credibility + ω2 × Spatial Verification Credibility, where ω1 and ω2 are preset weighting coefficients, and ω1 + ω2 = 1. The weighting coefficients can be set based on the relative importance of temporal and spatial diffusion patterns to fire early warning in specific application scenarios, combined with historical data verification results. For example, in environments with stable airflow and significant spatial diffusion patterns, ω2 can be set to 0.6 and ω1 to 0.4 to emphasize spatial verification; while in scenarios with obvious temporal patterns of electrical fault characteristics, ω1 can be set to 0.7 and ω2 to 0.3. The calculated fusion confidence level is a value between 0 and 1. The closer the value is to 1, the more the currently detected abnormal pattern is highly consistent with the characteristics of fire precursors in both time and space dimensions.

[0065] Next, a preset baseline confidence level is obtained. This baseline confidence level is an empirical or statistical threshold representing the minimum overall confidence level required to reliably trigger an early warning based on historical data. It is set based on the critical point distinguishing effective early warnings from interference noise, determined through analysis of numerous historical real fire alarm cases and false alarm cases. For example, the baseline confidence level can be set to 0.7. This value indicates that only when the spatiotemporal overall confidence assessment of the abnormal signal reaches this level or above is it considered to have sufficiently high reliability to support early warning decisions, thereby effectively controlling the system's false alarm rate while ensuring early warning sensitivity.

[0066] Furthermore, the basic warning parameters are corrected based on the fusion confidence level and the baseline confidence level. Preferably, the ratio of the fusion confidence level to the baseline confidence level is used as a correction factor, and this correction factor is multiplied by the basic warning parameters to obtain the final warning parameters. Specifically, the warning parameter = basic warning parameter * (fusion confidence level / baseline confidence level). When the fusion confidence level is higher than the baseline confidence level, the correction factor is greater than 1, amplifying the basic warning parameters and strengthening the warning signal; when the fusion confidence level is lower than the baseline confidence level, the correction factor is less than 1, suppressing the basic warning parameters and weakening the warning signal that may be caused by interference.

[0067] Finally, the calculated early warning parameters are compared with a pre-set early warning parameter threshold. This threshold is a decision-making critical value determined based on the risk level of the protected site and the acceptable false alarm rate. Its setting principle is to select the optimal discrimination point by balancing the risk of missed alarms and the frequency of false alarms, while ensuring the sensitivity of early fire detection, through historical data statistics and simulation tests. For example, in a typical industrial environment, this threshold can be set to 0.65. If the early warning parameter is greater than or equal to the threshold, a clear fire risk is determined, triggering the corresponding level of fire warning and initiating the preset linkage response procedure. If the early warning parameter is less than the threshold, the current state is determined not to meet the warning standard, and the monitoring status is maintained. Through this mechanism of correction combining spatiotemporal reliability, the accuracy and reliability of early warning decisions can be effectively improved, reducing unnecessary alarms caused by local interference or false alarms.

[0068] In summary, the embodiments of this application have at least the following technical effects: First, by flexibly deploying a multi-dimensional sensor array through wireless networking, synchronous acquisition and real-time transmission of multiple sources of electrical, gas, and temperature parameters were achieved. This solved the problems of inconvenient deployment and limited monitoring dimensions associated with traditional wired systems, providing a data foundation for comprehensive fire precursor detection. Second, after parameter anomalies are triggered, an innovative multi-source data evolution time verification mechanism was introduced. By analyzing the coordinated changes in electrical, gas, and temperature parameters over time, isolated anomalies caused by transient interference or equipment fluctuations were effectively identified, reducing false alarms caused by occasional sensor malfunctions.

[0069] Furthermore, by acquiring the airflow field and verifying the spatial evolution of gas parameters, the rationality of the spatial distribution of abnormal signals is verified from the perspective of physical diffusion principles, thereby further eliminating the influence of local environmental interference and enhancing the spatial reliability of the early warning. Finally, by fusing the credibility obtained from both temporal and spatial verification, the basic early warning parameters calculated based on the threshold exceedance ratio are dynamically corrected, generating the final early warning parameters after credibility calibration. This ensures that early warning decisions not only rely on the intensity and breadth of abnormal signals but also closely integrate their inherent spatiotemporal evolution rationality. In summary, this application can identify real fire hazards earlier, more accurately, and more reliably, while significantly reducing the overall false alarm rate and false negative rate of the system, achieving intelligent and precise hierarchical early warning and fire-fighting linkage.

[0070] Example 2, as Figure 2 As shown, based on the same inventive concept as the wireless network fire early warning method based on multi-source data provided in Embodiment 1, this embodiment of the invention also provides a wireless network fire early warning system based on multi-source data, including: The data acquisition module 11 is used to monitor multi-source data through a sensor array of multiple types of sensors deployed in the target space, obtain electrical parameter arrays, gas parameter arrays and temperature parameter arrays, and send them to the data center platform through wireless networking; Evolution time verification module 12 is used to perform multi-source data evolution time verification based on the electrical parameter array, gas parameter array and temperature parameter array when the trigger parameters are abnormal, and obtain a time verification credibility array. The spatial evolution verification module 13 is used to acquire the air flow field in the target space, perform spatial evolution verification on the gas parameter array, and obtain the spatial verification credibility. The early warning discrimination module 14 is used to classify and obtain basic early warning parameters based on the electrical parameter array, gas parameter array and temperature parameter array, and perform credibility correction by combining the time verification credibility array and spatial verification credibility to obtain early warning parameters and perform fire early warning discrimination.

[0071] Specifically, the data acquisition module 11 is used for: Multi-source data monitoring is performed using a sensor array of various types of sensors deployed within the target space to obtain electrical parameter arrays, gas parameter arrays, and temperature parameter arrays. This data is then transmitted to a data platform via a wireless network, including: Multi-source data is monitored by a sensor array of multiple types of sensors deployed in the target space. The sensor array includes a combination of sensors at multiple location coordinates in the target space. Each sensor combination includes an electrical sensor, a gas sensor, and a temperature sensor. According to the multi-source data type, the multi-source data array is divided to obtain an electrical parameter array, a gas parameter array, and a temperature parameter array, and then sent to the data platform via wireless networking.

[0072] The evolution time verification module 12 is specifically used for: When an abnormality is triggered, multi-source data evolution time verification is performed based on the electrical parameter array, gas parameter array, and temperature parameter array to obtain a time verification reliability array, including: When any electrical parameter, gas parameter, or temperature parameter exceeds the threshold value for that parameter, gas parameter, or temperature parameter, a parameter anomaly is triggered. The electrical parameter array, gas parameter array, and temperature parameter array are divided and combined according to multiple location coordinates to obtain multiple multi-source data combinations. Based on multiple multi-source data combinations, multi-source data evolution time verification is performed to obtain a time verification credibility array.

[0073] Based on multiple multi-source data combinations, multi-source data evolution time verification is performed to obtain a time verification confidence array, including: Multiple data evolution predictors corresponding to multiple location coordinates are obtained, wherein each data evolution predictor includes a gas evolution prediction branch and a temperature evolution prediction branch; The electrical and gas parameters in the multiple multi-source data combinations are input into multiple data evolution predictors, and multiple evolution prediction gas parameters and multiple evolution prediction temperature parameters are output. These are then combined to obtain multiple evolution prediction multi-source data combinations. Calculate the similarity between multiple evolution prediction multi-source data combinations and multiple multi-source data combinations to obtain a time-verified confidence array.

[0074] Obtain multiple data evolution predictors corresponding to multiple location coordinates, including: Obtain facility characteristics at multiple location coordinates, and index multiple similar location coordinates with similar facility characteristics within historical fire monitoring data; Based on historical monitoring data of multiple similar location coordinates, multiple sets of sample electrical parameters were collected when a fire alarm was issued, and gas parameters were collected when different sample electrical parameters were issued, to obtain multiple sets of sample evolution prediction gas parameters. Based on historical monitoring data of multiple similar location coordinates, multiple sets of sample gas parameters were collected when a fire alarm was issued, and temperature parameters were collected when different sample gas parameters appeared, to obtain multiple sets of sample evolution prediction temperature parameters. Based on multiple sets of sample electrical parameters and multiple sets of sample evolution prediction gas parameters, multiple gas evolution prediction branches are constructed. Based on multiple sets of sample gas parameters and multiple sets of sample evolution prediction temperature parameters, multiple temperature evolution prediction branches are constructed. By combining multiple gas evolution prediction branches and multiple temperature evolution prediction branches, multiple data evolution predictors are obtained.

[0075] The spatial evolution verification module 13 is specifically used for: Acquire the airflow field within the target space, perform spatial evolution verification on the gas parameter array, and obtain the spatial verification credibility, including: The test obtains the airflow direction within the target space and thus the airflow field. Based on the airflow field, several combinations of position coordinates within multiple position coordinates are selected. Each combination of position coordinates includes upstream position coordinates and downstream position coordinates. Each set of upstream and downstream position coordinates is distributed in the airflow field at the upstream and downstream positions of the airflow direction. Gas parameters are extracted from several multi-source data sets corresponding to several combinations of location coordinates to obtain several gas parameter combinations, where each gas parameter combination includes upstream gas parameters and downstream gas parameters; The system obtains the positional distances of several combinations of location coordinates and extracts several airflow features within the airflow field. These features, combined with the upstream gas parameters within each gas parameter combination, are input into a space gas evolution predictor, which outputs several predicted downstream gas parameters. Calculate the similarity between downstream gas parameters within several gas parameter combinations and several evolution-predicted downstream gas parameters, and calculate the mean to obtain the spatial verification credibility.

[0076] The steps for obtaining a space gas evolution predictor include: Based on the monitoring data of spatial gas parameter evolution over a historical period, the upstream gas parameter set, the sample air flow characteristic set, and the sample location distance set are collected, and the downstream gas parameters at the downstream location are collected as the downstream gas parameter set for sample evolution prediction. A space gas evolution predictor is constructed based on machine learning; The space gas evolution predictor is trained and tested using the upstream gas parameter set, the sample airflow feature set, the sample location distance set, and the sample evolution prediction downstream gas parameter set as training and testing data until convergence.

[0077] The early warning discrimination module 14 is specifically used for: Based on the electrical parameter array, gas parameter array, and temperature parameter array, basic early warning parameters are obtained by classification. These parameters are then combined with the time-verification reliability array and spatial verification reliability array for reliability correction to obtain early warning parameters. Fire early warning judgment is then performed, including: Based on the electrical parameter array, gas parameter array, and temperature parameter array, and combined with the electrical parameter threshold, gas parameter threshold, and temperature parameter threshold, the basic early warning parameters are calculated. Based on the time-verification credibility array and spatial verification credibility, the basic early warning parameters are reliably corrected to obtain early warning parameters for fire early warning judgment.

[0078] Based on the time-verification credibility array and spatial verification credibility, the basic early warning parameters are reliably corrected to obtain early warning parameters, and fire early warning discrimination is performed, including: The mean of the temporal verification confidence array is calculated, and combined with the spatial verification confidence, the fusion confidence is obtained. Obtain benchmark credibility; Based on the fusion credibility and the baseline credibility, the basic early warning parameters are calculated with credibility correction to obtain early warning parameters. It is then determined whether the parameters are greater than or equal to the early warning parameter threshold, and a fire warning is issued.

[0079] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0080] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0081] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A wireless network-based fire early warning method based on multi-source data, characterized in that, The method includes: By deploying a sensor array of multiple types of sensors in the target space, multi-source data monitoring is performed to obtain electrical parameter arrays, gas parameter arrays, and temperature parameter arrays, which are then transmitted to the data center via wireless networking. When the trigger parameters are abnormal, multi-source data evolution time verification is performed based on the electrical parameter array, gas parameter array, and temperature parameter array to obtain a time verification credibility array. The airflow field within the target space is acquired, and the spatial evolution of the gas parameter array is verified to obtain the spatial verification credibility. Based on the electrical parameter array, gas parameter array, and temperature parameter array, basic early warning parameters are obtained by classification. These parameters are then combined with the time verification credibility array and the spatial verification credibility array for credibility correction to obtain early warning parameters, which are then used for fire early warning judgment.

2. The wireless networking fire early warning method based on multi-source data according to claim 1, characterized in that, Multi-source data monitoring is performed using a sensor array of various types of sensors deployed within the target space to obtain electrical parameter arrays, gas parameter arrays, and temperature parameter arrays. This data is then transmitted to a data platform via a wireless network, including: Multi-source data is monitored by a sensor array of multiple types of sensors deployed in the target space. The sensor array includes a combination of sensors at multiple location coordinates in the target space. Each sensor combination includes an electrical sensor, a gas sensor, and a temperature sensor. According to the multi-source data type, the multi-source data array is divided to obtain an electrical parameter array, a gas parameter array, and a temperature parameter array, and then sent to the data platform via wireless networking.

3. The wireless networking fire early warning method based on multi-source data according to claim 1, characterized in that, When an abnormality is triggered, multi-source data evolution time verification is performed based on the electrical parameter array, gas parameter array, and temperature parameter array to obtain a time verification reliability array, including: When any electrical parameter, gas parameter, or temperature parameter exceeds the threshold value for that parameter, gas parameter, or temperature parameter, a parameter anomaly is triggered. The electrical parameter array, gas parameter array, and temperature parameter array are divided and combined according to multiple location coordinates to obtain multiple multi-source data combinations. Based on multiple multi-source data combinations, multi-source data evolution time verification is performed to obtain a time verification credibility array.

4. The wireless networking fire early warning method based on multi-source data according to claim 3, characterized in that, Based on multiple multi-source data combinations, multi-source data evolution time verification is performed to obtain a time verification confidence array, including: Multiple data evolution predictors corresponding to multiple location coordinates are obtained, wherein each data evolution predictor includes a gas evolution prediction branch and a temperature evolution prediction branch; The electrical and gas parameters in the multiple multi-source data combinations are input into multiple data evolution predictors, and multiple evolution prediction gas parameters and multiple evolution prediction temperature parameters are output. These are then combined to obtain multiple evolution prediction multi-source data combinations. Calculate the similarity between multiple evolution prediction multi-source data combinations and multiple multi-source data combinations to obtain a time-verified confidence array.

5. The wireless networking fire early warning method based on multi-source data according to claim 1, characterized in that, Obtain multiple data evolution predictors corresponding to multiple location coordinates, including: Obtain facility characteristics at multiple location coordinates, and index multiple similar location coordinates with similar facility characteristics within historical fire monitoring data; Based on historical monitoring data of multiple similar location coordinates, multiple sets of sample electrical parameters were collected when a fire alarm was issued, and gas parameters were collected when different sample electrical parameters were issued, to obtain multiple sets of sample evolution prediction gas parameters. Based on historical monitoring data of multiple similar location coordinates, multiple sets of sample gas parameters were collected when a fire alarm was issued, and temperature parameters were collected when different sample gas parameters appeared, to obtain multiple sets of sample evolution prediction temperature parameters. Based on multiple sets of sample electrical parameters and multiple sets of sample evolution prediction gas parameters, multiple gas evolution prediction branches are constructed. Based on multiple sets of sample gas parameters and multiple sets of sample evolution prediction temperature parameters, multiple temperature evolution prediction branches are constructed. By combining multiple gas evolution prediction branches and multiple temperature evolution prediction branches, multiple data evolution predictors are obtained.

6. The wireless networking fire early warning method based on multi-source data according to claim 1, characterized in that, Acquire the airflow field within the target space, perform spatial evolution verification on the gas parameter array, and obtain the spatial verification credibility, including: The test obtains the airflow direction within the target space and thus the airflow field. Based on the airflow field, several combinations of position coordinates within multiple position coordinates are selected. Each combination of position coordinates includes upstream position coordinates and downstream position coordinates. Each set of upstream and downstream position coordinates is distributed in the airflow field at the upstream and downstream positions of the airflow direction. Gas parameters are extracted from several multi-source data sets corresponding to several combinations of location coordinates to obtain several gas parameter combinations, where each gas parameter combination includes upstream gas parameters and downstream gas parameters; The system obtains the positional distances of several combinations of location coordinates and extracts several airflow features within the airflow field. These features, combined with the upstream gas parameters within each gas parameter combination, are input into a space gas evolution predictor, which outputs several predicted downstream gas parameters. Calculate the similarity between downstream gas parameters within several gas parameter combinations and several evolution-predicted downstream gas parameters, and calculate the mean to obtain the spatial verification credibility.

7. The wireless networking fire early warning method based on multi-source data according to claim 6, characterized in that, The steps for obtaining a space gas evolution predictor include: Based on the monitoring data of spatial gas parameter evolution over a historical period, the upstream gas parameter set, the sample air flow characteristic set, and the sample location distance set are collected, and the downstream gas parameters at the downstream location are collected as the downstream gas parameter set for sample evolution prediction. A space gas evolution predictor is constructed based on machine learning; The space gas evolution predictor is trained and tested using the upstream gas parameter set, the sample airflow feature set, the sample location distance set, and the sample evolution prediction downstream gas parameter set as training and testing data until convergence.

8. The wireless networking fire early warning method based on multi-source data according to claim 1, characterized in that, Based on the electrical parameter array, gas parameter array, and temperature parameter array, basic early warning parameters are obtained by classification. These parameters are then combined with the time-verification reliability array and spatial verification reliability array for reliability correction to obtain early warning parameters. Fire early warning judgment is then performed, including: Based on the electrical parameter array, gas parameter array, and temperature parameter array, and combined with the electrical parameter threshold, gas parameter threshold, and temperature parameter threshold, the basic early warning parameters are calculated. Based on the time-verification credibility array and spatial verification credibility, the basic early warning parameters are reliably corrected to obtain early warning parameters for fire early warning judgment.

9. The wireless networking fire early warning method based on multi-source data according to claim 8, characterized in that, Based on the time-verification credibility array and spatial verification credibility, the basic early warning parameters are reliably corrected to obtain early warning parameters, and fire early warning discrimination is performed, including: The mean of the temporal verification confidence array is calculated, and combined with the spatial verification confidence, the fusion confidence is obtained. Obtain benchmark credibility; Based on the fusion credibility and the baseline credibility, the basic early warning parameters are calculated with credibility correction to obtain early warning parameters. It is then determined whether the parameters are greater than or equal to the early warning parameter threshold, and a fire warning is issued.

10. A wireless networked fire early warning system based on multi-source data, characterized in that, The method for performing the wireless networking fire early warning method based on multi-source data according to any one of claims 1-9 includes: The data acquisition module is used to monitor multi-source data through a sensor array of multiple types of sensors deployed in the target space, obtain electrical parameter arrays, gas parameter arrays and temperature parameter arrays, and transmit them to the data platform through wireless networking; The evolution time verification module is used to perform multi-source data evolution time verification based on the electrical parameter array, gas parameter array, and temperature parameter array when the trigger parameters are abnormal, and obtain a time verification credibility array. The spatial evolution verification module is used to acquire the air flow field in the target space, perform spatial evolution verification on the gas parameter array, and obtain the spatial verification credibility. The early warning discrimination module is used to classify and obtain basic early warning parameters based on the electrical parameter array, gas parameter array, and temperature parameter array, and perform credibility correction by combining the time verification credibility array and spatial verification credibility to obtain early warning parameters for fire early warning discrimination.

Images