A fire sprinkler head thermal sensitivity automatic detection method and system

Abstract

The present invention relates to the technical field of automatic detection of sprinkler heads, and specifically to a method and system for automatic detection of thermal sensitivity of fire sprinkler heads. The present invention establishes a thermal detection model for fire sprinkler heads, analyzes temperature characteristics in historical fire data and suspected fire temperature data, obtains first abnormal change data and second abnormal change data, and performs data feature comparison to generate third abnormal change data; simultaneously, obtains temperature change characteristics during firefighting, and obtains fourth abnormal change data; based on ambient temperature data detected in real time by fire sprinkler heads, matches the first abnormal change data and the third abnormal change data in real time to match whether spraying is a real fire or a suspected fire, and performs real-time firefighting adjustment through the fourth abnormal change data during the process of the fire sprinkler heads; the method can automatically detect fire conditions, and perform intelligent spraying according to the fire conditions, thereby improving the accuracy of fire sprinkler spraying.

Description

Technical Field

[0001] The invention relates to the technical field of automatic detection of sprinkler heads, and in particular to a method and system for automatic detection of thermal sensitivity of fire sprinkler heads. Background Art

[0002] As a fire safety facility within a facility, fire sprinklers can start spraying water in time when a fire occurs indoors, so as to minimize the spread of fire and damage to the building, thereby reducing casualties and property losses caused by the fire.

[0003] In order to enable the fire sprinklers to start in time when a fire breaks out, the external temperature is generally sensed by hardware such as thermistors and fire detectors, so as to achieve the purpose of automatic start-up and rapid response of the fire sprinklers.

[0004] Fire sprinklers are usually triggered when they sense flames, smoke or temperatures exceeding a certain threshold. This triggering method is prone to some problems in the actual use of fire sprinklers. For example, false triggering may occur, causing the fire sprinklers to spray water when there is no fire. In the event of a fire, the fire intensity may not match the actual amount of water sprayed.

[0005] The reason for the above problems is that the fire sprinkler head's thermal sensitivity to the external temperature can only be judged through a single specific threshold. The intelligence level of thermal sensitivity sensing is low, and it is impossible to automatically detect the thermal sensitivity of the ambient temperature according to the actual environment, resulting in inaccurate triggering of the fire sprinkler head.

[0006] Therefore, a method and system for automatic detection of thermal sensitivity of fire sprinkler heads are proposed. Summary of the invention

[0007] The purpose of the present invention is to provide a method and system for automatically detecting the thermal sensitivity of fire sprinklers. By establishing a thermal sensitivity detection model for fire sprinklers, a first abnormal change feature and first abnormal change data in historical fire data are analyzed; a second abnormal change feature and second abnormal change data in suspected fire temperature data are analyzed; the first abnormal change data and the second abnormal change data are feature-compared to generate third abnormal change data; the third abnormal change feature and fourth abnormal change data in the firefighting process are simultaneously acquired; based on the ambient temperature data detected in real time by the fire sprinklers, the first abnormal change data and the third abnormal change data are matched in real time to match whether it is a real fire or a suspected fire for spraying, and real-time firefighting adjustments are performed through the fourth abnormal change data during the fire sprinkler process; this method can automatically detect fire conditions, and perform intelligent spraying according to the fire conditions, thereby improving the accuracy of fire sprinkler spraying.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for automatically detecting thermal sensitivity of a fire sprinkler head, comprising:

[0010] Acquire historical fire data of the fire sprinkler, and acquire first temperature change data according to the historical fire data; the first temperature change data includes the ambient temperature data when no fire occurs and the ambient temperature data when a fire occurs;

[0011] Establishing a fire sprinkler thermal detection model, analyzing the first abnormal change feature of the first temperature change data, and screening the first abnormal change data corresponding to the first abnormal change feature;

[0012] Furthermore, the fire sprinkler thermal detection model includes a detection data processing unit, a detection data feature extraction unit, an abnormal change feature mapping matching unit and a data storage marking unit;

[0013] The detection data processing unit preprocesses the first temperature change data, the second temperature change data and the third temperature change data to generate detection preprocessing data, wherein the preprocessing process includes data denoising, data segmentation and data standardization;

[0014] The detection data feature extraction unit includes an Attention mechanism, an LSTM network, a data feature fusion layer and a classifier, extracts the features of the detection preprocessing data, and generates the first abnormal change feature, the second abnormal change feature and the third abnormal change feature;

[0015] The abnormal change feature mapping matching unit performs feature mapping on the first abnormal change feature and the second abnormal change feature to obtain the corresponding first abnormal change data and the second abnormal change data; compares the trend change features of the first abnormal change data and the second abnormal change data, obtains corresponding data according to different trend change features, and generates the third abnormal change data; performs feature mapping on the third abnormal change feature to obtain the corresponding fourth abnormal change data;

[0016] Further, the abnormal change feature mapping matching unit compares the trend change features of the first abnormal change data and the second abnormal change data by using a DTW algorithm;

[0017] The data storage marking unit stores the first abnormal change data, the third abnormal change data and the fourth abnormal change data, and marks corresponding data feature tags;

[0018] Obtain suspected fire temperature data of the fire sprinkler, and obtain second temperature change data based on the suspected fire temperature data; the second temperature change data includes the ambient temperature data when no suspected fire occurs and the ambient temperature data when a suspected fire occurs;

[0019] Input the second temperature change data into the fire sprinkler thermal detection model to obtain a second abnormal change feature and a second abnormal change data; perform data feature comparison on the first abnormal change data and the second abnormal change data to generate third abnormal change data, and store the first abnormal change data and the third abnormal change data;

[0020] Acquire a plurality of third temperature change data when the fire sprinkler is spraying, wherein the third temperature change data is the ambient temperature data during firefighting;

[0021] Analyze a plurality of third abnormal change characteristics according to the fire sprinkler thermal detection model, and store fourth abnormal change data corresponding to each of the third abnormal change characteristics;

[0022] The fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the third abnormal change data from being matched;

[0023] Further, the real-time ambient temperature data is matched with the first abnormal change data and the third abnormal change data according to the DTW algorithm;

[0024] If the suspected fire threshold is met, the third abnormal change data and the first abnormal change data are matched according to the DTW algorithm; if the third abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the third abnormal change data; if the first abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the first abnormal change data;

[0025] Furthermore, the suspected fire threshold is obtained by clustering based on feature comparison data of the first abnormal change data and the second abnormal change data;

[0026] If neither the first abnormal change threshold nor the third abnormal change threshold is reached, an operation is performed according to the data feature label with the largest threshold similarity, and the corresponding data is stored;

[0027] When the first abnormal change data is identified as firefighting, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data. If the match fails, the spraying of the fire sprinkler is strengthened or weakened according to the matching result and the fourth abnormal change data;

[0028] Further, when the first abnormal change data is identified for fire fighting, the real-time environmental temperature data is detected in real time and matched with the fourth abnormal change data for feature matching through the DTW algorithm;

[0029] The present invention also proposes a fire sprinkler head thermal sensitivity automatic detection system, including a first abnormal change data acquisition module, a third abnormal change data acquisition module, a fourth abnormal change data acquisition module and a thermal sensitivity automatic detection module, specifically:

[0030] A first abnormal change data acquisition module is used to acquire historical fire data of fire sprinklers, and acquire first temperature change data based on the historical fire data; establish a fire sprinkler thermal detection model, analyze the first abnormal change characteristics of the first temperature change data, and select the first abnormal change data corresponding to the first abnormal change characteristics;

[0031] A third abnormal change data acquisition module is configured to acquire suspected fire temperature data of a fire sprinkler, and acquire second temperature change data based on the suspected fire temperature data; input the second temperature change data into the fire sprinkler thermal detection model to acquire second abnormal change characteristics and second abnormal change data; compare the first abnormal change data with the second abnormal change data to generate third abnormal change data, and store the first abnormal change data and the third abnormal change data;

[0032] a fourth abnormal change data acquisition module, which acquires a plurality of third temperature change data when the fire sprinkler is spraying, analyzes a plurality of third abnormal change characteristics according to the fire sprinkler thermal detection model, and stores fourth abnormal change data corresponding to each of the third abnormal change characteristics;

[0033] The automatic thermal sensitivity detection module, the fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the matching with the third abnormal change data;

[0034] Further, the real-time ambient temperature data is matched with the first abnormal change data and the third abnormal change data according to the DTW algorithm;

[0035] If the suspected fire threshold is met, the third abnormal change data and the first abnormal change data are matched according to the DTW algorithm; if the third abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the third abnormal change data; if the first abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the first abnormal change data;

[0036] The suspected fire threshold is obtained by clustering based on feature comparison data of the first abnormal change data and the second abnormal change data;

[0037] If neither the first abnormal change threshold nor the third abnormal change threshold is reached, an operation is performed according to the data feature label with the largest threshold similarity, and the corresponding data is stored;

[0038] Further, when the first abnormal change data is identified for fire fighting, the real-time environmental temperature data is detected in real time and matched with the fourth abnormal change data for feature matching through the DTW algorithm;

[0039] When the first abnormal change data is identified for fire fighting, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data. If no match is found, the spraying of the fire sprinkler is strengthened or weakened according to the matching result and the fourth abnormal change data.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. The present invention establishes a thermal detection model for fire sprinkler heads, extracts abnormal temperature change characteristics of historical fire data and suspected fire temperature data, extracts corresponding temperature change data based on each abnormal temperature change characteristic, and excludes abnormal change data of suspected fire temperature data through trend change characteristics of temperature change data; this method can intelligently exclude abnormal non-fire temperature changes of suspected fire during temperature detection of the real-time environment of fire sprinkler heads, so as to reduce interference with automatic detection of suspected fire temperature data by fire sprinkler heads and improve the accuracy of automatic detection.

[0042] 2. The present invention compares the suspected fire temperature data and historical fire data obtained by the fire sprinkler thermal detection model through the DTW algorithm and the clustering algorithm. The difference in the change data of the environmental thermal sensitivity when a real fire occurs and a suspected fire occurs can be grasped through the change law of the data itself, so that the fire sprinkler can perform intelligent and automated real-time detection of environmental thermal sensitivity, thereby improving the accuracy of identifying fire conditions. At the same time, through these screened data, the calculation degree of data analysis can be minimized during the analysis process, so as to provide rapid feedback on the implementation of fire sprinkler spraying.

[0043] 3. The present invention is based on the temperature of the environment when the fire sprinkler is spraying. By collecting real-time environmental temperature change data, the data is matched with the ideal environmental data during fire spraying based on the DTW algorithm. If the match cannot be made, the spraying amount of the sprinkler is adjusted according to the real-time environmental temperature change data. This method can automatically and intelligently detect and adjust the environmental thermal sensitivity during real-time spraying, and automatically adjust the spraying amount of the sprinkler according to the real-time environmental data to achieve more accurate intelligent fire extinguishing. In addition, the stored matching data can also be used as one of the data reference standards for subsequent fire sprinkler spraying quality inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0045] Figure 2 It is a schematic diagram of the flow chart of the fire sprinkler thermal detection model of the present invention;

[0046] Figure 3 It is a structural schematic diagram of a detection data feature extraction unit of a fire sprinkler thermal detection model of the present invention;

[0047] Figure 4 This is a schematic diagram of DTW results of the same data of the present invention;

[0048] Figure 5 Schematic diagram of DTW results of different data of the present invention;

[0049] Figure 6 It is a schematic diagram of the system flow of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] As the most basic and important fire-fighting measure in the room, fire sprinklers spray water by sensing the temperature change of the fire, so as to minimize the fire or extinguish the fire. Nowadays, the fire sprinklers sense the thermal sensitivity of the external temperature, and usually start when a specific threshold is reached. Because the intelligence level of the thermal sensitivity triggering method is low, it may cause false detection for some suspected fire situations. For the fire extinguishing of the actual fire, the fire may not match the actual water spraying amount, resulting in the problem of inaccurate triggering of the fire sprinklers. To this end, the present invention provides a method and system for automatic detection of the thermal sensitivity of fire sprinklers, referring to Figure 1 As shown, the technical solution is as follows:

[0052] Acquire historical fire condition data of the fire sprinkler, and acquire first temperature change data according to the historical fire condition data;

[0053] Establishing a fire sprinkler thermal detection model, analyzing the first abnormal change feature of the first temperature change data, and screening the first abnormal change data corresponding to the first abnormal change feature;

[0054] Acquire suspected fire temperature data of the fire sprinkler, and acquire second temperature change data according to the suspected fire temperature data;

[0055] Input the second temperature change data into the fire sprinkler thermal detection model to obtain a second abnormal change feature and a second abnormal change data; perform data feature comparison on the first abnormal change data and the second abnormal change data to generate third abnormal change data, and store the first abnormal change data and the third abnormal change data;

[0056] Acquire a plurality of third temperature change data when the fire sprinkler nozzle sprays, analyze a plurality of third abnormal change characteristics according to the fire sprinkler nozzle thermal sensitive detection model, and store fourth abnormal change data corresponding to each of the third abnormal change characteristics;

[0057] The fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the match with the third abnormal change data; when the first abnormal change data is identified for fire fighting, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data. If no match is found, the spraying of the fire sprinkler is strengthened or weakened according to the matching result and the fourth abnormal change data.

[0058] Through the historical fire data and suspected fire temperature data, the first temperature change data under the historical fire data is analyzed by establishing a fire sprinkler thermal detection model, and the first abnormal change feature of the normal ambient temperature changing to the fire ambient temperature when a fire occurs is extracted; based on the second temperature change data of the suspected fire temperature data, the second abnormal change feature of the normal ambient temperature changing to the suspected fire temperature when a suspected fire occurs is extracted, and the corresponding abnormal change data in these data is extracted through the fire sprinkler thermal detection model, so as to obtain the corresponding original data of the abnormal feature for data storage, and at the same time, the corresponding features extracted by the model are also convenient for subsequent data comparison, so as to improve the accuracy of data comparison and analysis;

[0059] According to the first abnormal change data corresponding to the first abnormal change feature and the second abnormal change data corresponding to the second abnormal change feature, by comparing the change trends of the first abnormal change data and the second abnormal change data, it is convenient to analyze the same data trends and different data trends of the temperature change trends between the real fire data and the suspected fire temperature data. Through the different data trends, the interference data of the suspected fire temperature data on the fire sprinkler can be excluded. At the same time, through the storage of different data trend characteristics, fast calculation and response can be performed in the actual use process, and intelligent automatic detection can be performed based on different data trend characteristics to reduce interference in the automatic detection situation and improve the accuracy of fire identification;

[0060] Collect temperature change data of fire sprinklers during actual use, and analyze the third abnormal change characteristics of ambient temperature data and the corresponding fourth abnormal change data through the fire sprinkler thermal detection model; match the fourth abnormal change data through real-time detection of temperature change data during the spraying process of fire sprinklers, so as to analyze whether the predetermined fire sprinkler effect is achieved, and automatically correct the spraying effect of the fire sprinklers for unmatched results, so as to achieve automated and intelligent adjustment during the spraying process and improve the accuracy of fire extinguishing when a fire occurs.

[0061] Embodiment 1

[0062] Obtain historical fire data of the fire sprinkler, and obtain first temperature change data based on the historical fire data; the first temperature change data includes the ambient temperature data when no fire occurs and the ambient temperature data when a fire occurs; by obtaining the temperature data when the normal ambient temperature data transitions to the temperature data when the fire occurs, it is convenient to analyze the change law and change characteristics of the data when the fire occurs; the temperature of the fire sprinkler is obtained by a temperature sensor;

[0063] Establishing a fire sprinkler thermal detection model, analyzing the first abnormal change feature of the first temperature change data, and screening the first abnormal change data corresponding to the first abnormal change feature; the first abnormal change feature is the change feature of the temperature data between the ambient temperature data without fire and the ambient temperature data with fire; the first abnormal change data is the original data corresponding to the first abnormal change feature;

[0064] Furthermore, the fire sprinkler thermal detection model includes a detection data processing unit, a detection data feature extraction unit, an abnormal change feature mapping matching unit and a data storage marking unit. Figure 2 As shown;

[0065] The detection data processing unit preprocesses the first temperature change data, the second temperature change data and the third temperature change data to generate detection preprocessing data, wherein the preprocessing process includes data denoising, data segmentation and data standardization;

[0066] The detection data feature extraction unit includes an Attention mechanism, an LSTM network, a data feature fusion layer and a classifier, extracts the features of the detection preprocessing data, and generates the first abnormal change feature, the second abnormal change feature and the third abnormal change feature; Figure 3 As shown, the number of Attention mechanisms is 2, the number of LSTM networks is 6, the number of data feature fusion layers is 1, and the data feature fusion layer in this embodiment adopts relu, and the number of classifiers is 1, which is a fully connected layer and a Softmax classifier in this embodiment;

[0067] The abnormal change feature mapping matching unit performs feature mapping on the first abnormal change feature and the second abnormal change feature to obtain the corresponding first abnormal change data and the second abnormal change data; compares the trend change features of the first abnormal change data and the second abnormal change data, obtains corresponding data according to different trend change features, and generates the third abnormal change data; performs feature mapping on the third abnormal change feature to obtain the corresponding fourth abnormal change data;

[0068] Further, the abnormal change feature mapping matching unit compares the trend change characteristics of the first abnormal change data and the second abnormal change data through the DTW algorithm; in this example, some data sets are obtained based on expert suggestions combined with real ambient temperature data to simulate the ambient temperature data when the fire occurs, and there are a total of 5 groups of data; the 5 groups of data contain changes in different temperatures under different scenarios, and each group of data is marked with abnormal data determined by experts. The corresponding time is obtained through model recognition, and the time marked by the model is compared with the abnormal data time. The recognition accuracy is determined according to how many abnormal data determined by experts are contained in the corresponding time recognized by the model. The recognition accuracy is calculated as follows:

[0069] ;

[0070] in, is the accuracy of model recognition, The number of abnormal data annotated by experts, The number of anomalies annotated by experts for the model to identify;

[0071] The results are shown in Table 1:

[0072] Table 1 Recognition accuracy of five data models

[0073]

[0074] From Table 1, we can see that the recognition accuracy of the models is above 90%, and the recognition results are good;

[0075] During actual fire extinguishing and fire occurrence, considering the time of model calculation and identification, the original data corresponding to a large number of data features extracted by the model are compared with the data features, and the DTW algorithm is used to compare the data for matching. In a short time, the suspected fire temperature data and the real fire data can be automatically distinguished. In the process of ensuring automatic detection, the suspected fire temperature data and the real fire data can be intelligently distinguished to ensure the accuracy of fire sprinkler extinguishing, while also improving the detection speed and achieving rapid response.

[0076] The data storage marking unit stores the first abnormal change data, the third abnormal change data and the fourth abnormal change data, and marks corresponding data feature labels; the data feature labels are mainly used to distinguish the stored first abnormal change data, the third abnormal change data and the fourth abnormal change data, and can be set according to specific implementation conditions, for example, the first abnormal change data is marked as 1, the third abnormal change data is marked as 2, and the fourth abnormal change data is marked as 3; different labels can also be marked according to different fire conditions in different environments, for example, the first abnormal change data of a small fire in an office building is marked as 11, the first abnormal change data of a medium fire in an office building is marked as 12, and the first abnormal change data of a large fire in an office building is marked as 13. The marking method can be combined with expert advice for marking;

[0077] According to the fire sprinkler thermal detection model, the temperature change characteristics of the data under suspected fire and real fire are extracted, and the characteristic data of these temperature changes are identified through the model, which is convenient for the subsequent comparative calculation of the corresponding data; at the same time, the temperature change data characteristics of the fire sprinkler during the fire extinguishing process can also be collected. By extracting these characteristic data, the actual target requirements to be achieved in the fire extinguishing process can be achieved. Through this model, the temperature change characteristics under different conditions can be extracted at the same time. While ensuring the reusability of the model and reducing the cost of model construction, the potential regular characteristics of the data can also be mined, which is convenient for extracting the original data for comparison and improving the accuracy of subsequent automatic detection;

[0078] Obtain suspected fire temperature data of fire sprinklers (for example, but not limited to large-scale smoking in the corridor, burning paper in the corridor, intentional triggering of the fire sprinkler thermal sensor, etc.), and obtain second temperature change data based on the suspected fire temperature data; the second temperature change data includes the ambient temperature data when no suspected fire occurs and the ambient temperature data when a suspected fire occurs; through the change characteristics of the suspected fire temperature data, it is convenient to provide data comparison and the change gap when the real fire occurs, improve the accuracy of data analysis, and provide a more reliable and accurate data basis for subsequent automated and intelligent detection;

[0079] The second temperature change data is input into the fire sprinkler thermal detection model to obtain a second abnormal change feature and a second abnormal change data; the second abnormal change feature is a change feature of the temperature data between the ambient temperature data without suspected fire and the ambient temperature data with suspected fire; the second abnormal change data is the original data corresponding to the second abnormal change feature;

[0080] Comparing data features of the first abnormal change data and the second abnormal change data to generate third abnormal change data, and storing the first abnormal change data and the third abnormal change data;

[0081] Acquire multiple third temperature change data when the fire sprinkler is spraying, wherein the third temperature change data is the ambient temperature data during firefighting; by collecting the temperature change characteristics during the fire extinguishing process, the temperature change conditions under different spraying schemes can be mastered, so that the water volume can reach the predetermined fire extinguishing scheme target during the real-time spraying detection process, thereby improving the accuracy of real-time automatic detection during the fire extinguishing process;

[0082] Analyze multiple third abnormal change characteristics according to the fire sprinkler thermal detection model, and store fourth abnormal change data corresponding to each third abnormal change characteristic; the third abnormal change characteristic is the change characteristic of the ambient temperature data when the fire sprinkler is extinguishing a fire; the fourth abnormal change data is the original data corresponding to the third abnormal change characteristic;

[0083] The fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the third abnormal change data from being matched;

[0084] Further, the real-time ambient temperature data is matched with the first abnormal change data and the third abnormal change data according to the DTW algorithm;

[0085] If the suspected fire threshold is met, the third abnormal change data and the first abnormal change data are matched according to the DTW algorithm; if the third abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the third abnormal change data; if the first abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the first abnormal change data;

[0086] Each data feature label of the first abnormal change data corresponds to a fire extinguishing scheme of a fire sprinkler, and a fire extinguishing scheme of a fire sprinkler corresponds to a data feature label of the fourth abnormal change data; when the first abnormal change data is detected, the data feature label of the fourth abnormal change data will be corresponded according to the corresponding feature label of the first abnormal change data, and the corresponding fire extinguishing scheme of the fire sprinkler will be adopted according to the data feature label of the fourth abnormal change data;

[0087] The third abnormal change threshold and the first abnormal change threshold are the data similarity obtained by the DTW algorithm, which are specifically calculated as follows:

[0088] ;

[0089] in, is the data similarity, DTW is the DTW algorithm, are the first abnormal change threshold and the third abnormal change threshold stored, is the real-time ambient temperature data, and count is the number of unit features consistent with the statistical data; refer to Figure 4 and Figure 5 As shown, Figure 4 and Figure 5 The black part of is the unit feature. Figure 4 The DTW algorithm calculates the results of two identical curves with consistent D1 features. Figure 5 The DTW algorithm calculates the inconsistent results of the two different curves D1 and D2, and count is the comparison Figure 5 and Figure 4 Same number of black parts;

[0090] Figure 4 and Figure 5 The actual number of grids is 4 times the number in the figure, that is Figure 4 and Figure 5 One grid in the grid is equal to four grids. In order to facilitate data storage and graphical display, the original grids are merged. The merging rule is that when the four grids have at least three of the same colors, they are recorded as the same color. For example, if three of the four grids are black, when the four grids are merged into one grid, the color of one grid is black, otherwise it is white.

[0091] The first abnormal change threshold and the third abnormal change threshold are set to 95% by default; the first abnormal change threshold and the third abnormal change threshold can be modified and set based on actual data and expert advice;

[0092] Furthermore, the suspected fire threshold is obtained by clustering based on the feature comparison data of the first abnormal change data and the second abnormal change data; specifically, a two-dimensional coordinate system is established, where the horizontal axis is the number of first abnormal change data and the vertical axis is the number of second abnormal change data, and the suspected fire threshold is obtained as follows: Figure 4 The number of identical black areas of the first abnormal change data and the second abnormal change data shown is used to generate a two-dimensional coordinate mark in a two-dimensional coordinate system, and k-means clustering is performed through a large amount of data, and the number of first abnormal change data at the cluster center point is used as the suspected fire threshold, where the number of k defaults to 1 and can be set and changed according to specific implementation conditions and expert experience;

[0093] Through clustering algorithm analysis, the DTW algorithm can be used to extract the same and different segmentation points of suspected fire temperature data and real fire data when comparing the change characteristics of data. These segmentation points are used as suspected fire thresholds to further improve the intelligence level of automatic detection from the data dimension, while also reducing the scope of misjudgment of fire data and improving the accuracy of fire identification.

[0094] If neither the first abnormal change threshold nor the third abnormal change threshold is reached, an operation is performed according to the data feature label with the largest threshold similarity, and the corresponding data is stored;

[0095] In the process of automatic detection, by matching the real-time detected temperature data with the first abnormal change data, the fire sprinkler can be intelligently tested for environmental thermal sensitivity during use. At the same time, by matching the third abnormal change data, the changes between the suspected fire temperature data and the real fire data can be automatically distinguished during the environmental thermal sensitivity detection process, thereby improving the accuracy of automatic detection.

[0096] When the first abnormal change data is identified as firefighting, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data. If the match fails, the spraying of the fire sprinkler is strengthened or weakened according to the matching result and the fourth abnormal change data;

[0097] Strengthen or weaken the rules Figure 4 and Figure 5 The data is used as a reference. Figure 4 The D1 data is used as the reference. Figure 5 The black area and Figure 4 Black area comparison, statistics Figure 5 Black area offset to Figure 4 The number of white areas below A1, count Figure 5 Black area offset to Figure 4 The number of white areas above A2, if A1>A2, then strengthen it, if A2>A1, then weaken it, if A1=A2, then mark and wait until A1>A2 or A2>A1 appears. If A1=A2 for a long time, there may be a system or data error, and it is necessary to notify the relevant technical personnel for maintenance;

[0098] Furthermore, when firefighting is carried out by identifying the first abnormal change data, the real-time ambient temperature data is detected in real time and the fourth abnormal change data is matched to perform feature matching through the DTW algorithm; through the DTW algorithm matching during the fire extinguishing process, the ambient temperature changes during the fire extinguishing process can be grasped, so as to achieve automatic and intelligent adjustment of the fire fighting head according to the development of the fire, so as to ensure the accuracy of spraying during the fire extinguishing process; at the same time, it is also possible to detect whether there is any abnormality in the spraying of the nozzle based on the matching results of the stored data, further ensuring the accuracy and safety of the nozzle spraying.

[0099] Embodiment 2

[0100] This embodiment also proposes a fire sprinkler head thermal sensitivity automatic detection system, including a first abnormal change data acquisition module, a third abnormal change data acquisition module, a fourth abnormal change data acquisition module and a thermal sensitivity automatic detection module. The system can enable the sprinkler head to perform automatic detection and data storage, facilitate automatic real-time monitoring and data analysis during use, ensure the correct operation of the fire sprinkler head, and improve the accuracy of automatic detection. Figure 6Specifically shown are:

[0101] A first abnormal change data acquisition module is used to acquire historical fire data of fire sprinklers, and acquire first temperature change data based on the historical fire data; establish a fire sprinkler thermal detection model, analyze the first abnormal change characteristics of the first temperature change data, and select the first abnormal change data corresponding to the first abnormal change characteristics;

[0102] In order to enable the fire sprinkler to respond quickly, this embodiment also compares the matching recognition speed of the DTW algorithm in the abnormal environment thermal sensitivity through the abnormal environment thermal sensitivity recognition speed of the detection data feature extraction unit of the fire sprinkler thermal detection model of this embodiment, and still takes the same 5 groups of data as in the first embodiment for comparison. The comparison results are shown in Table 2:

[0103] Table 2 Comparison of anomaly recognition speed between five data models and DTW algorithm

[0104]

[0105] From the results in Table 2, it can be seen that the recognition speed of the DTW algorithm is substantially faster than the recognition time of the fire sprinkler thermal detection model of this embodiment, taking only half the time;

[0106] A third abnormal change data acquisition module is configured to acquire suspected fire temperature data of a fire sprinkler, and acquire second temperature change data based on the suspected fire temperature data; input the second temperature change data into the fire sprinkler thermal detection model to acquire second abnormal change characteristics and second abnormal change data; compare the first abnormal change data with the second abnormal change data to generate third abnormal change data, and store the first abnormal change data and the third abnormal change data;

[0107] a fourth abnormal change data acquisition module, which acquires a plurality of third temperature change data when the fire sprinkler is spraying, analyzes a plurality of third abnormal change characteristics according to the fire sprinkler thermal detection model, and stores fourth abnormal change data corresponding to each of the third abnormal change characteristics;

[0108] The automatic thermal sensitivity detection module, the fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the matching with the third abnormal change data;

[0109] Further, the real-time ambient temperature data is matched with the first abnormal change data and the third abnormal change data according to the DTW algorithm;

[0110] If the suspected fire threshold is met, the third abnormal change data and the first abnormal change data are matched according to the DTW algorithm; if the third abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the third abnormal change data; if the first abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the first abnormal change data;

[0111] The suspected fire threshold is obtained by clustering based on feature comparison data of the first abnormal change data and the second abnormal change data;

[0112] If neither the first abnormal change threshold nor the third abnormal change threshold is reached, an operation is performed according to the data feature label with the largest threshold similarity, and the corresponding data is stored;

[0113] Further, when the first abnormal change data is identified for fire fighting, the real-time environmental temperature data is detected in real time and matched with the fourth abnormal change data for feature matching through the DTW algorithm;

[0114] When the first abnormal change data is identified for fire fighting, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data. If no match is found, the spraying of the fire sprinkler is strengthened or weakened according to the matching result and the fourth abnormal change data.

[0115] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for automatically detecting thermal sensitivity of fire sprinkler nozzles, characterized in that: include: Acquire historical fire condition data of the fire sprinkler, and acquire first temperature change data according to the historical fire condition data; Establishing a fire sprinkler thermal detection model, analyzing the first abnormal change feature of the first temperature change data, and screening the first abnormal change data corresponding to the first abnormal change feature; Acquire suspected fire temperature data of the fire sprinkler, and acquire second temperature change data according to the suspected fire temperature data; Input the second temperature change data into the fire sprinkler thermal detection model to obtain a second abnormal change feature and a second abnormal change data; perform data feature comparison on the first abnormal change data and the second abnormal change data to generate a third abnormal change data, and store the first abnormal change data and the third abnormal change data; Acquire multiple third temperature change data when the fire sprinkler nozzle sprays, analyze multiple third abnormal change characteristics according to the fire sprinkler nozzle thermal detection model, and store fourth abnormal change data corresponding to each of the third abnormal change characteristics; The fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the match with the third abnormal change data; when the first abnormal change data is identified for fire fighting, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data. If no match is found, the spraying of the fire sprinkler is strengthened or weakened according to the matching result and the fourth abnormal change data.

2. A method for automatically detecting thermal sensitivity of fire sprinkler nozzles according to claim 1, characterized in that: The first temperature change data includes the ambient temperature data when no fire occurs and the ambient temperature data when a fire occurs; the second temperature change data includes the ambient temperature data when no suspected fire occurs and the ambient temperature data when a suspected fire occurs; the third temperature change data is the ambient temperature data during firefighting.

3. The method for automatically detecting thermal sensitivity of fire sprinkler nozzles according to claim 1, characterized in that: Establishing a fire sprinkler thermal detection model, analyzing the first abnormal change feature of the first temperature change data, and screening the first abnormal change data corresponding to the first abnormal change feature; inputting the second temperature change data into the fire sprinkler thermal detection model to obtain the second abnormal change feature and the second abnormal change data; Acquire multiple third temperature change data when the fire sprinkler is spraying, and analyze multiple third abnormal change characteristics according to the fire sprinkler thermal detection model, including: The fire sprinkler thermal detection model includes a detection data processing unit, a detection data feature extraction unit, an abnormal change feature mapping matching unit and a data storage marking unit; The detection data processing unit preprocesses the first temperature change data, the second temperature change data and the third temperature change data to generate detection preprocessing data, wherein the preprocessing process includes data denoising, data segmentation and data standardization; The detection data feature extraction unit includes an Attention mechanism, an LSTM network, a data feature fusion layer and a classifier, extracts the features of the detection preprocessing data, and generates the first abnormal change feature, the second abnormal change feature and the third abnormal change feature; The abnormal change feature mapping matching unit performs feature mapping on the first abnormal change feature and the second abnormal change feature to obtain the corresponding first abnormal change data and the second abnormal change data; compares the trend change features of the first abnormal change data and the second abnormal change data, obtains corresponding data according to different trend change features, and generates the third abnormal change data; performs feature mapping on the third abnormal change feature to obtain the corresponding fourth abnormal change data; The data storage marking unit stores the first abnormal change data, the third abnormal change data and the fourth abnormal change data, and marks corresponding data feature tags.

4. A method for automatically detecting thermal sensitivity of fire sprinkler nozzles according to claim 3, characterized in that: The abnormal change feature mapping matching unit compares the trend change features of the first abnormal change data and the second abnormal change data by using a DTW algorithm.

5. The method for automatically detecting thermal sensitivity of fire sprinkler nozzles according to claim 1, characterized in that: The fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the matching with the third abnormal change data, including: According to the DTW algorithm, the real-time ambient temperature data is matched with the first abnormal change data and the third abnormal change data; If the suspected fire threshold is met, the third abnormal change data and the first abnormal change data are matched according to the DTW algorithm; if the third abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the third abnormal change data; if the first abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the first abnormal change data; If neither the first abnormal change threshold nor the third abnormal change threshold is reached, an operation is performed according to the data feature label with the largest threshold similarity, and the corresponding data is stored.

6. A method for automatically detecting thermal sensitivity of fire sprinkler nozzles according to claim 5, characterized in that: The suspected fire threshold is obtained by clustering based on feature comparison data of the first abnormal change data and the second abnormal change data.

7. The method for automatically detecting thermal sensitivity of fire sprinkler nozzles according to claim 1, characterized in that: When fire fighting is performed by identifying the first abnormal change data, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data to perform feature matching through the DTW algorithm.

8. An automatic detection system for thermal sensitivity of fire sprinkler heads, characterized in that: include: A first abnormal change data acquisition module is used to acquire historical fire data of fire sprinklers, and acquire first temperature change data based on the historical fire data; establish a fire sprinkler thermal detection model, analyze the first abnormal change characteristics of the first temperature change data, and select the first abnormal change data corresponding to the first abnormal change characteristics; A third abnormal change data acquisition module is configured to acquire suspected fire temperature data of a fire sprinkler, and acquire second temperature change data based on the suspected fire temperature data; input the second temperature change data into the fire sprinkler thermal detection model to acquire second abnormal change characteristics and second abnormal change data; compare the first abnormal change data with the second abnormal change data to generate third abnormal change data, and store the first abnormal change data and the third abnormal change data; a fourth abnormal change data acquisition module, which acquires a plurality of third temperature change data when the fire sprinkler is spraying, analyzes a plurality of third abnormal change characteristics according to the fire sprinkler thermal detection model, and stores fourth abnormal change data corresponding to each of the third abnormal change characteristics; The automatic thermal sensitivity detection module is used for the fire sprinkler to detect real-time ambient temperature data, match the real-time ambient temperature data with the first abnormal change data in real time, and automatically exclude the third abnormal change data; when the first abnormal change data is identified for fire fighting, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data. If the match fails, the spraying of the fire sprinkler is strengthened or weakened according to the matching result and the fourth abnormal change data.

9. The fire sprinkler head thermal sensitivity automatic detection system according to claim 8, characterized in that: The automatic thermal sensitivity detection module, the fire sprinkler detects real-time ambient temperature data, matches the real-time ambient temperature data with the first abnormal change data in real time, and automatically excludes the matching as the third abnormal change data, including: According to the DTW algorithm, the real-time ambient temperature data is matched with the first abnormal change data and the third abnormal change data; If the suspected fire threshold is met, the third abnormal change data and the first abnormal change data are matched according to the DTW algorithm; if the third abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the third abnormal change data; if the first abnormal change threshold is reached, an operation is performed according to the data feature label corresponding to the first abnormal change data; The suspected fire threshold is obtained by clustering based on feature comparison data of the first abnormal change data and the second abnormal change data; If neither the first abnormal change threshold nor the third abnormal change threshold is reached, an operation is performed according to the data feature label with the largest threshold similarity, and the corresponding data is stored.

10. The fire sprinkler head thermal sensitivity automatic detection system according to claim 8, characterized in that: When fire fighting is performed by identifying the first abnormal change data, the real-time ambient temperature data is detected in real time and matched with the fourth abnormal change data to perform feature matching through the DTW algorithm.

Images