Household self-control fire extinguishing system and method based on artificial intelligence

Abstract

The invention discloses a household self-control fire extinguishing system and method based on artificial intelligence. A smoke temperature sensor carries out smoke and temperature monitoring on a monitoring area where the smoke temperature sensor is located so as to generate smoke temperature data; the self-control fire extinguishing host calls a fire behavior prediction model deployed on the self-control fire extinguishing host to predict whether a fire behavior occurs in the monitoring area according to the smoke temperature data; the flame detection host performs flame detection on the monitoring area to generate first flame data under the condition that the self-control fire extinguishing host predicts that a fire occurs in the monitoring area, so that the self-control fire extinguishing host calls a fire behavior confirmation model deployed on the self-control fire extinguishing host to confirm the fire behavior according to the first flame data; a flame confirmation sensor is mounted on the fire extinguishing medium ejector; after the fire behavior is confirmed, flames are detected again to generate second flame data, so that the self-control fire extinguishing host system determines the position of the fire behavior in the monitoring area according to the smoke temperature data and the second flame data and controls the fire extinguishing medium ejector to be aligned with the position of the fire behavior to spray a fire extinguishing medium according to the position.

Description

Technical Field

[0001] The present invention relates to the field of edge computing, and in particular to an artificial intelligence-based household automatic fire extinguishing system and method. Background Art

[0002] In modern society, residential fires are one of the major hidden dangers that threaten the lives and property safety of residents. According to relevant statistics, a survey of a specific area found that a total of about 244,000 residential fires occurred, resulting in 979 deaths; although the number of residential fires dropped to about 217,000 in 2024, it still caused 999 deaths. Among them, the elderly group suffered more casualties in residential fires. In 2023, 476 people over the age of 60 died, accounting for 48.6% of the total number of deaths in residential fires; in 2024, 483 people over the age of 60 died, accounting for 48.3% of the total number. It can be seen that the elderly group is the main victim of residential fires.

[0003] At present, most of the existing household fire-fighting methods rely on residents to manually operate fire-fighting equipment, such as fire extinguishers and fire blankets. However, due to the decline in physical function, slow reaction speed and inconvenience in movement, the elderly often find it difficult to respond to fires in a timely and effective manner. Even if some houses are equipped with automatic alarm devices, they can only serve as reminders and cannot directly extinguish fires. If the fire cannot be controlled in time at the early stage of the fire, it is easy to cause the fire to spread, resulting in serious casualties and property losses. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a household automatic fire extinguishing system and method based on artificial intelligence to at least solve or alleviate the problems existing in the above-mentioned prior art.

[0005] An artificial intelligence-based household automatic fire extinguishing system, comprising: a smoke temperature sensor, a flame detection host, a fire extinguishing medium injector, and an automatic fire extinguishing host; a target household residential area is divided into a plurality of monitoring areas, so that the smoke temperature sensor, the flame sensor, the flame detection host, and the fire extinguishing medium injector are deployed in units of monitoring areas;

[0006] The smoke temperature sensor is used to monitor the smoke and temperature in the monitoring area to generate smoke temperature data;

[0007] The automatic fire extinguishing host is used to call the fire prediction model deployed on it to predict whether a fire occurs in the monitoring area based on the smoke temperature data;

[0008] The flame detection host is used to perform flame detection on the monitoring area to generate first flame data when the automatic fire extinguishing host predicts that a fire occurs in the monitoring area, so that the automatic fire extinguishing host calls the fire confirmation model deployed thereon to confirm the fire according to the first flame data;

[0009] A flame confirmation sensor is installed on the fire extinguishing medium injector, which is used to re-detect the flame after the fire is confirmed to generate the second flame data, so that the automatic fire extinguishing host system can determine the location of the fire in the monitoring area according to the smoke temperature data and the second flame data, and control the fire extinguishing medium injector to spray the fire extinguishing medium at the location of the fire.

[0010] A household automatic fire extinguishing method based on artificial intelligence, comprising:

[0011] The smoke and temperature sensors are used to monitor the smoke and temperature in the monitoring area to generate smoke and temperature data. The target household residential area is divided into several monitoring areas, and a smoke temperature sensor is deployed in each monitoring area.

[0012] The automatic fire extinguishing host calls the fire prediction model deployed on it to predict whether a fire has occurred in the monitoring area based on the smoke temperature data;

[0013] When the automatic fire extinguishing host predicts that a fire occurs in the monitoring area, the flame detection host deployed in the monitoring area performs flame detection in the monitoring area to generate first flame data;

[0014] The automatic fire extinguishing host calls the fire confirmation model deployed on it to confirm whether the fire has actually occurred based on the first flame data;

[0015] After the fire is confirmed, the flame is detected again by using the flame confirmation sensor installed on the fire extinguishing medium injector to generate second flame data;

[0016] The automatic fire extinguishing host determines the location of the fire in the monitoring area based on the smoke temperature data and the second flame data;

[0017] The automatic fire extinguishing host controls the fire extinguishing medium injector to spray the fire extinguishing medium at the determined fire location.

[0018] The solution provided in this application has the following technical advantages:

[0019] 1. Through the collaborative work of smoke temperature sensors, flame detection hosts and automatic fire extinguishing hosts, residents do not need to manually operate fire extinguishing equipment. The entire process from monitoring to fire extinguishing is automated. Once a fire occurs, the system automatically responds, effectively solving the problem that the elderly cannot respond to fires in a timely and effective manner due to decreased physical function, greatly improving fire extinguishing efficiency and reducing casualties and losses caused by fires.

[0020] 2. The fire prediction model can be called through the automatic fire extinguishing host to predict the fire situation based on the smoke temperature data provided by the smoke temperature sensor. It can make a judgment at the budding stage of the fire, which is more forward-looking than the traditional alarm device that only reminds after the fire occurs. It can find hidden dangers in the early stage of the fire and when the fire is small, buy precious time for fire extinguishing, effectively curb the spread of the fire, and reduce the losses caused by the fire.

[0021] 3. The flame detection host generates the first flame data, and the automatic fire extinguishing host calls the fire confirmation model to confirm the fire; the flame confirmation sensor on the fire extinguishing medium injector generates the second flame data, and the fire location is determined in combination with the smoke temperature data. After accurately locating the fire source, the fire extinguishing medium injector is controlled to spray in a directional manner to avoid blind fire extinguishing, improve the fire extinguishing effect, reduce the waste of fire extinguishing media, and reduce the impact of the fire on the surrounding area. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The drawings constituting a part of this application are used to provide a further understanding of this application, so that other features, purposes and advantages of this application become more obvious. The schematic embodiment drawings and their descriptions of this application are used to explain this application and do not constitute an improper limitation on this application. In the drawings:

[0023] Figure 1 This is a structural schematic diagram of an artificial intelligence-based household automatic fire extinguishing system according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0025] Figure 1 FIG. 1 is a schematic diagram of an artificial intelligence-based household automatic fire extinguishing system according to an embodiment of the present application. Figure 1 As shown, it includes: a smoke temperature sensor, a flame detection host, a fire extinguishing medium injector, and an automatic fire extinguishing host; the target household residential area is divided into several monitoring areas, so that the smoke temperature sensor, the flame sensor, the flame detection host, and the fire extinguishing medium injector are deployed in units of monitoring areas;

[0026] The smoke temperature sensor is used to monitor the smoke and temperature in the monitoring area to generate smoke temperature data;

[0027] The automatic fire extinguishing host is used to call the fire prediction model deployed on it to predict whether a fire occurs in the monitoring area based on the smoke temperature data;

[0028] The flame detection host is used to perform flame detection on the monitoring area to generate first flame data when the automatic fire extinguishing host predicts that a fire occurs in the monitoring area, so that the automatic fire extinguishing host calls the fire confirmation model deployed thereon to confirm the fire according to the first flame data;

[0029] A flame confirmation sensor is installed on the fire extinguishing medium injector, which is used to re-detect the flame after the fire is confirmed to generate the second flame data, so that the automatic fire extinguishing host system can determine the location of the fire in the monitoring area according to the smoke temperature data and the second flame data, and control the fire extinguishing medium injector to spray the fire extinguishing medium at the location of the fire.

[0030] For example, the monitoring area division principles are as follows:

[0031] 1. Zoning by function: Divide the target household living area according to its room function, such as the living room, bedroom, kitchen, bathroom, etc., each of which is an independent monitoring area. Because the causes and characteristics of fires in different functional areas are different, for example, kitchens are prone to fires caused by gas leaks, open flames for cooking, etc.; bedrooms are mostly caused by electrical equipment, bedding, etc., separate monitoring can make more accurate fire-related judgments.

[0032] 2. Consider the area factor: Set the maximum area of ​​each monitoring area, for example, no more than 50 square meters. For larger spaces, such as the open living room and dining room area, set up multiple monitoring points to subdivide it into multiple monitoring areas according to the actual situation, to ensure that the fire-related data in each area can be collected in a timely and accurate manner.

[0033] Equipment installation location and method

[0034] 1. Smoke temperature sensor

[0035] Installation location: In each monitoring area, the smoke temperature sensor should be installed in the center of the ceiling. Because hot air and smoke usually diffuse upward, the center of the ceiling can sense smoke and temperature changes most quickly and comprehensively. If there is a suspended ceiling in the monitoring area, the sensor needs to be installed below the suspended ceiling and flush with the surface of the suspended ceiling. This height can ensure that the sensor can sense effectively and avoid being disturbed by human activities due to being too low.

[0036] Installation method: Use the matching mounting bracket, which is fixed to the ceiling with screws. First, drill holes in the corresponding position of the ceiling, insert expansion bolts, and then tighten the bracket to the expansion bolts with screws. Finally, install the smoke temperature sensor on the bracket and ensure that the sensor is firmly connected to the bracket to avoid looseness affecting data collection.

[0037] 2. Flame sensor

[0038] Installation location: Installed on the wall of the monitoring area, 1.5-2 meters above the ground. The wall is chosen for installation because the wall can detect flame signals earlier when the flame spreads horizontally. At the same time, the flame sensor should be kept at least 1 meter away from the location where the fire may occur to prevent the fire from directly damaging the sensor, and avoid installing it near direct sunlight or strong light sources to avoid false alarms.

[0039] Installation method: Use wall-mounted installation. First, use expansion bolts to fix the sensor's mounting base on the wall. The location of the mounting base must be accurately determined according to the designed installation height and position. After fixing the base, install the flame sensor on the base. Use screws or buckles to ensure that the sensor is tightly connected to the base. The signal transmission line is connected to the sensor from the reserved wire groove on the base.

[0040] 3. Flame detection host

[0041] Installation location: Equipped with a flame detection host, installed in a corner of the area close to the sensor and convenient for line connection and operation and maintenance. The installation height is 1.2-1.5 meters from the ground, which is convenient for staff to view and operate. Choosing a corner location can reduce the occupation of normal activity space and ensure that the host can quickly receive sensor signals.

[0042] Installation method: Use a wall-mounted mounting bracket, first mark the installation position on the wall, drill holes, insert expansion bolts, and fix the mounting bracket on the wall. Then install the flame detection host on the bracket and fix it with screws to ensure that the host is stable and does not shake. The communication interface and power interface of the host should be easy to connect to the line, and certain protective measures should be taken to prevent dust and water vapor from entering.

[0043] 4. Fire extinguishing medium injector

[0044] Installation location: Installed in a location close to the exit of the monitoring area and easy to operate, with an installation height of 1-1.2 meters from the ground, so that personnel can quickly access and operate it in an emergency. At the same time, the spray direction of the ejector should cover the entire monitoring area to ensure that the fire extinguishing medium can be fully and effectively sprayed in the event of a fire. For example, for a rectangular monitoring area, the ejector can be installed in the middle of the long side, and the spray angle can be adjusted to cover the entire area.

[0045] Installation method: Use a special ejector fixing bracket, which is fixed to the wall with expansion bolts. The structure of the fixing bracket should be selected according to the model and weight of the ejector to ensure that it can bear the weight of the ejector and is firmly installed. After the ejector is installed on the bracket, adjust the angle adjustment device on the bracket to make the ejection angle of the ejector meet the requirements of covering the entire monitoring area, and finally use the locking device to fix the angle of the ejector.

[0046] Preferably, in a specific application scenario, the automatic fire extinguishing host is used to call the fire prediction model deployed thereon to perform the following steps to predict whether a fire occurs in the monitoring area based on the smoke temperature data:

[0047] (1) Preprocessing based on the data preprocessing layer:

[0048] The original smoke temperature data sequence X collected by the smoke temperature sensor is 1 ,x 2 ,…,x T ], where x t =[s t ,t t ] T ,s t is the smoke concentration at time t, t t is the temperature at time t, and T is the sequence length. The input data is normalized using the minimum and maximum normalization method to scale the data to [0,1].

[0049] For each feature dimension i (i=1 represents smoke concentration, i=2 represents temperature), we have:

[0050] in, is the value of the i-th feature at time t after normalization.

[0051] The temporal convolutional network layer is used to capture local features and short-term dependencies in smoke temperature data. It consists of multiple convolutional blocks, each of which contains a one-dimensional convolutional layer, a ReLU activation function, and an expanded convolution.

[0052] One-dimensional convolution: Let the input be H l-1 (l represents the number of layers of the convolution block), the convolution kernel is W l , with a bias of b l , then the convolution output Z l for: Among them, K is the size of the convolution kernel, d l is the dilation factor of the lth layer.

[0053] ReLU activation: Perform ReLU activation on the convolution output to get H l :H l =max(0,Z l )

[0054] Capturing long-term dependencies in smoke temperature data based on the gated recurrent unit layer: The output H of the TCN layer TCN .

[0055] GRU unit update gate: in, and is the weight matrix, b z is the bias vector and σ is the sigmoid function.

[0056] Reset the gate: Among them, W r and U r is the weight matrix, b r is the bias vector.

[0057] Candidate hidden states: in, and is the weight matrix, is the bias vector and ⊙ represents element-wise multiplication.

[0058] Hide status updates:

[0059] Based on the attention mechanism layer, different weights are given to the hidden state sequence output by the GRU layer to highlight important information.

[0060] Input: Output H of the GRU layer GRU =[h 1 ,h 2 ,…,h T ] Calculate the attention score: e t =v T tanh(W a h t +b a )W a is the weight matrix, b a is the bias vector and v is the learnable vector.

[0061] Attention weights: Weighted output:

[0062] The output c of the attention mechanism layer is processed based on the fully connected output layer:

[0063] Calculate the predicted probability: The probability p of a fire is obtained through a fully connected layer and a sigmoid function:

[0064] p=σ(W o c+b o )W o is the weight matrix, b o is the bias vector.

[0065] The process of the automatic fire extinguishing host calling the fire prediction model deployed on it to predict whether a fire has occurred in the monitoring area based on the smoke temperature data can be briefly summarized as follows:

[0066] The data preprocessing layer normalizes the original smoke temperature data and converts the data to a suitable range to facilitate subsequent model processing;

[0067] The temporal convolutional network layer performs convolution operations on the normalized data, extracts local features and short-term dependencies, and outputs feature sequences;

[0068] The gated recurrent unit layer receives the output of the TCN layer, captures the long-term dependencies in the data through the gating mechanism, and outputs a hidden state sequence;

[0069] The attention mechanism layer performs weighted processing on the output of the GRU layer, highlighting important time step information and obtaining a weighted feature vector;

[0070] The fully connected output layer maps the output of the attention mechanism layer to a probability value, indicating the possibility of a fire. The automatic fire extinguishing host compares this probability value with the preset threshold to determine whether a fire has occurred.

[0071] The specific implementation process of the above model has significant technical advantages such as high accuracy, high efficiency, and strong adaptability in the application scenario of home fire prediction, which can effectively ensure the fire safety of the family. The detailed principle is explained as follows:

[0072] By the minimum and maximum normalization formula Scale each feature dimension of the smoke temperature data to the $[0,1]$ interval. The core is to eliminate the differences in the dimensions and numerical ranges of different features (such as smoke concentration and temperature) so that all features have the same importance and influence in model training. In a home environment, the collection of smoke temperature data may be interfered by many factors, and the numerical ranges of smoke temperature at different times and in different rooms may vary greatly. Normalization can prevent the model from being overly sensitive or ignoring certain features due to differences in data ranges, and ensure that the model can run stably in various home scenarios. In addition, home scenarios require a quick response to fires and timely measures. The normalized data can allow the model to converge faster, reduce training time, enable the model to adapt to newly collected smoke temperature data more quickly, and improve the timeliness of early warnings.

[0073] Formula for 1D convolution in a TCN layer Sliding convolution on the input sequence through the convolution kernel, combined with the expansion factor d lExpand the receptive field, extract local features of smoke and temperature data, and capture longer-distance dependencies. This can meet the fact that home fires often develop from a local area, such as an overheated appliance in a corner causing smoke and temperature changes. TCN can automatically extract these local features and promptly detect potential fire hazards, such as a sudden increase in temperature or a short-term fluctuation in smoke concentration. In addition, in long-term smoke and temperature data monitoring, TCN's convolution operations can be calculated in parallel, and compared to traditional recurrent neural networks, it can process long sequence data more efficiently, meeting the needs of long-term continuous monitoring of households.

[0074] Through the ReLU activation formula H l =max(0,Z l ) Set the negative values ​​in the convolution output to 0, retain the positive values, introduce nonlinear factors, and enable the model to learn more complex feature representations. Therefore, in a complex home environment, the model may need to train a deeper network to learn the complex patterns of smoke temperature data. The ReLU function can effectively alleviate the gradient vanishing problem, ensure the stability of model training, and enable the model to learn more advanced features. Furthermore, the sparse activation characteristics of the ReLU function make the network sparse, reduce the mutual dependence between neurons, and enable the model to perform well on smoke temperature data in different home environments and different time periods, thereby improving generalization capabilities.

[0075] Through the Gated Recurrent Unit (GRU) layer through the update gate Reset Gate Formulas such as LSTM control the flow and forgetting of information, determine the impact of the hidden state of the previous moment on the current moment, and thus capture the long-term dependencies in the smoke and temperature data. It meets the requirement that it may take some time for a home fire to spread, and the smoke and temperature data will have a long-term change process. GRU can remember this historical information, accurately predict the development trend of the fire, and provide more reliable early warnings for families. Compared with LSTM, GRU has fewer parameters and faster training speed. When the resources of home devices are limited, GRU can reduce the computing cost while ensuring performance, and is more suitable for home automatic fire extinguishing systems.

[0076] Furthermore, the attention score e is calculated through the attention mechanism layer t =v T tanh(W a h t +b a ), attention weight And weighted sum to get Automatically identify the most important time step in smoke temperature data for fire prediction. The attention mechanism can highlight the key information that is truly related to fire, such as the sharp change in smoke temperature before the fire, to improve the model's attention to key information and enhance prediction accuracy. In addition, the characteristics of smoke temperature data will vary depending on the living habits and environment of different families. The attention mechanism can dynamically adjust the attention weight according to different input data, so that the model can adaptively select and utilize the most relevant features in various family scenarios.

[0077] Finally, through the fully connected output layer sigmoid and sigmoid function p = σ (W o c+b o ) maps the output of the attention mechanism layer to a probability value in the interval [0,1], indicating the possibility of a fire. This satisfies the need for simple and intuitive information to make decisions in home scenarios. The probability output allows the automatic fire extinguishing host to clearly determine whether a fire has occurred based on the preset threshold, making it easier to take measures such as fire extinguishing or alarming in a timely manner. Users can easily understand the meaning of the probability value, understand the degree of risk of fire currently facing the family, and enhance the user's trust in the system.

[0078] Specifically, the fire confirmation model includes: 1. Multi-feature extraction module: This module extracts multiple key features from the first flame data, including flame brightness, flickering frequency, color distribution, etc. 2. Fuzzy processing module: Fuzzy the extracted features and map continuous feature values ​​to different fuzzy sets. 3. Fuzzy rule reasoning module: According to the preset fuzzy rule library, the fuzzified features are reasoned to obtain a preliminary judgment on the possibility of fire. 4. Adaptive threshold adjustment module: Dynamically adjust the threshold used to judge the fire according to historical data and current environmental information. 5. Final decision module: Combine the results of fuzzy rule reasoning and the adaptive threshold to make the final fire confirmation decision.

[0079] To this end, the automatic fire extinguishing host calls the fire condition confirmation model deployed thereon to confirm the fire condition according to the first flame data through the following steps:

[0080] Assume that the first flame data is D, the brightness feature extracted from it is L, the flickering frequency feature is F, and the color distribution feature is C. The multi-feature extraction module extracts multiple key features from the first flame data, including flame brightness, flickering frequency, color distribution, etc., through the following formula:

[0081] Among them B i is the brightness value of the i-th pixel, and n is the number of pixels in the flame area. Where m is the number of flame flickers in time t. C can be obtained by counting the color histogram of the flame area.

[0082] Taking brightness feature as an example, let the domain of brightness be [L min ,L max ], and divide it into three fuzzy sets: “Low” (L low ), "Medium" (L medium ), "High" (L high ).

[0083] The fuzzification processing module uses the triangle membership function to fuzzify the extracted features and map the continuous feature values ​​to different fuzzy sets:

[0084]

[0085] Among them, L 1 , L 2 , L 3 , L 4 is the threshold for dividing the fuzzy set.

[0086] The same approach can be applied to blurring flicker frequency and color distribution features.

[0087] The fuzzy rule reasoning module uses the following formula to infer the fuzzified features according to the preset fuzzy rule library and obtain a preliminary judgment on the possibility of fire:

[0088] Assume that there are k rules in the fuzzy rule base, and the jth rule can be expressed as:

[0089] If L is L Aj And F is And C is The fire probability is P j .

[0090] in, is a fuzzy set, P j is the fire probability corresponding to this rule.

[0091] For each rule, calculate its activation α j :

[0092] Then, the fire probability of each rule is weighted and summed according to the activation degree to obtain the preliminary fire probability P pre :

[0093] The adaptive threshold adjustment module dynamically adjusts the threshold used to judge the fire situation based on historical data and current environmental information based on the following formula:

[0094] Assume the average brightness in the historical data is The brightness of the current environment is L env The threshold of the brightness feature can be dynamically adjusted according to the following formula:

[0095] in, is the initial brightness threshold, and β is the adjustment coefficient.

[0096] The same approach can be applied to threshold adjustment of flicker frequency and color distribution characteristics.

[0097] The final adaptive threshold is T = (T L ,T F ,T C ).

[0098] The final decision module makes the final fire confirmation decision based on the following formula combined with the results of fuzzy rule reasoning and adaptive thresholds:

[0099] The initial fire probability P pre Compared with the adaptive threshold T, if P pre If the value is greater than the threshold, it is confirmed that a fire has occurred; otherwise, it is confirmed that no fire has occurred.

[0100]

[0101] Preferably, the automatic fire extinguishing host system performs the following steps to determine the location of the fire in the monitoring area according to the smoke temperature data and the second flame data, and accordingly controls the fire extinguishing medium injector to spray the fire extinguishing medium at the location of the fire:

[0102] 1. Clean and normalize the smoke temperature data and the second flame data based on the following formula for data preprocessing:

[0103] Assume the smoke temperature data is S = [s 1 ,s 2 ,...,s n ], where s i represents the smoke temperature value collected by the i-th sensor; the second flame data is F = [f 1 ,f 2 ,...,f m ], where f j Represents the flame intensity value collected by the jth flame sensor.

[0104] Data cleaning: remove outliers. For smoke temperature data, if s i <μ S -kσ S or i >μ S +kσ S, then s i Replace with μ S , where μ S is the mean of the smoke temperature data, σ S is the standard deviation, k is a constant (usually 3). The flame data is processed in the same way.

[0105] Normalization: Use the minimum and maximum normalization method to scale the data to the [0,1] interval:

[0106] For smoke temperature data:

[0107] For the flame data:

[0108] The key features are extracted from the processed data based on the following formula, and the smoke temperature features and flame features are fused:

[0109] Feature extraction: Extract the temperature change rate ΔT from the smoke temperature data, and for the smoke temperature values ​​s at adjacent time points i and i+1 , Where Δt is the time interval. The flame intensity change rate ΔF is extracted from the flame data and the calculation method is similar.

[0110] Feature fusion: The smoke temperature feature and the flame feature are weighted and fused to obtain the fusion feature M. Suppose the weight of the smoke temperature feature is w S , the flame feature weight is w F , and w S +w F =1.

[0111] For each sensor location, the fused features

[0112] The location of the fire in the monitoring area is determined based on the following formula using the triangulation principle and fusion features:

[0113] N sensors (smoke temperature sensors and flame sensors) are arranged in the monitoring area, and the position coordinates of the sensors are P = [p 1 ,p 2 ,…,p N ], where p i =(x i ,y i ).

[0114] Construct a distance feature relationship model: the fire location is Q = (x, y), and the distance from sensor i to the fire location is Based on experimental data or experience, establish the distance d i And the fusion feature M i The relational model, for example Where a, b, and c are parameters to be determined.

[0115] Solving the fire location: By minimizing the error function To determine the parameters a, b, c and the fire location (x, y), a nonlinear optimization algorithm such as the Levenberg-Marquardt algorithm can be used for the solution.

[0116] The following formula is used to calculate the spray angle and time of the fire extinguishing medium injector according to the fire location, and control it to spray the fire extinguishing medium at the fire location:

[0117] The position of the extinguishing medium injector is R = (x R ,y R ), the maximum injection angle range of the injector is [θ min ,θ max ], the maximum spray distance is D max .

[0118] Calculate the injection angle: Based on the fire position Q = (x, y) and the injector position R = (x R ,y R ), calculate the injection angle θ: And limit it to [θ min ,θ max ] range.

[0119] Calculate the injection time: according to the distance from the fire location to the injector and the spray velocity v of the extinguishing medium, calculate the spray time At the same time, ensure that d ≤ D max .

[0120] Controlled injection: The automatic fire extinguishing host controls the fire extinguishing medium injector to spray the fire extinguishing medium at the fire location according to the calculated injection angle θ and injection time t.

[0121] The above algorithm for implementing the spraying of fire extinguishing media has the following technical benefits:

[0122] For smoke temperature data, by calculating the mean μ S and standard deviation σ S , set the normal data range to [μ S -kσ S ,μ S +kσ S ]. When data point s i When it is outside this range, it is considered an outlier and replaced with the mean μ SThis is based on statistical principles. In a normal distribution, about 99.7% of the data will fall within the interval [μ-3σ,μ+3σ]. Data outside this interval are likely to be outliers caused by sensor errors, external interference, and other factors. The flame data is processed in the same way. In a home fire monitoring scenario, the sensor may be interfered by various factors, such as electromagnetic interference, dust adhesion, etc., resulting in the collection of abnormal data. Removing these outliers can improve the quality and reliability of the data, avoid abnormal data from misleading subsequent fire location judgments and fire extinguishing decisions, and thus improve the accuracy and stability of the entire system.

[0123] The minimum and maximum normalization method is used to scale the smoke temperature data and flame data to the interval [0,1] respectively. By calculating the minimum and maximum values ​​of the data, each data point is mapped to the interval, so that the data collected by different sensors have the same scale, eliminating the differences in data dimensions and numerical ranges. The data collected by different types of sensors (smoke temperature sensors and flame sensors) may have different dimensions and numerical ranges. After normalization, these data have the same importance in subsequent feature fusion and calculation, avoiding the excessive or insufficient impact of certain features on the results due to differences in data scales, and improving the training efficiency and performance of the model.

[0124] The temperature change rate ΔT is extracted from the smoke temperature data. By calculating the ratio of the difference between the smoke temperature values ​​at adjacent time points to the time interval, the temperature change trend over time is reflected. Similarly, the flame intensity change rate ΔF is extracted from the flame data. These change rates can capture the dynamic change information of smoke temperature and flame intensity during the fire. During the development of a household fire, the change rate of temperature and flame intensity is an important indicator for judging the severity and development trend of the fire. For example, a sharp rise in temperature or a rapid increase in flame intensity may mean that the fire is spreading rapidly. Extracting these change rate features can more accurately reflect the real-time status of the fire and provide richer information for subsequent fire location positioning and fire extinguishing decisions.

[0125] The smoke temperature characteristics and flame characteristics are weighted and fused by setting the weight w S and w F , and satisfy w S +w F = 1, the two features are linearly combined according to the weights to obtain the fusion feature M i. The weight setting can be adjusted according to the actual situation to highlight the importance of different features. Both smoke temperature data and flame data can provide information about the fire situation, but using one type of data alone may have limitations. For example, an increase in smoke temperature may be due to other reasons (such as heating of electrical appliances) and does not necessarily mean that a fire has occurred; flame intensity may be affected by factors such as light. Through feature fusion, the advantages of the two types of data can be comprehensively utilized to describe the fire situation more comprehensively and accurately, thereby improving the accuracy of fire judgment.

[0126] Distance d from sensor to fire location i And the fusion feature M i There is an exponential decay relationship This is based on physical principles. Generally, the farther away from the fire source, the smaller the characteristic values ​​such as smoke temperature and flame intensity detected by the sensor. By determining the parameters a, b, and c through experimental data or experience, a quantitative relationship between distance and characteristics can be established. In the home monitoring area, by establishing this distance-feature relationship model, the location of the fire can be inferred based on the fused characteristic values ​​collected by the sensor. This method makes full use of the information of the sensor network and can more accurately locate the fire under the synergy of multiple sensors, avoiding the limitations of single sensor positioning.

[0127] By minimizing the error function To determine the parameters a, b, c and the fire location (x, y). The error function represents the difference between the fusion eigenvalues ​​actually collected and the eigenvalues ​​calculated according to the distance feature relationship model. Use a nonlinear optimization algorithm (such as the Levenberg-Marquardt algorithm) to continuously adjust the parameters and positions to minimize the value of the error function, thereby obtaining the optimal parameters and fire location estimation. In practical applications, due to factors such as sensor measurement errors and environmental interference, the distance feature relationship model may have certain errors. By minimizing the error function, the impact of these errors can be reduced to a certain extent, and the accuracy of fire location positioning can be improved. At the same time, the nonlinear optimization algorithm can find the optimal solution in a complex nonlinear relationship, ensuring the accuracy and reliability of the positioning results.

[0128] According to the principle of trigonometric function, by calculating the fire position Q = (x, y) and the injector position R = (x R ,y R ) and calculate the injection angle θ using the arctan function. At the same time, the injection angle is limited to [θ min ,θ max] range, ensuring that the injector can work within its adjustable angle range. Accurately calculating the injection angle can accurately aim the fire extinguishing medium injector at the fire location and improve fire extinguishing efficiency. In a home environment, the spatial layout may be more complicated. By reasonably controlling the injection angle, the fire extinguishing medium can be prevented from being sprayed into irrelevant areas and damage to other items can be reduced.

[0129] According to the principle of physical kinematics, by calculating the distance d from the fire location to the injector and the injection speed v of the fire extinguishing medium, the formula is used. Calculate the injection time. At the same time, ensure that the distance d does not exceed the maximum injection distance D of the injector max , ensuring that the ejector can effectively cover the fire area. Reasonable control of the ejection time can ensure that sufficient fire extinguishing medium is sprayed to the fire source in the shortest time and the fire is quickly controlled. Avoid the situation where the ejection time is too long, causing the waste of fire extinguishing medium, or the ejection time is too short to effectively extinguish the fire.

[0130] Optionally, the system further includes: a sub-control circuit and a distributor, the sub-control circuit is connected to the automatic fire extinguishing host, and the distributor is connected to the sub-control circuit, so that the automatic fire extinguishing host drives the sub-control circuit to control the distributor to control the fire situation and confirm that the corresponding fire extinguishing medium injector is aimed at the location of the fire to spray the fire extinguishing medium.

[0131] Optionally, the distributor includes a water distributor, which is deployed in the target household residential area in a manner close to a water source. The water distributor is connected to the water source so that after the fire is confirmed, liquid water can be transported through the water distributor to the water spray channel of the fire extinguishing medium injector corresponding to the fire confirmation by starting a water pump to use the liquid as the fire extinguishing medium.

[0132] Optionally, the distributor includes a fire extinguishing agent distributor, which is connected to a fire extinguishing agent storage tank to transport the fire extinguishing agent to a fire extinguishing agent channel of a fire extinguishing medium injector corresponding to the fire confirmation after the fire is confirmed so as to use the fire extinguishing agent as the fire extinguishing medium.

[0133] Preferably, in an application scenario, the system adds a sub-control circuit and a distributor on the original basis to realize the precise control of the fire extinguishing medium injector by the automatic fire extinguishing host. The automatic fire extinguishing host is responsible for receiving fire-related data, making fire judgments and decisions. The sub-control circuit, as an intermediate control link, receives the instructions of the automatic fire extinguishing host and performs signal conversion and amplification. The distributor delivers the corresponding fire extinguishing medium to the corresponding fire extinguishing medium injector according to the control signal of the sub-control circuit, so that it sprays the fire extinguishing medium at the fire location. The sub-control circuit is mainly composed of a microcontroller, a signal conversion module, a power amplifier module and a communication interface.

[0134] Microcontroller: Use high-performance single-chip microcomputer, such as STM32 series single-chip microcomputer. It has multiple general-purpose input and output ports (GPIO), timers and communication interfaces, which can realize the processing and control of various signals. The microcontroller communicates with the automatic fire extinguishing host through the communication interface and receives the fire confirmation signal and control instructions sent by the automatic fire extinguishing host.

[0135] Signal conversion module: It is composed of a level conversion chip and a filter circuit. Since the signal level output by the automatic fire extinguishing host may be incompatible with the microcontroller inside the sub-control circuit, the signal conversion module converts the signal output by the automatic fire extinguishing host to meet the input requirements of the microcontroller. At the same time, the filter circuit filters the signal to remove noise and interference in the signal, thereby improving the stability and reliability of the signal.

[0136] Power amplifier module: It is composed of power transistors and drive circuits. The control signal output by the microcontroller has low power and cannot directly drive the distributor to work. The power amplifier module amplifies the control signal output by the microcontroller so that it can provide enough power to drive the actuators such as water pumps and valves in the distributor.

[0137] Communication interface: RS485 communication interface or CAN communication interface is used, which has the advantages of long communication distance and strong anti-interference ability. The communication interface realizes data transmission and communication between the sub-control circuit and the automatic fire extinguishing host, ensuring that the control instructions can be accurately transmitted to the sub-control circuit.

[0138] When the automatic fire extinguishing host confirms the fire, it sends a control command to the sub-control circuit through the communication interface. After receiving the control command, the microcontroller parses and processes the command and generates a corresponding control signal according to the command content. After the control signal is level-converted and filtered by the signal conversion module, it is input to the power amplifier module for power amplification. The control signal after power amplification is output to the distributor to control the work of the distributor.

[0139] The water distributor is mainly composed of a water inlet, a water outlet, a valve and a shell. The water inlet is connected to a water source (such as a household water pipe or a fire water tank) through a pipe, and the water outlet is connected to the water spray channel of the fire extinguishing medium injector through a pipe. The valve adopts an electric ball valve or an electric butterfly valve, which can be quickly opened and closed according to the control signal of the sub-control circuit. The shell is made of stainless steel or plastic, which has good corrosion resistance and sealing.

[0140] The water distributor is deployed in the target household living area close to the water source, usually installed on the wall or on the ground near the water source. This can reduce the length of the pipeline, reduce the water flow resistance, and improve the water delivery efficiency. At the same time, the installation location of the water distributor should be convenient for maintenance and repair.

[0141] When the sub-control circuit receives the fire confirmation signal from the automatic fire extinguishing host, it sends an opening signal to the valve of the water distributor. After receiving the signal, the valve opens quickly and starts the water pump at the same time. The water pump extracts liquid water from the water source and transports it to the water distributor through the water inlet. The water distributor evenly distributes the water to each outlet, and then transports it to the water spray channel of the fire extinguishing medium injector corresponding to the fire confirmation through the pipeline, so that the injector can use liquid water as the fire extinguishing medium to extinguish the fire.

[0142] The fire extinguishing agent distributor is mainly composed of an inlet, an outlet, a valve, a pressure sensor and a mixer. The inlet is connected to the fire extinguishing agent storage tank through a pipeline, and the outlet is connected to the fire extinguishing agent channel of the fire extinguishing medium injector through a pipeline. The valve adopts an electric stop valve or an electric regulating valve, which can accurately control the flow of the fire extinguishing agent according to the control signal of the sub-control circuit. The pressure sensor is used to monitor the pressure of the fire extinguishing agent in real time to ensure the stability of the delivery pressure of the fire extinguishing agent. The mixer is used to mix the fire extinguishing agent with air or other additives to improve the fire extinguishing effect of the fire extinguishing agent.

[0143] When the sub-control circuit receives the fire confirmation signal from the automatic fire extinguishing host, it sends an opening signal to the valve of the fire extinguishing agent distributor. The valve opens after receiving the signal, and the fire extinguishing agent flows from the fire extinguishing agent storage barrel into the fire extinguishing agent distributor through the inlet. The pressure sensor monitors the pressure of the fire extinguishing agent in real time and feeds back the pressure signal to the sub-control circuit. The sub-control circuit adjusts the opening of the valve according to the pressure signal to ensure the stability of the delivery pressure of the fire extinguishing agent. At the same time, the mixer mixes the fire extinguishing agent with air or other additives, and the mixed fire extinguishing agent is delivered to the fire extinguishing agent channel of the fire extinguishing medium injector corresponding to the fire confirmation through the outlet, so that the injector can use the fire extinguishing agent as the fire extinguishing medium to extinguish the fire.

[0144] Optionally, the system further includes: a natural gas controller connected to the automatic fire extinguishing host, and the natural gas controller shuts off the natural gas valve under the control of the automatic fire extinguishing host after the fire is confirmed.

[0145] Specifically, the specific technical implementation of the above natural gas controller is as follows:

[0146] As a key component of the system, the core function of the natural gas controller is to quickly and reliably shut off the natural gas valve after receiving the fire confirmation signal from the automatic fire extinguishing host, thereby avoiding more serious safety accidents such as fire danger or explosion caused by natural gas leakage. The specific implementation principle of the natural gas controller is as follows:

[0147] Power supply circuit: Provides a stable working power supply for the entire natural gas controller. A switching power supply circuit is used to convert the input 220V AC mains into a DC voltage suitable for the operation of various chips and modules inside the controller, such as 5V and 3.3V. It is usually composed of a transformer, a rectifier bridge, a filter capacitor, a switch tube, and a voltage regulator chip. The transformer steps down the 220V AC voltage, the rectifier bridge converts the AC to DC, the filter capacitor removes the ripple in the DC, and the switch tube and the voltage regulator chip further stabilize the output voltage. For example, using a switching voltage regulator chip such as LM2596 can achieve efficient voltage conversion.

[0148] Communication circuit: realizes data communication between the natural gas controller and the automatic fire extinguishing host, and receives the fire confirmation signal and control command sent by the automatic fire extinguishing host. The RS485 communication interface circuit is adopted, which is composed of RS485 transceiver chip (such as MAX485), terminal resistor and protection circuit. The RS485 transceiver chip is responsible for converting the TTL level signal inside the controller into a differential signal for long-distance transmission. The terminal resistor is used to match the line impedance and reduce signal reflection. The protection circuit is generally composed of a transient suppression diode (TVS) to prevent static electricity and surges from damaging the chip.

[0149] Control circuit: Control the opening and closing action of the natural gas valve according to the received control instructions. With a microcontroller (such as Arduino or STM32 series single-chip microcomputer) as the core, the microcontroller receives the signal from the automatic fire extinguishing host through the communication circuit, and outputs the control signal to the relay drive circuit after processing. The relay drive circuit is composed of a transistor, a relay and a freewheeling diode. The transistor is used as a switching element to control the closing and disconnection of the relay. When the relay is closed, the valve drive power is turned on to realize the closing action of the valve.

[0150] Status detection circuit: monitors the switch status of the natural gas valve in real time and feeds back the status information to the automatic fire extinguishing host. Uses detection elements such as micro switches or Hall sensors. When the valve is in the open or closed state, the micro switch or Hall sensor outputs the corresponding level signal. The microcontroller determines the state of the valve by reading these signals and sends the status information to the automatic fire extinguishing host through the communication circuit.

[0151] Installation location: Install in a dry, well-ventilated location that is easy to operate and maintain. Avoid installing in humid, corrosive, high-temperature or collision-prone locations. Usually installed near the main valve of the natural gas pipeline, such as on the wall near the entrance of the natural gas pipeline in the kitchen. The installation height is generally 1.2-1.5 meters from the ground to facilitate personnel inspection and maintenance. At the same time, ensure that the controller maintains a certain safe distance from other surrounding equipment to avoid mutual interference.

[0152] If the natural gas valve is controlled by an electric actuator, the natural gas controller is connected to the drive shaft of the electric actuator of the valve through a mechanical transmission component (such as a coupling) to ensure that the controller can accurately transmit the torque to the valve to realize the opening and closing of the valve. If it is a pneumatic valve, the controller controls the pneumatic solenoid valve to adjust the air source of the valve, thereby realizing the opening and closing control of the valve. At this time, the controller and the pneumatic solenoid valve are connected through an air pipe. The natural gas controller is usually wall-mounted, and the mounting plate of the controller is fixed to the wall with expansion bolts or screws to ensure a firm and reliable installation.

[0153] Use shielded twisted pair cable to connect the natural gas controller to the automatic fire extinguishing host through RS485 interface. When connecting, pay attention to the polarity and signal definition of the interface to ensure normal communication. At the same time, the terminal resistors should be correctly connected at both ends of the communication line to ensure the stability of signal transmission.

[0154] Connect the natural gas controller to the 220V AC mains power supply through a power cord. A suitable fuse or circuit breaker should be installed on the power cord to prevent damage to the controller due to overcurrent or short circuit.

[0155] The output end of the control circuit of the controller is connected to the control end of the valve drive circuit (such as a relay) through a wire to realize the switch control of the valve. At the same time, the state detection signal of the valve is fed back to the input end of the state detection circuit of the controller through a wire.

[0156] Specifically, after the fire is confirmed, the natural gas controller shuts off the natural gas valve under the control of the automatic fire extinguishing host, which specifically includes the following steps:

[0157] The communication circuit of the natural gas controller monitors the signal from the automatic fire extinguishing host in real time. When receiving the fire confirmation signal sent by the automatic fire extinguishing host, the communication circuit converts the differential signal into a TTL level signal and transmits it to the microcontroller;

[0158] The microcontroller decodes and verifies the received signal to determine the validity of the signal and the content of the instruction. If it is confirmed to be an instruction to shut off the natural gas valve, the microcontroller enters the control process.

[0159] The microcontroller outputs a high-level signal to the relay drive circuit, the transistor is turned on, the relay is closed, the valve drive power is turned on, and the valve starts to close. At the same time, the microcontroller starts the timer to time the valve closing process.

[0160] The status detection circuit monitors the status of the valve in real time. When the valve is fully closed, the micro switch or Hall sensor outputs the corresponding status signal. After the microcontroller reads the signal, it confirms that the valve is closed and feeds back the valve closing status information to the automatic fire extinguishing host through the communication circuit.

[0161] If an abnormal situation occurs during the valve closing process, such as failure to close after timeout or abnormal status detection signal, the microcontroller will send a fault message to the automatic fire extinguishing host through the communication circuit, and try to re-control the valve or take other emergency measures.

[0162] Optionally, the system further includes: a power-off controller and a backup power supply, which are connected to the automatic fire extinguishing host. After the fire is confirmed, the power-off controller shuts down the target home living area under the control of the automatic fire extinguishing host, and keeps the backup power supply to power the artificial intelligence home automatic fire extinguishing system.

[0163] Preferably, the working principle of the above power failure controller and backup power supply is described as follows:

[0164] When a fire occurs in the target household residential area, the power-off controller and backup power supply in the system play an important role. After the automatic fire extinguishing host confirms the fire, the power-off controller will quickly cut off the regular power supply to the target household residential area to avoid secondary disasters caused by electrical line failures; at the same time, the backup power supply continues to supply power to the artificial intelligence household automatic fire extinguishing system, ensuring that the system can still operate normally in the event of a power outage and continue to perform fire-fighting related operations.

[0165] Power-off controller circuit structure:

[0166] Main control circuit: With a microcontroller (such as STM32 series single-chip microcomputer) as the core. The microcontroller receives the fire confirmation signal from the automatic fire extinguishing host, processes and judges the signal, and then outputs the control instruction. It has functions such as data processing, logic judgment and signal output, and controls the entire power-off controller through the internal program code.

[0167] Signal receiving circuit: Use photocouplers to isolate and convert the signal output by the automatic fire extinguishing host to meet the input requirements of the microcontroller. Photocouplers can effectively avoid interference from external signals to the microcontroller and improve the stability and reliability of the system. For example, use PC817 photocouplers to isolate and connect the signal of the automatic fire extinguishing host to the input port of the microcontroller.

[0168] Relay drive circuit: It consists of transistors and relays. The control signal output by the microcontroller is amplified by the transistor and drives the relay to operate. The contacts of the relay are used to control the power on and off of the target household living area. When the microcontroller outputs a high-level signal, the transistor is turned on and the relay is energized; when the microcontroller outputs a low-level signal, the transistor is turned off and the relay is disconnected.

[0169] Power supply circuit: Provides stable working power for each circuit module of the power-off controller. A switching power supply chip (such as LM2596) is used to convert the input AC power into a DC voltage suitable for the microcontroller and other circuit modules, such as 5V and 3.3V. The power supply circuit is also equipped with filter capacitors and voltage regulator diodes to ensure the stability and purity of the output voltage.

[0170] The power-off controller is installed near the distribution box in the target household living area to facilitate connection with the circuit in the distribution box. The installation location should be dry and well ventilated, and avoid installation in a humid, high-temperature or corrosive gas environment. The installation height is generally 1.5-1.8 meters from the ground to facilitate operator inspection and maintenance. At the same time, ensure that there is enough space around the controller for heat dissipation and wiring.

[0171] The power-off controller is fixed to the wall near the distribution box through a mounting bracket. The mounting bracket is made of metal and has sufficient strength and stability. The contacts of the relay are connected to the main switch or branch switch in the distribution box through wires to control the power supply of the target household living area. The wires should be of appropriate specifications to ensure that they can withstand the current in the circuit.

[0172] Use shielded twisted pair cables to connect the signal receiving circuit of the power-off controller to the signal output port of the automatic fire extinguishing host. The shielding layer should be reliably grounded to prevent electromagnetic interference. The contacts of the relay are connected to the power line in the distribution box through wires. When connecting, pay attention to the polarity and connection sequence of the wires to ensure correct connection. The power circuit of the power-off controller is connected to the mains power supply through the power cord. A fuse should be installed on the power cord to prevent damage to the controller in case of a short circuit.

[0173] The working process of the power-off controller and the backup power supply is as follows:

[0174] Signal reception: When the automatic fire extinguishing host confirms the fire, it sends a fire confirmation signal to the power off controller. After the signal receiving circuit receives the signal, it isolates and converts the signal through the photocoupler and transmits it to the microcontroller.

[0175] Signal processing: The microcontroller analyzes and judges the received signal. If it is confirmed to be a fire confirmation signal, the microcontroller executes the corresponding control program.

[0176] Control output: The microcontroller outputs a high-level signal to the relay drive circuit, the transistor is turned on, the relay is closed, and the power supply to the target household living area is cut off. At the same time, the microcontroller records the power-off time and related information.

[0177] Status feedback: The power failure controller feeds back the power failure status information to the automatic fire extinguishing host through the signal feedback circuit. The automatic fire extinguishing host can process and record the feedback information accordingly.

[0178] In addition, for the backup power supply, its circuit structure is:

[0179] Charging circuit: It is composed of a charging chip (such as TP4056) and related resistors and capacitors. The charging chip charges the battery with constant current and constant voltage according to the battery type (such as lithium battery) and charging requirements of the backup power supply. When the battery power is lower than the set value, the charging circuit automatically starts to charge the battery; when the battery is full, the charging circuit automatically stops charging to prevent overcharging.

[0180] Inverter circuit: Converts the DC power output by the battery into AC power to power the artificial intelligence home automatic fire extinguishing system. The inverter circuit is composed of an inverter chip (such as an SPWM inverter) and a power tube. The inverter chip generates a pulse width modulation (PWM) signal to control the conduction and cutoff of the power tube and convert DC power into AC power.

[0181] Battery management circuit: real-time monitoring of battery voltage, current, temperature and other parameters. When the battery is over-voltage, under-voltage, over-current or over-heated, the battery management circuit automatically takes protective measures, such as cutting off the charging or discharging circuit, to prevent battery damage.

[0182] The backup power supply should be installed in a dry, well-ventilated and easy-to-maintain location. It can be installed near the automatic fire extinguishing host or in a dedicated backup power supply cabinet. The installation location should be away from fire sources and flammable items to avoid fire-related safety issues. At the same time, ensure that there is enough space around the backup power supply for heat dissipation and operation.

[0183] The backup power supply is fixed to the installation location by a mounting bracket. The mounting bracket must have sufficient strength and stability to bear the weight of the backup power supply. The battery is fixed inside the backup power supply by a battery rack, which must be made of insulating material to prevent the battery from short-circuiting.

[0184] The output end of the backup power supply is connected to the power input port of the automatic fire extinguishing host through a power cord. When connecting, pay attention to the polarity and voltage matching of the power supply to ensure that a stable power supply is provided to the automatic fire extinguishing host. The charging circuit of the backup power supply is connected to the mains power supply through a power cord. A fuse should be installed on the power cord to prevent damage to the backup power supply in the event of a short circuit.

[0185] Preferably, the principle of backup power supply participating in power supply is described as follows:

[0186] Charging control: When the mains power is supplied normally, the charging circuit automatically charges the backup power battery. The battery management circuit monitors the battery charging status in real time and automatically stops charging when the battery is fully charged.

[0187] Inverter power supply: When the power outage controller cuts off the power supply to the target home living area, the inverter circuit of the backup power supply automatically starts to convert the DC power output by the battery into AC power to power the artificial intelligence home automatic fire extinguishing system.

[0188] Status monitoring: The battery management circuit monitors the battery voltage, current, temperature and other parameters in real time, and transmits the monitoring information to the automatic fire extinguishing host. The automatic fire extinguishing host can judge the working status of the backup power supply based on the monitoring information and take corresponding measures.

[0189] Optionally, the system further includes: an alarm connected to the automatic fire extinguishing host, which is used to issue a fire alarm under the control of the automatic fire extinguishing host after the fire is confirmed.

[0190] Specifically, the implementation principle of the above alarm is as follows:

[0191] As an important part of the household automatic fire extinguishing system, the alarm has the core function of sending out an obvious alarm signal quickly after the automatic fire extinguishing host confirms the fire, so as to remind the residents to take timely response measures to ensure the safety of life and property. The alarm circuit includes:

[0192] (I) Power supply circuit: Provide stable working power for each module of the alarm. Use a transformer to reduce the voltage of 220V AC mains, for example, use a step-down transformer with a suitable transformation ratio to reduce the voltage to about 12V. Then convert the AC power into DC power through a rectifier bridge (such as a full-wave rectifier bridge composed of four diodes). Then pass through a filter capacitor (such as a large-capacity electrolytic capacitor) to remove the ripple in the DC power, and finally use a voltage regulator chip (such as 7812 or LM317, etc.) to further stabilize the output voltage to ensure the output of 12V stable DC power for subsequent circuits.

[0193] (ii) Signal receiving circuit: Receive the fire confirmation signal from the automatic fire extinguishing host. Use a photocoupler (such as PC817) for signal isolation and conversion. The signal output by the automatic fire extinguishing host passes through the input end of the photocoupler. When there is a signal input, the light-emitting diode inside the photocoupler emits light, turning on the phototransistor at the output end and passing the signal to the subsequent control circuit. This can effectively avoid the influence of external interference on the signal of the automatic fire extinguishing host, and prevent the alarm circuit failure from causing damage to the automatic fire extinguishing host.

[0194] (III) Control circuit: Control the alarm to sound and light according to the received signal. The microcontroller (such as 51 single-chip microcomputer or Arduino, etc.) is the core. The microcontroller receives the signal from the signal receiving circuit, and after judgment by the internal program, outputs the control signal to the sound and light driving circuit. For example, when a fire confirmation signal is received, the microcontroller outputs high and low level signals according to the preset program to control the sound frequency and light mode of the alarm.

[0195] (IV) Sounding circuit: emits a loud alarm sound. Usually a buzzer or speaker is used as the sounding element. For an active buzzer, it only needs to apply a suitable voltage signal (usually a high level) to its control end to make a sound; for a speaker, it is necessary to amplify the audio signal output by the microcontroller through a power amplifier circuit (such as the LM386 audio power amplifier) ​​and then drive the speaker to make a loud sound.

[0196] (V) Lighting circuit: emits eye-catching warning lights. Use light-emitting diodes (LEDs) as light-emitting elements. Connect the LED to the output port of the microcontroller through a current-limiting resistor. The high and low level signals output by the microcontroller control the on and off of the LED. In order to enhance the warning effect, multiple LEDs of different colors can be used to form a flashing or alternating light-emitting mode.

[0197] It should be installed in a location that is easily noticed by residents and ensure that signal transmission is not obstructed. Avoid installing it in an environment that may affect its normal operation.

[0198] Specific location

[0199] Living room: Installed in the center of the living room ceiling or on the wall near the door, the height is generally 2.2-2.5 meters from the ground. This allows the entire living room to clearly hear the alarm sound and see the alarm light, making it easier for residents to find fire in time when they are in the living room.

[0200] Bedroom: Installed above the bedside or on the wall near the bedroom door, at a height of 2-2.2 meters from the ground. Ensure that the occupants can be awakened by the alarm signal even when they are sleeping.

[0201] Kitchen: Install it on the kitchen ceiling, but keep it away from stoves, range hoods and other equipment to prevent oil smoke and high temperature from damaging the alarm.

[0202] Alarms are usually installed on the wall or in the ceiling. For wall-mounted installation, use expansion bolts or screws to fix the alarm's mounting plate to the wall, and then firmly connect the alarm body to the mounting plate with buckles or screws. For ceiling-mounted installation, first reserve suitable mounting holes on the ceiling, and then use a hanger or screws to fix the alarm to the ceiling. The alarm itself generally has no direct mechanical connection to other components, but make sure there are no obstacles around its installation location to prevent sound and light from being transmitted.

[0203] Use shielded twisted pair cables to connect the signal receiving circuit of the alarm to the signal output port of the automatic fire extinguishing host. The shielding layer should be reliably grounded to prevent electromagnetic interference. Pay attention to the polarity and definition of the signal when connecting to ensure that the signal can be transmitted correctly.

[0204] Connect the power circuit of the alarm to the 220V AC mains through the power cord. A suitable fuse or circuit breaker should be installed on the power cord to prevent damage to the alarm in case of short circuit or overload.

[0205] Preferably, the steps for implementing the above alarm are as follows:

[0206] When the automatic fire extinguishing host confirms the fire, it sends a fire confirmation signal to the alarm. The photocoupler in the signal receiving circuit receives the signal and converts it into a signal suitable for microcontroller processing and transmits it to the microcontroller.

[0207] The microcontroller analyzes and judges the received signal. If it is confirmed to be a fire confirmation signal, the microcontroller starts to execute the alarm control logic according to the preset program.

[0208] Sound control: The microcontroller outputs a control signal to the sound circuit, and controls the buzzer or speaker to emit a corresponding alarm sound according to the preset alarm mode (such as continuous sounding, interval sounding, etc.).

[0209] Lighting control: The microcontroller simultaneously outputs a control signal to the lighting circuit to control the LED to emit an alarm light according to a preset lighting mode (such as flashing, alternating lighting, etc.).

[0210] The alarm can feed back the alarm status information to the automatic fire extinguishing host through the signal feedback circuit. For example, when the alarm is normally activated, a feedback signal is sent to the automatic fire extinguishing host to inform the automatic fire extinguishing host that the alarm is working properly.

[0211] Sample code (using Arduino as an example)

[0212]

[0213]

[0214] The embodiment of the present application also provides a household automatic fire extinguishing method based on artificial intelligence, which includes:

[0215] The smoke and temperature sensors are used to monitor the smoke and temperature in the monitoring area to generate smoke and temperature data. The target household residential area is divided into several monitoring areas, and a smoke temperature sensor is deployed in each monitoring area.

[0216] The automatic fire extinguishing host calls the fire prediction model deployed on it to predict whether a fire has occurred in the monitoring area based on the smoke temperature data;

[0217] When the automatic fire extinguishing host predicts that a fire occurs in the monitoring area, the flame detection host deployed in the monitoring area performs flame detection in the monitoring area to generate first flame data;

[0218] The automatic fire extinguishing host calls the fire confirmation model deployed on it to confirm whether the fire has actually occurred based on the first flame data;

[0219] After the fire is confirmed, the flame is detected again by using the flame confirmation sensor installed on the fire extinguishing medium injector to generate second flame data;

[0220] The automatic fire extinguishing host determines the location of the fire in the monitoring area based on the smoke temperature data and the second flame data;

[0221] The automatic fire extinguishing host controls the fire extinguishing medium injector to spray the fire extinguishing medium at the determined fire location.

[0222] In the above method embodiment, the exemplary description of each step refers to the description of the above system.

[0223] Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A household automatic fire extinguishing system based on artificial intelligence, characterized in that: include: Smoke temperature sensor, flame detection host, fire extinguishing medium injector, automatic fire extinguishing host; The target household residential area is divided into several monitoring areas, so that the smoke temperature sensor, flame sensor, flame detection host, and fire extinguishing medium injector are deployed in units of monitoring areas; The smoke temperature sensor is used to monitor the smoke and temperature in the monitoring area to generate smoke temperature data; The automatic fire extinguishing host is used to call the fire prediction model deployed on it to predict whether a fire occurs in the monitoring area based on the smoke temperature data; The flame detection host is used to perform flame detection on the monitoring area to generate first flame data when the automatic fire extinguishing host predicts that a fire occurs in the monitoring area, so that the automatic fire extinguishing host calls the fire confirmation model deployed thereon to confirm the fire according to the first flame data; A flame confirmation sensor is installed on the fire extinguishing medium injector, which is used to re-detect the flame after the fire is confirmed to generate the second flame data, so that the automatic fire extinguishing host system can determine the location of the fire in the monitoring area according to the smoke temperature data and the second flame data, and control the fire extinguishing medium injector to spray the fire extinguishing medium at the location of the fire.

2. The artificial intelligence household automatic fire extinguishing system according to claim 1 is characterized in that: The system also includes: a sub-control circuit and a distributor, the sub-control circuit is connected to the automatic fire extinguishing host, and the distributor is connected to the sub-control circuit, so that the automatic fire extinguishing host drives the sub-control circuit to control the distributor to control the fire situation and confirm that the corresponding fire extinguishing medium injector is aimed at the location of the fire to spray the fire extinguishing medium.

3. The artificial intelligence household automatic fire extinguishing system according to claim 2 is characterized in that: The distributor includes a water distributor, which is deployed in the target household residential area in a manner close to a water source. The water distributor is connected to the water source so that after the fire is confirmed, liquid water can be transported through the water distributor to the water spray channel of the fire extinguishing medium injector corresponding to the fire confirmation by starting a water pump to use the liquid as the fire extinguishing medium.

4. The artificial intelligence household automatic fire extinguishing system according to claim 1 is characterized in that: The distributor includes a fire extinguishing agent distributor, which is connected to the fire extinguishing agent storage barrel to transport the fire extinguishing agent to the fire extinguishing agent channel of the fire extinguishing medium injector corresponding to the fire confirmation after the fire is confirmed so as to use the fire extinguishing agent as the fire extinguishing medium.

5. The artificial intelligence household automatic fire extinguishing system according to claim 1 is characterized in that: The system also includes: a natural gas controller connected to the automatic fire extinguishing host. After the fire is confirmed, the natural gas controller shuts off the natural gas valve under the control of the automatic fire extinguishing host.

6. The artificial intelligence household automatic fire extinguishing system according to claim 1 is characterized in that: The system also includes: a power-off controller and a backup power supply, which are connected to the automatic fire extinguishing host. After the fire is confirmed, the power-off controller shuts down the target home living area under the control of the automatic fire extinguishing host, and keeps the backup power supply to power the artificial intelligence home automatic fire extinguishing system.

7. The artificial intelligence household automatic fire extinguishing system according to claim 1 is characterized in that: The system also includes: an alarm connected to the automatic fire extinguishing host, which is used to issue a fire alarm under the control of the automatic fire extinguishing host after the fire is confirmed.

8. A household automatic fire extinguishing method based on artificial intelligence, characterized in that: include: The smoke and temperature sensors are used to monitor the smoke and temperature in the monitoring area to generate smoke and temperature data. The target household residential area is divided into several monitoring areas, and a smoke temperature sensor is deployed in each monitoring area. The automatic fire extinguishing host calls the fire prediction model deployed on it to predict whether a fire has occurred in the monitoring area based on the smoke temperature data; When the automatic fire extinguishing host predicts that a fire occurs in the monitoring area, the flame detection host deployed in the monitoring area performs flame detection in the monitoring area to generate first flame data; The automatic fire extinguishing host calls the fire confirmation model deployed on it to confirm whether the fire has actually occurred based on the first flame data; After the fire is confirmed, the flame is detected again by using the flame confirmation sensor installed on the fire extinguishing medium injector to generate second flame data; The automatic fire extinguishing host determines the location of the fire in the monitoring area based on the smoke temperature data and the second flame data; The automatic fire extinguishing host controls the fire extinguishing medium injector to spray the fire extinguishing medium at the determined fire location.

Images