Fire safety early warning system based on Internet of Things

Abstract

The invention discloses a fire safety early warning system based on the Internet of Things, and relates to the technical field of safety early warning, a signal change coefficient is calculated by acquiring optical signal data around a flame detector, and whether open fire appears around the flame detector is judged by combining a preset threshold value, so that false alarm of fire early warning is reduced; when open fire does not appear, temperature data and humidity data are obtained, a temperature fluctuation abnormal coefficient and a humidity change abnormal coefficient are calculated, a fire initial evaluation index is generated in combination with a signal change coefficient, the fire initial evaluation index is compared with a preset threshold value, and early warning is carried out on fire safety; whether a fire safety problem exists or not can be judged in combination with other data, the limitation of a single detector is made up, and missing alarm or false alarm caused by the fact that the flame detector does not detect open fire is avoided; comprehensive analysis and risk judgment of multi-dimensional data can be realized, and the timeliness of early warning is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of safety early warning, and in particular to a fire safety early warning system based on the Internet of Things. Background Art

[0002] A flame detector is a sensor device used to detect flames or combustion processes. It usually works by sensing characteristic signals in the ultraviolet (UV), infrared (IR) or visible spectrum released by flames. It can quickly identify the presence of flames and their changing characteristics, and is widely used in fire safety early warning systems. Flame detectors can detect open flames in time at the early stage of a fire and trigger an alarm, thereby shortening response time and reducing fire losses. Flame detectors are more sensitive to open flames, so they are widely used in fire safety early warning systems.

[0003] The working principle of flame detectors is to identify the presence of flames by detecting specific spectral features released during the combustion process of flames, such as ultraviolet, infrared or visible light signals; its sensitive photoelectric sensors capture the intensity and changes of these spectra and convert them into electronic signals to determine whether there is a fire; however, in actual applications, flame detectors may be interfered with by environmental factors; for example, strong sunlight may misjudge strong solar radiation as flames, resulting in false alarms of fire warnings; and some fires may not have open flames in the early stages, and relying solely on the detection results of flame detectors for fire safety warnings may lead to delays or false alarms of fire safety warnings, causing greater dangers or losses. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned problems and provide a fire safety early warning system based on the Internet of Things.

[0005] The present invention proposes a fire safety early warning system based on the Internet of Things, the system comprising:

[0006] The first judgment and early warning module: obtains the light signal data around the flame detector, calculates the signal variation coefficient based on the light signal data, and judges whether there is an open flame around the flame detector in combination with the preset signal variation coefficient threshold;

[0007] Temperature anomaly module: When there is no open flame around the flame detector, the temperature data around the flame detector is obtained, and the temperature fluctuation anomaly coefficient is calculated based on the temperature data;

[0008] Humidity anomaly module: obtains humidity data around the flame detector and calculates humidity change anomaly coefficient based on the humidity data;

[0009] The second judgment and early warning module: constructs a fire initial assessment model based on the signal change coefficient, temperature fluctuation abnormal coefficient and humidity change abnormal coefficient, generates a fire initial assessment index, compares the fire initial assessment index with the preset fire initial assessment index threshold, and issues a fire safety early warning based on the comparison results.

[0010] Optionally, calculating the signal variation coefficient according to the optical signal data includes:

[0011] S1: The photoelectric sensor inside the flame detector continuously obtains the light signal around the flame detector at multiple times and converts it into a digital signal that changes with time. , that is, time Signal strength on

[0012] S2: From digital signal Find all local maxima and local minima in , and fit the upper envelope and lower envelope by spline interpolation respectively;

[0013] S3: Calculate the envelope mean based on the upper and lower envelopes ;

[0014] S4: From digital signal Subtract the envelope mean Get the preliminary intrinsic mode function IMF; express the preliminary intrinsic mode function IMF as ;

[0015] S5: Check Whether the IMF conditions are met: that is, the signal Fluctuates around zero mean, and the difference between the number of maximum and minimum values ​​of the envelope does not exceed 1; if If the conditions are not met, Consider it as a new signal and repeat steps S2–S4 until the IMF condition is met;

[0016] S6: Extracting the IMF from the digital signal Eliminate it and get the residual signal ; :right Repeat the above decomposition process to continue extracting subsequent IMFs until the remaining signal The final signal is decomposed into several IMFs. and a residual term ; ; is the total number of IMFs;

[0017] S7: For each IMF component Perform Hilbert transform and construct the analytical signal: ; is the analytical signal after Hilbert transform, for The result after Hilbert transform is the imaginary part after Hilbert transform. is an imaginary unit;

[0018] S8: Calculate instantaneous amplitude , ; Calculate the instantaneous phase , ; Calculate instantaneous frequency , ;

[0019] S9: Using instantaneous amplitude and instantaneous frequency The variance over time of , calculate the signal variation coefficient;

[0020] S9.1: Calculate the variance of the instantaneous frequency ;

[0021] S9.2: Calculate the variance of the instantaneous amplitude ;

[0022] S9.3: Calculate the signal variation coefficient , .

[0023] Optionally, judging whether an open flame is present around the flame detector in combination with a preset signal variation coefficient threshold includes:

[0024] The signal variation coefficient is compared with the preset signal variation coefficient threshold. If the signal variation coefficient is not less than the preset signal variation coefficient threshold, it means that the signal detected by the flame detector is a flame signal, indicating that an open fire has appeared around the flame detector, and a fire safety warning signal is immediately issued;

[0025] If the signal variation coefficient is less than the preset signal variation coefficient threshold, it means that the signal detected by the flame detector is not a flame signal and there is no open flame around the flame detector. However, it is necessary to further combine other data to determine whether the area around the flame detector is in the early stages of a fire and whether to issue a fire safety warning signal.

[0026] Optionally, calculating the temperature fluctuation anomaly coefficient according to the temperature data includes:

[0027] The temperature sensor around the flame detector continuously obtains the current temperature data around the flame detector at multiple times to obtain a time-temperature series data set; and the time-temperature series data set is divided according to the time window, each time window contains data points, a total of time windows;

[0028] Arrange the temperature data for each time window into a matrix , the size is ;

[0029] Pair Matrix Perform singular value decomposition, the decomposition formula is: ; for The left singular vector matrix of represents the main mode of the time window; for A diagonal matrix with singular values ​​on the diagonal, arranged in descending order, for The right singular vector matrix of represents the contribution of the data at each time point;

[0030] from Extract singular values ​​from , represents the order number of the singular values, , represents the total number of singular values, and is a positive integer;

[0031] Calculate the temperature fluctuation anomaly coefficient around the flame detector by singular value , the calculation formula is: , is the maximum singular value.

[0032] Optionally, obtaining humidity data around the flame detector and calculating the humidity change anomaly coefficient according to the humidity data includes:

[0033] The humidity sensor around the flame detector continuously obtains the current humidity data around the flame detector at multiple times to obtain the time humidity series data set , , where Indicates The humidity value at a time point, represents the total number of time points, and is a positive integer;

[0034] Set a neighborhood size parameter , the number of reference points used to calculate the local density of each point; The value range is 10-20;

[0035] For each data point , calculate its relationship with the time humidity series dataset The distance between other data points in , and select the The nearest data point, get v-distance: ;

[0036] Calculate data points and its neighboring points The distance between ; ; In the formula, for and The distance between

[0037] Calculate data points The local reachable density ; , where yes The number of neighboring points, Indicate point The sum of the reachable distances to all points in its neighborhood;

[0038] Calculate the data points based on the local reachable density The local anomaly factor ; ;

[0039] The time humidity series dataset The local anomaly factor of the humidity value at each time point in the is compared with the preset threshold, and the humidity value at the time point corresponding to the local anomaly factor being greater than the preset threshold is recorded as abnormal humidity. The total number of abnormal humidity in the time humidity series dataset is divided by The total number of humidity values ​​in the equation is used to obtain the humidity change anomaly coefficient.

[0040] Optionally, a fire initial stage assessment model is constructed based on the signal variation coefficient, the temperature fluctuation anomaly coefficient, and the humidity variation anomaly coefficient to generate a fire initial stage assessment index including:

[0041] A fire initial assessment model is constructed based on the signal variation coefficient, temperature fluctuation anomaly coefficient, and humidity variation anomaly coefficient to generate a fire initial assessment index , based on the following formula , where represent the preset proportional coefficients of the signal variation coefficient, the temperature fluctuation anomaly coefficient and the humidity variation anomaly coefficient, respectively, and Both are greater than 0.

[0042] Optionally, comparing the fire initial stage assessment index with a preset fire initial stage assessment index threshold, and providing an early warning for fire safety according to the comparison result includes:

[0043] The initial fire assessment index is compared with the preset initial fire assessment index threshold. If the initial fire assessment index is not less than the preset initial fire assessment index threshold, it means that there is a high possibility that the area around the flame detector is in the early stage of fire, and a fire safety warning signal is issued at this time; otherwise, no fire safety warning signal is issued.

[0044] Beneficial effects of the present invention:

[0045] 1. The present invention proposes a fire safety early warning system based on the Internet of Things, which obtains light signal data around the flame detector, calculates the signal variation coefficient based on the light signal data, and judges whether there is an open fire around the flame detector in combination with a preset signal variation coefficient threshold, and issues an early warning in a timely manner. In this way, it is possible to eliminate the misjudgment of flames caused by strong sunlight or other interference factors, thereby reducing false alarms of fire early warnings and ensuring the accuracy and reliability of fire early warnings;

[0046] 2. When there is no open flame around the flame detector, the temperature data and humidity data around the flame detector are obtained to obtain the temperature fluctuation anomaly coefficient and the humidity change anomaly coefficient, and the initial fire assessment index is generated in combination with the signal change coefficient. The initial fire assessment index is compared with the preset initial fire assessment index threshold, and a fire safety warning is issued based on the comparison result. In this way, when there is no open flame in the early stage of the fire, other data can be combined to determine whether there are fire safety problems, make up for the limitations of a single detector, and avoid underreporting or false alarms due to the flame detector not detecting an open flame; and by comparing with the preset assessment index threshold, comprehensive analysis and risk judgment of multi-dimensional data can be achieved, thereby improving the timeliness of the warning and providing a more comprehensive basis for firefighting decision-making, thereby effectively reducing fire losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The present invention will be further described below in conjunction with the accompanying drawings.

[0048] Figure 1 This is a framework diagram of a fire safety early warning system based on the Internet of Things. DETAILED DESCRIPTION

[0049] 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.

[0050] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0051] The embodiment of the present invention provides a fire safety early warning system based on the Internet of Things. Figure 1 , Figure 1 A framework diagram of a fire safety early warning system based on the Internet of Things provided by an embodiment of the present invention, the system includes:

[0052] The first judgment and early warning module: obtains the light signal data around the flame detector, calculates the signal variation coefficient based on the light signal data, and judges whether there is an open flame around the flame detector in combination with the preset signal variation coefficient threshold;

[0053] Temperature anomaly module: When there is no open flame around the flame detector, the temperature data around the flame detector is obtained, and the temperature fluctuation anomaly coefficient is calculated based on the temperature data;

[0054] Humidity anomaly module: obtains humidity data around the flame detector and calculates humidity change anomaly coefficient based on the humidity data;

[0055] The second judgment and early warning module: constructs a fire initial assessment model based on the signal change coefficient, temperature fluctuation abnormal coefficient and humidity change abnormal coefficient, generates a fire initial assessment index, compares the fire initial assessment index with the preset fire initial assessment index threshold, and issues a fire safety early warning based on the comparison results.

[0056] According to an embodiment of the present invention, a fire safety early warning system based on the Internet of Things is provided. By acquiring light signal data around a flame detector, calculating a signal variation coefficient based on the light signal data, and combining a preset signal variation coefficient threshold value to determine whether an open fire appears around the flame detector, and issuing an early warning in a timely manner, in this way, flame misjudgment caused by strong sunlight or other interference factors can be eliminated, thereby reducing false alarms of fire early warnings and ensuring the accuracy and reliability of fire early warnings; when there is no open fire around the flame detector, temperature data and humidity data around the flame detector are acquired to obtain temperature fluctuation anomaly coefficients and humidity variation anomaly coefficients, and combining the signal variation coefficient to generate an initial fire assessment index, and the initial fire assessment index is compared with a preset initial fire assessment index threshold value, and an early warning of fire safety is issued according to the comparison result. In this way, when there is no open fire at the initial stage of a fire, it is possible to determine whether there is a fire safety problem in combination with other data, make up for the limitations of a single detector, and avoid underreporting or false reporting due to the flame detector not detecting an open fire; and by comparing with the preset assessment index threshold value, it is possible to achieve comprehensive analysis and risk determination of multi-dimensional data, improve the timeliness of early warnings, and provide a more comprehensive basis for fire decision-making, thereby effectively reducing fire losses.

[0057] In one embodiment, calculating the signal variation coefficient according to the optical signal data includes:

[0058] S1: The photoelectric sensor inside the flame detector continuously obtains the light signal around the flame detector at multiple times and converts it into a digital signal that changes with time. , that is, time Signal strength on

[0059] S2: From digital signal Find all local maxima and local minima in the , and fit the upper envelope (connecting the maximum points) and the lower envelope (connecting the minimum points) by spline interpolation respectively;

[0060] S3: Calculate the envelope mean , the calculation formula is: , where (Upper envelope): It is a curve obtained by connecting the maximum points of the signal, indicating the trend of the local maximum value of the signal; (Lower envelope): It is a curve obtained by connecting the minimum points of the signal, indicating the trend of the local minimum value of the signal;

[0061] S4: From digital signal Subtract the envelope mean Get the preliminary intrinsic mode function (IMF) ; ;

[0062] S5: Check Whether the IMF conditions are met: that is, the signal Fluctuates around zero mean, and the difference between the number of maximum and minimum values ​​of the envelope does not exceed 1; if If the conditions are not met, Consider it as a new signal and repeat steps S2–S4 until the IMF condition is met;

[0063] S6: Extracting the IMF from the digital signal Eliminate it and get the residual signal ; :right Repeat the above decomposition process to continue extracting subsequent IMFs until the remaining signal Until it becomes monotonically changing or noise; finally, the signal is decomposed into several IMFs and a residual term ; ; is the total number of IMFs;

[0064] S7: For each IMF component Perform Hilbert transform and construct the analytical signal: ; is the analytical signal after Hilbert transform, for The result after Hilbert transform is the imaginary part after Hilbert transform. is an imaginary unit; the function of Hilbert transform is to extract the phase information of the signal and form a Complementary signals; the Hilbert transform generates a signal with the same frequency as the original signal but with a phase shift;

[0065] S8: Calculate instantaneous amplitude , instantaneous amplitude Reflects the strength of the signal; ; Calculate the instantaneous phase , instantaneous phase Describe the local phase change of a signal: ; Calculate instantaneous frequency , instantaneous frequency is the instantaneous phase Derivative with respect to time: ;

[0066] It should be noted that is the double inverse tangent function: it is used to calculate the angle (i.e. phase) of a point on a plane; given two numbers and The return value is from the origin to the point The straight line and The angle between the axes is in the range ; Corresponds to , that is, the signal after Hilbert transformation (imaginary part); Corresponds to , i.e. the original signal (real part); This formula represents the signal after Hilbert transform and original signal The ratio of The instantaneous phase of an eigenmode function can help analyze the local phase changes of the signal and then perform deeper analysis of the signal, such as calculating the instantaneous frequency.

[0067] S9: Using instantaneous amplitude and instantaneous frequency The variance over time of , calculate the signal variation coefficient;

[0068] S9.1: Calculate the variance of the instantaneous frequency , instantaneous frequency variance ;

[0069] S9.2: Calculate the variance of the instantaneous amplitude , instantaneous amplitude variance ;

[0070] S9.3: Calculate the signal variation coefficient , .

[0071] It should be noted that the function of the Hilbert transform is to make the signal get a phase shift, which helps to extract the instantaneous frequency, instantaneous amplitude and instantaneous phase of the signal, so as to more accurately capture the changing characteristics of the signal, especially when processing non-stationary signals, it can effectively extract the local characteristics of the signal.

[0072] It should be noted that the light signals around the flame detector are usually obtained through the photoelectric sensor inside the detector; the photoelectric sensor uses the change in light intensity to sense the light in the surrounding environment and converts it into an electrical signal; the electrical signal is then converted into a digital signal through an analog-to-digital converter (ADC) for further processing and analysis; by monitoring the intensity changes of these signals, the presence and changing characteristics of the flame can be judged in real time.

[0073] It should be noted that the signal variation coefficient reflects the change in the instantaneous amplitude and instantaneous frequency of the optical signal; when the calculated signal variation coefficient is large, it means that the amplitude and frequency of the signal have fluctuated greatly, which is usually a characteristic of a flame signal, because the fluctuation of the flame will cause a drastic change in the optical signal. This sudden change may mean that a fire has started or a flame has suddenly appeared; when the signal variation coefficient is small, it means that the signal changes relatively smoothly, which is usually caused by environmental factors such as solar radiation, meteorological changes or other background noise. The changes caused by these factors are relatively slow and uniform, and will not suddenly fluctuate greatly like the flame signal; therefore, by analyzing the signal variation coefficient, it is possible to effectively distinguish the interference of the flame signal and other environmental factors, and then determine whether there is a fire risk.

[0074] In one implementation, the advantage of calculating the signal variation coefficient by the above calculation method is that it can accurately capture the fluctuation of the signal in instantaneous amplitude and instantaneous frequency, so as to effectively distinguish the interference of flame signal and other environmental factors (such as solar radiation, etc.); when the signal variation coefficient is large, it usually indicates the sudden change of flame signal, which helps to quickly identify the occurrence of fire; and by combining the intrinsic mode decomposition (EMD) and Hilbert transform, different frequency components can be effectively extracted from complex signals and the instantaneous characteristics of signals can be accurately captured. EMD can decompose the original signal into multiple intrinsic mode functions (IMFs), each IMF represents a local oscillation mode of the signal, avoiding the assumptions and filtering problems that may be introduced by traditional methods, and Hilbert transform further reveals the dynamic changes of the signal by extracting the instantaneous phase and instantaneous frequency of the signal. The combination of these two methods can more accurately identify the sudden changes in the signal, reduce noise interference and improve the accuracy and reliability of fire detection; this method can more sensitively reflect the characteristics of the flame signal and ensure the timeliness and accuracy of the fire alarm system.

[0075] In one embodiment, judging whether an open flame is present around the flame detector in combination with a preset signal variation coefficient threshold comprises:

[0076] The signal variation coefficient is compared with the preset signal variation coefficient threshold. If the signal variation coefficient is not less than the preset signal variation coefficient threshold, it means that the signal detected by the flame detector is a flame signal, indicating that an open fire has appeared around the flame detector, and a fire safety warning signal is immediately issued;

[0077] If the signal variation coefficient is less than the preset signal variation coefficient threshold, it means that the signal detected by the flame detector is not a flame signal and there is no open flame around the flame detector. However, it is necessary to further combine other data to determine whether the area around the flame detector is in the early stages of a fire and whether to issue a fire safety warning signal.

[0078] In one implementation, by comparing the signal variation coefficient with a preset threshold, it is possible to accurately determine whether there is an open flame around the flame detector; when the signal variation coefficient reaches or exceeds the threshold, the flame signal can be quickly identified and an early warning can be issued in time, thereby improving the efficiency and accuracy of fire response; on the contrary, if the signal variation coefficient is small, the flame signal can be excluded to avoid false alarms, while providing space for further data analysis and early warning judgment. This method combines the dynamic characteristics of signal changes with preset thresholds, which can effectively reduce the impact of environmental interference and improve the reliability and responsiveness of the fire monitoring system.

[0079] In one embodiment, calculating the temperature fluctuation anomaly coefficient according to the temperature data includes:

[0080] The temperature sensor around the flame detector continuously obtains the current temperature data around the flame detector at multiple times to obtain a time-temperature series data set , , where Indicates the sequence number of the time in the time temperature series data set, Indicates the total number of moments;

[0081] The time-temperature series data set is split according to time windows, and each time window contains data points, a total of time windows;

[0082] Arrange the temperature data for each time window into a matrix , the size is ;

[0083]

[0084] Pair Matrix Perform singular value decomposition to extract the principal components of the data and separate different modes.

[0085] It should be noted that singular value decomposition can separate the main mode (principal component) and noise mode of the data, and the size of the singular value reflects the importance of each mode in the data;

[0086] ; for The left singular vector matrix of represents the main mode of the time window; for A diagonal matrix with singular values ​​on the diagonal, arranged in descending order, for The right singular vector matrix of represents the contribution of the data at each time point; this step is to extract the main mode of the data in order to distinguish between normal mode and abnormal mode;

[0087] from Extract singular values ​​from the data, analyze their distribution, and identify possible abnormal changes in temperature data; abnormal temperature changes usually significantly affect the size of singular values, especially the largest singular value , and mark the singular values ​​as , represents the order number of the singular values, , represents the total number of singular values, and is a positive integer;

[0088] Calculate the temperature fluctuation anomaly coefficient around the flame detector by singular value , which is used to quantify the abnormal degree of temperature change around the flame detector. The calculation formula is: , is the largest singular value. The anomaly coefficient reflects the proportion of the largest singular value in the overall singular value. Abnormal data usually causes this value to increase significantly.

[0089] It should be noted that the temperature fluctuation anomaly coefficient around the flame detector is used to measure the degree of abnormal temperature change in the area, especially the abnormal fluctuation in the temperature signal near the flame detector. In general, the temperature data around the flame detector can be obtained through the temperature sensor around the flame detector. Under normal circumstances, the temperature change around the flame detector should have a certain stability or regularity, and is less affected by environmental factors (such as airflow and seasonal changes). However, when the temperature fluctuation anomaly coefficient increases, it usually means that there are violent fluctuations or irregular patterns in the local temperature change that are different from the normal state. This may be caused by the sudden appearance of potential heat sources or energy release. Even if there is no obvious flame, it may also indicate some potential signs of the early stage of the fire, such as electrical overload, slow combustion of materials, etc., which may cause abnormal temperature signals. The larger the temperature fluctuation anomaly coefficient around the flame detector, the higher the proportion of abnormal patterns contained in the main mode of the temperature data. The early stage of a fire is often manifested as high temperature or rapid temperature rise in a small area. Although the flame is not fully manifested, heat conduction and convection will quickly cause changes in the surrounding temperature. At this time, the proportion of the maximum singular value to the total singular value will increase significantly, which is reflected in the increase of the anomaly coefficient. Even if there is no open flame in the environment, the early signs of fire, such as hidden combustion, heat accumulation, or chemical reactions, will significantly change the distribution pattern of the temperature signal. Therefore, the increase in the anomaly coefficient is a key parameter for early warning of fire, which can capture risk signals before the flame appears and gain valuable time for fire prevention and control.

[0090] In one implementation, the main reason for calculating the temperature fluctuation anomaly coefficient around the flame detector by the above calculation method is to quantitatively evaluate the degree of anomaly in temperature changes and extract key features from complex temperature data. The use of singular value decomposition (SVD) can not only effectively separate the main mode and noise of temperature data, but also intuitively reflect whether there are abnormal changes in the temperature signal through the distribution of singular values; the benefits of this method are: on the one hand, singular value decomposition can compress data dimensions, extract key information, and make anomaly detection more efficient and robust; on the other hand, the calculated anomaly coefficient can quantify the significance of temperature anomalies, avoid relying on the judgment of a single moment or local temperature changes, and improve the sensitivity to the hidden signs of the initial fire. In addition, this method can also adapt to different environmental conditions and noise interference, has strong versatility, and helps to achieve accurate early warning, thereby providing a scientific basis for fire prevention and control.

[0091] In one embodiment, obtaining humidity data around the flame detector and calculating the humidity change abnormality coefficient according to the humidity data includes:

[0092] The humidity sensor around the flame detector continuously obtains the current humidity data around the flame detector at multiple times to obtain the time humidity series data set , , where Indicates The humidity value at a time point, represents the total number of time points, and is a positive integer; this step collects and prepares all humidity data for subsequent calculations;

[0093] Set a neighborhood size parameter (usually 10-20), which indicates the number of reference points used to calculate the local density of each point; Determines the sensitivity of anomaly detection; smaller More attention is paid to local mutations, larger More suitable for global trends; generally The value range is 10-20. The specific setting depends on the actual situation and is not limited.

[0094] For each data point , calculate its relationship with the time humidity series dataset The distance between other data points in (can be calculated using Euclidean distance or Manhattan distance); and select the The nearest data point, get v-distance: ; By calculating the v-distance of each data point, the neighborhood range of each data point can be determined to prepare for the subsequent calculation of local density; data points farther away may be outliers, so identifying the nearest neighbors helps detect anomalies;

[0095] Calculate data points and its neighboring points The distance between ; ; In the formula, for and The distance between points (such as Euclidean distance or Manhattan distance, and the above distances correspond to each other, either Euclidean distance or Manhattan distance); the reachable distance ensures the stability of local density, not only considering the direct distance between points, but also ensuring that the density calculation is not affected by distant points through v-distance; it can better reflect the density changes in local areas in the data and avoid misjudgment of anomalies;

[0096] Calculate data points The local reachable density ; , where yes The number of neighboring points, Indicate point The sum of the reachable distances to all points in its neighborhood; through this calculation, the local density of each point can be obtained. High density means that there are more data points clustered around the point, and low density may mean that there is a large difference between the point and other points in the neighborhood, which may be an outlier; the local reachable density reflects the density of data points in its neighborhood. If the local density of a point is significantly lower than the density of its neighborhood, it may be an outlier; by calculating the local density of each point, the sparseness of its local area can be accurately reflected.

[0097] According to the local reachable density, calculate the point The local anomaly factor ; ; LOF value is used to quantify the point If the LOF value of a point is much greater than 1, it means that the density difference between it and its neighbors is very large and it may be an anomaly. This process helps identify sudden changes in humidity data.

[0098] The time humidity series dataset The local anomaly factor of the humidity value at each time point in the is compared with the preset threshold, and the humidity value at the time point corresponding to the local anomaly factor being greater than the preset threshold is recorded as abnormal humidity. The total number of abnormal humidity in the time humidity series dataset is divided by The total number of humidity values ​​in the equation is used to obtain the humidity change anomaly coefficient.

[0099] It should be noted that the data acquisition method involved in the above steps is mainly to collect real-time data through humidity sensors around the flame detector. These humidity sensors measure and record the humidity values ​​of the surrounding environment at continuous time nodes to form a time humidity series data set. The frequency and duration of data collection can be set according to actual needs to ensure that the data is detailed enough to capture the changing trends and mutations of humidity, especially in the early stages of a fire that may cause rapid changes in ambient humidity. These data provide the basis for subsequent anomaly detection and calculation of humidity change anomaly coefficients.

[0100] It should be noted that the humidity change anomaly coefficient around the flame detector is used to measure the degree of abnormal fluctuation of the environmental humidity at the initial stage of a fire. In the early stage of a fire, especially in a high temperature environment, the humidity tends to drop rapidly. When a fire occurs, the evaporation caused by heat and the oxygen consumption will cause the humidity to change sharply. Therefore, the calculation of the humidity change anomaly coefficient can effectively reflect these rapid changes, thus serving as one of the potential indicators of the occurrence of a fire. The larger the humidity change anomaly coefficient, the more drastic the humidity change around the flame detector, which may indicate that the fire is entering the initial stage. Even if there is no obvious open flame around the flame detector at this time, due to the high temperature effect at the initial stage of the fire, a sharp change in humidity may still occur, and this change can sometimes serve as an early signal of a fire. Therefore, by monitoring the humidity change anomaly coefficient, potential fire risks can be discovered in time before the fire has fully broken out, thereby improving the response speed and accuracy of fire warnings.

[0101] In one implementation, the benefits of calculating the humidity change anomaly coefficient around the flame detector by the above-mentioned calculation method are: it can effectively identify abnormal fluctuations in humidity in the initial stage of a fire, thereby improving the sensitivity and accuracy of fire detection; drastic changes in humidity are often a signal of an early fire, especially in a closed environment, where humidity changes caused by high temperature can indicate the occurrence of a fire; by utilizing algorithms such as the local anomaly factor (LOF), it is possible to more accurately analyze local density changes in humidity data and identify mutation points that deviate from normal humidity fluctuations; compared with traditional smoke detection, the humidity change anomaly coefficient provides a supplementary monitoring method that can capture potential signs of fire in advance, providing an important basis for taking timely measures and reducing losses.

[0102] In one embodiment, a fire initial stage assessment model is constructed based on the signal variation coefficient, the temperature fluctuation abnormal coefficient, and the humidity variation abnormal coefficient, and the fire initial stage assessment index is generated, including:

[0103] A fire initial assessment model is constructed based on the signal variation coefficient, temperature fluctuation anomaly coefficient, and humidity variation anomaly coefficient to generate a fire initial assessment index , based on the following formula , where represent the preset proportional coefficients of the signal variation coefficient, the temperature fluctuation anomaly coefficient and the humidity variation anomaly coefficient, respectively, and All are greater than 0;

[0104] It should be noted that before building the fire initial assessment model, the signal variation coefficient, temperature fluctuation anomaly coefficient and humidity variation anomaly coefficient need to be normalized. Commonly used normalization methods include Min-Max normalization and Z-Score normalization. It is set according to actual conditions. For example, the expert empowerment method is adopted, that is, experts in relevant fields are invited to determine the preset proportion coefficients of various indicators through professional opinion surveys and comprehensive evaluations.

[0105] In one embodiment, comparing the fire initial stage assessment index with a preset fire initial stage assessment index threshold, and providing an early warning for fire safety according to the comparison result includes:

[0106] The initial fire assessment index is compared with the preset initial fire assessment index threshold. If the initial fire assessment index is not less than the preset initial fire assessment index threshold, it means that there is a high possibility that the area around the flame detector is in the early stage of fire, and a fire safety warning signal is issued at this time; otherwise, no fire safety warning signal is issued.

[0107] It should be noted that the preset fire initial assessment index threshold is set by professionals based on actual conditions and is not limited or elaborated on in detail.

[0108] In one implementation, the accuracy and reliability of early fire warnings are improved through the above method. Although the flame detector does not detect an open flame, other environmental parameters such as temperature fluctuations and humidity changes may have become abnormal, and these changes are often potential signs of an early fire. By integrating these factors and the signal change coefficient of the light signal into an early fire assessment index, the limitations of a single detector can be compensated, and the omission or false alarm of the flame detector due to the failure to detect an open flame can be avoided. By comparing with the preset assessment index threshold, comprehensive analysis and risk determination of multi-dimensional data can be achieved, the timeliness of the warning can be improved, and a more comprehensive basis can be provided for fire fighting decisions, thereby effectively reducing fire losses and protecting life and property.

[0109] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be used to artificially limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A fire safety early warning system based on the Internet of Things, characterized in that: The system comprises: The first judgment and early warning module: obtains the light signal data around the flame detector, calculates the signal variation coefficient based on the light signal data, and judges whether there is an open flame around the flame detector in combination with the preset signal variation coefficient threshold; Temperature anomaly module: When there is no open flame around the flame detector, the temperature data around the flame detector is obtained, and the temperature fluctuation anomaly coefficient is calculated based on the temperature data; Humidity anomaly module: obtains humidity data around the flame detector and calculates humidity change anomaly coefficient based on the humidity data; The second judgment and early warning module: constructs a fire initial assessment model based on the signal change coefficient, temperature fluctuation abnormal coefficient and humidity change abnormal coefficient, generates a fire initial assessment index, compares the fire initial assessment index with the preset fire initial assessment index threshold, and issues a fire safety early warning based on the comparison results.

2. According to the Internet of Things-based fire safety early warning system according to claim 1, it is characterized in that: Calculating the signal variation coefficient based on the optical signal data includes: S1: The photoelectric sensor inside the flame detector continuously obtains the light signal around the flame detector at multiple times and converts it into a digital signal that changes with time. , that is, time Signal strength on S2: From digital signal Find all local maxima and local minima in , and fit the upper envelope and lower envelope by spline interpolation respectively; S3: Calculate the envelope mean based on the upper and lower envelopes ; S4: From digital signal Subtract the envelope mean Get the preliminary intrinsic mode function IMF; express the preliminary intrinsic mode function IMF as ; S5: Check Whether the IMF conditions are met: that is, the signal Fluctuates around zero mean, and the difference between the number of maximum and minimum values ​​of the envelope does not exceed 1; if If the conditions are not met, Consider it as a new signal and repeat steps S2–S4 until the IMF condition is met; S6: Extracting the IMF from the digital signal Eliminate it and get the residual signal ; :right Repeat the above decomposition process to continue extracting subsequent IMFs until the remaining signal The final signal is decomposed into several IMFs. and a residual term ; ; is the total number of IMFs; S7: For each IMF component Perform Hilbert transform and construct the analytical signal: ; is the analytical signal after Hilbert transform, for The result after Hilbert transform is the imaginary part after Hilbert transform. is an imaginary unit; S8: Calculate instantaneous amplitude , ; Calculate the instantaneous phase , ; Calculate instantaneous frequency , ; S9: Using instantaneous amplitude and instantaneous frequency The variance over time of , calculate the signal variation coefficient; S9.1: Calculate the variance of the instantaneous frequency ; S9.2: Calculate the variance of the instantaneous amplitude ; S9.3: Calculate the signal variation coefficient , .

3. According to the Internet of Things-based fire safety early warning system of claim 1, it is characterized in that: Combining the preset signal change coefficient threshold to determine whether there is an open flame around the flame detector includes: The signal variation coefficient is compared with the preset signal variation coefficient threshold. If the signal variation coefficient is not less than the preset signal variation coefficient threshold, it means that the signal detected by the flame detector is a flame signal, indicating that an open fire has appeared around the flame detector, and a fire safety warning signal is immediately issued; If the signal variation coefficient is less than the preset signal variation coefficient threshold, it means that the signal detected by the flame detector is not a flame signal and there is no open flame around the flame detector. However, it is necessary to further combine other data to determine whether the area around the flame detector is in the early stage of a fire and whether to issue a fire safety warning signal.

4. The fire safety early warning system based on the Internet of Things according to claim 1 is characterized in that: Calculating the temperature fluctuation anomaly coefficient based on temperature data includes: The temperature sensor around the flame detector continuously obtains the current temperature data around the flame detector at multiple times to obtain a time-temperature series data set; and the time-temperature series data set is divided according to the time window, each time window contains data points, a total of time windows; Arrange the temperature data for each time window into a matrix , the size is ; Pair Matrix Perform singular value decomposition, the decomposition formula is: ; for The left singular vector matrix of represents the main mode of the time window; for A diagonal matrix with singular values ​​on the diagonal, arranged in descending order, for The right singular vector matrix of represents the contribution of the data at each time point; from Extract singular values ​​from , represents the order number of the singular values, , represents the total number of singular values, and is a positive integer; Calculate the temperature fluctuation anomaly coefficient around the flame detector by singular value , the calculation formula is: , is the maximum singular value.

5. The fire safety early warning system based on the Internet of Things according to claim 1 is characterized in that: Obtaining humidity data around the flame detector and calculating humidity change anomaly coefficient based on humidity data include: The humidity sensor around the flame detector continuously obtains the current humidity data around the flame detector at multiple times to obtain the time humidity series data set , , where Indicates The humidity value at a time point, represents the total number of time points, and is a positive integer; Set a neighborhood size parameter , the number of reference points used to calculate the local density of each point; The value range is 10-20; For each data point , calculate its relationship with the time humidity series dataset The distance between other data points in , and select the The nearest data point, get v-distance: ; Calculate data points and its neighboring points The distance between ; ； In the formula, for and The distance between Calculate data points The local reachable density ; , where yes The number of neighboring points, Indicate point The sum of the reachable distances to all points in its neighborhood; According to the local reachable density, calculate the point The local anomaly factor ; ; The time humidity series dataset The local anomaly factor of the humidity value at each time point in the is compared with the preset threshold, and the humidity value at the time point corresponding to the local anomaly factor being greater than the preset threshold is recorded as abnormal humidity. The total number of abnormal humidity in the time humidity series dataset is divided by The total number of humidity values ​​in the equation is used to obtain the humidity change anomaly coefficient.

6. The fire safety early warning system based on the Internet of Things according to claim 1 is characterized in that: The fire initial assessment model is constructed based on the signal variation coefficient, temperature fluctuation anomaly coefficient and humidity variation anomaly coefficient, and the fire initial assessment index is generated, including: A fire initial assessment model is constructed based on the signal variation coefficient, temperature fluctuation anomaly coefficient, and humidity variation anomaly coefficient to generate a fire initial assessment index , based on the following formula , where represent the preset proportional coefficients of the signal variation coefficient, the temperature fluctuation anomaly coefficient and the humidity variation anomaly coefficient, respectively, and Both are greater than 0.

7. The fire safety early warning system based on the Internet of Things according to claim 1 is characterized in that: The fire early assessment index is compared with the preset fire early assessment index threshold, and early warning of fire safety is given according to the comparison results, including: The initial fire assessment index is compared with the preset initial fire assessment index threshold. If the initial fire assessment index is not less than the preset initial fire assessment index threshold, it means that there is a high possibility that the area around the flame detector is in the early stage of fire, and a fire safety warning signal is issued at this time; otherwise, no fire safety warning signal is issued.

Images