Intelligent building fire safety control method and system based on big data

Abstract

The invention discloses an intelligent building fire safety control method and system based on big data, relates to the technical field of fire safety control, breaks through the smoldering detection bottleneck, realizes early capture of slow smoldering based on a cable trend gradient temperature rise model, and overcomes the defect of traditional threshold alarm fault; the 2 DEG C / h temperature rise failure problem is solved; thermal fluctuation is quantified through wavelet packet energy entropy, and the pyrolysis characteristics of the insulating material are identified in the open-fire-free stage; according to the method, the false alarm rate is essentially reduced, and double mechanisms of spatial correlation filtering and frequency domain energy analysis are innovatively fused; in a spatial layer, non-uniform interference such as air conditioner hot air is eliminated; according to the frequency domain layer, a weighting function suppresses extremely low frequency thermal drift, and entropy change purely reflects the negative-set heat release instability; and dual-mechanism cooperation cuts off a false alarm source from a system level.

Description

Technical Field

[0001] The present invention relates to the field of fire safety control technology, and in particular to a big data-based smart building fire safety control method and system. Background Art

[0002] Current smart building fire protection systems generally use distributed fiber-optic temperature measurement (DTS) and multi-sensor threshold alarms as distribution room monitoring solutions. By collecting real-time cable surface temperature data and combining it with preset thresholds to trigger warnings, they provide stable performance in scenarios involving open flames or rapid temperature rises. Some newly built buildings have introduced thermal imaging cameras to assist in monitoring and identify local hot spots.

[0003] However, in the case of cable smoldering in distribution rooms, smoldering is a slow, flameless combustion phenomenon caused by aging, moisture, and other factors in the cable insulation. The temperature rise rate is extremely low, less than 2°C per hour, and the heat distribution is diffuse. Existing systems rely on fixed threshold alarm mechanisms that are completely ineffective. The system does not respond until the threshold is reached, and by the time the threshold is triggered, the smoldering has already spread and become a disaster. While thermal imaging technology can capture temperature distribution, it lacks resolution for gradual temperature changes with a difference of less than 1°C and is susceptible to interference from ambient heat sources. More critically, current solutions lack the ability to mine weak correlation signals between multiple sensors, making it impossible to extract the spatial heat conduction characteristics unique to smoldering from the noise.

[0004] To address this limitation, some solutions increase sensor density or lower alarm thresholds, but this results in a surge in false alarm rates, such as false triggering of hot air from air conditioners. A few high-end systems use weighted fusion algorithms to integrate temperature and smoke data, but there is neither significant smoke nor temperature changes in the early stages of smoldering, and multiple sources of data still operate independently. In practice, smoldering hazards in distribution rooms are mostly discovered through manual inspections, but the frequency and accuracy of manual coverage are insufficient. Therefore, there is an urgent need for a smart building fire safety control solution based on big data to solve such problems. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a smart building fire safety control method and system based on big data to solve the problems of existing fire protection systems such as delayed response to cable smoldering, high false alarm rate, reliance on manual inspections, and lack of effective prediction methods.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a smart building fire safety control method based on big data, which includes:

[0009] Step S1, collecting a data sequence of the surface temperature of cables in a power distribution room in real time through a distributed temperature sensor array;

[0010] Step S2, performing spatial correlation filtering on the temperature data sequence to extract temperature gradient distribution characteristics between adjacent sensors;

[0011] Step S3, performing energy entropy calculation on the filtered data based on the wavelet packet transform algorithm to generate a temperature fluctuation feature vector;

[0012] Step S4: When the characteristic vector continuously exceeds the preset baseline range and reaches a first time threshold, a smoldering risk warning is triggered;

[0013] Step S5: Linking the ventilation system to execute the cooling plan.

[0014] As a preferred solution of the big data-based smart building fire safety control method described in the present invention, the spatial correlation filtering process in step S2 includes:

[0015] Divide the sensor groups according to the cable direction, and each group contains at least three linearly arranged sensors;

[0016] Calculate the real-time temperature difference between adjacent sensors in the group and generate a temperature difference sequence;

[0017] When the temperature difference sequence meets the following conditions: a monotonically increasing / decreasing distribution along the cable direction, and the gradient change rate is within a preset range, it is determined to be a valid smoldering characteristic signal.

[0018] As a preferred solution of the big data-based smart building fire safety control method described in the present invention, in step S2, in the process of determining the effective smoldering characteristic signal, a temperature difference sequence is first constructed according to the actual temperature measured by the sensors in the group. When ΔT i =T i+1 (t)-T i (t), i=1,…,N-1, judge the consistency of temperature difference sign, where ΔT i is the temperature difference between the i-th pair of adjacent sensors, in °C, T i+1 (t), T i (t) are the temperatures of sensor No. i+1 and No. i at time t, respectively. t is the sampling time in seconds, and N is the total number of sensors in the group.

[0019] Define μ:

[0020]

[0021] Where μ is the sign consistency factor, sgn(·) is the sign function;

[0022] Definition of g i Quantify gradient smoothness:

[0023]

[0024] Among them, g i is the local gradient change rate, in °C·m -1 , d is the center distance between adjacent sensors, in m;

[0025] |g i |Compare with the threshold and get

[0026] Where γ is the gradient consistency factor, H(·) is the Heaviside step function, and λ max is the gradient change rate threshold, in °C·m -1 ;

[0027] The threshold adopts the historical statistical adaptive form, expressed as λ max =k s σ T , where k s is the empirical coefficient, σ T is the standard deviation of 24h temperature difference;

[0028] The two factors are combined to obtain the judgment value S=μγ, where S is the smoldering feature judgment result, 1 is valid and 0 is invalid.

[0029] As a preferred solution of the big data-based smart building fire safety control method described in the present invention, the energy entropy calculation in step S3 includes:

[0030] Decompose the temperature data sequence into multiple frequency band nodes using wavelet packets;

[0031] Extract the energy value of the node in the 0.01-0.1Hz frequency band and calculate its probability distribution entropy;

[0032] When the entropy value exceeds the historical mean by 3 times the standard deviation continuously, the warning is activated.

[0033] As a preferred solution of the big data-based smart building fire safety control method described in the present invention, in step S3, the step of calculating the probability distribution entropy includes:

[0034] Perform M-layer wavelet packet decomposition on the filtered temperature sequence X(t) output in step S2:

[0035]

[0036] Where X(t) is the filtered temperature sequence in °C, M is the number of decomposition layers, j is the node index, k is the time displacement index, and c j,k is the decomposition coefficient of node j at displacement k, in °C, w j,k (t) is the corresponding wavelet packet basis function;

[0037] Calculate the node center frequency using the formula:

[0038]

[0039] Among them, f j is the center frequency of node j, in Hz, f s is the sampling frequency in Hz;

[0040] Construct the target frequency band node set B:

[0041] B={j|0.01≤f j ≤0.1},

[0042] Among them, B is the frequency band node index set;

[0043] Corresponding node energy:

[0044]

[0045] Among them, E j is the unweighted energy of node j, in °C 2 ;

[0046] A sensitive weight is introduced to the energy to suppress the very low frequency thermal drift, which is expressed as:

[0047]

[0048] Among them, w j is the sensitivity weight of node j, α is the amplitude adjustment coefficient, and β is the frequency attenuation coefficient;

[0049] Weighted Energy in, is the weighted energy of node j, in °C 2 ;

[0050] Energy probability distribution:

[0051]

[0052] Among them, p j is the energy probability of node j, the denominator is the total energy of the target frequency band, and h is the traversal index;

[0053] Band energy entropy H b =-∑ j∈B p j ln p j , where H b It is the energy entropy of the 0.010.1Hz frequency band;

[0054] Save σ H and Havg For subsequent early warning calls,

[0055] σ H =std(H b ),H avg =mean(H b ),

[0056] Among them, σ H is the standard deviation of historical energy entropy, H avg is the historical mean energy entropy.

[0057] As a preferred solution of the big data-based smart building fire safety control method of the present invention, step S4 includes:

[0058] Retrieve the load current timing data of the target cable;

[0059] If the current fluctuation trend and the temperature characteristic vector match the preset correlation pattern in the time domain, the warning level is raised to the second level response.

[0060] In a second aspect, the present invention provides a smart building fire safety control system based on big data, comprising:

[0061] Temperature sensor array, distributed along the cable surface;

[0062] Edge computing nodes, including:

[0063] A spatial filtering unit connected to the sensor array;

[0064] An entropy analysis unit receives an output of the filtering unit;

[0065] The early warning decision module generates instructions based on the entropy analysis results;

[0066] Linkage execution module, connected to the building ventilation control system.

[0067] As a preferred solution of the big data-based smart building fire safety control system described in the present invention, the entropy analysis unit has a built-in:

[0068] The wavelet packet decomposition subunit is configured to select the 0.01-0.1 Hz frequency band;

[0069] The entropy calculation subunit generates a feature vector based on the frequency band energy distribution;

[0070] The baseline comparison subunit stores the historical entropy mean and standard deviation data.

[0071] As a preferred solution of the big data-based smart building fire safety control system described in the present invention, the early warning decision module executes:

[0072] When the entropy value is abnormal for 5 minutes and there is no current correlation, a level 1 warning log is generated;

[0073] When the entropy value is abnormal and there is current correlation, a secondary response instruction is generated and the linkage module is started.

[0074] As a preferred solution of the big data-based smart building fire safety control system described in the present invention, the linkage execution module includes:

[0075] Local strong wind control interface, locate the ventilation equipment closest to the smoldering point;

[0076] Air conditioning return air blocking unit, closing the return air duct of adjacent areas;

[0077] Operation and maintenance report generator, outputs cable location identification and health score.

[0078] The beneficial effects of the present invention are as follows: the present invention breaks through the bottleneck of smoldering detection, realizes early capture of slow smoldering based on the gradient temperature rise model of the cable direction, and overcomes the failure problem of traditional threshold alarm for temperature rise <2℃ / h; quantifies thermal fluctuations through wavelet packet energy entropy, and identifies the thermal decomposition characteristics of insulating materials in the non-open flame stage; the false alarm rate of the present invention is essentially reduced, and innovatively integrates the dual mechanisms of spatial correlation filtering and frequency domain energy analysis: spatial layer: eliminates non-uniform interference such as hot air from air conditioners; frequency domain layer: weight function suppresses extremely low-frequency thermal drift, so that the entropy value change purely reflects the instability of smoldering heat release; the dual mechanisms work together to cut off the source of false alarms at the system level; in addition, the introduction of the load current-temperature correlation model provides cross-validation for cable health: when the entropy value anomaly and current harmonics appear synchronously, a secondary response is automatically triggered and a health score is generated, which promotes fire management from post-fault repair to predictive maintenance; the existing temperature sensors and ventilation systems of the building are reused, and the core functions can be realized by upgrading the edge node algorithm, and the renovation of old distribution rooms does not require large-scale wiring. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0080] Figure 1 This is a flow chart of the big data-based smart building fire safety control method in Example 1.

[0081] Figure 2 This is a schematic diagram of the framework of the big data-based smart building fire safety control system in Example 1. DETAILED DESCRIPTION

[0082] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0083] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0084] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0085] Example 1, reference Figure 1 and Figure 2 This embodiment provides a smart building fire safety control method based on big data, including the following steps:

[0086] Step S1, collecting a data sequence of the surface temperature of cables in a power distribution room in real time through a distributed temperature sensor array;

[0087] Step S2, performing spatial correlation filtering on the temperature data sequence to extract the temperature gradient distribution characteristics between adjacent sensors;

[0088] The spatial correlation filtering process in step S2 includes:

[0089] Divide the sensor groups according to the cable direction, and each group contains at least three linearly arranged sensors;

[0090] Calculate the real-time temperature difference between adjacent sensors in the group and generate a temperature difference sequence;

[0091] When the temperature difference sequence satisfies the following conditions: a monotonically increasing / decreasing distribution along the cable direction, and the gradient change rate is within the preset range, it is determined to be a valid smoldering characteristic signal;

[0092] In step S2, in the process of determining the effective smoldering characteristic signal, firstly, a temperature difference sequence is constructed based on the actual temperature measured by the sensors in the group. i =T i+1 (t)-T i (t), i=1,…,N-1, judge the consistency of temperature difference sign, where ΔT i is the temperature difference between the i-th pair of adjacent sensors, in °C, T i+1 (t), T i(t) are the temperatures of sensor No. i+1 and No. i at time t, respectively. t is the sampling time in seconds, and N is the total number of sensors in the group.

[0093] Define μ:

[0094]

[0095] Where μ is the sign consistency factor, sgn(·) is the sign function;

[0096] Definition of g i Quantify gradient smoothness:

[0097]

[0098] Among them, g i is the local gradient change rate, in °C·m -1 , d is the center distance between adjacent sensors, in m;

[0099] |g i |Compare with the threshold and get

[0100] Where γ is the gradient consistency factor, H(·) is the Heaviside step function, and λ max is the gradient change rate threshold, in °C·m -1 ;

[0101] The threshold adopts the historical statistical adaptive form, expressed as λ max =k s σ T , where k s The empirical coefficient is 1.5-2.0, σ T is the standard deviation of 24h temperature difference;

[0102] The two factors are combined to obtain the judgment value S = μγ, where S is the smoldering feature judgment result, 1 is valid, and 0 is invalid;

[0103] Specifically, the model uses μ to eliminate accidental temperature differences with mixed signs, and then uses γ to limit false triggering caused by gradient mutations. The multiplication of the two takes 1 only when the temperature difference sequence is monotonous along the cable direction and the fluctuation is gentle, corresponding to the real smoldering heat conduction path, and the threshold λ max It self-adjusts based on real-time statistics, eliminating the need for regular manual recalibration. It adaptively converges with the noise level on site. The computational effort is linearly related to the number of sensors, adapting to the computing power of edge nodes and enabling judgments to be made within seconds. Field verification has shown a significant improvement in the recognition rate of early smoldering signals, while maintaining a low false alarm rate.

[0104] Step S3, performing energy entropy calculation on the filtered data based on the wavelet packet transform algorithm to generate a temperature fluctuation feature vector;

[0105] The energy entropy calculation in step S3 includes:

[0106] Decompose the temperature data sequence into multiple frequency band nodes using wavelet packets;

[0107] Extract the energy value of the node in the 0.01-0.1Hz frequency band and calculate its probability distribution entropy;

[0108] When the entropy value exceeds the historical mean by 3 times the standard deviation continuously, the warning is activated;

[0109] In step S3, the step of calculating the probability distribution entropy includes:

[0110] Perform M-layer wavelet packet decomposition on the filtered temperature sequence X(t) output in step S2:

[0111]

[0112] Where X(t) is the filtered temperature sequence in °C, M is the number of decomposition layers, j is the node index, k is the time displacement index, and c j,k is the decomposition coefficient of node j at displacement k, in °C, w j,k (t) is the corresponding wavelet packet basis function;

[0113] Calculate the node center frequency using the formula:

[0114]

[0115] Among them, f j is the center frequency of node j, in Hz, f s is the sampling frequency in Hz;

[0116] Construct the target frequency band node set B:

[0117] B={j|0.01≤f j ≤0.1},

[0118] Among them, B is the frequency band node index set;

[0119] Corresponding node energy:

[0120]

[0121] Among them, E j is the unweighted energy of node j, in °C 2 ;

[0122] A sensitive weight is introduced to the energy to suppress the very low frequency thermal drift, which is expressed as:

[0123]

[0124] Among them, w j is the sensitivity weight of node j, α is the amplitude adjustment coefficient, 0.3, β is the frequency attenuation coefficient, 10;

[0125] Weighted Energy in, is the weighted energy of node j, in °C 2 ;

[0126] Energy probability distribution:

[0127]

[0128] Among them, p j is the energy probability of node j, the denominator is the total energy of the target frequency band, and h is the traversal index;

[0129] Band energy entropy H b =-∑ j∈B p j ln p j , where H b It is the energy entropy of the 0.010.1Hz frequency band;

[0130] Save σ H and H avg For subsequent early warning calls,

[0131] σ H =std(H b ),H avg =mean(H b ),

[0132] Among them, σ H is the standard deviation of historical energy entropy, H avg is the mean historical energy entropy;

[0133] Specifically, wavelet packets are used to decompose temperature series at multiple scales, preserving both instantaneous fluctuations and long-period background components. The target bandwidth is precisely located using the node center frequency formula, and the influence of very low-frequency thermal drift is weakened with exponential weights, making the energy distribution more sensitive to unstable heat release signals in the early stages of smoldering. Energy entropy measures the uniformity of node energy. An increase in entropy indicates energy concentration in a local frequency band, corresponding to the intermittent heat release characteristics of smoldering. The historical mean and standard deviation are updated in real time, providing an adaptive benchmark for subsequent threshold judgments, thereby improving the versatility and robustness of the model under different environments.

[0134] Step S4: When the characteristic vector continuously exceeds the preset baseline range and reaches a first time threshold, a smoldering risk warning is triggered;

[0135] Step S4 includes:

[0136] Retrieve the load current timing data of the target cable;

[0137] If the current fluctuation trend and the temperature characteristic vector match the preset correlation pattern in the time domain, the warning level is raised to the second level response;

[0138] Step S5: Linking the ventilation system to execute the cooling plan.

[0139] This embodiment also provides a smart building fire safety control system based on big data, including:

[0140] Temperature sensor array, distributed along the cable surface;

[0141] Edge computing nodes, including:

[0142] A spatial filtering unit connected to the sensor array;

[0143] An entropy analysis unit receives an output of the filtering unit;

[0144] The early warning decision module generates instructions based on the entropy analysis results;

[0145] Linkage execution module, connected to the building ventilation control system;

[0146] Entropy analysis unit built-in:

[0147] The wavelet packet decomposition subunit is configured to select the 0.01-0.1 Hz frequency band;

[0148] The entropy calculation subunit generates a feature vector based on the frequency band energy distribution;

[0149] Baseline comparison subunit, which stores historical entropy mean and standard deviation data;

[0150] Early warning decision module execution:

[0151] When the entropy value is abnormal for 5 minutes and there is no current correlation, a level 1 warning log is generated;

[0152] When the entropy value is abnormal and there is a current correlation, a secondary response instruction is generated and the linkage module is activated;

[0153] The linkage execution module includes:

[0154] Local strong wind control interface, locate the ventilation equipment closest to the smoldering point;

[0155] Air conditioning return air blocking unit, closing the return air duct of adjacent areas;

[0156] Operation and maintenance report generator, outputs cable location identification and health score.

[0157] In summary, the bottleneck of smoldering detection has been broken through. Based on the gradient temperature rise model of the cable direction, early capture of slow smoldering is achieved, and the failure problem of traditional threshold alarm for temperature rise <2℃ / h is overcome; thermal fluctuations are quantified by wavelet packet energy entropy, and the thermal decomposition characteristics of insulating materials are identified in the non-open flame stage; the false alarm rate of the present invention is essentially reduced, and the dual mechanisms of spatial correlation filtering and frequency domain energy analysis are innovatively integrated: spatial layer: eliminates non-uniform interference such as hot air from air conditioners; frequency domain layer: weight function suppresses extremely low-frequency thermal drift, so that the entropy value change purely reflects the instability of smoldering heat release; the dual mechanisms work together to cut off the source of false alarms at the system level; in addition, the load current-temperature correlation model is introduced to provide cross-validation for cable health: when the entropy value anomaly and current harmonics appear synchronously, a secondary response is automatically triggered and a health score is generated, which promotes fire management from post-fault repair to predictive maintenance; the existing temperature sensors and ventilation systems of the building are reused, and the core functions can be realized by upgrading the edge node algorithm, and the renovation of old distribution rooms does not require large-scale wiring.

[0158] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A smart building fire safety control method based on big data, characterized in that: include, Step S1, collecting a data sequence of the surface temperature of cables in a power distribution room in real time through a distributed temperature sensor array; Step S2, performing spatial correlation filtering on the temperature data sequence to extract temperature gradient distribution characteristics between adjacent sensors; Step S3, performing energy entropy calculation on the filtered data based on the wavelet packet transform algorithm to generate a temperature fluctuation feature vector; Step S4: When the characteristic vector continuously exceeds the preset baseline range and reaches a first time threshold, a smoldering risk warning is triggered; Step S5: Linking the ventilation system to execute the cooling plan.

2. A smart building fire safety control method based on big data as claimed in claim 1, characterized in that: The spatial correlation filtering process in step S2 includes: Divide the sensor groups according to the cable direction, and each group contains at least three linearly arranged sensors; Calculate the real-time temperature difference between adjacent sensors in the group and generate a temperature difference sequence; When the temperature difference sequence meets the following conditions: a monotonically increasing / decreasing distribution along the cable direction, and the gradient change rate is within a preset range, it is determined to be a valid smoldering characteristic signal.

3. The method for controlling fire safety of smart buildings based on big data according to claim 2, characterized in that: In step S2, in the process of determining the effective smoldering characteristic signal, firstly, a temperature difference sequence is constructed based on the actual temperature measured by the sensors in the group. i =T i+1 (T)-T i (t), i=1,…,N-1, judge the consistency of temperature difference sign, where ΔT i is the temperature difference between the i-th pair of adjacent sensors, in °C, T i+1 (t), T i (t) are the temperatures of sensor No. i+1 and No. i at time t, respectively. t is the sampling time in seconds, and N is the total number of sensors in the group. Define μ: Where μ is the sign consistency factor, sgn(·) is the sign function; Definition of g i Quantify gradient smoothness: Among them, g i is the local gradient change rate, in °C·m -1 , d is the center distance between adjacent sensors, in m; |g i |Compare with the threshold and get Where γ is the gradient consistency factor, H(·) is the Heaviside step function, and λ max is the gradient change rate threshold, in °C·m -1 ; The threshold adopts the historical statistical adaptive form, expressed as λ max =k s σ T , where k s is the empirical coefficient, σ T is the standard deviation of 24h temperature difference; The two factors are combined to obtain the judgment value S=μγ, where S is the smoldering feature judgment result, 1 is valid and 0 is invalid.

4. The method for controlling fire safety of smart buildings based on big data according to claim 1, characterized in that: The energy entropy calculation in step S3 includes: Decompose the temperature data sequence into multiple frequency band nodes using wavelet packets; Extract the energy value of the node in the 0.01-0.1Hz frequency band and calculate its probability distribution entropy; When the entropy value exceeds the historical mean by 3 times the standard deviation continuously, the warning is activated.

5. The method for controlling fire safety of smart buildings based on big data according to claim 4, characterized in that: In step S3, the step of calculating the probability distribution entropy includes: Perform M-layer wavelet packet decomposition on the filtered temperature sequence X(t) output in step S2: Where X(t) is the filtered temperature sequence in °C, M is the number of decomposition layers, j is the node index, k is the time displacement index, and c j,k is the decomposition coefficient of node j at displacement k, in °C, w j,k (t) is the corresponding wavelet packet basis function; Calculate the node center frequency using the formula: Among them, f j is the center frequency of node j, in Hz, f s is the sampling frequency in Hz; Construct the target frequency band node set B: B={j|0.01≤f j ≤0.1}, Among them, B is the frequency band node index set; Corresponding node energy: Among them, E j is the unweighted energy of node j, in °C 2 ; A sensitive weight is introduced to the energy to suppress the very low frequency thermal drift, which is expressed as: Among them, w j is the sensitivity weight of node j, α is the amplitude adjustment coefficient, and β is the frequency attenuation coefficient; Weighted Energy in, is the weighted energy of node j, in °C 2 ; Energy probability distribution: Among them, p j is the energy probability of node j, the denominator is the total energy of the target frequency band, and h is the traversal index; Band energy entropy H b =-∑ j∈B p j ln p j , where H b It is the energy entropy of the 0.010.1Hz frequency band; Save σ H and H avg For subsequent early warning calls, σ H =std(H b ),H avg =mean(H b ), Among them, σ H is the standard deviation of historical energy entropy, H avg is the historical mean energy entropy.

6. The method for controlling fire safety of smart buildings based on big data according to claim 1, characterized in that: The step S4 comprises: Retrieve the load current timing data of the target cable; If the current fluctuation trend and the temperature characteristic vector match the preset correlation pattern in the time domain, the warning level is raised to the second level response.

7. A smart building fire safety control system based on big data, based on a smart building fire safety control method based on big data according to any one of claims 1 to 6, characterized in that: include: Temperature sensor array, distributed along the cable surface; Edge computing nodes, including: A spatial filtering unit connected to the sensor array; An entropy analysis unit receives an output of the filtering unit; The early warning decision module generates instructions based on the entropy analysis results; Linkage execution module, connected to the building ventilation control system.

8. The big data-based smart building fire safety control system according to claim 7, characterized in that: The entropy analysis unit has built-in: The wavelet packet decomposition subunit is configured to select the 0.01-0.1 Hz frequency band; The entropy calculation subunit generates a feature vector based on the frequency band energy distribution; The baseline comparison subunit stores the historical entropy mean and standard deviation data.

9. The big data-based smart building fire safety control system according to claim 7, characterized in that: The early warning decision module performs: When the entropy value is abnormal for 5 minutes and there is no current correlation, a level 1 warning log is generated; When the entropy value is abnormal and there is current correlation, a secondary response instruction is generated and the linkage module is started.

10. The big data-based smart building fire safety control system according to claim 7, characterized in that: The linkage execution module includes: Local strong wind control interface, locate the ventilation equipment closest to the smoldering point; Air conditioning return air blocking unit, closing the return air duct of adjacent areas; Operation and maintenance report generator, outputs cable location identification and health score.

Images