Real-time processing method and system for fire alarm data

Abstract

This invention relates to the field of public safety and smart fire protection technology, specifically to a real-time fire alarm data processing method and system, including: environmental steady-state reconstruction: accessing dynamic parameters of a non-fire environment; presetting the thermal static parameters of the building space; constructing a spatial thermal inertia differential equation; generating an ideal baseline; knowledge-driven simulation: receiving the ideal baseline; calling the synthesis operator in a pre-set disaster interference knowledge base; performing dynamic superposition injection; generating a virtual sensor state stream with physical characteristics and environmental background characteristics; homomorphism determination: performing double difference calculation; obtaining the real residual feature vector; performing geometric homomorphism verification; outputting fire alarm triggering, interference filtering, or fault indication commands. This invention solves the false alarm problem caused by non-stationary fluctuations in the environmental background in the prior art, and realizes the dynamic fusion and scenario adaptation of the standard signal model and the real environmental background.

Description

Technical Field

[0001] This invention relates to the field of public safety and smart fire protection technology, specifically to a method and system for real-time processing of fire alarm data. Background Technology

[0002] In existing fire alarm monitoring scenarios, sensors not only receive fire signals, but are also continuously subjected to complex interference from non-fire environmental dynamic parameters such as air conditioning start-stop, diurnal temperature variation, and seasonal radiation differences, resulting in non-stationary fluctuations in the environmental background.

[0003] Existing data processing solutions generally rely on static threshold judgment or simple statistical fitting, lacking physical modeling of the static thermal parameters and thermal inertia mechanism of the monitoring space. This processing method makes it difficult to track the drift of the environmental zero point in real time, resulting in the system being unable to effectively distinguish between environmental background noise, interference sources and real fire signals when faced with sudden changes in water vapor, equipment heat dissipation or gradual environmental changes. This easily leads to missed alarms and false alarms, and it is also difficult to accurately identify the faults of the sensor's own components. Therefore, how to reconstruct the dynamic environmental benchmark based on physical mechanisms, achieve real-time stripping of environmental common-mode interference and accurate matching of signal evolution patterns, so as to improve the timeliness and accuracy of fire alarm judgment under complex working conditions, has become an urgent technical problem to be solved. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for real-time processing of fire alarm data. This method and system can reconstruct a dynamic environmental baseline based on physical mechanisms, achieving real-time removal of common-mode interference and precise matching of signal evolution patterns. This effectively distinguishes between environmental background noise, interference sources, and actual fire signals, reducing false alarms and reducing false alarm rates, while improving the accuracy of judgment under complex operating conditions. Specifically, the technical solution of this invention is as follows:

[0005] Real-time processing methods for fire alarm data include:

[0006] Step 1, Environmental Steady-State Reconstruction: Access dynamic parameters of the non-fire environment; pre-set the thermal static parameters of the building space; construct the spatial thermal inertia differential equation using thermodynamic equilibrium deduction logic; substitute the dynamic parameters of the non-fire environment as energy input boundary conditions into the spatial thermal inertia differential equation; calculate the numerical evolution trajectory that the sensor should output at the current moment in a fire-free state; generate an ideal baseline.

[0007] Step 2, Knowledge-Driven Simulation: Receive the ideal baseline; call the synthesis operator in the pre-set disaster interference knowledge base; perform dynamic overlay injection; overlay the synthesis operator onto the ideal baseline in real time; generate a virtual sensor state stream with physical characteristics and environmental background characteristics;

[0008] Step 3, Homomorphism Determination: Receive real-time data acquired from the real sensor, the ideal baseline, and the virtual sensor state flow; perform dual difference calculation; calculate the real-time acquired data minus the ideal baseline to obtain the real residual feature vector; calculate the virtual sensor state flow minus the ideal baseline to obtain the theoretical residual feature vector; perform geometric homomorphism verification; compare the overlap between the real residual feature vector and the theoretical residual feature vector in terms of evolutionary morphology; output fire alarm triggering, interference filtering, or fault indication commands.

[0009] Preferably, the non-fire environment dynamic parameters mentioned in step 1 include the start / stop status of the HVAC system, the set temperature, the current time, and the seasonal factor; the thermal static parameters include the building space volume, the heat transfer coefficient of the wall envelope, and the combined specific heat capacity of indoor air and furnishings.

[0010] The thermodynamic equilibrium deduction logic includes: calculating the energy flow equilibrium state in the enclosed space based on the first law of thermodynamics and the aforementioned thermal static parameters; and calculating the theoretical temperature decrease or increase curve based on thermal capacity inertia when the air conditioner is turned on or the environment changes.

[0011] Preferably, the synthesis operator in step 2 is a parameterized expression of expert knowledge, including defining smoldering as a functional relationship in which the carbon monoxide concentration increases exponentially and the temperature change lags behind the smoke change, and defining water vapor as a functional relationship in which the infrared scattering rate changes abruptly but the chemical gas concentration remains constant.

[0012] The dynamic overlay injection includes: using the fluctuation of the ideal baseline to change the basis of the synthesis operator; making the generated virtual sensor state stream automatically include environmental background features; the virtual sensor state stream includes a real fire virtual stream and a water mist virtual stream.

[0013] Preferably, the dual difference calculation in step 3 includes the actual difference of the first path and the theoretical difference of the second path;

[0014] The first path real-world difference includes: calculating the real-time acquired data minus the ideal baseline; obtaining the real-world residual feature vector; the real-world residual feature vector is stripped of normal day-night temperature differences and air conditioning effects, while retaining abnormal fluctuations in reality;

[0015] The second path theoretical difference includes: calculating the virtual sensor state flow of each group minus the ideal baseline; obtaining multiple groups of theoretical residual feature vectors; the theoretical residual feature vectors represent the pure signal characteristics theoretically presented under the current environment if a specific disaster or interference occurs.

[0016] Preferably, the geometric homomorphism verification in step 3 includes: comparing the consistency of the curvature change, inflection point position and change trend of the actual residual feature vector and the theoretical residual feature vector on the time axis;

[0017] The consistency is determined by calculating cosine similarity or dynamic time-normalized distance.

[0018] Preferably, the rules for determining the output fire alarm trigger, interference filtering, or fault indication command include:

[0019] If the actual residual feature vector and the theoretical residual feature vector corresponding to the real fire highly overlap in evolutionary form, a fire alarm command is triggered.

[0020] If the actual residual feature vector coincides with the theoretical residual feature vector corresponding to the water mist, a filtering instruction is triggered.

[0021] If the actual residual feature vector exhibits numerical fluctuations but there is no matching model, a fault warning is triggered.

[0022] Based on a real-time fire alarm data processing system, including:

[0023] The environmental steady-state reconstruction unit is used to access dynamic parameters and thermal static parameters of non-fire environment in real time, and outputs the ideal baseline to the knowledge-driven simulation synthesis unit and the dual-difference geometric homomorphism determination unit through thermodynamic equilibrium deduction logic.

[0024] The knowledge-driven simulation synthesis unit is used to receive the ideal baseline and call the synthesis operator to generate multiple sets of virtual sensor state streams and send them to the dual-difference geometric homomorphism determination unit.

[0025] The dual-difference geometric homomorphism determination unit is used to receive real sensor data, the ideal baseline and the virtual sensor state stream, perform dual-difference calculation and geometric homomorphism verification, and output the determination result.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] 1. This invention utilizes thermodynamic equilibrium deduction logic to construct a spatial thermal inertia differential equation, taking the dynamic parameters of the non-fire environment as energy input boundary conditions, and calculates the ideal baseline in the fire-free state in real time; this mechanism establishes a zero-point reference surface that dynamically follows environmental changes, effectively eliminating environmental common-mode interference caused by day-night temperature differences, air conditioning start-up and shutdown, etc., and solving the false alarm problem caused by non-stationary fluctuations of the environmental background in the background technology.

[0028] 2. This invention employs knowledge-driven simulation technology, using synthesis operators to parameterize expert experience regarding fires or interference, and dynamically superimposes it onto an ideal baseline to generate a virtual sensor state stream containing current environmental background characteristics. This mechanism enables virtual signals to automatically adapt to the current complex working conditions, ensuring that the comparison model possesses an environmental fingerprint, and realizing the dynamic fusion and scenario-based adaptation of standard signal models and real-world environmental backgrounds.

[0029] 3. This invention constructs a clean signal space with zero background interference through dual difference calculation and performs geometric homomorphism verification, focusing on comparing the topological consistency between the actual residual and the theoretical residual in terms of curvature change, inflection point position and evolution trend; this method abandons the traditional static threshold judgment and can keenly identify early fire signals with weak amplitude but consistent with combustion dynamics characteristics, thus improving the ability to capture weak signals and strong feature events.

[0030] 4. This invention constructs a three-element classification judgment rule that includes fire alarm triggering, interference filtering, and fault indication. By comparing real signals with interference models such as real fire and water mist in a full library, it accurately distinguishes between real fires and known environmental interference. At the same time, it triggers fault indication for abnormal situations with numerical fluctuations but no matching model, realizing online self-diagnosis of sensor zero drift or component damage, and avoiding false alarms of equipment failure as fires. Attached Figure Description

[0031] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0032] Figure 1 This is a flowchart of the method of the present invention;

[0033] Figure 2 This is a structural diagram of the system of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0035] Example 1:

[0036] Please see Figure 1 Real-time processing methods based on fire alarm data include:

[0037] Environmental steady-state reconstruction: Integrate dynamic parameters of the non-fire environment; pre-set the thermal static parameters of the building space; construct the spatial thermal inertia differential equation using thermodynamic equilibrium deduction logic; substitute the dynamic parameters of the non-fire environment as energy input boundary conditions into the spatial thermal inertia differential equation; calculate the numerical evolution trajectory that the sensor should output at the current moment under non-fire conditions; generate an ideal reference baseline;

[0038] Knowledge-driven simulation: Receives an ideal baseline; invokes a synthesis operator from a pre-set disaster disturbance knowledge base; performs dynamic overlay injection; overlays the synthesis operator onto the ideal baseline in real time; generates a virtual sensor state stream with physical characteristics and environmental background features;

[0039] Homomorphism determination: Receive real-time data acquired by real sensors, ideal baseline, and virtual sensor state flow; perform dual difference calculation; calculate the real residual feature vector by subtracting the ideal baseline from the real-time acquired data; calculate the theoretical residual feature vector by subtracting the ideal baseline from the virtual sensor state flow; perform geometric homomorphism verification; compare the overlap between the real residual feature vector and the theoretical residual feature vector in terms of evolutionary morphology; output fire alarm trigger, interference filtering, or fault indication commands.

[0040] This embodiment details the execution mechanism of the environmental steady-state reconstruction steps. The core objective is to establish a zero-point reference surface that can dynamically follow environmental changes. Upon system startup, initial state synchronization is performed, reading the current sensor measurements as the initial conditions for the thermodynamic differential equation, i.e., the state value at t=0. Immediately, dynamic parameters of the non-fire environment are input; these parameters serve as external excitation sources for the system, driving subsequent physical deductions. The system retrieves pre-stored thermal static parameters of the building space, which define the physical properties of the monitored space. Based on these two types of input data, the system runs thermodynamic equilibrium deduction logic. This logic is not based on simple statistical fitting, but rather on constructing a spatial thermal inertia differential equation based on the principle of energy conservation. During the calculation process, the system substitutes the real-time changing dynamic parameters of the non-fire environment as boundary conditions for energy input into the differential equation to solve, thereby calculating the theoretical physical field data within the space at the current moment, such as theoretical temperature or theoretical smoke concentration. A pre-set sensor response mapping model is invoked to map the aforementioned theoretical physical field data into numerical values ​​in the sensor electrical signal domain. The evolution trajectory of these values ​​is defined as the ideal baseline, whose physical meaning is to characterize the pure state of the environmental background.

[0041] The knowledge-driven simulation step aims to construct a dynamic standard model for comparison; after receiving the ideal baseline, the system immediately calls the synthesis operator from the disaster interference knowledge base; the synthesis operator here is a mathematical parameterized encapsulation of the fire expert's experience knowledge; by performing a dynamic superposition injection operation, the system superimposes the abnormal signal features described by the synthesis operator onto the ideal baseline that is fluctuating with the environment in a nonlinear manner.

[0042] The nonlinear superposition operation is performed using the following formula: ;in, For virtual sensor state flow, For an ideal reference baseline, The standard signal output by the synthesis operator; This is an environmental fluctuation extraction function used to calculate the normalized fluctuation index of the baseline within the current sliding window, which is the dimensionless value obtained by dividing the variance by the square of the mean. The preferred data segment is a time interval of 30 to 60 seconds. The environmental coupling coefficient is set to a value between 0.1 and 0.5. This formula enables the amplitude of the injected virtual signal to automatically scale with the intensity of the environmental background noise, achieving dynamic fusion of the signal and the background. The product of this operation is the virtual sensor state stream, which not only has the typical physical characteristics of fire or interference, but more importantly, it automatically integrates the environmental background fluctuations at the current moment, achieving scene-specific adaptation of the signal.

[0043] The homomorphic determination step serves as the decision-making center; the system synchronously receives three signals: real-time data from real sensors, an ideal baseline, and virtual sensor state streams; the calculation engine performs dual differential calculations: the first calculation subtracts the ideal baseline from the real data to generate a real residual feature vector, which filters out environmental background noise and retains only the abnormal fluctuation components present in reality; the second calculation subtracts the ideal baseline from the virtual sensor state stream to generate a theoretical residual feature vector, which represents the pure form that a standard fire or interference signal should theoretically present under the current specific environment;

[0044] Based on this, the system performs geometric homomorphism verification, which ignores the difference in absolute signal amplitude and focuses on comparing the topological consistency of the actual residual feature vector and the theoretical residual feature vector in terms of evolutionary form, and accurately outputs fire alarm triggering, interference filtering or fault indication commands accordingly. Through the above mechanism, the present invention achieves physical stripping of environmental noise at the logical level, and solves the problem of missed and false alarms caused by environmental drift in the prior art.

[0045] Example 2:

[0046] Non-fire environment dynamic parameters include The system's start / stop status, set temperature, current time, and seasonal factors; thermal static parameters include building space volume, heat transfer coefficient of the wall envelope, and the combined specific heat capacity of indoor air and furnishings;

[0047] The thermodynamic equilibrium deduction logic includes: calculating the energy flow equilibrium state in a closed space based on the first law of thermodynamics and combined with thermal static parameters; and calculating the theoretical temperature decrease or increase curve based on thermal capacity inertia when the air conditioner is turned on or the environment changes.

[0048] In this specific configuration, the dynamic parameters of the non-fire environment are strictly defined as variables that significantly disturb the indoor thermal field; including The real-time start / stop status signals of the air conditioning system, the set temperature value on the thermostat panel, the current time indicating the day-night cycle, and the seasonal factor reflecting changes in solar radiation intensity are all collected in real time through the standard communication interface of the Building Information Modeling (BMS) or IoT sensing terminals deployed on-site. Simultaneously, considering the openness of the real environment, the model introduces a random disturbance tolerance factor to characterize occasional thermal disturbances not explicitly modeled, such as random window opening, personnel movement, and equipment heat dissipation, ensuring the baseline has a certain dynamic tolerance. The static thermal parameters are determined by the Building Information Modeling (BIM). Or on-site survey data, covering the determined building space volume, the thermal conductivity and heat transfer coefficient of the wall enclosure structure used to quantify the rate of heat loss, and the comprehensive specific heat capacity of indoor air and furnishings used to characterize the heat absorption and release capacity of indoor objects.

[0049] The specific execution flow of the thermodynamic equilibrium derivation logic follows the first law of thermodynamics; the core calculation uses the aforementioned thermodynamic static parameters to construct the system's internal energy function and calculates the flow equilibrium state of energy input and output within the enclosed space in real time; the specifically constructed lumped-parameter thermal inertia differential equation is as follows: ;

[0050] in, The combined heat capacity of indoor air and furnishings. For the theoretical temperature to be determined, This refers to the outdoor ambient temperature or the temperature at the outer boundary of the wall. The heating and cooling load power varies with the start and stop status of the air conditioner; the value is specified as positive for heating mode and negative for cooling mode. The radiative heat gain varies with time and seasonal factors. and The heat transfer coefficients and areas of different enclosure structures are given respectively; the system uses the fourth-order Runge-Kutta method to discretize and solve the equation to obtain... The time-series solution; when the air conditioner is turned on for cooling or the external ambient temperature undergoes a step change, the logic excludes the physical possibility of instantaneous jumps in sensor values. Instead, based on the principle of thermal capacity inertia, it calculates the specific slope of the theoretical temperature's downward or upward curve through numerical integration. This calculation method based on physical mechanisms ensures that the generated ideal baseline has continuity and differentiability in the time domain, accurately reproducing the gradual environmental change process, such as "boiling a frog in warm water," thus providing a high-precision background reference for subsequent identification of minor fire signals.

[0051] Example 3:

[0052] The synthesis operators are parameterized expressions of expert knowledge, including defining smoldering as a function of exponentially increasing carbon monoxide concentration with temperature lags behind smoke changes, and defining water vapor as a function of abrupt changes in infrared scattering rate but constant chemical gas concentration.

[0053] Dynamic overlay injection includes: using fluctuations in the ideal baseline to change the basis of the synthesis operator; automatically incorporating environmental background features into the generated virtual sensor state stream; the virtual sensor state stream includes a real fire virtual stream and a water mist virtual stream.

[0054] This embodiment provides a clear technical definition of the synthesis operator, serving as an interface for translating expert experience into machine language. For the extremely difficult-to-detect smoldering phenomenon, the synthesis operator parameterizes it as a multidimensional functional relationship: carbon monoxide concentration increases exponentially with time, while the rate of change of temperature parameter significantly lags behind the rate of change of smoke concentration on the time axis. Specifically, the parameterized function is defined as: carbon monoxide concentration... ,in, This is the starting point of the smoldering phenomenon. This represents the initial background concentration of carbon monoxide. This is the concentration increase coefficient. This is a preset smoldering growth rate coefficient, whose unit is the reciprocal of time, for example... To ensure that the independent variable of the exponential function is dimensionless, for example, its value is taken to be between 0.05 and 0.2; the rate of change of the temperature parameter With the rate of change of smoke concentration Satisfying the convolution relation: ,in, This is a preset smoke-temperature coupling conversion coefficient, with dimensions [K / %Obs], used to balance the physical units on both sides of the equation. The impulse response is a first-order hysteresis transfer function. , The thermal convection lag time constant is determined by the spatial height; for common false alarm sources such as water vapor, the synthesis operator defines it as a function of the infrared scattering rate undergoing high-frequency abrupt changes while the concentrations of carbon monoxide and other combustion product gases remain constant.

[0055] In the process of generating virtual signals, the dynamic superposition injection mechanism plays a crucial coupling role. This mechanism abandons the traditional simple linear addition and instead uses the real-time fluctuations of the ideal baseline generated in the preceding steps as a carrier to modulate the basis of the synthesis operator. If the current environmental baseline shows a decreasing temperature trend due to the strong cooling of the air conditioner, the generated real fire virtual flow will appear as an abnormal temperature rise superimposed on this decreasing curve. Its synthesis result may only show a flat or slight temperature rise, rather than a standard sharp temperature rise. The generated water mist virtual flow will also carry the temperature and humidity characteristics of the current environment. This mechanism makes the virtual sensor state flow no longer an ideal waveform in the laboratory environment, but automatically includes the environmental background characteristics of the current complex working conditions. This significantly improves the robustness of the system in extreme or dynamic environments, ensuring that the comparison benchmark always keeps in sync with the real scene.

[0056] Example 4:

[0057] The dual-difference calculation includes the actual difference of the first path and the theoretical difference of the second path;

[0058] The first path of real-world difference includes: calculating the real-time collected data minus the ideal baseline; obtaining the real-world residual feature vector; and removing the normal day-night temperature difference and air conditioning effects from the real-world residual feature vector, while retaining the abnormal fluctuations in reality.

[0059] The second path theoretical difference includes: calculating the state flow of each group of virtual sensors minus the ideal baseline; obtaining multiple sets of theoretical residual feature vectors; the theoretical residual feature vectors represent the pure signal characteristics theoretically presented under the current environment if a specific disaster or interference occurs.

[0060] This embodiment employs dual-difference computation as the core method for signal feature extraction, aiming to construct a clean signal space with zero background interference. The processing flow is divided into two parallel paths. In the first path, real-world difference, the computing unit receives the raw data stream from the sensor in real time and subtracts the ideal baseline synchronously derived from the physical model. Since this baseline has accurately simulated normal environmental behaviors such as day-night temperature drift and air conditioning start-stop impact, the real-world residual feature vector obtained by subtraction essentially removes all legitimate environmental background influences. If the environment is normal, the amplitude of this vector should fluctuate around zero. If there is an anomaly, the vector retains the abnormal fluctuation components that violate physical predictions in reality.

[0061] The synchronously executed second-path theoretical difference subtracts the same ideal baseline from each group of virtual sensor state flows containing environmental features. This mathematical operation aims to extract the theoretical residual feature vector. This vector has a clear physical orientation, representing the purely abnormal form that the signal should theoretically exhibit relative to the normal background under the current specific environmental background, assuming a specific type of disaster or specific interference. Through the dual difference, the system successfully eliminates environmental common-mode interference, enabling subsequent feature comparison to focus entirely on the essential differences in the signal, rather than being affected by environmental temperature or equipment operating status.

[0062] Example 5:

[0063] Step 3, geometric homomorphism verification, includes comparing the consistency of curvature changes, inflection point positions, and trends of the actual residual eigenvectors and the theoretical residual eigenvectors on the time axis.

[0064] Consistency is determined by calculating cosine similarity or dynamic time-warped distance.

[0065] This embodiment introduces a geometric homomorphism verification mechanism to replace the traditional threshold comparison method. This verification mechanism focuses on the topological similarity of the signal in the time dimension. The system places the actual residual feature vector and each group of theoretical residual feature vectors in the same vector space for in-depth comparison. The core evaluation indicators include: curvature change, that is, the second derivative feature in the signal evolution process, which reflects the acceleration of temperature rise or smoke diffusion; inflection point position, that is, the time node when the signal properties undergo a fundamental change; and the overall trend of change.

[0066] To quantify this morphological consistency, the calculation module employs a cosine similarity algorithm. This algorithm determines the degree of directional similarity by calculating the cosine of the angle between two high-dimensional vectors; the closer the values ​​are, the more similar the vectors appear. The more similar the waveforms, the better; or dynamic time warping can be used. The distance algorithm allows for non-linear elastic deformation of two time series along the time axis, thereby effectively identifying signal patterns that are highly similar in shape but have slight differences in occurrence rate. This technique enables the system to keenly identify early fire signals with weak amplitudes but consistent with combustion dynamics, achieving accurate capture of weak signals and strong characteristic events.

[0067] Example 6:

[0068] The rules for determining whether to output fire alarm trigger, interference filtering, or fault indication commands include:

[0069] If the actual residual eigenvector and the theoretical residual eigenvector corresponding to the real fire highly overlap in evolutionary form, a fire alarm command will be triggered.

[0070] If the actual residual feature vector coincides with the theoretical residual feature vector corresponding to the water mist, a filtering instruction is triggered.

[0071] If the actual residual feature vector exhibits numerical fluctuations but there is no matching model, a fault warning will be triggered.

[0072] This embodiment constructs a strict set of ternary classification judgment rules to ensure the accuracy and interpretability of the system output instructions; the judgment logic executes fire confirmation: if the calculation results show that the similarity between the actual residual feature vector and the theoretical residual feature vector corresponding to the real fire in geometric evolution exceeds the preset high confidence threshold, it indicates that the current abnormal fluctuation strictly conforms to the laws of combustion physics, and the system triggers a fire alarm command;

[0073] Judgment logic execution interference elimination: If the comparison finds that the actual residual feature vector mainly coincides with the theoretical residual feature vector corresponding to the water mist, the system determines that the signal is a known environmental interference, triggers the filtering instruction, and keeps silent while recording the event in the background to avoid disturbing the user;

[0074] Fault diagnosis of the judgment logic execution: If the system detects significant numerical fluctuations in the actual residual feature vector, that is, the amplitude or variance of the feature vector exceeds the preset silent environment noise threshold. This eliminates normal circuit noise, but after a full comparison with all preset real fire models and interference models, no matching model was found. This indicates that the anomaly does not conform to either fire characteristics or known interference characteristics. At this point, the system infers that the sensor itself may have experienced zero-point drift, circuit failure, or encountered unknown types of physical damage, thus triggering a fault prompt to guide maintenance personnel to conduct targeted troubleshooting. This closed-loop logic ensures that the system will not mistakenly report unknown equipment failures as fires.

[0075] Example 7:

[0076] Please see Figure 2 A real-time fire alarm data processing system includes:

[0077] The environmental steady-state reconstruction unit is used to access dynamic parameters and thermal static parameters of non-fire environment in real time, and outputs the ideal baseline to the knowledge-driven simulation synthesis unit and the dual-difference geometric homomorphism determination unit through thermodynamic equilibrium deduction logic.

[0078] The knowledge-driven simulation synthesis unit is used to receive the ideal baseline and call the synthesis operator to generate multiple sets of virtual sensor state streams and send them to the dual-difference geometric homomorphism determination unit.

[0079] The dual-difference geometric homomorphism determination unit is used to receive real sensor data, ideal baseline and virtual sensor state stream, perform dual-difference calculation and geometric homomorphism verification, and output determination results.

[0080] This embodiment illustrates the hardware architecture and data flow of the system, which consists of three highly decoupled functional units working together. The environmental steady-state reconstruction unit, as the logical foundation of the system, is equipped with a high-frequency data acquisition interface and a floating-point arithmetic core. Its function is to receive dynamic parameters and thermal static parameters of the non-fire environment in real time, and continuously run the thermodynamic equilibrium deduction logic to output an ideal baseline representing the theoretical state of the fire-free environment with high-precision numerical calculations. This baseline signal is used as a synchronization clock signal and is sent to the subsequent simulation unit and decision unit respectively.

[0081] The knowledge-driven simulation synthesis unit serves as the system's dynamic adversarial sample generator. It integrates a disaster interference knowledge base that stores disaster feature models. After receiving the ideal baseline, the unit calls the synthesis operator in real time according to the current environmental state to generate multiple sets of virtual sensor state streams containing real fire features and specific interference features, and sends these simulated data streams carrying environmental fingerprints to the downstream.

[0082] The dual-difference geometric homomorphic decision unit, as the final decision-making and execution mechanism of the system, integrates real sensor data from the physical world, ideal baseline data from physical deduction, and virtual sensor state stream data from knowledge simulation. This unit integrates a high-performance microprocessor and a pattern recognition algorithm engine, performing dual-difference calculations in parallel to extract residual features from each signal, then performing geometric homomorphism verification to complete waveform matching, and finally outputting a determined decision command based on the matching result. This system architecture, through the deep integration of the physical model and the expert knowledge base, significantly improves fire detection efficiency in complex environments.

[0083] 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 it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for real-time processing of fire alarm data, characterized in that, include: Step 1, Environmental Steady-State Reconstruction: Access dynamic parameters of the non-fire environment; Pre-set the static thermal parameters of the building space; The differential equation of space thermal inertia is constructed using the deductive logic of thermodynamic equilibrium. Substitute the dynamic parameters of the non-fire environment as energy input boundary conditions into the spatial thermal inertia differential equation; calculate the numerical evolution trajectory that the sensor should output at the current moment in a fire-free state; Generate an ideal baseline; Step 2, Knowledge-driven simulation: Receive the ideal baseline; Invoke the composition operators in the pre-set disaster interference knowledge base; perform dynamic overlay injection; and overlay the composition operators onto the ideal baseline in real time. Generate a virtual sensor state stream that has physical characteristics and includes environmental background characteristics; Step 3, Homomorphism Determination: Receive real-time data acquired by the real sensor, the ideal baseline, and the virtual sensor state stream; perform dual difference calculation; The real-time acquired data is subtracted from the ideal baseline to obtain the actual residual feature vector; the virtual sensor state flow is subtracted from the ideal baseline to obtain the theoretical residual feature vector; geometric homomorphism verification is performed; the overlap between the actual residual feature vector and the theoretical residual feature vector in terms of evolutionary morphology is compared; and fire alarm triggering, interference filtering, or fault indication commands are output. The synthesis operator described in step 2 is a parameterized expression of expert knowledge, including defining smoldering as a function in which carbon monoxide concentration increases exponentially and temperature changes lag behind smoke changes, and defining water vapor as a function in which infrared scattering rate changes abruptly but chemical gas concentration remains constant. The dynamic overlay injection includes: using the fluctuation of the ideal baseline to change the basis of the synthesis operator; making the generated virtual sensor state stream automatically include environmental background features; the virtual sensor state stream includes a real fire virtual stream and a water mist virtual stream; The geometric homomorphism verification in step 3 includes: comparing the consistency of the curvature changes, inflection point positions, and trends of the actual residual feature vector and the theoretical residual feature vector on the time axis; The consistency is determined by calculating cosine similarity or dynamic time-normalized distance.

2. The method for real-time processing of fire alarm data according to claim 1, characterized in that, The non-fire environment dynamic parameters mentioned in step 1 include the start / stop status of the HVAC system, the set temperature, the current time, and the seasonal factor; the thermal static parameters include the building space volume, the heat transfer coefficient of the wall envelope, and the combined specific heat capacity of indoor air and furnishings. The thermodynamic equilibrium deduction logic includes: calculating the energy flow equilibrium state in the enclosed space based on the first law of thermodynamics and the aforementioned thermal static parameters; and calculating the theoretical temperature decrease or increase curve based on thermal capacity inertia when the air conditioner is turned on or the environment changes.

3. The method for real-time processing of fire alarm data according to claim 1, characterized in that, The dual difference calculation described in step 3 includes the actual difference of the first path and the theoretical difference of the second path; The first path real-world difference includes: calculating the real-time acquired data minus the ideal baseline; obtaining the real-world residual feature vector; the real-world residual feature vector is stripped of normal day-night temperature differences and air conditioning effects, while retaining abnormal fluctuations in reality; The second path theoretical difference includes: calculating the virtual sensor state flow of each group minus the ideal baseline; obtaining multiple groups of theoretical residual feature vectors; the theoretical residual feature vectors represent the pure signal characteristics theoretically presented under the current environment if a disaster or interference occurs.

4. The method for real-time processing of fire alarm data according to claim 1, characterized in that, The rules for determining the output fire alarm trigger, interference filtering, or fault indication commands include: If the actual residual feature vector and the theoretical residual feature vector corresponding to the real fire highly overlap in evolutionary form, a fire alarm command is triggered. If the actual residual feature vector coincides with the theoretical residual feature vector corresponding to the water mist, a filtering instruction is triggered. If the actual residual feature vector exhibits numerical fluctuations but there is no matching model, a fault warning is triggered.

5. A real-time fire alarm data processing system, applied to the real-time fire alarm data processing method according to any one of claims 1 to 4, characterized in that, include: The environmental steady-state reconstruction unit is used to access dynamic parameters and thermal static parameters of non-fire environment in real time, and outputs the ideal baseline to the knowledge-driven simulation synthesis unit and the dual-difference geometric homomorphism determination unit through thermodynamic equilibrium deduction logic. The knowledge-driven simulation synthesis unit is used to receive the ideal baseline and call the synthesis operator to generate multiple sets of virtual sensor state streams and send them to the dual-difference geometric homomorphism determination unit. The dual-difference geometric homomorphism determination unit is used to receive real sensor data, the ideal baseline and the virtual sensor state stream, perform dual-difference calculation and geometric homomorphism verification, and output the determination result.

Images