Multi-sensor linkage fire monitoring system based on image recognition

Abstract

The invention discloses a multi-sensor linkage fire monitoring system based on image recognition, particularly relates to the field of multi-sensor fire monitoring, is used for solving the problem of harmonic oscillation and spatial data harmonic mismatch in a complex environment, and aims at solving the problem of harmonic oscillation and spatial data harmonic mismatch through frequency domain conversion and spatial harmonicity analysis of multi-sensor data in combination with seismic frequency coupling degree calculation. And the harmonic oscillation risk in the feedback loop is quantified, and the system stability and the data consistency are evaluated by using the seismic frequency harmonic index and the atlas complex range index. Under different risk states, the system can be flexibly switched to a harmonic attenuation atlas mode or a complex shock control mode, shock is inhibited, space harmony is optimized, and forest fire monitoring precision is improved. And by dynamically adjusting the feedback frequency and a spatial harmonic algorithm, data harmonic mismatch is reduced, sensor node data coordination is ensured, and the stability and reliability of the system are enhanced. The system can quickly adapt to environmental changes in the early stage of a fire, ensures accurate capture and quick response of fire signals, and improves the fire detection reliability.

Description

Technical Field

[0001] The present invention relates to the field of multi - sensor fire monitoring. More specifically, the present invention relates to a multi - sensor linkage monitoring fire system based on image recognition. Background Art

[0002] In the specific scenario of forest fire monitoring, the sensor network is usually arranged in areas with complex terrain and dense vegetation, such as valleys, hills, and high - altitude forest areas. Due to the different geographical locations of different sensor nodes, the received signals such as temperature, humidity, and smoke vary greatly. For example, the sensor data arranged in low - lying areas may lag behind the nodes at higher places, and the barrier effect of the terrain may cause uneven signal propagation. Such environmental characteristics make it more difficult to capture fire signals, especially in the early stage of a fire. Due to sudden changes in wind speed or abnormal temperature and humidity, the monitoring system may face the situation of inconsistent multi - sensor data and unstable feedback. In addition, the change in vegetation coverage also directly affects the signal transmission and response, making it difficult for the sensor network to achieve stable data collection and transmission in forest areas with complex terrain. These factors together lead to the need for a more sensitive and stable real - time response ability of the fire monitoring system when facing a complex natural environment.

[0003] Existing fire monitoring systems have exposed some significant technical deficiencies when dealing with these complex situations. First, in terms of the spatial reconciliation of multi - sensor data, existing systems usually rely on simple weighted average or mean processing methods, which are difficult to fully cope with data deviations caused by terrain and vegetation. The data reconciliation between sensor nodes is poor, especially in areas with severe data mismatch. Existing methods cannot effectively identify and process these out - of - tune areas, resulting in the system being unable to accurately judge the true change of fire signals. Second, existing systems lack a real - time suppression mechanism for harmonic oscillations in the sensor feedback loop. When harmonic oscillations occur in the system, the oscillation frequency may accumulate continuously with environmental changes. Existing systems fail to provide an effective real - time adjustment strategy and cannot adjust and suppress the oscillation frequency in a timely manner. This instability in the feedback loop directly affects the accuracy of fire monitoring. Especially in an environment with drastic terrain undulations, the oscillation effect further increases the risk of false alarms and missed alarms of the system for fire signals. Therefore, existing technologies are insufficient in dealing with multi - sensor feedback oscillations and data reconciliation problems in complex environments and urgently need more accurate real - time adjustment and processing solutions.

[0004] To solve the above problems, a technical solution is provided as follows. Summary of the Invention

[0005] To overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a multi-sensor linkage fire monitoring system based on image recognition. By performing frequency-domain conversion and spatial harmonicity analysis on multi-sensor data, combining the calculation of seismic frequency coupling degree, quantifying the harmonic oscillation risk in the feedback loop, and evaluating the system stability and spatial data consistency through the seismic frequency harmonic control index and the harmonic adjustment complex mountain index. The feedback frequency and spatial harmonicity are adjusted in real time, and the harmonic attenuation harmonic adjustment mode and the complex seismic adjustment control mode are switched respectively in high-risk and medium-risk states to suppress the oscillation caused by high-energy harmonics, enhance the spatial data harmonicity, and effectively improve the fire monitoring accuracy in complex forest environments. Dynamically adjust the frequencies with high seismic frequency coupling degree in the feedback loop, optimize the spatial data harmonicity, reduce the problem of data harmonic mismatch, ensure the data coordination among sensor nodes, and improve the stability of the system and the reliability of monitoring. Through real-time data feedback and multi-level information fusion, combined with the verification of the geographic information system, the system can quickly adapt to environmental changes, especially in the early stage of a fire, ensure the accurate capture and rapid response of fire signals, and effectively improve the reliability of fire detection and the accuracy of system response to solve the problems proposed in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A multi-sensor linkage fire monitoring system based on image recognition, comprising: a data acquisition module, a harmonic analysis module, a stability evaluation module, a harmonic attenuation regulation module, and a complex seismic adjustment control module;

[0008] The data acquisition module collects data from multiple sensors in real time and filters the noise through a Kalman filter; the filtered data is used as the input for harmonic analysis and spatial correlation analysis and is transmitted to the harmonic analysis module.

[0009] Based on the collected sensor data, the harmonic analysis module first performs harmonic analysis to identify the energy distribution and attenuation characteristics of the frequency components in the feedback loop, calculates the seismic frequency harmonic control index for quantifying the resonance oscillation that may be caused by the sensor feedback loop during the fire monitoring process; at the same time, through spatial correlation analysis, combined with the spatial distribution characteristics of multi-sensor data in the forest fire environment, evaluates the harmonic consistency between the data, and calculates the harmonic adjustment complex mountain index for measuring the spatial harmonicity in the sensor network; transmits the seismic frequency harmonic control index and the harmonic adjustment complex mountain index to the stability evaluation module.

[0010] After obtaining the seismic frequency harmonic control index and the harmonic adjustment and complex mountain index, the stability assessment module evaluates the stability of the current fire monitoring system according to their real-time values; when both the seismic frequency harmonic control index and the harmonic adjustment and complex mountain index exceed the set high threshold, it switches to the harmonic attenuation and harmonic adjustment mode; when the two indicators are in the interval between their respective set high and low thresholds, it switches to the complex seismic adjustment and control mode; and it transmits the switching instructions to the harmonic attenuation and harmonic adjustment module or the complex seismic adjustment and control module respectively.

[0011] After the harmonic attenuation control module switches to the harmonic attenuation and harmonic adjustment mode, it dynamically adjusts the calibrated frequency components in the feedback loop, applies phase inversion processing using an adaptive filter to suppress oscillations; at the same time, it optimizes the spatial data harmonic algorithm, identifies the harmonic failure areas based on the spatial harmonic deviation degree, and uses local regression and smoothing processing, combined with terrain weighting adjustment, to achieve data consistency.

[0012] After the complex seismic adjustment and control module switches to the complex seismic adjustment and control mode, it dynamically adjusts the intermediate frequency harmonic frequencies with high seismic frequency coupling degree in the feedback loop and adjusts the frequency response using a band-pass filter; it optimizes the spatial data harmonic degree, identifies the areas with poor data harmonic degree, and applies the local adaptive harmonic algorithm to adjust the data fusion weights.

[0013] In a preferred embodiment, the operation process of the harmonic analysis module includes the following:

[0014] S2.1, Based on the sensor data collected in real time in step S1 and processed by Kalman filtering, first perform the frequency domain conversion of the data; use the fast Fourier transform to convert the time domain data into the frequency domain to obtain the amplitude and phase information of each frequency component.

[0015] S2.2, Based on the frequency domain analysis results, identify the main frequency components in the feedback loop and record their harmonic distribution; by calculating the energy attenuation rate of these frequency components, evaluate whether the harmonic energy gradually decays during the feedback process, so as to calculate the seismic frequency harmonic control index.

[0016] The process of obtaining the seismic frequency harmonic control index is as follows:

[0017] S2.2.1, From the sensor data processed by filtering in step S1, perform the frequency domain conversion using wavelet transform instead of the conventional fast Fourier transform; the converted frequency domain data is W(f,t), where f represents frequency and t represents time; by analyzing the wavelet transform coefficients, identify the main harmonic frequency components f h , h represents different harmonic frequency components.

[0018] Calculate the energy distribution of the identified main harmonic frequencies, and at the same time consider the interaction between harmonics; for each main harmonic frequency, first calculate its energy E(f h), the formula is: E(f h ) = ∫ T |W(f h , t)| 2 dt; here, T is the time window of the wavelet transform, and W(f h , t) is the coefficient at frequency f h at time t; the coefficient of interaction between harmonics β h is used to evaluate the coupling effect between each harmonic, and the calculation method is: where, W(f k , t) is the wavelet coefficient at other frequency f k , and β h measures the energy interaction between harmonics.

[0019] S2.2.2, to quantify the resonance effect of harmonic frequencies in the quantization feedback loop, calculate the seismic frequency coupling degree ξ h , and its formula is: ξ h = E(f h ) · (1 + β h ).

[0020] S2.2.3, finally, integrate the seismic frequency coupling degrees of each main harmonic frequency to calculate the overall seismic frequency harmonic control index THAI, and its formula is: Here, represents the frequency change rate, which is used to capture the dynamic change of frequency over time, and f min and f max are the lowest and highest harmonic frequencies respectively.

[0021] In a preferred embodiment, S2.3, to evaluate the data consistency of multi-sensors in the forest fire environment, a spatial correlation analysis method is adopted to analyze the data correlation between each sensor node.

[0022] S2.4, based on the results of the spatial correlation analysis, calculate the global adjustment and complex mountain index.

[0023] The acquisition process of the global adjustment and complex mountain index is as follows:

[0024] S2.4.1, map the preliminarily processed multi-sensor data into a two-dimensional space to form a spatial data matrix D(x, y), where x and y are the coordinates of the sensors in the geographical space, and the spatial data matrix is the measurement value of the sensors at the corresponding positions.

[0025] S2.4.2, based on the spatial data matrix, calculate the data gradient around each sensor node to form a spatial gradient vector field G(x, y); the calculation method of the gradient vector is: where, and They respectively represent the change rates of data in the x - direction and y - direction.

[0026] S2.4.3. To evaluate the spatial harmony between different sensor nodes, a spatial interaction potential field Φ(x, y) is constructed to quantify the spatial coupling effect in the sensor network; the calculation formula is: Among them, (x′, y′) represents the coordinates of adjacent sensor nodes, G(x, y)·G(x′, y′) represents the dot - product of the gradient vectors between two nodes, and r(x, y, x′, y′) is the Euclidean distance between the two nodes.

[0027] S2.4.4. To quantify the degree of deviation of spatial harmony, the spatial harmony deviation δ(x, y) is used, and its calculation formula is: Among them, G avg (x, y) is the average gradient vector of all adjacent nodes around the node (x, y), and ||Gavg(x, y)|| is the modulus of the corresponding average gradient vector.

[0028] S2.4.5. Finally, by combining the spatial interaction potential field and the spatial harmony deviation, the overall harmonic complex index is calculated; the formula is as follows: SRMI = ∫ A Φ(x, y)·δ(x, y)dxdy; where A represents the geographical area covered by the sensor network.

[0029] In a preferred embodiment, the operation process of the stability evaluation module includes the following:

[0030] S3.1. Based on the seismic frequency harmonic control index and the harmonic complex index calculated in step S2, the changes of these two indexes are monitored in real - time.

[0031] S3.2. Combining the characteristics of the forest fire monitoring scenario, high thresholds and medium - range intervals of the seismic frequency harmonic control index and the harmonic complex index are preset.

[0032] S3.3. When the seismic frequency harmonic control index and the harmonic complex index monitored in real - time both exceed their respective set high thresholds, it is identified that the current state is a high - risk state; immediately switch to the harmonic attenuation and harmonic adjustment mode, and suppress the oscillation caused by high - energy harmonics by adjusting the feedback frequency and the spatial data harmonic algorithm.

[0033] S3.4. When the seismic frequency harmonic control index and the harmonic complex index monitored in real - time are in the interval between their respective set high thresholds and low thresholds, it is identified that the current state is a medium - risk state, indicating that although there are certain fluctuations in the resonance and spatial harmony in the feedback loop, they have not reached the out - of - control level; at this time, switch to the complex seismic adjustment mode.

[0034] S3.5, after the mode is switched, the corresponding adjustment strategy is immediately executed.

[0035] S3.6, after the mode is switched, continue to monitor the changes of the seismic frequency harmonic control index and the atmospheric modulation complex index in real time, and adjust the mode maintenance time or switch back to the regular monitoring mode according to the actual situation.

[0036] In a preferred embodiment, the operation process of the harmonic decay control module includes the following contents:

[0037] S4.1, after switching to the harmonic attenuation modulation mode, the frequency parameters in the feedback loop are dynamically adjusted to suppress the oscillation caused by high-energy harmonics; first, based on the seismic frequency harmonic control index calculated in step S2, the main harmonic frequency components that cause system oscillations in the feedback loop are identified; the frequency component with the highest seismic frequency coupling, that is, the specific frequency component that produces strong resonance, is marked as the calibration frequency component, and the calibration frequency component is attenuated in real time through an adaptive filter; the specific method is to apply phase inversion technology to the calibration frequency component to adjust the phase of the feedback signal.

[0038] S4.2, in order to ensure the data consistency of multiple sensors in the spatial dimension, a multi-level spatial data reconciliation algorithm is used for processing; first, based on the reconciliation index in step S3, the spatial reconciliation between different sensor nodes is analyzed to identify the reconciliation failure area; for the reconciliation failure area, a data reconciliation method based on local regression is applied, and the reconciliation data of the adjacent nodes are used to smooth the data of the disordered area; at the same time, the terrain weighted reconciliation technology is used to consider the attenuation and deformation of the sensor signal by the terrain factors, and the sensor data in the complex terrain area is weighted and adjusted to make it consistent with the overall network data.

[0039] In a preferred embodiment, the identification process of the blending failure area is specifically as follows:

[0040] After calculating the spatial harmony deviation of all sensor nodes, a global analysis of the deviation of the entire sensor network is performed; specifically, when the spatial harmony deviation of a node is greater than the harmony deviation threshold, and the spatial harmony deviation of its surrounding adjacent nodes exceeds the harmony deviation threshold by more than 20%, it is considered that the area where the corresponding node is located may have harmony failure, and is identified as a harmony failure area.

[0041] In a preferred embodiment, the operation process of the secondary shock control module includes the following contents:

[0042] S5.1, After switching to the complex vibration control mode, first dynamically adjust the frequency of the feedback loop to control and suppress medium-level oscillations. Based on the frequency components in step S4 where the vibration-frequency coupling degree is higher than 70% but lower than the resonance threshold itself, identify them as medium oscillation risk frequencies and adjust the frequency response characteristics of the feedback signal. Specifically, dynamically adjust the intermediate-frequency harmonics through a band-pass filter to separate its frequency response range from the resonance frequency of the system.

[0043] S5.2, In the complex vibration control mode, optimize the spatial harmony degree of multi-sensor data. Based on the calculated global adjustment complex mountain index and its real-time monitoring results in step S4, identify the regions with poor data harmony degree. Adopt a local adaptive harmony algorithm to gradually adjust the data fusion weights of adjacent sensor nodes for these regions with poor harmony degree. Specifically, weight the data of the sensor nodes by using the terrain factor and vegetation density factor.

[0044] In a preferred embodiment, the process of identifying the regions with poor data harmony degree is specifically as follows:

[0045] First, calculate the global adjustment complex mountain index value of each node and compare it with the data harmony degree threshold. If the global adjustment complex mountain index value of a certain node is lower than the data harmony degree threshold and the global adjustment complex mountain index values of its surrounding adjacent sensor nodes are all lower than 80% of the data harmony degree threshold, the corresponding region is identified as the region with poor data harmony degree.

[0046] The technical effects and advantages of a multi-sensor linkage fire monitoring system based on image recognition according to the present invention:

[0047] 1. Through the frequency-domain conversion and spatial harmony analysis of multi-sensor data, the present invention first identifies and calculates the harmonic frequency components, their energy distribution, and the interaction intensity in the feedback loop, and then calculates the vibration-frequency harmonic control index through the comprehensive analysis of the vibration-frequency coupling degree to quantify the oscillation risk that may be caused by the resonance effect in the feedback loop. At the same time, construct a spatial gradient vector field and analyze the spatial interaction potential field between sensor nodes, and combine the spatial harmony deviation degree to calculate the global adjustment complex mountain index to evaluate the spatial harmony degree and data consistency in the multi-sensor network. Based on the real-time values of these two indicators, conduct dynamic stability evaluation, and switch to the harmonic attenuation global adjustment mode and the complex vibration control mode respectively in high-risk and medium-risk states. By adjusting the feedback frequency, suppressing the oscillations caused by high-energy harmonics, and enhancing the spatial data harmony, effectively improve the fire monitoring accuracy and stability of the system in a complex forest environment, ensure the accurate capture and timely response to fire signals, avoid monitoring errors caused by harmonic resonance and spatial data disorder, and thus improve the reliability of fire detection and the accuracy of system response.

[0048] 2. After the present invention switches to the complex shock regulation mode, it dynamically adjusts the intermediate frequency harmonic frequencies with high shock frequency coupling degrees in the feedback loop, and uses a band-pass filter to adjust the frequency response. It optimizes the spatial data reconciliation degree, identifies areas with poor data reconciliation degree, and applies a local adaptive reconciliation algorithm to adjust the data fusion weights. It evaluates the shock frequency regulation index and the cosmic adjustment complex mountain index in real time, continuously optimizes the feedback control strategy, strengthens multi-level information fusion, and combines with a geographic information system for monitoring and verification.

[0049] 3. By switching to the complex shock regulation mode, the present invention can dynamically adjust the feedback frequency, suppress medium-level oscillations, and maintain the stability of the feedback loop. At the same time, it optimizes the spatial reconciliation degree of multi-sensor data, effectively reduces the data reconciliation mismatch problems caused by terrain complexity or vegetation density changes, and ensures the data coordination among sensor nodes. Through real-time data feedback and a continuously optimized feedback control strategy, the system can quickly adapt to environmental changes. Especially in the early stage of a fire, through multi-level information fusion and geographic information system verification, it improves the detection accuracy of fire signals and the overall reliability of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic structural diagram of a multi-sensor linkage fire monitoring system based on image recognition according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0052] Embodiment 1

[0053] Figure 1 A multi-sensor linkage fire monitoring system based on image recognition according to the present invention is provided, including: a data acquisition module, a harmonic analysis module, a stability evaluation module, a harmonic attenuation regulation module, and a complex shock regulation module;

[0054] The data acquisition module collects data from multiple sensors in real time and filters the noise through a Kalman filter; the filtered data is used as the input for harmonic analysis and spatial correlation analysis and is transmitted to the harmonic analysis module.

[0055] Based on the sensor data collected, the harmonic analysis module first conducts harmonic analysis to identify the energy distribution and attenuation characteristics of the frequency components in the feedback loop, calculates the seismic frequency harmonic control index for quantifying the resonance oscillations that may be caused by the sensor feedback loop during the forest fire monitoring process. Meanwhile, through spatial correlation analysis, combined with the spatial distribution characteristics of multi-sensor data in the forest fire environment, it evaluates the harmonic consistency between the data and calculates the global harmonic complex index for measuring the spatial harmony degree in the sensor network. Then it transmits the seismic frequency harmonic control index and the global harmonic complex index to the stability evaluation module.

[0056] After obtaining the seismic frequency harmonic control index and the global harmonic complex index, the stability evaluation module evaluates the stability of the current forest fire monitoring system according to their real-time values. When both the seismic frequency harmonic control index and the global harmonic complex index exceed the set high threshold simultaneously, it switches to the harmonic attenuation global harmony mode. When the two indexes are within the intervals between their respective set high thresholds and low thresholds, it switches to the complex seismic harmonic control mode, and respectively transmits the switching instructions to the harmonic attenuation global harmony module or the complex seismic harmonic control module.

[0057] After switching to the harmonic attenuation global harmony mode, the harmonic attenuation control module dynamically adjusts the calibrated frequency components in the feedback loop, applies phase inversion processing using an adaptive filter to suppress oscillations. Meanwhile, it optimizes the spatial data harmonic algorithm, identifies the harmonic failure regions based on the spatial harmony deviation degree, and adopts local regression and smoothing processing, combined with terrain weighting adjustment, to achieve data consistency.

[0058] After switching to the complex seismic harmonic control mode, the complex seismic control module dynamically adjusts the intermediate frequency harmonic frequencies with high seismic frequency coupling degree in the feedback loop, and adjusts the frequency response using a band-pass filter. It optimizes the spatial data harmony degree, identifies the regions with poor data harmony degree, and applies the local adaptive harmonic algorithm to adjust the data fusion weights.

[0059] The early warning and monitoring of forest fires are extremely challenging, especially in mountainous or forest areas with complex terrain and dense vegetation. The fire monitoring system needs to process multi-sensor data from different locations in real time. These sensors include temperature, humidity, smoke, image recognition, etc., and are distributed in high-fire-risk areas. Due to the wide coverage of the sensor network and the changing environmental factors, the monitoring system must have a highly sensitive and stable response ability, and be able to adjust in real time to cope with the risks that may cause false alarms, missed alarms or oscillation effects.

[0060] In mountainous areas, due to the differences in altitude, the terrain undulations between valleys and peaks, and the dense distribution of vegetation, the data reconciliation among sensor nodes faces great challenges. For example, some sensors may be located in valleys where the propagation speeds of smoke and temperature changes are relatively slow, while some other sensors are located on peaks where data changes may be faster and more frequent. In such cases, the monitoring system often generates errors due to spatial data reconciliation mismatches, affecting the accurate capture of fire signals. Therefore, the fire monitoring system not only needs to control oscillations through high-precision frequency analysis but also must address the issue of sensor data reconciliation caused by complex terrain and environmental changes.

[0061] For example, in a specific area, when the wind speed suddenly increases, the propagation rate and range of fire signals increase significantly. Sensors at different heights will receive different temperature, humidity, and smoke signals, and this difference will cause changes in the harmonic frequencies in the sensor feedback loop. Under the cumulative effect of frequencies, the system may generate resonant oscillations, further exacerbating the monitoring instability. If these harmonic frequencies are not effectively controlled, the monitoring system may not only miss the initial fire signals but also cause false alarms or errors due to the oscillation effect.

[0062] Meanwhile, the differences in vegetation coverage have a direct impact on the data reconciliation in the sensor network. Some sensors are arranged in dense forest areas, and the smoke or temperature change signals received by the sensors may be weakened or even distorted due to the obstruction of vegetation; while some other sensors are arranged in relatively open areas where the signal propagation is relatively fast and there are no obvious obstacles. As a result, problems occur in the data reconciliation among different nodes in the sensor network, and the sensor data at different positions are out of tune with each other, making it difficult for the system to accurately judge the fire signal as a whole. This mismatch of spatial data further exacerbates the monitoring error, posing a severe challenge to the accuracy of fire signal capture.

[0063] Therefore, in this scenario, the multi-sensor joint monitoring system proposed by the present invention effectively addresses the problems in complex forest fire monitoring through innovative frequency adjustment and spatial data reconciliation algorithms.

[0064] The operation process of the data acquisition module includes the following:

[0065] S1.1, Multiple sensors deployed in the forest environment, including temperature sensors, humidity sensors, smoke sensors, and cameras for image recognition, continuously and real-time collect environmental data. Each sensor is arranged in fire-prone areas to ensure coverage of key positions where fires may occur. The data of the sensors is transmitted to the central processing unit through wired or wireless networks to ensure the timeliness and integrity of data transmission.

[0066] S1.2, After receiving the sensor data, first perform time synchronization on the data to ensure the consistency of the collected data from different sensors on the same time axis. The time synchronization is achieved through calibration signals and timestamp technology to ensure the effective fusion of data in subsequent processing.

[0067] S1.3, For the synchronized sensor data, use a Kalman filter for noise filtering. The Kalman filter is a recursive algorithm that can dynamically estimate data during real-time processing and effectively remove high-frequency noise and instantaneous interference in the measurement data. Specifically, the Kalman filter compares the current measurement value of the sensor with the prior estimate value, calculates a new estimate value, and adjusts the gain of the estimation process according to the covariance of the current and historical data to minimize the impact of noise on the data.

[0068] The operation process of the harmonic analysis module includes the following:

[0069] S2.1, Based on the sensor data collected in real-time and processed by the Kalman filter in step S1, first perform a frequency-domain conversion of the data. Use the Fast Fourier Transform (FFT) to convert the time-domain data into the frequency domain to obtain the amplitude and phase information of each frequency component. Especially for the data of temperature and smoke sensors, these data are easily affected by environmental fluctuations and terrain in the forest fire scenario, showing complex frequency characteristics. The frequency-domain data is used to identify the resonant frequencies and their energy distributions that may exist in the feedback loop, with a focus on whether there is a resonance effect for high-frequency perturbations in the early stage of the fire.

[0070] S2.2, Based on the frequency-domain analysis results, identify the main frequency components in the feedback loop, especially those with significant energy peaks, and record their harmonic distribution. By calculating the energy decay rate of these frequency components, evaluate whether the harmonic energy gradually decays during the feedback process, and thus calculate the Tremor Frequency Harmonic Resistance Index (THAI). This index quantifies the oscillation risk that may be caused by the resonance effect between sensors during the fire monitoring process by measuring the energy ratio and decay characteristics of high-frequency components in the feedback loop. The calculation of the Tremor Frequency Harmonic Resistance Index is mainly based on the unique harmonic characteristics of the forest environment, reflecting the feedback resonance characteristics under specific terrain and vegetation conditions. This calculation method is specifically applicable to such complex natural environments and avoids the general harmonic analysis method.

[0071] The process of obtaining the Tremor Frequency Harmonic Resistance Index is as follows:

[0072] S2.2.1. Perform frequency domain conversion on the sensor data filtered in step S1 using wavelet transform (Wavelet Transform) instead of the conventional fast Fourier transform (FFT). Wavelet transform can not only provide frequency information but also retain time localization features, which helps capture instantaneous frequency changes in complex forest fire scenarios. The converted frequency domain data is W(f, t), where f represents frequency and t represents time. By analyzing the wavelet transform coefficients, identify the main harmonic frequency components f h , where h represents different harmonic frequency components.

[0073] Calculate the energy distribution of the identified main harmonic frequencies, taking into account the interaction between harmonics. For each main harmonic frequency, first calculate its energy E(f h ), and the formula is: E(f h ) = ∫ T |W(f h , t)| 2 dt; here, T is the time window of the wavelet transform, and W(f h , t) is the coefficient of frequency f h at time t. Further, introduce the harmonic interaction coefficient β h to evaluate the coupling effect between harmonics, and the calculation method is: where, W(f k , t) is the wavelet coefficient of other frequencies f k , and β h measures the energy interaction between harmonics, which is particularly important in multi-sensor feedback because energy transfer between different frequencies may cause resonance.

[0074] S2.2.2. To quantify the resonance effect of harmonic frequencies in the feedback loop, calculate the seismic frequency coupling degree ξ h , and its formula is: ξ h = E(f h )·(1 + β h );

[0075] This formula combines the energy of a single main harmonic frequency with its harmonic interaction coefficient, reflecting the coupling strength of this frequency in the feedback loop. The higher the seismic frequency coupling degree, the greater the resonance risk of this frequency. Especially in forest fires, the complex environmental characteristics may lead to strong coupling between high-frequency harmonic frequencies, increasing the oscillation risk of the system.

[0076] S2.2.3. Finally, integrate the seismic frequency coupling degrees of each main harmonic frequency to calculate the overall seismic frequency harmonic resistance index THAI, and its formula is: Here, Denotes the rate of change of frequency, used to capture the dynamic change of frequency over time, f min and f max are the lowest and highest harmonic frequencies respectively. Through the integration operation, the coupling degree within the entire frequency range is accumulated to obtain the overall resonance oscillation intensity in the feedback loop. This index is specifically designed for fire monitoring in complex environments, taking into account the interaction between harmonics and the dynamic change of frequency, and is particularly suitable for forest fire scenarios with complex terrain.

[0077] S2.3. To evaluate the data consistency of multi-sensors in the forest fire environment, a spatial correlation analysis method is adopted to analyze the data correlation between each sensor node. Combining the layout of sensors in the forest, especially in terrain areas with significant height differences, by calculating the spatial correlation coefficient between the data of each sensor, the anomalies or inconsistencies existing in the spatial distribution are identified. This analysis focuses on identifying the non-linear distribution characteristics of data under complex terrain and capturing the spatial harmonic mismatch problems caused by terrain influence in the sensor network.

[0078] S2.4. Based on the results of the spatial correlation analysis, the Spatial Resonance and Mismatch Index (SRMI) is calculated. This index quantifies the spatial harmony in the sensor network by evaluating the data correlation and harmonic consistency between each sensor node. The calculation of the Spatial Resonance and Mismatch Index takes into account the special terrain and vegetation coverage in the forest fire scenario, especially weighting the spatial data inconsistencies caused by terrain undulations. By quantifying the spatial harmony, the Spatial Resonance and Mismatch Index can reveal the consistency level of multi-sensor data in the complex forest environment and ensure the spatial accuracy of fire monitoring data.

[0079] The process of obtaining the Spatial Resonance and Mismatch Index is as follows:

[0080] S2.4.1. In the forest fire monitoring scenario, the spatial distribution of the sensor network is complex, with significant differences in terrain undulation and vegetation density. First, the pre-processed multi-sensor data is mapped into a two-dimensional space to form a spatial data matrix D(x,y), where x and y are the coordinates of the sensors in the geographical space, and the spatial data matrix is the sensor measurement values (such as temperature, smoke concentration, etc.) at the corresponding positions. This mapping method can capture the spatial distribution characteristics of data in complex terrain areas.

[0081] S2.4.2. Based on the spatial data matrix, calculate the data gradient around each sensor node to form a spatial gradient vector field G(x,y). The calculation method of the gradient vector is: where and They represent the rates of change of data in the x - direction and y - direction respectively. This vector field describes the spatial variation trend of data, which can reveal the data differences between different regions in the forest. Such differences are particularly obvious in areas with complex terrain or dense vegetation.

[0082] S2.4.3. To evaluate the spatial harmony between different sensor nodes, a spatial interaction potential field Φ(x,y) is constructed to quantify the spatial coupling effect in the sensor network. The calculation formula is as follows: Among them, (x′,y′) represents the coordinates of adjacent sensor nodes, G(x,y)·G(x′,y′) represents the dot product of the gradient vectors between two nodes, and r(x,y,x′,y′) is the Euclidean distance between the two nodes. This potential field reflects the intensity of the interaction of sensor data in space. The closer the distance and the more consistent the gradient directions between nodes, the stronger the influence on each other. Especially in the forest fire scenario, this interaction will become more complex due to the obstruction or guidance of the terrain.

[0083] S2.4.4. To further quantify the degree of deviation from spatial harmony, a spatial harmony deviation δ(x,y) is introduced, and its calculation formula is as follows: Among them, G avg (x,y) is the average gradient vector of all adjacent nodes around the node (x,y), and ||G avg (x,y)|| is the modulus of this average gradient vector. This formula measures the degree of deviation of the gradient of a single sensor node from the gradient directions of its surrounding nodes. The larger the deviation, the more inconsistent the data of this node with its surrounding environment, which may indicate abnormal signs of fire.

[0084] S2.4.5. Finally, by combining the spatial interaction potential field and the spatial harmony deviation, the overall Spatial Harmony and Complexity Index (SRMI) is calculated. The formula is as follows: SRMI = ∫ A Φ(x,y)·δ(x,y)dxdy; where A represents the geographical area covered by the sensor network. By integrating the potential field and deviation degree over the entire area, the Spatial Harmony and Complexity Index quantifies the spatial harmony of the entire sensor network. This index is particularly suitable for evaluating the impact of complex terrain on the consistency of sensor data in forest fire monitoring, and can effectively identify data inconsistencies caused by terrain obstruction, uneven sensor deployment, or environmental complexity, thus providing an accurate assessment of spatial harmony during the monitoring process.

[0085] The operation process of the stability evaluation module includes the following content:

[0086] S3.1, Based on the shock frequency harmonic control index and the ring modulation complex mountain index calculated in step S2, monitor the changes of these two indicators in real time. The shock frequency harmonic control index is used to measure the oscillation risk that may be caused by the resonance effect in the sensor feedback loop, while the ring modulation complex mountain index is used to evaluate the spatial harmony in the multi-sensor network. In the complex forest fire monitoring scenario, these two indicators jointly reflect the stability and monitoring accuracy of the system, and are the core basis for the system to perform dynamic adjustment and decision-making.

[0087] S3.2, Combining the characteristics of the forest fire monitoring scenario, preset the high thresholds and medium-range intervals of the shock frequency harmonic control index and the ring modulation complex mountain index. The high threshold indicates that the system is in a high-risk state, and there may be significant oscillations and spatial harmony failures; the medium-range interval indicates that the system is in a relatively stable state, but fine-tuning is still required to ensure the accuracy of monitoring. The setting of the thresholds is based on the comprehensive analysis of historical data and environmental characteristics, with particular attention to the impact of complex terrain and variable climate on the monitoring system.

[0088] S3.3, When the shock frequency harmonic control index and the ring modulation complex mountain index monitored in real time simultaneously exceed their respective set high thresholds, the system immediately identifies that it is in a high-risk state. This state is usually manifested in the early stage of the fire, due to factors such as sudden changes in wind speed and terrain obstruction, the resonance oscillation in the sensor feedback loop intensifies, and the spatial harmony failure occurs in the sensor network. In this case, the system immediately switches to the harmonic decay ring modulation mode (HDMM), and by adjusting the feedback frequency and the spatial data harmonic algorithm, suppresses the oscillation caused by high-energy harmonics and enhances the consistency of the spatial data to ensure that the fire signal can still be accurately captured in a complex environment.

[0089] The harmonic decay ring modulation mode refers to a response mode that the fire monitoring system switches to when the shock frequency harmonic control index (THAI) and the ring modulation complex mountain index (SRMI) simultaneously exceed the high threshold. The purpose of this mode is to suppress the oscillation caused by high-energy harmonics by adjusting the frequency parameters in the feedback loop and enhance the harmony of the spatial data, so as to reduce the monitoring error caused by the resonance effect and spatial harmony failure, and ensure the accurate capture and stable monitoring of the fire signal in a complex forest environment.

[0090] S3.4, When the shock frequency harmonic control index and the ring modulation complex mountain index monitored in real time are in the interval between their respective set high thresholds and low thresholds, the system identifies that it is in a medium-risk state. This state usually occurs in the initial stage of the fire development or when the environmental changes are not significant. Although there are certain fluctuations in the resonance and spatial harmony in the feedback loop, they have not reached the out-of-control level. At this time, the system switches to the resonance recovery harmonic control mode (RHAM), and by dynamically adjusting the feedback frequency and optimizing the spatial harmony degree, ensures that the monitoring system maintains high sensitivity and accuracy in fire detection while maintaining stability.

[0091] The complex earthquake adjustment mode is a mode switched by the fire monitoring system when the earthquake frequency harmonic adjustment index (THAI) and the global adjustment complex mountain index (SRMI) are in the medium range. This mode aims to maintain the stability of the system and the high sensitivity of monitoring by dynamically adjusting the feedback frequency and optimizing the spatial harmony of multi-sensor data, ensuring that in the medium-risk state, the accuracy of fire detection and the response ability of the system are effectively regulated, and preventing false alarms or missed alarms caused by environmental changes during the monitoring process.

[0092] S3.5, after the mode switch, the system immediately executes the corresponding adjustment strategy. For the harmonic decay global adjustment mode (HDMM), the focus is on adjusting the feedback frequency, suppressing the harmonic energy inside the system, preventing the further intensification of oscillations, and performing harmonic processing on the spatial data to reduce data inconsistency. For the complex earthquake adjustment mode (RHAM), the system is mainly based on dynamic adjustment, focusing on ensuring a smooth transition of the system in a changing environment, continuously monitoring the data changes of sensor feedback to prevent monitoring errors caused by environmental changes.

[0093] S3.6, after the mode switch, continue to monitor the changes of the earthquake frequency harmonic adjustment index and the global adjustment complex mountain index in real time, and adjust the maintenance time of the mode or switch back to the normal monitoring mode according to the actual situation. If the system operates stably for a period of time and the indicators return to the normal level, consider switching back to the normal monitoring mode to restore the low-power and efficient operation state of the system. If the indicators continue to be in a high-risk state, stay in the harmonic decay global adjustment mode and make further adjustments and optimizations to cope with the challenges in the complex fire environment.

[0094] Through the frequency-domain conversion and spatial harmony analysis of multi-sensor data, the present invention first identifies and calculates the harmonic frequency components, their energy distribution, and the interaction strength in the feedback loop, and then calculates the earthquake frequency harmonic adjustment index through the comprehensive analysis of the earthquake frequency coupling degree, which is used to quantify the oscillation risk that may be caused by the resonance effect in the feedback loop. At the same time, a spatial gradient vector field is constructed and the spatial interaction potential field between sensor nodes is analyzed. Combining the spatial harmony deviation degree, the global adjustment complex mountain index is calculated to evaluate the spatial harmony and data consistency in the multi-sensor network. Based on the real-time values of these two indicators, dynamic stability evaluation is carried out, and the harmonic decay global adjustment mode and the complex earthquake adjustment mode are switched respectively in the high-risk and medium-risk states. By adjusting the feedback frequency, suppressing the oscillations caused by high-energy harmonics, and enhancing the spatial data harmony, the fire monitoring accuracy and stability of the system in the complex forest environment are effectively improved, ensuring the accurate capture and timely response of fire signals, avoiding monitoring errors caused by harmonic resonance and spatial data imbalance, and thus enhancing the reliability of fire detection and the accuracy of system response.

[0095] The operation process of the harmonic decay control module includes the following:

[0096] S4.1, after switching to the harmonic attenuation modulation mode, the frequency parameters in the feedback loop need to be dynamically adjusted to suppress the oscillation caused by high-energy harmonics. In a complex forest fire environment, due to differences in terrain undulation and vegetation density, sensors in different areas may produce different harmonic frequencies and intensities. First, based on the seismic harmonic control index (THAI) calculated in step S2, identify the main harmonic frequency components in the feedback loop that cause system oscillations. Specifically, the seismic harmonic control index analyzes the harmonic frequencies f in the multi-sensor feedback data. h The energy E(f h ) and the interaction coefficient β h Calculate the frequency coupling degree ξ of each harmonic frequency h , the formula is: h =E(f h )·(1+β h ); among them, the frequency component with the highest degree of seismic frequency coupling, that is, the specific frequency component that produces strong resonance in the system, is marked as the calibration frequency component. Usually, these calibration frequency components are concentrated on the harmonic frequencies with higher energy and strong interaction in the feedback loop. In fire monitoring, these frequencies may cause the resonance effect of the calibration frequency components to be amplified due to the complexity of the environment, such as wind speed changes, terrain undulations, etc., thereby causing system instability. After identifying these high-coupling frequency components, these calibration frequency components are attenuated in real time through adaptive filters. The specific method is to apply phase inversion technology to the calibration frequency components, adjust the phase of the feedback signal, and reduce the gain of the frequency in the system, thereby suppressing the further aggravation of the oscillation and ensuring the signal stability during the fire monitoring process.

[0097] S4.2, in forest fire monitoring, due to complex terrain and diverse vegetation coverage, the data of each node in the sensor network may be significantly inconsistent in the spatial dimension. In order to ensure the data consistency of multiple sensors in the spatial dimension, a multi-level spatial data reconciliation algorithm is used for processing. The algorithm first analyzes the spatial reconciliation between different sensor nodes based on the SRMI in step S3 to identify the reconciliation failure area. For the reconciliation failure area, a data reconciliation method based on local regression (Local Regression) is applied to smooth the data in the misaligned area using the reconciliation data of adjacent nodes to reduce the error caused by data inconsistency. At the same time, the terrain weighted reconciliation technology is used to consider the attenuation and deformation of the sensor signal by the terrain factor, and the sensor data in the complex terrain area is weighted and adjusted to make it consistent with the overall network data.

[0098] The specific process for identifying the harmonic failure area is as follows:

[0099] After calculating the spatial harmonic deviation degrees of all sensor nodes, a global analysis of the deviation degrees of the entire sensor network is carried out. Specifically, when the spatial harmonic deviation degree of a certain node is greater than the harmonic deviation degree threshold, and the spatial harmonic deviation degrees of its surrounding adjacent nodes (for example, all nodes within a certain range of distance) exceed 20% of the harmonic deviation degree threshold, it is considered that there may be a harmonic failure in the area where the node is located, and it is identified as a harmonic failure area.

[0100] S4.3, After completing the frequency adjustment and spatial data harmonization, the adjusted feedback signal and spatial data are verified in real time to ensure that the adjusted data has high stability and consistency. Specifically, by monitoring the changes in the adjusted seismic frequency harmonic control index and the global adjustment complex index, it is verified whether the frequency adjustment effectively suppresses the influence of high-energy harmonics, and whether the spatial data harmonization algorithm successfully eliminates the data inconsistency caused by terrain complexity. These verification results are continuously recorded and analyzed to ensure that the system can maintain the best response state during the entire fire monitoring process and reduce false alarms or missed alarms caused by environmental changes.

[0101] S4.4, To cope with the possible environmental changes during the forest fire monitoring process, the system continuously performs dynamic adaptive adjustment in the harmonic attenuation and global adjustment mode. According to the data feedback from real-time monitoring, the parameters of the adaptive filter are automatically adjusted, and the weight distribution of the spatial data harmonization algorithm is adjusted in real time according to the terrain changes. Through this mechanism of continuous monitoring and dynamic adjustment, it is ensured that the system can cope with various uncertain factors in the forest fire environment and maintain the monitoring accuracy and stability.

[0102] After the present invention switches to the complex seismic adjustment mode, the intermediate frequency harmonic frequency with high seismic frequency coupling degree in the feedback loop is dynamically adjusted, and the frequency response is adjusted using a band-pass filter. The spatial data harmonization degree is optimized, the area with poor data harmonization degree is identified, and the local adaptive harmonization algorithm is applied to adjust the data fusion weight. The seismic frequency harmonic control index and the global adjustment complex index are evaluated in real time, the feedback control strategy is continuously optimized, the multi-level information fusion is strengthened, and the monitoring verification is carried out in combination with the geographic information system.

[0103] The operation process of the complex seismic adjustment module includes the following contents:

[0104] S5.1. After switching to the complex oscillation adjustment mode, first dynamically adjust the frequency of the feedback loop to control and suppress moderate oscillations. In the forest fire monitoring scenario, due to the influence of wind speed changes, temperature fluctuations, and terrain undulations, harmonic oscillations with moderate frequencies may occur in the feedback loop. Based on the frequency components identified in step S4 where the oscillation-frequency coupling degree is higher than 70% but lower than the resonance threshold itself, they are identified as frequencies with moderate oscillation risks, and the frequency response characteristics of the feedback signal are adjusted. Specifically, the intermediate-frequency harmonics are dynamically regulated through a band-pass filter (Band-Pass Filter) to separate its frequency response range from the resonance frequency of the system, reducing the energy accumulation of these frequencies in the feedback loop, thereby suppressing the further development of moderate oscillations. The bandwidth and center frequency of the filter are adaptively adjusted according to real-time environmental data to ensure that the oscillations are effectively controlled while maintaining the overall stability of the system.

[0105] S5.2. In the complex oscillation adjustment mode, optimize the spatial harmonization degree of multi-sensor data to reduce the data harmonization mismatch problems caused by terrain complexity or vegetation density changes. Based on the calculated Spherical-Regional Multi-Index (SRMI) in step S4 and its real-time monitoring results, identify areas with poor data harmonization, especially in terrains with high and low drops or areas covered by dense vegetation in the forest. Adopt a local adaptive harmonization algorithm (LocalAdaptiveHarmonizationAlgorithm) to gradually adjust the data fusion weights of adjacent sensor nodes for these areas with poor harmonization. Specifically, by introducing terrain factors and vegetation density factors, the data of sensor nodes are weighted, enabling the sensor data in complex terrains or vegetation-dense areas to be better harmonized with the data of other nodes, reducing the harmonization mismatch phenomenon. This optimization process ensures the coordination and consistency of spatial data among sensor nodes. Especially during a fire, each sensor can provide a consistent and accurate spatial data distribution, improving the accuracy of fire monitoring.

[0106] The process of identifying areas with poor data harmonization is specifically as follows:

[0107] First, calculate the Spherical-Regional Multi-Index value of each node and compare it with the data harmonization threshold. If the Spherical-Regional Multi-Index value of a certain node is lower than the data harmonization threshold, and the Spherical-Regional Multi-Index values of its surrounding adjacent sensor nodes are all lower than 80% of the data harmonization threshold, then this area is identified as an area with poor data harmonization.

[0108] S5.3. After completing the feedback frequency adjustment and spatial harmonicity optimization, conduct an assessment of real-time data feedback and monitoring accuracy. By monitoring the dynamic changes of the tremor frequency harmonic control index (THAI) and the harmonic adjustment and complex landscape index (SRMI), evaluate the stability and monitoring accuracy of the system in the complex earthquake adjustment and control mode. Pay particular attention to whether the monitored tremor frequency harmonic control index and harmonic adjustment and complex landscape index after system processing show the expected oscillation suppression and harmonicity improvement. Through continuous monitoring of these indicators, ensure that the system can promptly capture any signals of fire occurrence and avoid false alarms or missed alarms caused by oscillations or harmonic mismatches.

[0109] S5.4. According to the real-time monitoring results, continuously optimize the feedback control strategy to adapt to the dynamic changes of the environment. By analyzing the data after feedback frequency adjustment and spatial harmonicity optimization, adjust the parameters of the band-pass filter and the weighting factors of the local adaptive harmonic algorithm to ensure that the system can cope with the changing forest fire monitoring environment. This adaptive optimization process can adjust the response mode of the system according to the changes in the environment. Especially when meteorological factors such as wind speed, humidity, and temperature change suddenly, the system can quickly adjust the monitoring strategy and maintain high sensitivity to fire signals.

[0110] S5.5. In the complex earthquake adjustment and control mode, further strengthen multi-level information fusion, comprehensively analyze the data from multiple sensors such as temperature, humidity, smoke, and image recognition to ensure the consistency and complementarity among various types of data. Combine the terrain and vegetation data in the geographic information system (GIS) to verify the monitoring results and ensure that the final fire signal detection results have high accuracy and stability. Especially in the early stage of fire occurrence, the system can timely detect the early signs of fire through multi-level information fusion and avoid monitoring errors caused by the failure of a single data source.

[0111] By switching to the complex earthquake adjustment and control mode, the present invention can dynamically adjust the feedback frequency, suppress medium-level oscillations, and maintain the stability of the feedback loop. At the same time, it optimizes the spatial harmonicity of multi-sensor data, effectively reduces the problem of data harmonic mismatch caused by terrain complexity or vegetation density changes, and ensures the coordination of data among sensor nodes. Through real-time data feedback and continuously optimized feedback control strategies, the system can quickly adapt to environmental changes. Especially in the early stage of fire, through multi-level information fusion and geographic information system verification, the detection accuracy of fire signals and the overall reliability of monitoring are improved.

[0112] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0113] Only some exemplary embodiments of the present invention have been described by way of illustration. Undoubtedly, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0114] It should be noted that in this text, if there are relational terms such as first and second, they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0115] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A multi-sensor linkage fire monitoring system based on image recognition, characterized in that: include: Data acquisition module, harmonic analysis module, stability assessment module, harmonic attenuation control module and re-seismic control module; The data acquisition module collects data from multiple sensors in real time and filters noise through the Kalman filter; the filtered data is used as the input of harmonic analysis and spatial correlation analysis and is transmitted to the harmonic analysis module; Based on the collected sensor data, the harmonic analysis module first performs harmonic analysis to identify the energy distribution and attenuation characteristics of the frequency components in the feedback loop, and calculates the frequency harmonic index, which is used to quantify the resonant oscillation that may be caused by the sensor feedback loop during the fire monitoring process; at the same time, through spatial correlation analysis, combined with the spatial distribution characteristics of multi-sensor data in the forest fire environment, the harmonic consistency between the data is evaluated, and the harmonic complex index is calculated to measure the spatial harmony in the sensor network; the frequency harmonic index and the harmonic complex index are transmitted to the stability assessment module; After obtaining the seismic frequency harmonic control index and the attenuation modulation complex index, the stability evaluation module evaluates the stability of the current fire monitoring system according to their real-time values; when the seismic frequency harmonic control index and the attenuation modulation complex index both exceed the set high threshold, it switches to the harmonic attenuation modulation mode; when the two indicators are between the respective set high threshold and low threshold, it switches to the complex seismic modulation control mode; and transmits the switching command to the complex attenuation modulation module or the complex seismic modulation module respectively; After the harmonic attenuation control module switches to the harmonic attenuation modulation mode, it dynamically adjusts the calibration frequency component in the feedback loop, and uses an adaptive filter to apply phase reversal processing to suppress oscillation. At the same time, it optimizes the spatial data reconciliation algorithm, identifies the reconciliation failure area based on the spatial reconciliation deviation, and uses local regression and smoothing processing combined with terrain weighted adjustment to achieve data consistency. After the re-seismic control module switches to the re-seismic control mode, it dynamically adjusts the intermediate frequency harmonic frequency with high seismic frequency coupling in the feedback loop, and uses a bandpass filter to adjust the frequency response; optimizes the spatial data harmony, identifies areas with poor data harmony, and applies a local adaptive harmony algorithm to adjust the data fusion weight.

2. The multi-sensor linkage fire monitoring system based on image recognition according to claim 1 is characterized in that: The operation process of the harmonic analysis module includes the following: S2.1, based on the sensor data collected in real time and processed by Kalman filtering in step S1, first perform frequency domain conversion on the data; use fast Fourier transform to convert the time domain data into the frequency domain to obtain the amplitude and phase information of each frequency component; S2.2, based on the frequency domain analysis results, identify the main frequency components in the feedback loop and record their harmonic distribution; by calculating the energy attenuation rate of these frequency components, evaluate whether the harmonic energy gradually decays during the feedback process, and thus calculate the frequency harmonic control index; The process of obtaining the frequency harmonic control index is as follows: S2.2.1, the sensor data after filtering in step S1 is converted to the frequency domain using wavelet transform instead of conventional fast Fourier transform; the converted frequency domain data is W(f,t), where f represents frequency and t represents time; by analyzing the wavelet transform coefficients, the main harmonic frequency component f with significant energy in the feedback loop is identified h , h represents different harmonic frequency components; Calculate the energy distribution of the identified main harmonic frequencies, taking into account the interaction between harmonics; for each main harmonic frequency, first calculate its energy E(f h ), the formula is: E(f h )=∫ T |W(f h ,t)| 2 dt; here, T is the time window of wavelet transform, W(f h ,t) is the frequency f h coefficient at time t; using the harmonic interaction coefficient β h , which is used to evaluate the coupling effect between harmonics, is calculated as: Among them, W(f k ,t) is other frequency f k The wavelet coefficients, β h The energy interactions between harmonics are measured; S2.2.2, to quantify the resonance effect of the harmonic frequencies in the feedback loop, calculate the frequency coupling ξ h , the formula is: h =E(f h )·(1+β h ); S2.2.3, finally, the seismic frequency coupling of each main harmonic frequency is integrated to calculate the overall seismic frequency harmonic index THAI, the formula of which is: Here, Represents the frequency change rate, which is used to capture the dynamic change of frequency over time, f min and f max are the lowest and highest harmonic frequencies respectively.

3. The multi-sensor linkage fire monitoring system based on image recognition according to claim 2 is characterized in that: S2.3, in order to evaluate the data consistency of multiple sensors in forest fire environment, the spatial correlation analysis method is used to analyze the data correlation between each sensor node; S2.4, based on the results of spatial correlation analysis, calculate the atlantic complex index; The process of obtaining the global adjustment complex index is as follows: S2.4.1, mapping the preliminarily processed multi-sensor data into a two-dimensional space to form a spatial data matrix D(x, y), where x and y are the coordinates of the sensors in geographic space, and the spatial data matrix is ​​the sensor measurements at the corresponding locations; S2.4.2, based on the spatial data matrix, calculate the data gradient around each sensor node to form a spatial gradient vector field G(x,y); the gradient vector is calculated as follows: in, and Represent the rate of change of data in the x direction and y direction respectively; S2.4.3, in order to evaluate the spatial compatibility between different sensor nodes, a spatial interaction potential field Φ(x,y) is constructed to quantify the spatial coupling effect in the sensor network; the calculation formula is: Where (x′, y′) represents the coordinates of adjacent sensor nodes, G(x, y) · G(x′, y′) represents the dot product of the gradient vector between two nodes, and r(x, y, x′, y′) is the Euclidean distance between two nodes; S2.4.4, in order to quantify the degree of deviation from spatial harmony, the spatial harmony deviation δ(x,y) is used, and its calculation formula is: Among them, G avg (x, y) is the average gradient vector of all neighboring nodes around node (x, y), ||G avg (x,y)|| is the modulus of the corresponding average gradient vector; S2.4.5, finally, the spatial interaction potential field is combined with the spatial harmonic deviation to calculate the overall harmonic complex index; the formula is as follows: SRMI = ∫ A Φ(x,y)·δ(x,y)dxdy; where A represents the geographical area covered by the sensor network.

4. The multi-sensor linkage fire monitoring system based on image recognition according to claim 3 is characterized in that: The operation process of the stability assessment module includes the following: S3.1, based on the seismic frequency harmonic control index and the atmospheric harmonic complex index calculated in step S2, real-time monitoring of the changes of these two indicators; S3.2, based on the characteristics of forest fire monitoring scenarios, pre-set the high threshold and medium range of the frequency harmonic index and the frequency harmonic index; S3.3, when the real-time monitored seismic frequency harmonic control index and the harmonic complex index exceed their respective set high thresholds at the same time, it is recognized that the current state is in a high-risk state; immediately switch to the harmonic attenuation harmonic mode, and suppress the oscillation caused by high-energy harmonics by adjusting the feedback frequency and spatial data harmonic algorithm; S3.4, when the real-time monitored frequency harmonic control index and spatial harmonic control index are between the high threshold and the low threshold respectively set, it is recognized that the current state is in a medium risk state, indicating that although there is a certain degree of fluctuation in the resonance and spatial harmonicity in the feedback loop, it has not yet reached the level of being out of control; at this time, switch to the vibration control mode; S3.5, after the mode is switched, the corresponding adjustment strategy is immediately executed; S3.6, after the mode is switched, continue to monitor the changes of the seismic frequency harmonic control index and the atmospheric modulation complex index in real time, and adjust the mode maintenance time or switch back to the regular monitoring mode according to the actual situation.

5. The multi-sensor linkage fire monitoring system based on image recognition according to claim 4 is characterized in that: The operation process of the harmonic decay control module includes the following: S4.1, after switching to the harmonic attenuation modulation mode, the frequency parameters in the feedback loop are dynamically adjusted to suppress the oscillation caused by high-energy harmonics; first, based on the seismic frequency harmonic control index calculated in step S2, the main harmonic frequency components that cause system oscillation in the feedback loop are identified; the frequency component with the highest seismic frequency coupling, that is, the specific frequency component that produces strong resonance, is marked as the calibration frequency component, and the calibration frequency component is attenuated in real time through an adaptive filter; the specific method is to apply a phase inversion technique to the calibration frequency component to adjust the phase of the feedback signal; S4.2, in order to ensure the data consistency of multiple sensors in the spatial dimension, a multi-level spatial data reconciliation algorithm is used for processing; first, based on the reconciliation index in step S3, the spatial reconciliation between different sensor nodes is analyzed to identify the reconciliation failure area; for the reconciliation failure area, a data reconciliation method based on local regression is applied, and the reconciliation data of the adjacent nodes are used to smooth the data of the disordered area; at the same time, the terrain weighted reconciliation technology is used to consider the attenuation and deformation of the sensor signal by the terrain factors, and the sensor data in the complex terrain area is weighted and adjusted to make it consistent with the overall network data.

6. The multi-sensor linkage fire monitoring system based on image recognition according to claim 5 is characterized in that: The identification process of the harmonic failure area is as follows: After calculating the spatial harmony deviation of all sensor nodes, a global analysis of the deviation of the entire sensor network is performed; specifically, when the spatial harmony deviation of a node is greater than the harmony deviation threshold, and the spatial harmony deviation of its surrounding adjacent nodes exceeds the harmony deviation threshold by more than 20%, it is considered that the area where the corresponding node is located may have harmony failure, and is identified as a harmony failure area.

7. The multi-sensor linkage fire monitoring system based on image recognition according to claim 4 is characterized in that: The operation process of the re-shock control module includes the following: S5.1, after switching to the re-vibration control mode, first dynamically adjust the frequency of the feedback loop to control and suppress moderate oscillations; based on the frequency component whose frequency coupling degree is higher than 70% of the resonance threshold but lower than the resonance threshold itself identified in step S4, it is identified as a medium oscillation risk frequency, and the frequency response characteristics of the feedback signal are adjusted; specifically, the intermediate frequency harmonics are dynamically adjusted through a bandpass filter so that its frequency response range is separated from the resonant frequency of the system; S5.2, in the re-seismic control mode, optimize the spatial harmony of multi-sensor data; based on the spatial harmony index calculated in step S4 and its real-time monitoring results, identify areas with poor data harmony; use a local adaptive harmony algorithm to gradually adjust the data fusion weights of adjacent sensor nodes for these areas with poor harmony; specifically, perform weighted processing on the data of the sensor nodes by using terrain factors and vegetation density factors.

8. The multi-sensor linkage fire monitoring system based on image recognition according to claim 7 is characterized in that: The specific process of identifying areas with poor data harmony is as follows: First, the attunement complex index value of each node is calculated and compared with the data harmony threshold. If the attunement complex index value of a node is lower than the data harmony threshold, and the attunement complex index values ​​of the surrounding neighboring sensor nodes are all lower than 80% of the data harmony threshold, the corresponding area is identified as an area with poor data harmony.

Images