Explosion source flame detection method and system

Abstract

The embodiment of the invention discloses an explosion source flame detection method and system, and the method comprises the steps: collecting a multi-mode signal of an explosion source flame, and the multi-mode signal comprises an acoustic vibration type signal and an optical type signal; preprocessing the acoustic vibration type signal and the optical type signal to generate an acoustic vibration feature set and an optical feature set; performing topology analysis on the acoustic vibration feature set by using a persistent coherence algorithm to identify a dynamic behavior mode of the explosion source flame, and obtaining a dynamic behavior mode identification result of the explosion source flame; the dynamic behavior pattern recognition result and the optical feature set are converted into a symbol sequence through a symbolic dynamics algorithm, a cross-modal symbol pattern is analyzed through a transition probability matrix, and comprehensive feature representation of the explosion flame is generated; and based on the comprehensive feature representation, judging whether an explosion source flame exists through a classification algorithm, and outputting a detection result. According to the invention, high-precision explosion source flame detection suitable for a complex environment is realized.

Description

Technical Field

[0001] The present application relates to the technical field of flame detection technology, and in particular to a method and system for detecting explosion source flames. Background Art

[0002] Explosion flames are characterized by their transient nature, high energy, and presence of smoke or strong light interference. Traditional flame detection methods (such as infrared and ultraviolet detection, or the image processing-based YOLO algorithm) have limitations in complex environments (e.g., low visibility and high noise). For example, infrared and ultraviolet detection are susceptible to background light interference, while image processing methods fail under smoke obstruction. Furthermore, existing methods often rely on single-modal signals, making it difficult to capture the dynamic behavior of explosion flames, resulting in high false detection rates and slow response times.

[0003] Therefore, there is an urgent need for a high-precision explosion source flame detection method that can integrate multimodal signals and adapt to complex environments. Summary of the Invention

[0004] The present application provides a method for detecting explosion source flames. By staged processing of multimodal signals (acoustic-vibration and optical), combined with persistent coherence algorithms and symbolic dynamics algorithms, the method realizes the recognition and comprehensive feature representation of the dynamic behavior patterns of explosion flames, significantly improving the detection accuracy and environmental adaptability.

[0005] This application provides the following solutions: According to a first aspect, a method for detecting an explosion source flame is provided, the method comprising: collecting multimodal signals of an explosion source flame, the multimodal signals comprising acoustic vibration signals and optical signals; preprocessing the acoustic vibration signals and the optical signals to generate an acoustic vibration feature set and an optical feature set; topologically analyzing the acoustic vibration feature set using a persistent coherence algorithm to identify a dynamic behavior pattern of the explosion source flame, and obtaining a dynamic behavior pattern recognition result of the explosion source flame; converting the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence using a symbolic dynamics algorithm, analyzing the cross-modal symbol pattern through a transition probability matrix, and generating a comprehensive feature representation of the explosion flame; based on the comprehensive feature representation, determining whether an explosion source flame exists through a classification algorithm, and outputting a detection result.

[0006] According to an achievable method in an embodiment of the present application, the acoustic vibration signals and the optical signals are preprocessed to generate an acoustic vibration feature set and an optical feature set, including: using discrete wavelet transform to perform multi-scale decomposition of the acoustic wave signal and the micro-vibration signal, extracting the time-frequency features at each scale, and removing environmental noise interference through adaptive threshold filtering to generate an acoustic vibration feature set; using fast Fourier transform to analyze the frequency distribution of the multi-spectral signal, extracting the spectral line features in the wavelength range of 400 nanometers to 2500 nanometers, and generating an optical feature set.

[0007] According to an achievable method in an embodiment of the present application, the use of a persistent coherence algorithm to perform a topological analysis on the acoustic vibration feature set to identify the dynamic behavior pattern of the explosion source flame includes: mapping the acoustic vibration feature set to a high-dimensional point cloud representation, constructing a persistent coherence barcode, and extracting the topological features of the dynamic behavior pattern of the explosion flame, wherein the topological features include the duration of the 0-dimensional connected component and the 1-dimensional ring structure.

[0008] According to an achievable method in an embodiment of the present application, the use of a symbolic dynamics algorithm to convert the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence, and analyzing the cross-modal symbol pattern through a transition probability matrix to generate a comprehensive feature representation of the explosion flame includes: discretizing the dynamic behavior pattern recognition result and the optical feature set into finite symbol sequences, respectively, and constructing a joint transition probability matrix; calculating the conditional entropy of the cross-modal symbol sequence to generate a comprehensive feature representation of the explosion flame, wherein the conditional entropy is used to quantify the correlation between the two types of signals in the instantaneous dynamics of the explosion flame.

[0009] According to an achievable method in an embodiment of the present application, the dynamic behavior pattern recognition result and the optical feature set are respectively discretized into finite symbol sequences, and a joint transition probability matrix is ​​constructed, including: discretizing the dynamic behavior pattern recognition result and the optical feature set into finite symbol sequences through an adaptive partition coding method, wherein the adaptive partition coding dynamically adjusts the symbol division threshold according to the instantaneous intensity of the explosion flame.

[0010] According to an achievable method in an embodiment of the present application, the determining whether there is an explosion source flame through a classification algorithm includes: using a classifier based on sparse representation, sparsely encoding the comprehensive feature representation, constructing a shared dictionary of explosion flame features, and determining whether there is an explosion source flame through dictionary reconstruction error.

[0011] According to an achievable method in an embodiment of the present application, the determining whether an explosion source flame exists through a classification algorithm includes: employing a classification method based on rough set theory to perform attribute simplification on the comprehensive feature representation, extracting a core feature subset of the explosion flame, and determining whether an explosion source flame exists through a rough set decision rule, wherein the decision rule is dynamically updated to adapt to detection requirements in a high-noise environment.

[0012] According to a second aspect, an explosion source flame detection system is provided, the system comprising: a signal acquisition unit configured to acquire multimodal signals of the explosion source flame, the multimodal signals comprising acoustic vibration signals and optical signals; a feature set generation unit configured to preprocess the acoustic vibration signals and the optical signals to generate an acoustic vibration feature set and an optical feature set; a pattern recognition unit configured to perform topological analysis on the acoustic vibration feature set using a persistent homology algorithm to identify the dynamic behavior pattern of the explosion source flame and obtain a dynamic behavior pattern recognition result of the explosion source flame; a comprehensive feature generation unit configured to convert the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence using a symbolic dynamics algorithm, analyze the cross-modal symbol pattern through a transition probability matrix, and generate a comprehensive feature representation of the explosion flame; a detection result generation unit configured to determine whether an explosion source flame exists based on the comprehensive feature representation through a classification algorithm and output a detection result.

[0013] According to a third aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the steps of any one of the methods according to the first aspect are implemented.

[0014] According to a fourth aspect, an electronic device is provided, comprising: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions, which, when read and executed by the one or more processors, execute the steps of the method described in any one of the above-mentioned first aspects.

[0015] According to the specific embodiments provided in this application, this application discloses the following technical effects: This application achieves high-precision, real-time detection through multimodal signal fusion. The method combines acoustic-vibration signals and optical signals, uses a persistent coherence algorithm to identify the dynamic behavior pattern of the explosion flame, captures its nonlinear topological characteristics, and then uses a symbolic dynamics algorithm to convert the two types of signals into a symbol sequence. Based on the transition probability matrix, the cross-modal association is analyzed to generate a comprehensive feature representation, and finally the presence or absence of the flame is determined by a classification algorithm. This method breaks through the limitations of traditional single-modal detection and significantly improves the robustness in low visibility or high interference environments. The adaptive partition coding and conditional entropy calculation of symbolic dynamics enhance the dynamic adaptability of feature fusion, which is suitable for the instantaneous and complexity of explosion flames.

[0016] Of course, any product implementing the present application does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 A diagram of the system architecture applicable to the embodiments of the present application; Figure 2 A flow chart of a method for detecting explosion source flames provided in an embodiment of the present application; Figure 3 A structural block diagram of an explosion source flame detection system provided in an embodiment of the present application; Figure 4 A schematic block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of this application.

[0020] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "an", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise.

[0021] It should be understood that the term "and / or" as used herein is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0022] The word "if," as used herein, may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0023] In recent years, multimodal signal fusion technology (such as the combination of infrared and visible light) has been applied to flame detection to a certain extent, but the following problems still exist: fusion methods are mostly simple feature splicing or linear weighting, which makes it difficult to handle the nonlinear dynamic characteristics of explosion flames; insufficient utilization of acoustic or vibration signals limits detection capabilities in low-visibility environments; and there is a lack of adaptive processing mechanisms for the instantaneous nature of explosion flames.

[0024] In view of this, the present application provides a new approach. To facilitate understanding of the present application, the system architecture on which the present application is based is first described. Figure 1 An exemplary system architecture to which the embodiments of the present application can be applied is shown. Figure 1 As shown in , the system architecture may include: user equipment and an explosion source flame detection system located on the server side.

[0025] A user can input a multimodal signal through a user device, which then transmits it to a server-side explosion source flame detection system. The explosion source flame detection system can employ the methods provided in the embodiments of this application to obtain detection results. The server-side can then transmit the detection results to the user terminal, which then uses the detection results to perform subsequent operations.

[0026] User devices include, but are not limited to, smart mobile terminals, smart home devices, wearable devices, and personal computers (PCs). Smart mobile devices include mobile phones, tablets, laptops, PDAs (Personal Digital Assistants), and internet-connected cars. Smart home devices include smart TVs and smart refrigerators. Wearable devices include smart watches, smart glasses, virtual reality devices, augmented reality devices, and mixed reality devices.

[0027] The explosion source flame detection system can be set up as an independent server, a server group, or a cloud server. A cloud server, also known as a cloud computing server or cloud host, is a host product in the cloud computing service system. It solves the problems of difficult management and weak service scalability in traditional physical hosts and virtual private servers (VPS). Figure 1 In addition to the shown architecture, the explosion source flame detection system can also be set up on a computer terminal with strong computing capabilities.

[0028] It should be understood that Figure 1 The user equipment and explosion source flame detection system in the figure are only illustrative. According to the implementation requirements, any number of user equipment and explosion source flame detection systems can be provided.

[0029] Figure 2 This is a flow chart of the explosion source flame detection method provided in the embodiment of the present application. Figure 2 As shown in , the method may include the following steps: Step 201: Collect multimodal signals of explosion source flames, where the multimodal signals include acoustic vibration signals and optical signals.

[0030] Step 202: Preprocess the acoustic vibration signal and the optical signal to generate an acoustic vibration feature set and an optical feature set.

[0031] Step 203: performing a topological analysis on the acoustic vibration feature set using a persistent homology algorithm to identify the dynamic behavior pattern of the explosion source flame, and obtaining a dynamic behavior pattern recognition result of the explosion source flame.

[0032] Step 204: using a symbolic dynamics algorithm to convert the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence, analyzing the cross-modal symbol pattern through a transition probability matrix, and generating a comprehensive feature representation of the explosion flame.

[0033] Step 205: Based on the comprehensive feature representation, determine whether there is an explosion source flame through a classification algorithm and output the detection result.

[0034] It can be seen from the above process that the present application achieves high-precision and real-time detection through multimodal signal fusion. The method combines acoustic vibration signals and optical signals, uses the persistent coherence algorithm to identify the dynamic behavior pattern of the explosion flame, captures its nonlinear topological characteristics, and then converts the two types of signals into symbol sequences through the symbolic dynamics algorithm. Based on the transition probability matrix, the cross-modal association is analyzed to generate a comprehensive feature representation, and finally the presence or absence of the flame is determined by the classification algorithm. This method breaks through the limitations of traditional single-modal detection and significantly improves the robustness in low visibility or high interference environments. The adaptive partition coding and conditional entropy calculation of symbolic dynamics enhance the dynamic adaptability of feature fusion, which is suitable for the instantaneous and complexity of explosion flames.

[0035] The following describes in detail each step in the above process and the effects that can be further produced in conjunction with the embodiments. First, the above step 201, namely "collecting multimodal signals of the explosion source flame, the multimodal signals including acoustic vibration signals and optical signals", is described in detail in conjunction with the embodiments.

[0036] Explosion flames are characterized by their transient nature, high energy, and complex environmental interference. A single signal type is insufficient to fully characterize their behavior. Therefore, the present invention employs a multimodal signal acquisition strategy, combining acoustic, vibration, and optical signals to achieve comprehensive detection of explosion flames. This multimodal approach effectively addresses low visibility or high noise environments, ensuring robust and reliable detection.

[0037] Acoustic vibration signals can include specific frequency acoustic signals generated by explosion flames and ground or air micro-vibration signals caused by explosion shock waves. Specific frequency acoustic signals originate from the unique acoustic characteristics generated during the combustion process of explosion flames, such as the shock wave sound formed by the rapid expansion of high-temperature gases or the acoustic pressure fluctuations of the combustion reaction. These acoustic wave signals are typically concentrated in a specific frequency range and are significantly different from conventional flames or other environmental noise (such as wind noise or mechanical vibrations), providing key clues for identifying explosion flames. Micro-vibration signals arise from the disturbance of the explosion shock wave on the surrounding medium, such as ground vibrations or changes in air pressure. These vibration signals are highly sensitive and instantaneous, and can provide supplementary information in scenarios where optical signals are limited (such as smoke obstruction), enhancing detection stability.

[0038] Optical signals can cover multispectral signals in the visible, near-infrared and mid-infrared bands. Explosion flames release a wide range of spectral energy during the combustion process, covering a variety of characteristics from visible light to infrared bands. Visible light signals can capture the morphology and color characteristics of the flame, such as bright white light or orange-red light, which are closely related to the high temperature characteristics of the explosion flame. Near-infrared and mid-infrared signals can reflect the thermal radiation characteristics of the flame, especially in smoke or dust environments. The mid-infrared signal has a strong penetrating ability and can effectively capture the obscured flame spectral characteristics. Through the comprehensive acquisition of multi-spectral signals, the present invention can extract the wavelength-specific characteristics of the explosion flame, thereby improving the detection discrimination.

[0039] Multimodal signal acquisition is achieved by deploying a highly sensitive sensor array. For example, high-frequency acoustic sensors capture acoustic signals, micro-vibration sensors detect ground or air vibrations, and multispectral cameras acquire optical signals. These sensors work together to form a multi-dimensional representation of the explosion flame.

[0040] The above step 202, namely "pre-processing the acoustic vibration signal and the optical signal to generate an acoustic vibration feature set and an optical feature set" is described in detail below in conjunction with an embodiment.

[0041] Preprocessing acoustic vibration and optical signals to generate acoustic vibration feature sets and optical feature sets can extract features that characterize the characteristics of the explosion flame from the original multimodal signals, providing high-quality input for subsequent dynamic behavior pattern recognition and comprehensive analysis. This preprocessing process uses specific signal processing techniques to transform complex acoustic vibration and optical signals into structured feature sets that adapt to the transient, nonlinear, and complex environmental interference characteristics of the explosion flame. The preprocessing step can extract key features and remove environmental noise interference by applying time-frequency analysis and spectral analysis techniques to the acoustic vibration and optical signals, respectively, to generate acoustic vibration and optical feature sets.

[0042] As an implementable method, acoustic vibration signals and optical signals are preprocessed to generate acoustic vibration feature sets and optical feature sets, including: using discrete wavelet transform method to perform multi-scale decomposition of acoustic wave signals and micro-vibration signals, extracting time-frequency features at each scale, and removing environmental noise interference through adaptive threshold filtering to generate acoustic vibration feature sets; using fast Fourier transform to analyze the frequency distribution of multi-spectral signals, extracting spectral line features in the wavelength range of 400 nanometers to 2500 nanometers, and generating optical feature sets.

[0043] Acoustic vibration signals are preprocessed using the discrete wavelet transform (DWT) method, which performs a multi-scale decomposition of the acoustic and micro-vibration signals to extract time-frequency features at each scale. The acoustic signal of an explosion flame typically contains transient shock waves of specific frequencies, while the micro-vibration signal reflects the ground or air disturbances caused by the explosion shockwave. These signals are non-stationary and highly dynamic. By decomposing the signal into subbands of varying frequencies and time scales, the discrete wavelet transform (DWT) can accurately capture these transient variations, such as the frequency peaks of the acoustic wave or the energy distribution of the vibration. To further improve feature quality, adaptive threshold filtering is introduced into the preprocessing process. This dynamically adjusts the filtering threshold based on the signal's energy level, effectively removing environmental noise interference, such as wind noise or mechanical vibration, and generating a high-quality acoustic vibration feature set. Compared to traditional filtering techniques, this approach is more adaptable to the complex noise environments of explosion scenes.

[0044] For optical signal preprocessing, the Fast Fourier Transform (FFT) method is used to analyze the frequency distribution of multispectral signals in the visible (400-700 nm), near-infrared (700-1100 nm), and mid-infrared (1100-2500 nm) bands. Spectral line features are extracted from the wavelength range of 400 to 2500 nm, which covers typical spectral characteristics of explosive flames, such as the visible light signature of high-temperature combustion and the infrared signature of thermal radiation. Spectral signals of explosive flames contain unique wavelength features, such as the 780 nm or 1600 nm spectral lines produced by high-temperature combustion. These features reflect the thermal radiation and chemical composition of the flame in different wavelength bands. Fast Fourier Transform (FFT) precisely extracts these spectral line features by converting the spectral signals from the time domain to the frequency domain, generating an optical feature set. Compared to traditional spectral analysis methods, this method focuses on the broad spectral characteristics of explosive flames and maintains feature validity despite interference from smoke or strong light.

[0045] The above step 203, i.e., "performing a topological analysis on the acoustic vibration feature set using a persistent coherence algorithm to identify the dynamic behavior pattern of the explosion source flame and obtaining a dynamic behavior pattern recognition result of the explosion source flame" is described in detail below in conjunction with an embodiment.

[0046] The acoustic vibration signals of explosive flames typically consist of transient shock waves and ground or air micro-vibrations. These signals are highly nonlinear and nonstationary, making traditional time-frequency analysis difficult to fully characterize their complex dynamics. The Persistent Homology algorithm, a topological data analysis tool, reveals the topological structure of signals and is well-suited for processing the dynamic characteristics of explosive flames, significantly improving detection discrimination and robustness.

[0047] Specifically, the acoustic vibration feature set consists of the time-frequency characteristics of the acoustic signal and the energy distribution of the microvibration signal. These features reflect the instantaneous shock wave and vibration patterns of the explosion flame. The persistent homology algorithm first maps this feature set into a high-dimensional point cloud representation, forming an abstract geometric structure in which each point corresponds to a data point in the feature set, such as the peak frequency of the acoustic wave or the amplitude of the vibration. By constructing a RIPS complex of the point cloud, the algorithm generates a series of scale-dependent topological structures, recording the persistence of these structures at different scales, generating the so-called persistent homology barcode. The topological features in the barcode include 0-dimensional connected components and 1-dimensional ring structures, which represent the connectivity and periodic pattern of the signal, respectively. For example, the 0-dimensional connected component can reflect the instantaneous propagation range of the explosion shock wave, while the 1-dimensional ring structure may correspond to the periodic fluctuations of the acoustic wave or vibration.

[0048] The duration information of persistent coherence barcodes is used to quantify the dynamic behavior patterns of explosive flames. For example, the acoustic signal of an explosive flame may exhibit brief, high-intensity peaks within a specific frequency range. The persistent coherence algorithm can capture this transient characteristic through the short duration of its 0-dimensional connected component. Microvibration signals may also contain periodic shock wave patterns, characterized by the duration of a 1-dimensional ring structure. These topological features can effectively distinguish the dynamic patterns of explosive flames from those of conventional flames or other interference sources (such as mechanical vibration or ambient noise), generating dynamic behavior pattern recognition results that provide highly discriminative input for subsequent steps.

[0049] The acoustic vibration feature set consists of the time-frequency characteristics of the acoustic wave signal and the energy distribution of the micro-vibration signal, which is represented as a feature vector set , where each is a dimensional feature vector, containing time-frequency features such as frequency peak, energy intensity or vibration amplitude. First, the feature set is normalized to eliminate dimensional differences and enhance the stability of topological analysis. The normalization formula is: (1) in, is the mean of each dimension of the feature set, is the standard deviation. The standardized feature set Construct a high-dimensional point cloud representation, each point Represents a feature vector of the explosion flame acoustic-vibration signal, mapped to Space, forming a point cloud structure.

[0050] Point cloud based , constructing a RIPS complex to capture the topological structure. The RIPS complex is a distance-based simplex complex that is constructed by a scale parameter Definition. For any point pair in the point cloud , calculate the Euclidean distance: (2) when When and Connect to form edges, construct 0-dimensional simplex (points), 1-dimensional simplex (edges) and higher-dimensional simplex. RIPS complex Defined as: (3) By gradually increasing From 0 to the maximum value (such as the maximum distance of the point cloud), a series of RIPS complexes are generated to form a filtering process .

[0051] Filter ,calculate Maintain long-term homology and extract topological features. Persistent homology is achieved through homology groups. Describes the topological structure of the point cloud, where represents the connected component, Represents a ring structure. For each dimensional homology group, calculate its generator with Date of birth and death: Time of birth : Topological features (such as connected components or rings) are appears.

[0052] Time of death :Topological features in The persistent homology barcode consists of the birth-death pairs of all topological features. Composition, expressed as: (4) in, for The 0-dimensional barcode represents the duration of the connected component, reflecting the instantaneous propagation range of the explosion flame shock wave; the 1-dimensional barcode represents the duration of the ring structure, representing the periodic pattern of the sound wave or vibration.

[0053] Extract the dynamic behavior pattern characteristics of the explosion flame from the persistent coherence barcode. For the 0-dimensional connected component, calculate the duration: (5) The 0-dimensional features with a longer duration reflect how the instantaneous shock wave of the explosion flame quickly connects the point cloud and characterizes its propagation range.

[0054] For a 1D ring structure, calculate the duration: (6) The duration of the 1D feature reflects the periodic dynamics of the explosion flame, such as the frequency fluctuation of the acoustic signal or the periodic impact of micro-vibration. The 0D and 1D durations are combined into a topological feature vector: (7) This vector characterizes the dynamic behavior pattern of the explosion flame, such as the instantaneous propagation and periodic vibration characteristics of the shock wave.

[0055] For topological eigenvectors Perform post-processing to filter out data with a duration exceeding the threshold Features to remove noise effects: (8) in, Dynamically set according to the signal strength of the explosion flame, for example based on the sound wave peak or vibration amplitude. The final output of the dynamic behavior pattern recognition result is , used for subsequent symbolic dynamics analysis and comprehensive feature representation.

[0056] The above-mentioned step 204, namely, "using a symbolic dynamics algorithm to convert the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence, analyzing the cross-modal symbol pattern through a transition probability matrix, and generating a comprehensive feature representation of the explosion flame" is described in detail below in conjunction with an embodiment.

[0057] This step uses a symbolic dynamics algorithm to nonlinearly fuse the dynamic behavior pattern recognition results of acoustic vibration signals with the feature set of optical signals to generate a comprehensive feature representation that can fully characterize the characteristics of explosive flames. Explosive flames are characterized by strong instantaneousness, nonlinear dynamics, and high environmental interference. Traditional feature fusion methods such as linear concatenation or weighted averaging struggle to effectively capture their complex patterns. Symbolic dynamics discretizes continuous signals into symbolic sequences and analyzes their dynamic transition patterns, providing a niche and efficient feature fusion method that is well-suited to the instantaneous and complex nature of explosive flames.

[0058] Specifically, the dynamic behavior pattern recognition results are derived from the topological analysis of the acoustic vibration feature set using the persistent homology algorithm. These features include the duration of the 0-dimensional connected components and the 1-dimensional ring structure, which characterize the instantaneous shock wave propagation and periodic vibration patterns of the explosion flame. The optical feature set includes spectral line features in the visible, near-infrared, and mid-infrared bands, reflecting the thermal radiation and chemical composition characteristics of the explosion flame. The symbolic dynamics algorithm first converts these two types of features into a symbol sequence. Through partition coding, the continuous feature vector is discretized into a finite set of symbols. For example, the feature value is divided into several intervals, each corresponding to a symbol such as A, B, and C. This discretization process simplifies high-dimensional features into a discrete symbol sequence, preserving the dynamic pattern of the explosion flame while reducing computational complexity, making it suitable for real-time detection needs.

[0059] After the symbol sequence is generated, the cross-modal symbol pattern is analyzed by constructing a joint transition probability matrix. This joint transition probability matrix records the transition probabilities between the acoustic vibration symbol sequence and the optical symbol sequence, for example, the probability of transitioning from acoustic vibration symbol A to optical symbol B. This matrix captures the dynamic correlation between the two types of signals in time or feature space, reflecting the instantaneous behavior of the explosion flame, such as the synchronization of the shock wave and the spectral peak. Using a Markov model optimized based on information entropy, the conditional entropy of the cross-modal symbol sequence is calculated to quantify the correlation between the acoustic vibration and optical features. As the core component of the comprehensive feature representation, the conditional entropy value can effectively distinguish explosion flames from other flame types or environmental interference, ensuring high discrimination of the feature representation.

[0060] As an implementable method, the dynamic behavior pattern recognition results and the optical feature set are respectively discretized into finite symbol sequences, and constructing a joint transition probability matrix includes: discretizing the dynamic behavior pattern recognition results and the optical feature set into finite symbol sequences through an adaptive partition coding method, wherein the adaptive partition coding dynamically adjusts the symbol division threshold according to the instantaneous intensity of the explosion flame.

[0061] By monitoring the instantaneous intensity of the explosion flame in real time, such as sound pressure peaks or spectral amplitudes, the algorithm dynamically adjusts the symbol partitioning threshold. For example, in high-intensity explosions, where the acoustic signal may exhibit rapidly changing frequency peaks, the algorithm increases the number of symbol intervals to capture these details. In low-visibility environments, such as those caused by smoke obstruction that weakens the optical signal, the algorithm reduces the threshold sensitivity to ensure the stability of the symbol sequence. This adaptability significantly improves the method's adaptability to complex environments such as smoke, strong light, or high noise, outperforming traditional fixed-partition coding methods.

[0062] Preferably, the dynamic behavior pattern recognition result T' and the optical feature set S' are normalized feature vectors, representing topological features such as duration and spectral features such as intensity, respectively. The symbol set is defined as , the number of symbols Instantaneous intensity Calculated by weighting the peak sound pressure and spectral amplitude: (9) in, is the maximum amplitude of the sound wave signal, is the maximum intensity of the spectral signal. Partition threshold according to Dynamic Adjustment: (10) For the eigenvalue or , assign symbols: if ,but ;like ,but symbol ;like ,but Similarly, Generate symbol sequence by allocating the same threshold Based on symbol sequence and , construct the joint transition probability matrix M, dimension . Matrix elements Indicates the symbol for acoustic vibration to optical symbols The transition probability is: (11) For symbols Followed by The number of times, For symbols The total number of times is counted using a 10ms sliding window to capture the instantaneous dynamic correlation of the explosion flame.

[0063] Calculate the entropy of the symbol sequence and verify the amount of information: (12) in, and is the probability of the symbol appearing. If or Below 0.5, increase Or adjust the scaling parameter such as 0.8 to 0.9 and re-encode. Final output and , used for conditional entropy calculation.

[0064] The above step 205, namely "based on the comprehensive feature representation, judging whether there is an explosion source flame by a classification algorithm and outputting a detection result" will be described in detail below in conjunction with an embodiment.

[0065] This step utilizes the comprehensive feature representation generated in the previous step to identify the characteristics of the explosion flame through a classification algorithm, determining whether it is the explosion source flame and outputting a clear detection result. The comprehensive feature representation combines the dynamic behavior patterns of acoustic vibration signals with the spectral characteristics of optical signals, encompassing the transient, nonlinear, and cross-modal correlation characteristics of the explosion flame. By analyzing these characteristics, the classification algorithm distinguishes explosion flames from other flame types or environmental interference, providing highly accurate detection results suitable for real-time monitoring of complex scenarios such as chemical explosions and industrial blasting.

[0066] The comprehensive feature representation is derived from a symbolic dynamics algorithm. Through adaptive partition coding and a joint transition probability matrix, it captures the dynamic correlation between acoustic vibration and optical signals, generating conditional entropy and transition probability features. These features, represented as vectors, encompass the unique patterns of explosive flames, such as the topological properties of the instantaneous shock wave and the synchronization of spectral peaks. The classification algorithm's task is to map this high-dimensional feature vector into a binary or multi-class decision space to determine whether an explosive flame is present. The output is typically "explosion flame detected" or "not detected" and may include a confidence score to support real-time decision-making and subsequent early warning systems.

[0067] Specifically, the classification algorithm can employ a sparse representation classifier. This sparse representation classifier is a machine learning method based on sparse signal decomposition and is particularly well-suited for processing high-dimensional, noisy data. The core concept is to represent the input feature vector as a linear combination of a predefined dictionary, where the coefficients are as sparse as possible, meaning that most coefficients are zero. This sparsity highlights the core patterns in the data and suppresses noise interference. In explosive flame detection, the comprehensive feature representation, which includes the topological features of the acoustic vibration signal and the spectral features of the optical signal, is high-dimensional and nonlinear. The sparse representation classifier constructs a shared dictionary of explosive flame characteristics, decomposing the comprehensive feature vector into sparse coefficients and then determining whether it represents an explosive flame based on the reconstruction error. The sparse representation classifier constructs a shared dictionary of explosive flame characteristics and decomposes the comprehensive feature representation into sparse coefficients. The shared dictionary can be optimized using training data and contains typical characteristic patterns of explosive flames, such as instantaneous acoustic frequency variations or spectral intensity distribution. The algorithm determines whether the input features match the explosive flame pattern by calculating the reconstruction error between the comprehensive feature vector and the dictionary. If the reconstruction error is below a set threshold, the signal is considered an explosive flame; otherwise, it is considered a non-target signal. This method is robust to high-noise environments and can effectively distinguish explosion flames from regular flames or environmental interference.

[0068] In implementation, the sparse representation classifier first builds a shared dictionary through training data , each column of the dictionary is called an atom, which represents a typical characteristic pattern of explosion flame or non-flame signal. The training data includes known explosion flame feature vectors such as conditional entropy value, topological duration and spectral peak. The dictionary can be constructed using the K-SVD algorithm, and iterative optimization is used to ensure that the dictionary atoms can effectively represent the training data. For the input comprehensive feature vector , the classifier solves the sparse coefficient , making ,in is sparse. The solution process uses the orthogonal matching pursuit algorithm, and the optimization objective is: (13) in, is the sparsity constraint, the reconstruction error Used for classification. If the error is lower than the set threshold, it indicates If the dictionary pattern matches the explosion flame, it is determined to be an explosion flame; otherwise, it is determined to be a non-target signal.

[0069] The advantage of sparse representation in explosion flame detection lies in its ability to effectively handle high-noise environments, such as smoke or strong light interference. A shared dictionary is trained to capture the transient and nonlinear characteristics of explosion flames, such as the topological pattern of shock waves and the transient peaks of the spectrum, enhancing classification discriminability. Furthermore, sparse representation has low computational complexity, making it suitable for real-time detection requirements, such as completing classification within 100 milliseconds. Compared to traditional classifiers such as support vector machines, sparse representation more fully utilizes the sparsity of high-dimensional data and is well-suited to the complex characteristics of explosion flames.

[0070] Another implementation approach is a classification method based on rough set theory. Rough set theory is a mathematical tool for dealing with uncertainty and incomplete data. Its core concept is to extract core features from complex data through attribute reduction and decision rule generation, thereby constructing a concise classification model. In explosive flame detection, the comprehensive feature representation includes multiple dimensions such as conditional entropy, topological features, and spectral features, which may contain redundancy or noise. Rough set theory reduces redundant attributes to extract a core feature subset, which is then used to generate dynamic decision rules to determine the presence of explosive flames. Through attribute reduction, rough set theory extracts a core feature subset from the comprehensive feature representation, reducing redundant information and improving computational efficiency. The core feature subset may include key indicators such as conditional entropy, topological features, or spectral peaks. The algorithm constructs rough set decision rules based on these feature subsets to determine the presence of explosive flames. The decision rules are dynamically updated to adapt to the noise level in different scenarios, for example, adjusting the rule threshold in a smoky environment to reduce the false alarm rate. This approach is particularly suitable for dealing with the uncertainty of explosive flame characteristics and complex environmental interference.

[0071] In practice, the rough set method first organizes the comprehensive feature representation into an information table, with rows representing samples and columns representing features such as conditional entropy, duration of 0-dimensional connected components, and spectral peaks. The target is the classification label, i.e., explosive flame or non-explosive flame. Attribute reduction is performed by calculating the discriminability matrix or information entropy to remove redundant features irrelevant to the classification, retaining a subset of core features. For example, conditional entropy may be crucial for distinguishing explosive flames, while certain spectral features may be redundant. After reduction, a decision rule is generated based on the core feature subset in the form of: (14) in, and The threshold is optimized using training data. Rules are dynamically updated to adapt to environmental changes, such as adjusting the threshold in a smoky environment to reduce false positives. During classification, the input feature vector is matched against the rule, and the detection result is output.

[0072] The advantage of rough set theory in explosion and flame detection lies in its ability to handle feature uncertainty and noise, eliminating assumptions about data distribution and making it suitable for complex scenarios. The dynamic rule update mechanism enhances the method's adaptability to high-noise environments, such as dealing with smoke interference during chemical explosions. Compared to methods such as neural networks, rough set theory offers computational simplicity and highly interpretable rules, making it suitable for embedded system applications.

[0073] The above method provided in the embodiment of the present application can be applied to a variety of application scenarios, including but not limited to: in chemical plant explosion monitoring, the method collects acoustic vibration signals and multi-spectral signals, and uses persistent coherence algorithms and symbolic dynamics fusion features to quickly identify the dynamic behavior patterns of explosion flames, output detection results in real time, and effectively deal with smoke or strong light interference to ensure safe production. In industrial blasting scenarios, such as mining or tunnel blasting operations, the method uses topological analysis of sound waves and micro-vibrations, combined with optical features, to accurately distinguish explosion flames from environmental interference, adapt to complex terrain and dust environments, provide reliable flame detection support, reduce false alarm rates and improve response speed. The robustness and real-time nature of this method make it significantly practical in the fields of chemical industry, mining, etc., providing an efficient solution for safety monitoring and emergency response.

[0074] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0075] According to an embodiment of another aspect, an explosion source flame detection system is provided. Figure 3 FIG. 1 shows a schematic block diagram of the explosion source flame detection system according to one embodiment. Figure 3 As shown, the system 300 includes: The signal acquisition unit 301 is configured to acquire multimodal signals of the explosion source flame, wherein the multimodal signals include acoustic vibration signals and optical signals.

[0076] The feature set generating unit 302 is configured to pre-process the acoustic vibration signal and the optical signal to generate an acoustic vibration feature set and an optical feature set.

[0077] The pattern recognition unit 303 is configured to perform topological analysis on the acoustic vibration feature set using a persistent homology algorithm to identify the dynamic behavior pattern of the explosion source flame and obtain a dynamic behavior pattern recognition result of the explosion source flame.

[0078] The comprehensive feature generation unit 304 is configured to convert the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence using a symbolic dynamics algorithm, analyze the cross-modal symbol pattern through a transition probability matrix, and generate a comprehensive feature representation of the explosion flame.

[0079] The detection result generating unit 305 is configured to determine whether there is an explosion source flame based on the comprehensive feature representation through a classification algorithm and output a detection result.

[0080] As an implementable manner, the feature set generation unit 302 can be configured to: use discrete wavelet transform to perform multi-scale decomposition of the acoustic wave signal and the micro-vibration signal, extract the time-frequency features at each scale, and remove environmental noise interference through adaptive threshold filtering to generate an acoustic vibration feature set; use fast Fourier transform to analyze the frequency distribution of the multi-spectral signal, extract the spectral line features in the wavelength range of 400 nanometers to 2500 nanometers, and generate an optical feature set when preprocessing the acoustic vibration signal and the optical signal to generate an acoustic vibration feature set and an optical feature set.

[0081] As an implementable manner, when the pattern recognition unit 303 uses the persistent coherence algorithm to perform topological analysis on the acoustic vibration feature set to identify the dynamic behavior pattern of the explosion source flame, it can be configured as follows: mapping the acoustic vibration feature set to a high-dimensional point cloud representation, constructing a persistent coherence barcode, and extracting the topological features of the dynamic behavior pattern of the explosion flame, wherein the topological features include the duration of the 0-dimensional connected component and the 1-dimensional ring structure.

[0082] As an implementable manner, the comprehensive feature generation unit 304 can be configured to: discretize the dynamic behavior pattern recognition result and the optical feature set into a finite symbol sequence respectively, and construct a joint transition probability matrix when converting the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence using a symbolic dynamics algorithm, analyzing the cross-modal symbol pattern through a transition probability matrix, and generating a comprehensive feature representation of the explosion flame, wherein the conditional entropy of the cross-modal symbol sequence is calculated to generate a comprehensive feature representation of the explosion flame, wherein the conditional entropy is used to quantify the correlation between the two types of signals in the instantaneous dynamics of the explosion flame.

[0083] As an implementable manner, the comprehensive feature generation unit 304 can be configured to discretize the dynamic behavior pattern recognition result and the optical feature set into a finite symbol sequence respectively and construct a joint transition probability matrix as follows: discretize the dynamic behavior pattern recognition result and the optical feature set into a finite symbol sequence through an adaptive partition coding method, wherein the adaptive partition coding dynamically adjusts the symbol division threshold according to the instantaneous intensity of the explosion flame.

[0084] As an implementable method, when determining whether there is an explosion source flame through a classification algorithm, the detection result generation unit 305 can be configured as follows: using a classifier based on sparse representation, sparsely encoding the comprehensive feature representation, constructing a shared dictionary of explosion flame features, and determining whether there is an explosion source flame through dictionary reconstruction error.

[0085] As an implementable manner, when determining whether an explosion source flame exists through a classification algorithm, the detection result generation unit 305 can be configured as follows: adopting a classification method based on rough set theory, performing attribute simplification on the comprehensive feature representation, extracting a core feature subset of the explosion flame, and determining whether an explosion source flame exists through a rough set decision rule, wherein the decision rule is dynamically updated to adapt to detection requirements in a high noise environment.

[0086] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiment. The system embodiment described above is only exemplary, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without making any creative efforts.

[0087] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0088] In addition, an embodiment of the present application further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of any one of the methods in the aforementioned method embodiments are implemented.

[0089] And an electronic device comprising: One or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions, which, when read and executed by the one or more processors, execute the steps of the method described in any one of the aforementioned method embodiments.

[0090] The present application also provides a computer program product, comprising a computer program, which implements the steps of any one of the methods described in the aforementioned method embodiments when executed by a processor.

[0091] in, Figure 4 The electronic device architecture is shown as an example, and may include a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, and a memory 420. The processor 410, the video display adapter 411, the disk drive 412, the input / output interface 413, the network interface 414, and the memory 420 may be communicatively connected via a communication bus 430.

[0092] The processor 410 may be implemented as a general-purpose CPU, a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and may be used to execute relevant programs to implement the technical solutions provided in this application.

[0093] The memory 420 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 420 can store an operating system 421 for controlling the operation of the electronic device 400 and a basic input and output system (BIOS) 422 for controlling the low-level operations of the electronic device 400. In addition, a web browser 423, a data storage management system 424, and an explosion source flame detection system 425 can also be stored. The above-mentioned explosion source flame detection system 425 can be an application program that specifically implements the operations of the aforementioned steps in the embodiment of the present application. In short, when the technical solution provided by the present application is implemented by software or firmware, the relevant program code is stored in the memory 420 and is called and executed by the processor 410.

[0094] The input / output interface 413 is used to connect to input / output modules to enable information input and output. The input / output modules can be configured as components within the device (not shown) or externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, and various sensors, while output devices may include a display, speaker, vibrator, indicator light, and the like.

[0095] The network interface 414 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.).

[0096] The bus 430 comprises a pathway for transmitting information between the various components of the device, such as the processor 410 , the video display adapter 411 , the disk drive 412 , the input / output interface 413 , the network interface 414 , and the memory 420 .

[0097] It should be noted that although the above device only shows a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, a memory 420, a bus 430, etc., in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may also include only the components necessary to implement the solution of the present application, and does not necessarily include all the components shown in the figure.

[0098] Through the description of the above embodiments, it can be seen that those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer program product. The computer program product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments of the present application or certain parts of the embodiments.

[0099] The above is a detailed introduction to the technical solutions provided by this application. Specific examples are used herein to illustrate the principles and implementation methods of this application. The description of the above embodiments is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the contents of this specification should not be understood as limiting this application.

Claims

1. A method for detecting explosion source flame, characterized in that: The method comprises: Collecting multimodal signals of the explosion source flame, wherein the multimodal signals include acoustic vibration signals and optical signals; Preprocessing the acoustic vibration signal and the optical signal to generate an acoustic vibration feature set and an optical feature set; Performing a topological analysis on the acoustic vibration feature set using a persistent homology algorithm to identify a dynamic behavior pattern of the explosion source flame, thereby obtaining a dynamic behavior pattern recognition result of the explosion source flame; The dynamic behavior pattern recognition results and the optical feature set are converted into a symbol sequence using a symbolic dynamics algorithm, and a cross-modal symbol pattern is analyzed through a transition probability matrix to generate a comprehensive feature representation of the explosion flame; Based on the comprehensive feature representation, a classification algorithm is used to determine whether there is an explosion source flame, and a detection result is output.

2. The explosion source flame detection method according to claim 1, characterized in that: Preprocessing the acoustic vibration signal and the optical signal to generate an acoustic vibration feature set and an optical feature set includes: The acoustic and micro-vibration signals are decomposed into multiple scales using discrete wavelet transform to extract the time-frequency features at each scale. The environmental noise interference is removed through adaptive threshold filtering to generate an acoustic vibration feature set. Fast Fourier transform is used to analyze the frequency distribution of multispectral signals, extract the spectral line features in the wavelength range of 400 nm to 2500 nm, and generate an optical feature set.

3. The explosion source flame detection method according to claim 1, characterized in that: The method of performing topological analysis on the acoustic vibration feature set using a persistent coherence algorithm to identify a dynamic behavior pattern of the explosion source flame includes: The acoustic vibration feature set is mapped to a high-dimensional point cloud representation, a persistent coherent barcode is constructed, and the topological features of the dynamic behavior pattern of the explosion flame are extracted. The topological features include the duration of the 0-dimensional connected component and the 1-dimensional ring structure.

4. The explosion source flame detection method according to claim 1, characterized in that: The method of converting the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence by using a symbolic dynamics algorithm, analyzing the cross-modal symbol pattern by a transition probability matrix, and generating a comprehensive feature representation of the explosion flame includes: Discretizing the dynamic behavior pattern recognition result and the optical feature set into finite symbol sequences respectively, and constructing a joint transition probability matrix; The conditional entropy of the cross-modal symbol sequence is calculated to generate a comprehensive feature representation of the explosion flame, where the conditional entropy is used to quantify the correlation between the two types of signals in the instantaneous dynamics of the explosion flame.

5. The explosion source flame detection method according to claim 4, characterized in that: Discretizing the dynamic behavior pattern recognition result and the optical feature set into finite symbol sequences respectively and constructing a joint transition probability matrix includes: The dynamic behavior pattern recognition result and the optical feature set are discretized into a finite symbol sequence through an adaptive partition coding method, wherein the adaptive partition coding dynamically adjusts the symbol division threshold according to the instantaneous intensity of the explosion flame.

6. The explosion source flame detection method according to claim 1, characterized in that: The determining whether there is an explosion source flame by a classification algorithm includes: A classifier based on sparse representation is used to perform sparse coding on the comprehensive feature representation, and a shared dictionary of explosion flame features is constructed. The presence of explosion source flame is judged based on dictionary reconstruction error.

7. The explosion source flame detection method according to claim 1, characterized in that: The determining whether there is an explosion source flame by a classification algorithm includes: A classification method based on rough set theory is adopted to simplify the attributes of the comprehensive feature representation, extract the core feature subset of the explosion flame, and judge whether the explosion source flame exists by rough set decision rules. The decision rules are dynamically updated to adapt to the detection needs in high noise environments.

8. An explosion source flame detection system, characterized in that: The system comprises: A signal acquisition unit is configured to acquire multimodal signals of the explosion source flame, wherein the multimodal signals include acoustic vibration signals and optical signals; a feature set generating unit configured to pre-process the acoustic vibration signal and the optical signal to generate an acoustic vibration feature set and an optical feature set; a pattern recognition unit configured to perform a topological analysis on the acoustic vibration feature set using a persistent coherence algorithm to identify a dynamic behavior pattern of the explosion source flame, and obtain a dynamic behavior pattern recognition result of the explosion source flame; a comprehensive feature generation unit configured to convert the dynamic behavior pattern recognition result and the optical feature set into a symbol sequence using a symbolic dynamics algorithm, analyze the cross-modal symbol pattern through a transition probability matrix, and generate a comprehensive feature representation of the explosion flame; The detection result generating unit is configured to determine whether there is an explosion source flame based on the comprehensive feature representation through a classification algorithm and output a detection result.

9. An electronic device, characterized in that: include: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions, wherein the program instructions, when read and executed by the one or more processors, execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images