Earthquake activity real-time monitoring method and device based on deep learning multi-modal fusion
Through deep learning technology, the data fusion and monitoring of multimodal seismic data is solved, and the data fusion problem in traditional technology is achieved, high-precision and efficient seismic monitoring is achieved, adapting to different environmental changes, and providing accurate and timely earthquake early warning support.
Patent Information
- Application Number
- CN202410926250.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-11
- Publication Date
- 2024-11-08
AI Technical Summary
Traditional seismic monitoring technology is difficult to effectively integrate data from different sources and types, resulting in limited accuracy and timeliness of monitoring and inability to adapt to changing environmental changes.
Using a multimodal fusion method based on deep learning, data is collected through seismic wave detectors, geological radars and satellite remote sensing, data preprocessing, feature extraction and fusion, adaptive weight adjustment and integrated learning algorithms are designed, and multi-input neural network architecture is used Perform data fusion and monitoring.
It significantly improves the monitoring accuracy and real-time nature of earthquake activities, can provide timely information on the time, location and magnitude of earthquakes, has flexibility and high anti-interference ability, and improves user experience and computing efficiency.
Smart Images

Figure CN118915136A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of earthquake monitoring technology, and in particular to a method and device for real-time monitoring of earthquake activities based on deep learning multimodal fusion. Background Art
[0002] Real-time monitoring of seismic activity is crucial to preventing earthquake disasters and reducing casualties and property losses. Traditional earthquake monitoring technology is mainly based on the detection and analysis of seismic waves. However, these methods have obvious limitations: First, seismic waves are easily affected by multiple factors when propagating in the earth's crust, such as the complexity of geological structures and environmental noise, which can cause attenuation and interference of seismic wave signals, thereby affecting the accuracy and timeliness of monitoring. Second, traditional methods are usually unable to effectively integrate data from different sources and types, such as geological, meteorological, and human activity data, which limits the comprehensiveness of prediction models and the early identification capabilities of warnings.
[0003] In recent years, with the development of big data technology and artificial intelligence, real-time monitoring methods for earthquake activity based on multimodal data fusion have begun to be studied and applied. These methods improve the response speed and accuracy of monitoring systems by integrating data from multiple fields such as seismology, meteorology, and artificial intelligence. However, existing multimodal data fusion methods still face some technical challenges, such as how to efficiently process and integrate large-scale heterogeneous data and how to design intelligent algorithms that can adapt to various environmental changes. Summary of the invention
[0004] The present invention provides a real-time monitoring method and device for earthquake activity based on deep learning multimodal fusion, so as to solve the defects of the multimodal data fusion method in the prior art and improve the monitoring accuracy and real-time performance of earthquake activity.
[0005] The present invention provides a method for real-time monitoring of seismic activity based on deep learning multimodal fusion, comprising the following steps: Collecting data, including collecting seismic wave signals at multiple times, geological structure information at multiple times, and surface deformation information at multiple times; Preprocess the collected data; Perform feature extraction on the preprocessed data; Perform data fusion; Monitor seismic activity; Output monitoring results.
[0006] According to a real-time monitoring method for seismic activity based on deep learning multimodal fusion provided by the present invention, the data collection includes: real-time collection of seismic wave signals at multiple moments by a seismic wave detector; real-time collection of geological structure information at multiple moments by a geological radar; and real-time collection of surface deformation information at multiple moments by satellite remote sensing.
[0007] According to a real-time monitoring method for seismic activity based on deep learning multimodal fusion provided by the present invention, the preprocessing of the collected data includes: subjecting the seismic wave signals at multiple moments to data cleaning, denoising, and standardization processing to obtain preprocessed seismic wave signals at multiple moments; subjecting the geological structure information at multiple moments to data cleaning, denoising, and standardization processing to obtain preprocessed geological structure information at multiple moments; subjecting the surface deformation information at multiple moments to data cleaning, denoising, and standardization processing to obtain preprocessed surface deformation information at multiple moments.
[0008] According to a real-time monitoring method for seismic activity based on deep learning multimodal fusion provided by the present invention, the feature extraction of preprocessed data includes: extracting features of preprocessed seismic wave signals at multiple moments, preprocessed geological structure information at multiple moments, and preprocessed surface deformation information at multiple moments using a deep learning algorithm, respectively, to obtain seismic wave features, geological structure features, and terrain change features in turn.
[0009] According to a method for real-time monitoring of seismic activity based on deep learning multimodal fusion provided by the present invention, the data fusion comprises: at the feature level, determining the weights of different modal features through an adaptive weight adjustment algorithm; at the decision level, using an integrated learning method, output by multiple classifiers; model-level fusion is achieved by designing a multi-input neural network architecture, which simultaneously receives features from seismic wave signals, geological structure information, and surface deformation information, and learns the complex relationship between different modalities through internal connections.
[0010] According to a method for real-time monitoring of earthquake activities based on deep learning multimodal fusion provided by the present invention, the monitoring of earthquake activities includes: based on the feature vector after data fusion, using a classification algorithm or a regression algorithm to monitor the earthquake activities in real time, so as to obtain predictions and judgments on the time, location and magnitude of the earthquake.
[0011] According to a real-time monitoring method for seismic activity based on deep learning multimodal fusion provided by the present invention, the output monitoring results include: outputting the monitoring results in graphical or textual form.
[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a real-time monitoring method for seismic activity based on deep learning multimodal fusion as described in any one of the above-mentioned methods is implemented.
[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for real-time monitoring of seismic activities based on deep learning multimodal fusion as described in any one of the above is implemented.
[0014] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for real-time monitoring of seismic activities based on deep learning multimodal fusion.
[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention uses deep learning technology to comprehensively analyze multimodal data collected by multiple sensors, realizes comprehensive capture of seismic activity characteristics, and significantly improves monitoring accuracy. Compared with the traditional seismic wave amplitude prediction method, the present invention can monitor the occurrence of earthquakes in real time, and promptly provide key information such as the time, location and magnitude of the earthquake, providing accurate and timely data support for earthquake early warning and emergency response. The data processing technology of the present invention not only adapts to the needs of different sensors and data formats, but also has a high degree of flexibility and scalability, and can cope with changing application scenarios. In addition, the deep learning model of the present invention has been specially optimized and has stronger generalization ability, which can adapt to the earthquake monitoring needs of different regions. At the same time, it effectively identifies and filters noise in the process of multimodal data fusion, and enhances the anti-interference ability of the monitoring system. In terms of user experience, the present invention provides an intuitive visual display of monitoring results, which makes the monitoring data easy to understand and analyze, thereby improving the user's data interpretation efficiency and user experience. By optimizing the computing efficiency, the present invention reduces the demand for computing resources while ensuring the monitoring accuracy and real-time performance, making the monitoring system more efficient and practical. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 It is a flowchart of real-time monitoring of seismic activity based on deep learning multi-modal fusion provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0018] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0019] The present embodiment provides a real-time monitoring method for earthquake activity based on deep learning multimodal fusion, which realizes real-time monitoring and early warning of earthquake activity by fusing earthquake-related data of different sources and different types, and extracting and fusing multimodal data using deep learning technology. The method is based on a deep learning algorithm, and realizes real-time monitoring of earthquake activity by extracting and fusing multimodal earthquake data. Multimodal data includes seismic wave signals, geological environment data, earthquake history data, etc. By preprocessing and extracting features from these data, feature representations of each data modality are obtained. Then, these features are fused using a deep learning model to obtain a comprehensive feature representation of earthquake activity. Finally, real-time monitoring and early warning of earthquake activity are performed based on the comprehensive feature representation. Preprocessing of multimodal earthquake data includes steps such as data cleaning and data standardization to ensure data quality and consistency. The real-time monitoring method for earthquake activity based on multimodal data fusion of deep learning of the present invention has the advantages of multimodal data fusion, deep learning technology and real-time monitoring, and provides a new solution for real-time monitoring and early warning of earthquake activity.
[0020] This embodiment is implemented by the following technical solution, a real-time monitoring method of seismic activity based on deep learning multimodal fusion, comprising the following steps: 1. Data collection: The seismic wave signals at multiple moments are collected in real time by using a seismic wave detector; Collect geological structure information at multiple times in real time through geological radar; Collect surface deformation information at multiple times in real time through satellite remote sensing; 2. Data preprocessing: preprocess the collected multimodal data; The seismic wave signals at multiple moments are processed by data cleaning, denoising and standardization to obtain pre-processed seismic wave signals at multiple moments; The geological structure information at multiple times is processed by data cleaning, denoising and standardization to obtain the pre-processed geological structure information at multiple times; The surface deformation information at multiple times is processed by data cleaning, denoising and standardization to obtain the pre-processed surface deformation information at multiple times; 3. Feature extraction: The pre-processed seismic wave signals at multiple moments, the pre-processed geological structure information at multiple moments, and the pre-processed surface deformation information at multiple moments are respectively extracted using a deep learning algorithm to obtain seismic wave features, geological structure features, and terrain change features in turn; 4. Data Fusion: At the feature level, the weights of different modal features are determined through an adaptive weight adjustment algorithm to ensure that key information is enhanced; At the decision level, ensemble learning methods are used to combine the outputs of multiple classifiers to improve the stability and accuracy of predictions; Model-level fusion is achieved by designing a multi-input neural network architecture that can simultaneously receive features from seismic wave signals, geological structure information, and surface deformation information, and learn the complex relationship between different modes through internal connections; 5. Earthquake activity monitoring: Based on the fused feature vector, appropriate classification or regression algorithms are used to monitor earthquake activities in real time to predict and judge the time, location, and magnitude of earthquakes.
[0021] 6. Result output: Display the monitoring results in graphical or textual form to facilitate user viewing and analysis.
[0022] This embodiment realizes real-time monitoring of earthquake activity through deep learning algorithm and multimodal data fusion technology, and improves the accuracy and real-time performance of monitoring. At the same time, the present invention also has the advantages of simple operation and good visualization effect, and has good application prospects.
[0023] When implementing it, Figure 1 As shown, a real-time monitoring method for earthquake activity based on deep learning multimodal fusion includes the following steps: Step 1: Use the obspy function of the Python library for the seismic field to read the ".mseed" file that stores the original seismic waveform data (i.e., seismic velocity waveform data); at the same time, read the data files of ground displacement and groundwater level related to the earthquake. Synchronize these data files to ensure time alignment for subsequent data fusion analysis; Step 2: Preprocess the seismic waveform data, ground displacement and groundwater level data obtained in step 1. First, the seismic waveform data is divided by the sensitivity of the seismic instrument for correction, and then all data are normalized to eliminate the impact of the dimension, so that data from different sources can be compared and integrated under the same standard; Step 3: Extract features from the preprocessed multimodal data through a deep neural network. Construct a composite neural network model that integrates a convolutional neural network (CNN) and a long short-term memory network (LSTM). CNN is used to extract the spatial features of seismic waveform data, while LSTM is used to analyze the temporal features of ground displacement and groundwater level data. Step 4: The data fusion link is implemented through a carefully designed deep learning architecture. First, the features extracted in step 3 are input into a customized fusion network, which consists of CNN and LSTM to process spatial and time series features, respectively. CNN focuses on the spatial characteristics of seismic waveform data, while LSTM analyzes the temporal dynamics of ground displacement and groundwater level data. The fusion network effectively integrates the features of different modalities through feature mapping and weight adjustment to enhance the accuracy of seismic activity prediction. In addition, the attention mechanism is used to identify key features, and the fusion effect is evaluated through cross-validation to ensure the generalization ability of the model. Finally, the fusion network outputs a comprehensive feature vector, which provides strong data support for real-time monitoring of seismic activity; Step 5: After the deep neural network model is built, it is trained with the training data set, and the network weights are fine-tuned using the back-propagation algorithm. During the training process, we use the loss function to quantify the difference between the model prediction and the actual observation value, and use this to guide the optimization of the weights. Through iterative training, the model learns the characteristic representation of earthquake activity, thereby improving its prediction accuracy. To ensure the generalization ability of the model, we implement cross-validation, divide the data set into multiple subsets, and use one of the subsets as the test set and the rest as the training set in turn to evaluate the performance of the model on different data subsets. In addition, we monitor the overfitting phenomenon during the training process and control the model complexity by adjusting the network structure or introducing regularization terms. Ultimately, the trained and verified model will be able to receive new earthquake monitoring data in real time, and quickly and accurately predict earthquake activity, realizing the timely identification of potential earthquake risks.
[0024] The earthquake monitoring method proposed in this embodiment realizes the efficient fusion and analysis of earthquake waveforms, ground displacement and groundwater level data through deep learning technology. This method optimizes the data fusion strategy through synchronous processing and feature extraction, and improves the accuracy and real-time performance of earthquake prediction. The model training adopts cross-validation and regularization technology to ensure good generalization ability and avoid overfitting. Finally, this method can receive monitoring data in real time, quickly predict earthquake activities, and effectively identify potential risks, which has important practical application value and promotion potential.
[0025] Figure 2 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 2As shown, the electronic device may include: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus. The processor may call the logic instructions in the memory to execute the real-time monitoring method of seismic activity based on deep learning multimodal fusion.
[0026] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.
[0027] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the real-time monitoring method of seismic activity based on deep learning multimodal fusion provided by the above-mentioned methods.
[0028] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the real-time monitoring method for seismic activity based on deep learning multimodal fusion provided by the above-mentioned methods.
[0029] The device embodiments described above are merely illustrative, 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 may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0030] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0031] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A real-time monitoring method for earthquake activity based on deep learning multimodal fusion, characterized in that: The following steps are involved: Collecting data, including collecting seismic wave signals at multiple times, geological structure information at multiple times, and surface deformation information at multiple times; Preprocess the collected data; Perform feature extraction on the preprocessed data; Perform data fusion; Monitor seismic activity; Output monitoring results.
2. The real-time monitoring method for earthquake activity based on deep learning multimodal fusion according to claim 1 is characterized in that: The data collection includes: collecting seismic wave signals at multiple moments in real time through a seismic wave detector; collecting geological structure information at multiple moments in real time through a geological radar; and collecting surface deformation information at multiple moments in real time through satellite remote sensing.
3. The real-time monitoring method for earthquake activity based on deep learning multimodal fusion according to claim 1 is characterized in that: The preprocessing of the collected data includes: performing data cleaning, denoising and standardization on seismic wave signals at multiple moments to obtain preprocessed seismic wave signals at multiple moments; performing data cleaning, denoising and standardization on geological structure information at multiple moments to obtain preprocessed geological structure information at multiple moments; performing data cleaning, denoising and standardization on surface deformation information at multiple moments to obtain preprocessed surface deformation information at multiple moments.
4. The real-time monitoring method for earthquake activity based on deep learning multimodal fusion according to claim 3 is characterized in that: The feature extraction of the preprocessed data includes: extracting features of the preprocessed seismic wave signals at multiple moments, the preprocessed geological structure information at multiple moments, and the preprocessed surface deformation information at multiple moments using a deep learning algorithm, respectively, to obtain seismic wave features, geological structure features, and terrain change features in turn.
5. The real-time monitoring method for earthquake activity based on deep learning multimodal fusion according to claim 3 is characterized in that: The data fusion process includes: at the feature level, determining the weights of different modal features through an adaptive weight adjustment algorithm; at the decision level, using an integrated learning method, output by multiple classifiers; and model-level fusion is achieved by designing a multi-input neural network architecture that simultaneously receives features from seismic wave signals, geological structure information, and surface deformation information, and learns the complex relationship between different modalities through internal connections.
6. The real-time monitoring method for earthquake activity based on deep learning multimodal fusion according to claim 5 is characterized in that: The monitoring of earthquake activity includes: based on the feature vector after data fusion, using a classification algorithm or a regression algorithm to monitor the earthquake activity in real time, so as to predict and judge the time, location and magnitude of the earthquake.
7. The real-time monitoring method for earthquake activity based on deep learning multimodal fusion according to claim 6 is characterized in that: The outputting of the monitoring results includes: outputting the monitoring results in a graphical or textual form.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the real-time monitoring method of seismic activity based on deep learning multimodal fusion as described in any one of claims 1 to 6 is implemented.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the real-time monitoring method of seismic activity based on deep learning multimodal fusion as described in any one of claims 1 to 6 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the real-time monitoring method of seismic activity based on deep learning multimodal fusion as described in any one of claims 1 to 6 is implemented.
Citation Information
Cited By
Earthquake information summary generation method and device, terminal and storage medium
CN119471791A
Multi-source monitoring data analysis method and system based on earthquake active source
CN119805556A
Multi-source monitoring data analysis method and system based on active earthquake source
CN119805556B
Earthquake monitoring and early warning method and system based on GNSS multi-source information fusion
CN121348406A
Microseismic time sequence prediction method based on multi-modal feature fusion
CN122568599A
