Seismic station monitoring method and device based on edge calculation and federated learning

Abstract

The invention provides a seismic station monitoring method and device based on edge calculation and federated learning, and relates to the technical field of data monitoring. The method comprises the following steps: acquiring a communication protocol of a non-standard instrument and monitoring data of a seismic station instrument; inputting a communication protocol of the non-standard instrument into the trained language model, extracting communication protocol characteristics by adopting the trained language model, and generating a rule corresponding to protocol conversion according to the communication protocol characteristics; inputting the monitoring data of the seismic station instrument into the space-time fusion deep learning model, and performing feature extraction on the monitoring data of the seismic station instrument by adopting the space-time fusion deep learning model to obtain a space-time feature map; and inputting the spatial-temporal characteristic pattern into the trained lightweight SVM model, and performing fault diagnosis according to the spatial-temporal characteristic pattern by adopting the trained lightweight SVM model to obtain a fault diagnosis result. The intelligent level of earthquake monitoring is remarkably improved, and reliable technical support is provided for disaster prevention and reduction.

Description

Technical Field

[0001] This invention relates to the field of data monitoring technology, and more specifically, to a method and apparatus for monitoring seismic stations based on edge computing and federated learning. Background Technology

[0002] In recent years, my country's earthquake monitoring system has gradually developed from single-discipline observation to a comprehensive observation network covering multiple disciplines such as seismometry, deformation, electromagnetics, and underground fluids, and has achieved a technological upgrade from analog to digital and networked systems. However, existing seismic stations still face significant technical bottlenecks in actual operation:

[0003] 1) Insufficient protocol compatibility: Many imported instruments use proprietary communication protocols that are incompatible with earthquake industry standard protocols (such as the Fifteenth Protocol and the TSF Protocol), resulting in difficulties in data access and high costs for manual adaptation; 2) Limited data processing capabilities: Traditional stations rely on centralized servers for data fusion and analysis, which makes it difficult to meet the low-latency processing requirements of massive real-time data. In addition, multi-disciplinary data is stored independently, lacking spatiotemporal correlation mining. 3) Low level of intelligence: Instrument fault diagnosis relies on human experience, lacks automated anomaly detection and early warning mechanisms, and the environmental monitoring dimensions are limited (such as monitoring only temperature, humidity and power supply status), making it difficult to fully perceive the environmental risks of the station. 4) Data privacy and collaboration conflict: The serious data silo phenomenon between stations and the risk of data leakage when training models across regions jointly face the problem, which restricts the improvement of the overall earthquake early warning capability.

[0004] To address the aforementioned issues, while existing technologies attempt to optimize local data processing efficiency through edge computing frameworks, their protocol conversion modules rely on static rule bases, making them unable to dynamically adapt to new non-standard devices. Multi-source data fusion often employs simple weighting or threshold judgments, failing to fully consider the deep correlations of spatiotemporal features. Fault diagnosis algorithms are limited to independent analysis of single nodes, lacking federated collaboration and privacy protection mechanisms. Furthermore, environmental monitoring networks lack sufficient accuracy in sensing environmental parameters, and anomaly responses rely on manual intervention, making it difficult to meet high reliability requirements.

[0005] Therefore, there is an urgent need for a new seismic station monitoring method that integrates edge computing, federated learning, and multimodal sensing technologies to overcome key technical bottlenecks such as protocol compatibility, real-time data processing, intelligent diagnosis, and data privacy collaboration, and to provide a more efficient and secure solution for earthquake disaster early warning and emergency response. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of the prior art by providing a seismic station monitoring method and device based on edge computing and federated learning, so as to improve the intelligence level of seismic monitoring and provide reliable technical support for disaster prevention and mitigation.

[0007] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a seismic station monitoring method based on edge computing and federated learning, comprising: acquiring the communication protocol of a non-standard instrument and the monitoring data of the seismic station instrument; inputting the communication protocol of the non-standard instrument into a trained language model, extracting communication protocol features using the trained language model, and generating rules corresponding to protocol conversion based on the communication protocol features; inputting the monitoring data of the seismic station instrument into a spatiotemporal fusion deep learning model, extracting features from the monitoring data of the seismic station instrument using the spatiotemporal fusion deep learning model to obtain a spatiotemporal feature map; inputting the spatiotemporal feature map into a trained lightweight SVM model, performing fault diagnosis based on the spatiotemporal feature map using the trained lightweight SVM model to obtain a fault diagnosis result.

[0008] In one embodiment, the step of using the spatiotemporal fusion deep learning model to extract features from the monitoring data of the seismic station instruments to obtain a spatiotemporal feature map includes: extracting temporal dependency features using the time channel in the spatiotemporal fusion deep learning model, extracting physical correlation features between the seismic station instruments using the spatial channel in the spatiotemporal fusion deep learning model, and fusing the temporal dependency features and the physical correlation features to obtain the spatiotemporal feature map.

[0009] In one implementation, the time channel extracts the temporal dependency features through a long short-term memory network.

[0010] In one implementation, the spatial channel extracts the physical correlation features between seismic station instruments using a graph convolutional network.

[0011] In one implementation, the update of the trained lightweight SVM model is shown in the following formula: in, This represents the model parameters in the t-th iteration; Indicates the learning rate; This represents the gradient of the local loss function of the i-th edge node with respect to the parameters; Indicates the number of participating nodes; This represents Gaussian noise with a mean of 0 and a variance of . Used for differential privacy protection; This indicates noise intensity; the higher the value, the stronger the privacy protection.

[0012] In one embodiment, after acquiring the communication protocol of the non-standard instrument and the monitoring data of the seismic station instrument, the method further includes: constructing a multi-dimensional environmental monitoring system by using a microseismic sensor array and a barometric gradiometer, combined with the sensors in the seismic station instrument, and correcting the monitoring data of the seismic station instrument in real time through Kalman filtering.

[0013] In one implementation, the Kalman filter formula is as follows: in, This represents the Kalman filter state estimate at time k; Represents the state transition matrix; This represents the Kalman filter state estimate at time k-1; Indicates Kalman gain; This represents the monitoring data from the seismic station instruments at time k. This represents the monitoring matrix corresponding to the monitoring data from the seismic station instruments.

[0014] In one embodiment, after obtaining the fault diagnosis result, the method further includes: uploading the fault diagnosis result to a central server, wherein the central server periodically distributes updated versions of the trained lightweight SVM model.

[0015] In one embodiment, after obtaining the fault diagnosis result, the method further includes: pushing early warning information when the fault diagnosis result shows that a fault exists.

[0016] Secondly, embodiments of this application also provide a seismic station monitoring device based on edge computing and federated learning, comprising: an acquisition module configured to acquire the communication protocol of a non-standard instrument and the monitoring data of the seismic station instrument; a protocol conversion module configured to input the communication protocol of the non-standard instrument into a trained language model, extract communication protocol features using the trained language model, and generate rules corresponding to protocol conversion based on the communication protocol features; a feature extraction module configured to input the monitoring data of the seismic station instrument into a spatiotemporal fusion deep learning model, extract features from the monitoring data of the seismic station instrument using the spatiotemporal fusion deep learning model, and obtain a spatiotemporal feature map; and a fault diagnosis module configured to input the spatiotemporal feature map into a trained lightweight SVM model, perform fault diagnosis using the trained lightweight SVM model based on the spatiotemporal feature map, and obtain a fault diagnosis result.

[0017] In one embodiment, the feature extraction module is configured to: extract temporal dependency features using the time channel in the spatiotemporal fusion deep learning model, extract physical correlation features between seismic station instruments using the spatial channel in the spatiotemporal fusion deep learning model, and fuse the temporal dependency features and the physical correlation features to obtain the spatiotemporal feature map.

[0018] In one implementation, the time channel extracts the temporal dependency features through a long short-term memory network.

[0019] In one implementation, the spatial channel extracts the physical correlation features between seismic station instruments using a graph convolutional network.

[0020] In one implementation, the update of the trained lightweight SVM model is shown in the following formula: in, This represents the model parameters in the t-th iteration; Indicates the learning rate; This represents the gradient of the local loss function of the i-th edge node with respect to the parameters; Indicates the number of participating nodes; This represents Gaussian noise with a mean of 0 and a variance of . Used for differential privacy protection; This indicates noise intensity; the higher the value, the stronger the privacy protection.

[0021] In one embodiment, the seismic station monitoring device based on edge computing and federated learning further includes a data correction module, which is configured to: construct a multi-dimensional environmental monitoring system by using a microseismic sensor array and a barometric gradient meter in combination with the sensors in the seismic station instrument, and correct the monitoring data of the seismic station instrument in real time through Kalman filtering.

[0022] In one implementation, the Kalman filter formula is as follows: in, This represents the Kalman filter state estimate at time k; Represents the state transition matrix; This represents the Kalman filter state estimate at time k-1; Indicates Kalman gain; This represents the monitoring data from the seismic station instruments at time k. This represents the monitoring matrix corresponding to the monitoring data from the seismic station instruments.

[0023] In one embodiment, the seismic station monitoring device based on edge computing and federated learning further includes a data transmission module, which is configured to upload the fault diagnosis results to a central server, and the central server periodically distributes updated trained lightweight SVM models.

[0024] In one embodiment, the seismic station monitoring device based on edge computing and federated learning further includes an information push module, which is configured to push early warning information when the fault diagnosis result shows that a fault exists.

[0025] Thirdly, embodiments of this application provide a computer device, including: a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the computer device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to perform the steps of any of the above methods.

[0026] Fourthly, embodiments of this application provide a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of any of the methods described above.

[0027] The beneficial effects of this application are as follows: First, it acquires the communication protocols of non-standard instruments and the monitoring data of seismic station instruments; second, it inputs the communication protocols of non-standard instruments into a trained language model, extracts communication protocol features using the trained language model, and generates rules corresponding to protocol conversion based on the communication protocol features; third, it inputs the monitoring data of seismic station instruments into a spatiotemporal fusion deep learning model, extracts features from the monitoring data of the seismic station instruments using the spatiotemporal fusion deep learning model, and obtains a spatiotemporal feature map; finally, it inputs the spatiotemporal feature map into a trained lightweight SVM model, uses the trained lightweight SVM model to perform fault diagnosis based on the spatiotemporal feature map, and obtains the fault diagnosis results. Thus, through the deep integration of edge computing, federated learning, and multimodal sensing technologies, the intelligence level of earthquake monitoring is significantly improved, providing reliable technical support for disaster prevention and mitigation. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1A flowchart illustrating a seismic station monitoring method based on edge computing and federated learning, provided for an embodiment of this application; Figure 2 A flowchart illustrating a seismic station monitoring method based on edge computing and federated learning, provided for an embodiment of this application; Figure 3 A schematic diagram of the structure of a seismic station monitoring device based on edge computing and federated learning, provided for an embodiment of this application; Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0031] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0032] In the description of this application, it should be noted that if the terms "upper", "lower", etc. appear to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of this application is usually placed in, it is only for the convenience of describing this application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0033] Furthermore, the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Additionally, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0034] It should be noted that, where there is no conflict, the features in the embodiments of this application can be combined with each other.

[0035] Figure 1 A flowchart illustrating a seismic station monitoring method based on edge computing and federated learning, provided as an embodiment of this application; Figure 1 As shown, the method includes: Step 110: Obtain the communication protocol of non-standard instruments and the monitoring data of seismic station instruments.

[0036] The monitoring data from seismic stations include: geophysical instrument monitoring data, environmental monitoring data, and superconducting gravimeter monitoring data; geophysical instrument monitoring data includes: fluid monitoring data, deformation monitoring data, electromagnetic monitoring data, and meteorological instrument monitoring data; environmental monitoring data includes: video surveillance data, temperature and humidity sensor monitoring data, barometric pressure sensor monitoring data, noise sensor monitoring data, UPS power supply monitoring data, network status monitoring data, lightning and thunderstorm monitoring data, personnel intrusion monitoring data, and water immersion monitoring data.

[0037] The seismic station monitoring method based on edge computing and federated learning provided in this application embodiment can correct data in real time through Kalman filtering; specifically, after step 110 above, the seismic station monitoring method based on edge computing and federated learning provided in this application embodiment can further include the following steps: A multi-dimensional environmental monitoring system was constructed by using a microseismic sensor array and a barometric gradient meter, combined with sensors in the seismic station instruments. The monitoring data of the seismic station instruments were corrected in real time by Kalman filtering.

[0038] The Kalman filter is shown in the following formula (1): (1) in, This represents the Kalman filter state estimate at time k; Represents the state transition matrix; This represents the Kalman filter state estimate at time k-1; Indicates Kalman gain; This represents the monitoring data from the seismic station instruments at time k. This represents the monitoring matrix corresponding to the monitoring data from the seismic station instruments.

[0039] Step 120: Input the communication protocol of the non-standard instrument into the trained language model, extract the communication protocol features using the trained language model, and generate the corresponding rules for protocol conversion based on the communication protocol features.

[0040] The seismic station monitoring method based on edge computing and federated learning provided in this application introduces a dynamic protocol parsing engine on the basis of existing protocol conversion. It automatically identifies the communication protocol characteristics of non-standard instruments through a trained language model (such as the BERT model) and generates an adaptation rule base to support real-time protocol conversion.

[0041] The protocol matching degree can be calculated using the following formula (2): (2) in, This represents the protocol matching score, used to measure the degree of compatibility between non-standard and standard protocols; This represents the weight of the i-th protocol feature; This is a similarity function used to calculate the input protocol. With standard protocol semantic similarity; This represents the total number of protocol features.

[0042] Step 130: Input the monitoring data of the seismic station instruments into the spatiotemporal fusion deep learning model, and use the spatiotemporal fusion deep learning model to extract features from the monitoring data of the seismic station instruments to obtain a spatiotemporal feature map.

[0043] Among them, the spatiotemporal fusion deep learning model is a dual-channel spatiotemporal convolutional network, including temporal and spatial channels; specifically, such as Figure 2 As shown, step 130 above may further include steps 210 to 230: Step 210: Extract temporal dependency features using the temporal channel in the spatiotemporal fusion deep learning model.

[0044] The time channel can be a long short-term memory network; furthermore, the time channel extracts temporal dependency features through the long short-term memory network.

[0045] Long Short-Term Memory (LSTM) is a type of recurrent neural network designed to address the long-term dependency problem inherent in general RNNs (Recurrent Neural Networks). All RNNs have a chain-like structure of repeating neural network modules.

[0046] Step 220: Use the spatial channels in the spatiotemporal fusion deep learning model to extract the physical correlation features between seismic station instruments.

[0047] The spatial channel can be a graph convolutional network; furthermore, the spatial channel extracts the physical correlation features between seismic station instruments through the graph convolutional network.

[0048] Graph Convolution Neural Networks (GCNNs) methods are divided into two categories: spectral domain-based methods and spatial domain-based methods. Spectral domain-based methods define graph convolution by introducing filters from the perspective of graph signal processing, where the graph convolution operation is interpreted as removing noise from the graph signal. Spatial domain-based methods represent graph convolution as aggregating feature information from neighbors.

[0049] Step 230: Fuse temporal dependency features and physical correlation features to obtain a spatiotemporal feature map.

[0050] In practice, the two-channel features can be fused at the output layer, namely, the temporal dependency features and the physical correlation features, to generate a unified geophysical state matrix (i.e., the spatiotemporal feature map mentioned above).

[0051] Step 140: Input the spatiotemporal feature map into the trained lightweight SVM model, use the trained lightweight SVM model to perform fault diagnosis based on the spatiotemporal feature map, and obtain the fault diagnosis result.

[0052] This step involves fault diagnosis driven by federated learning.

[0053] The trained lightweight SVM model can be trained locally on each edge node; in practice, the central server can aggregate model parameters to solve the problem of data silos.

[0054] In this step, the trained lightweight SVM model introduces a differential privacy mechanism to protect the privacy of station data; specifically, the update of the trained lightweight SVM model is shown in the following formula (3): (3) in, This represents the model parameters in the t-th iteration; Indicates the learning rate; This represents the gradient of the local loss function of the i-th edge node with respect to the parameters; Indicates the number of participating nodes; This represents Gaussian noise with a mean of 0 and a variance of . Used for differential privacy protection; This indicates noise intensity; the higher the value, the stronger the privacy protection.

[0055] In practice, after obtaining the fault diagnosis results, the fault diagnosis results can be uploaded to the central server, and the central server will periodically distribute the updated trained lightweight SVM model.

[0056] After obtaining the fault diagnosis results, an early warning message can be pushed when the fault diagnosis results show that a fault exists.

[0057] The aforementioned warning information can be pushed via MQTT.

[0058] The seismic station monitoring method based on edge computing and federated learning provided in this application first acquires the communication protocol of a non-standard instrument and the monitoring data of the seismic station instrument. Second, the communication protocol of the non-standard instrument is input into a trained language model, which extracts communication protocol features and generates corresponding rules for protocol conversion based on these features. Third, the monitoring data of the seismic station instrument is input into a spatiotemporal fusion deep learning model, which extracts features from the monitoring data to obtain a spatiotemporal feature map. Finally, the spatiotemporal feature map is input into a trained lightweight SVM model, which performs fault diagnosis based on the spatiotemporal feature map to obtain the fault diagnosis result. Thus, through the deep integration of edge computing, federated learning, and multimodal sensing technologies, the intelligence level of seismic monitoring is significantly improved, providing reliable technical support for disaster prevention and mitigation.

[0059] After introducing the seismic station monitoring method based on edge computing and federated learning according to exemplary embodiments of this disclosure, the following will refer to... Figure 3 An exemplary embodiment of the seismic station monitoring device 300 based on edge computing and federated learning will be described.

[0060] refer to Figure 3 A seismic station monitoring device 300 based on edge computing and federated learning includes: an acquisition module 310 configured to acquire the communication protocol of a non-standard instrument and the monitoring data of the seismic station instrument; a protocol conversion module 320 configured to input the communication protocol of the non-standard instrument into a trained language model, extract the communication protocol features using the trained language model, and generate rules corresponding to the protocol conversion based on the communication protocol features; a feature extraction module 330 configured to input the monitoring data of the seismic station instrument into a spatiotemporal fusion deep learning model, extract features from the monitoring data of the seismic station instrument using the spatiotemporal fusion deep learning model, and obtain a spatiotemporal feature map; and a fault diagnosis module 340 configured to input the spatiotemporal feature map into a trained lightweight SVM model, perform fault diagnosis based on the spatiotemporal feature map using the trained lightweight SVM model, and obtain fault diagnosis results.

[0061] In one implementation, the feature extraction module 330 is configured to: extract temporal dependency features using the temporal channel in the spatiotemporal fusion deep learning model, extract physical correlation features between seismic station instruments using the spatial channel in the spatiotemporal fusion deep learning model, and fuse the temporal dependency features and physical correlation features to obtain a spatiotemporal feature map.

[0062] In one implementation, the time channel extracts temporal dependency features through a long short-term memory network.

[0063] In one implementation, the spatial channel extracts physical correlation features between seismic station instruments using a graph convolutional network.

[0064] In one implementation, the update of the trained lightweight SVM model is shown in the following formula: in, This represents the model parameters in the t-th iteration; Indicates the learning rate; This represents the gradient of the local loss function of the i-th edge node with respect to the parameters; Indicates the number of participating nodes; This represents Gaussian noise with a mean of 0 and a variance of . Used for differential privacy protection; This indicates noise intensity; the higher the value, the stronger the privacy protection.

[0065] In one embodiment, the seismic station monitoring device 300 based on edge computing and federated learning also includes a data correction module, which is configured to: construct a multi-dimensional environmental monitoring system by using a microseismic sensor array and a barometric gradient meter in combination with the sensors in the seismic station instruments, and correct the monitoring data of the seismic station instruments in real time through Kalman filtering.

[0066] In one implementation, the Kalman filter formula is as follows: in, This represents the Kalman filter state estimate at time k; Represents the state transition matrix; This represents the Kalman filter state estimate at time k-1; Indicates Kalman gain; This represents the monitoring data from the seismic station instruments at time k. This represents the monitoring matrix corresponding to the monitoring data from the seismic station instruments.

[0067] In one implementation, the seismic station monitoring device 300 based on edge computing and federated learning also includes a data transmission module, which is configured to upload fault diagnosis results to a central server, and the central server periodically distributes updated trained lightweight SVM models.

[0068] In one implementation, the seismic station monitoring device 300 based on edge computing and federated learning also includes an information push module, which is configured to push early warning information when the fault diagnosis results show that a fault exists.

[0069] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0070] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more microprocessors, or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0071] Figure 4 This is a schematic diagram of a computer device provided in an embodiment of this application. The device can be integrated into a terminal device or a chip of a terminal device. The terminal can be a computing device with data processing capabilities.

[0072] The device includes: processor 401, storage medium 402, and bus 403.

[0073] Storage medium 402 stores program instructions executable by processor 401. When computer device 400 is running, processor 401 communicates with storage medium 402 via bus 403, and processor 401 executes the program instructions to perform the above-described method embodiment. The specific implementation and technical effects are similar and will not be described in detail here.

[0074] Optionally, the present invention also provides a program product, such as a computer-readable storage medium, including a program that, when executed by a processor, is used to perform the above-described method embodiments.

[0075] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0076] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0077] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0078] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0079] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for seismic station monitoring based on edge computing and federated learning, characterized in that, The method comprises the following steps: acquiring the communication protocol of non-standard instruments and the monitoring data of seismic station instruments; inputting the communication protocol of the non-standard instruments into a trained language model, extracting communication protocol features by using the trained language model, and generating protocol conversion corresponding rules according to the communication protocol features; inputting the monitoring data of the seismic station instruments into a spatio-temporal fusion deep learning model, extracting features of the monitoring data of the seismic station instruments by using the spatio-temporal fusion deep learning model, and obtaining a spatio-temporal feature map; inputting the spatio-temporal feature map into a trained lightweight SVM model, performing fault diagnosis according to the spatio-temporal feature map by using the trained lightweight SVM model, and obtaining a fault diagnosis result.

2. The method of claim 1, wherein, The method comprises the following steps: extracting time sequence dependent features by using a time channel in the spatio-temporal fusion deep learning model, extracting physical correlation features between seismic station instruments by using a space channel in the spatio-temporal fusion deep learning model, and fusing the time sequence dependent features and the physical correlation features to obtain the spatio-temporal feature map.

3. The method of claim 2, wherein, The time channel extracts the time sequence dependent features by using a long short-term memory network.

4. The method of claim 2, wherein, The space channel extracts the physical correlation features between seismic station instruments by using a graph convolution network.

5. The method of claim 1, wherein, The trained lightweight SVM model is updated according to the following formula: where, denotes the model parameters of the t-th iteration; denotes the learning rate; denotes the gradient of the i-th edge node local loss function with respect to the parameters; denotes the number of participating nodes; denotes the Gaussian noise with mean 0 and variance for differential privacy protection; denotes the noise intensity, the larger the privacy protection is stronger.

6. The method of claim 1, wherein, After acquiring the communication protocol of non-standard instruments and the monitoring data of seismic station instruments, the method further comprises the following steps: constructing a multi-dimensional environmental monitoring system by using a microseismic sensor array, a barometric gradient meter, and sensors in the seismic station instruments, and correcting the monitoring data of the seismic station instruments in real time by using Kalman filtering.

7. The method of claim 6, wherein, The Kalman filtering formula is as follows: wherein, represents a Kalman filter state estimation value at the kth time; represents a state transition matrix; represents a Kalman filter state estimation value at the (k-1)th time; represents a Kalman gain; represents monitoring data of a seismic station instrument at the kth time; represents a monitoring matrix corresponding to the monitoring data of the seismic station instrument.

8. The method of claim 1, wherein, After obtaining the fault diagnosis result, the method further comprises the following steps: uploading the fault diagnosis result to a central server, and periodically issuing an updated trained lightweight SVM model by the central server.

9. The method of claim 1, wherein, After obtaining the fault diagnosis result, the method further comprises the following steps: when the fault diagnosis result shows that there is a fault, pushing an early warning message.

10. An edge computing and federated learning based seismic station monitoring device, characterized in that, The method comprises the following steps: an acquisition module configured to acquire the communication protocol of non-standard instruments and the monitoring data of seismic station instruments; a protocol conversion module configured to input the communication protocol of the non-standard instruments into a trained language model, extract communication protocol features by using the trained language model, and generate protocol conversion corresponding rules according to the communication protocol features; a feature extraction module configured to input the monitoring data of the seismic station instruments into a spatio-temporal fusion deep learning model, extract features of the monitoring data of the seismic station instruments by using the spatio-temporal fusion deep learning model, and obtain a spatio-temporal feature map; a fault diagnosis module configured to input the spatio-temporal feature map into a trained lightweight SVM model, perform fault diagnosis according to the spatio-temporal feature map by using the trained lightweight SVM model, and obtain a fault diagnosis result.

Images