Simultaneous decision-making method, device and equipment based on multi-sensing unit model
By configuring different model parameters for each sensing unit and making joint decisions, the problem of insufficient accuracy in identifying disturbance events in complex environments by sensing optical cables is solved, and high-precision and stable event identification is achieved.
Patent Information
- Application Number
- CN202411495454.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In large-scale and complex monitoring scenarios, when the sensing optical cable passes through different geological structures, a single sensing unit is difficult to adapt to the changing environment, resulting in insufficient accuracy in identifying disturbance event types.
A multi-sensor unit model is adopted. By configuring different model parameters for each sensor unit and combining the joint decision-making method, the disturbance signals of each sensor unit are processed and regional prediction is performed.
It improves the accuracy and robustness of disturbance event identification in complex environments, reduces the uncertainty of individual sensing unit models, and enhances the accuracy and stability of identification.
Smart Images

Figure CN119025923A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of pattern recognition technology, and more specifically, relates to a joint decision-making method, device and equipment based on a multi-sensor unit model. Background Art
[0002] The external disturbance event pattern recognition technology based on distributed fiber optic sensing is widely used in pipeline leakage detection, perimeter security, power transmission cable monitoring, urban infrastructure safety monitoring, underwater detection and other application fields. Based on the distributed fiber optic sensing technology, a large range of optical fibers are laid to detect abnormal events in the monitoring area in real time. When a disturbance event occurs in the outside world, the vibration signal generated by it is transmitted to the optical fiber through the medium in the form of a mechanical wave, and the optical signal in the optical fiber is modulated. The detection of external disturbance events can be achieved through the corresponding demodulation technology.
[0003] In order to achieve high-precision pattern recognition of external disturbance event types, artificial intelligence-based pattern recognition technology is integrated with distributed fiber optic sensing technology. The pattern recognition technology is used to process and analyze different categories of feature information corresponding to various target events to achieve classification and recognition of different types of event patterns.
[0004] However, in actual engineering applications, especially in large-scale and complex monitoring scenarios, sensor optical cables will pass through various complex geological structures during the actual laying process, including soft soil, cement ground structure, and sand and gravel structure. In addition, due to the large differences in the external monitoring environment, the sensor optical cables at different locations have large differences. If only the data detected by a single sensor unit is considered or only the data detected by the sensor unit is analyzed and processed according to a single recognition algorithm model, it will be difficult to adapt to the high-precision recognition requirements of various types of disturbance events in different complex environmental scenarios. Summary of the invention
[0005] In view of the defects of the prior art, the purpose of this application is to achieve accurate identification of various disturbance events in different complex environmental scenarios.
[0006] To achieve the above objectives, in a first aspect, the present application provides a joint decision-making method based on a multi-sensor unit model, comprising: Based on the sensor unit model corresponding to each sensor unit, the disturbance signal at each sensor unit is processed to obtain the unit prediction result corresponding to each sensor unit. The sensor unit is determined by dividing the sensing optical fiber of the target monitoring area into multiple sections. The unit prediction result is used to characterize the event occurring at the corresponding sensor unit. The sensor unit models corresponding to different sensor units have different model parameters. Based on the unit prediction results corresponding to each sensor unit, a joint decision is made to determine the regional prediction results, which are used to characterize the events occurring in the target monitoring area.
[0007] In a possible implementation, the sensor unit model corresponding to the sensor unit is obtained through the following steps: By collecting disturbance signal samples of the sensing unit, a first sensing data set corresponding to the sensing unit is constructed, wherein the first sensing data set includes a plurality of disturbance signal samples and event labels corresponding to each disturbance signal sample; Model training is performed based on a first sensor data set corresponding to the sensor unit to obtain a sensor unit model corresponding to the sensor unit.
[0008] In a possible implementation, performing model training based on a first sensor data set corresponding to the sensor unit to obtain a sensor unit model corresponding to the sensor unit includes: Based on the pre-trained sensor unit model and the first sensor data set corresponding to the sensor unit, transfer learning is performed to obtain the sensor unit model corresponding to the sensor unit.
[0009] In one possible implementation, the pre-trained sensor unit model is obtained by the following steps: constructing a second sensing data set based on disturbance signal samples and corresponding event labels of a plurality of monitoring areas, wherein the plurality of monitoring areas do not overlap with the target monitoring area; Based on the second sensor data set, the initial sensor unit model is trained to obtain a pre-trained sensor unit model.
[0010] In a possible implementation, the unit prediction result includes the confidence of each event category, and the regional prediction result is determined by joint decision-making based on the unit prediction results corresponding to each sensor unit, including: Based on the confidence threshold, a target unit prediction result is selected from the unit prediction results corresponding to the plurality of sensor units, and the maximum confidence value of the target unit prediction result is greater than the confidence threshold; Based on the prediction results of each screened target unit, a joint decision is made to determine the regional prediction results.
[0011] In a possible implementation, the joint decision making is a weighted average method based on the confidence of the predicted event.
[0012] In a second aspect, the present application provides a joint decision-making device based on a multi-sensor unit model, comprising: A prediction module is used to process disturbance signals at each sensing unit based on a sensing unit model corresponding to each sensing unit, and obtain a unit prediction result corresponding to each sensing unit. The sensing unit is determined by dividing the sensing optical fiber of the target monitoring area into multiple sections. The unit prediction result is used to characterize the event occurring at the corresponding sensing unit. The sensing unit models corresponding to different sensing units have different model parameters. The joint decision module is used to make joint decisions based on the unit prediction results corresponding to each sensor unit to determine the regional prediction results, and the regional prediction results are used to characterize the events occurring in the target monitoring area.
[0013] In a third aspect, the present application provides an electronic device comprising: at least one memory for storing programs; and at least one processor for executing the programs stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method described in the first aspect or any possible implementation of the first aspect.
[0014] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method described in the first aspect or any possible implementation of the first aspect.
[0015] In a fifth aspect, the present application provides a computer program product. When the computer program product runs on a processor, the processor executes the method described in the first aspect or any possible implementation of the first aspect.
[0016] In general, the above technical solutions conceived by this application have the following beneficial effects compared with the prior art: (1) The external environment scenes at different sensor units are different. By configuring different sensor unit models for different sensor units, it can be ensured that the sensor unit model corresponding to the sensor unit can adapt to the external environment scene in which it is located, thereby enhancing the adaptability of the prediction method to complex environments.
[0017] (2) Based on the unit prediction results corresponding to each sensor unit, joint decision-making can reduce the uncertainty of the prediction of a single sensor unit model, improve the accuracy of the regional prediction results, and improve the robustness of the prediction method, thereby achieving accurate identification of various disturbance events in different complex environmental scenarios.
[0018] (3) By migrating the pre-trained sensor unit model to the sensor unit model corresponding to each sensor unit, the computational complexity of learning and training the common parameters in each sensor unit model can be effectively reduced, thereby improving the training efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 It is a flowchart of a joint decision-making method based on a multi-sensor unit model provided in an embodiment of the present application; Figure 2 It is a structural schematic diagram of a joint decision-making device based on a multi-sensor unit model provided in an embodiment of the present application; Figure 3It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0020] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0021] The terms "first", "second", etc. in the specification and claims herein are used to distinguish different objects rather than to describe a specific order of the objects. For example, a first sensor data set and a second sensor data set are used to distinguish different sensor data sets rather than to describe a specific order of response messages.
[0022] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.
[0023] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc.; multiple elements refer to two or more elements, etc.
[0024] The embodiments of the present application are described below in conjunction with the drawings in the embodiments of the present application.
[0025] Figure 1 is a flow chart of a joint decision-making method based on a multi-sensor unit model provided in an embodiment of the present application, such as Figure 1 As shown, the execution subject of the method may be an electronic device, such as a server, etc. The method includes the following steps S101 and S102.
[0026] Step S101, based on the sensor unit model corresponding to each sensor unit, the disturbance signal at each sensor unit is processed to obtain the unit prediction result corresponding to each sensor unit. The sensor unit is determined by dividing the sensing optical fiber in the target monitoring area into multiple sections. The unit prediction result is used to characterize the event occurring at the corresponding sensor unit. The sensor unit models corresponding to different sensor units have different model parameters.
[0027] Step S102, based on the unit prediction results corresponding to each sensor unit, a joint decision is made to determine a regional prediction result, and the regional prediction result is used to characterize the events occurring in the target monitoring area.
[0028] The above-mentioned sensing unit is explained here. In order to detect the disturbance condition of the target monitoring area, a sensing optical fiber can be laid in the target monitoring area and divided into multiple sections, each section of the optical fiber being a sensing unit.
[0029] For example, the length of the sensing fiber is L, and the sensing fiber is divided into N segments. In this case, the spatial distance corresponding to each sensing unit is .
[0030] By demodulating the phase of the backward Rayleigh scattered light signal at different sensor units through a phase demodulation algorithm, the external disturbance signal corresponding to each sensor unit can be obtained, and the disturbance signal can characterize the vibration and stress changes at the sensor unit.
[0031] The above-mentioned unit prediction results are explained here. The external disturbance signals corresponding to each sensor unit are input into the sensor unit model corresponding to each sensor unit. The sensor unit model can extract and analyze the features of the external disturbance signals to predict the events occurring at the sensor unit, and then the unit prediction results corresponding to each sensor unit can be obtained.
[0032] For example, when the environment in which the sensor unit is located is a wall, the sensor unit model corresponding to the sensor unit can predict whether an event of climbing over the wall occurs at the sensor unit. The unit prediction result can be the confidence corresponding to the event category of climbing over the wall and the confidence corresponding to other event categories. The higher the confidence, the greater the probability of the event category occurring.
[0033] For example, when the environment in which the sensor unit is located is a road, the sensor unit model corresponding to the sensor unit can predict whether a vehicle passing event occurs at the sensor unit. The unit prediction result can be the confidence corresponding to the vehicle passing event category and the confidence corresponding to other event categories. The higher the confidence, the greater the probability of the event category occurring.
[0034] The sensor unit model can specifically be a convolutional neural network (CNN), a long short-term memory network (LSTM), a network constructed by combining CNN and LSTM, etc.
[0035] It is understandable that there are differences in the external environmental scenes at different sensor units. By configuring different sensor unit models for different sensor units (the sensor unit models corresponding to different sensor units have different model parameters), it can be ensured that the sensor unit model corresponding to the sensor unit can adapt to the external environmental scene in which it is located, so as to enhance the adaptability of the prediction method to complex environments; at the same time, based on the unit prediction results corresponding to each sensor unit, joint decision-making is made, which can reduce the uncertainty of the prediction of a single sensor unit model, improve the accuracy of the regional prediction results and improve the robustness of the prediction method, so as to achieve accurate identification of various disturbance events in different complex environmental scenes.
[0036] Therefore, the joint decision-making method based on the multi-sensing unit model provided in the present application can identify various vibration source events in the surrounding area based on distributed fiber optic sensing technology using sensor optical cables laid over a large area. It can be applicable to the fields of urban infrastructure, communication optics, cables, long-distance oil and gas pipelines, and security monitoring of regional perimeters. It has the characteristics of a wide monitoring area, high recognition accuracy, high real-time performance, and high stability.
[0037] In a possible implementation, the sensor unit model is obtained through the following steps: By collecting disturbance signal samples of the sensing unit, a first sensing data set corresponding to the sensing unit is constructed, wherein the first sensing data set includes a plurality of disturbance signal samples and event labels corresponding to each disturbance signal sample; Model training is performed based on a first sensor data set corresponding to the sensor unit to obtain a sensor unit model corresponding to the sensor unit.
[0038] Specifically, the disturbance signal of the sensor unit collected within a specified time period can be used as a signal sample, and the events corresponding to each disturbance signal sample can be marked manually to obtain an event label corresponding to each disturbance signal sample, and the event label is used to indicate the event represented by the disturbance signal sample. Then, based on multiple disturbance signal samples and the event labels corresponding to each disturbance signal sample, a first sensor data set corresponding to the sensor unit can be constructed, and then, based on the first sensor data set, supervised training can be performed to obtain a sensor unit model corresponding to the sensor unit.
[0039] For example, assume that the number of event categories at the sensing unit is c , you can j The disturbance signal samples of the event-like type are put into the sample set , , the sample sets corresponding to various events are grouped together to construct the first sensor data set corresponding to the sensor unit ,in i Indicates the number of the sensor unit.
[0040] In a possible implementation, the above-mentioned performing model training based on the first sensor data set corresponding to the sensor unit to obtain the sensor unit model corresponding to the sensor unit includes: Based on the pre-trained sensor unit model and the first sensor data set corresponding to the sensor unit, transfer learning is performed to obtain the sensor unit model corresponding to the sensor unit.
[0041] Specifically, the pretrained sensor unit model includes multiple network layers, and the first several layers in the pretrained sensor unit model can be used as feature extractors, and the network layers other than the feature extractors in the pretrained sensor unit model can be used as classifiers. For example, the pretrained sensor unit model includes N network layers, and the first k (1<k<N) layers of the pretrained sensor unit model can be used as feature extractors, and the (k+1)th layer to the Nth layer of the pretrained sensor unit model can be used as classifiers. The classifier can be adaptively adjusted so that the number of classifications of the classifier is equal to the number of event categories of the environment in which the sensor unit is located, and the model parameters of the classifier can be (randomly) initialized. After adjusting and initializing the classifier, the pretrained sensor unit model can be fine-tuned based on the first sensor data set corresponding to the sensor unit.
[0042] Optionally, during the fine-tuning process, the layers of the classifier can be trained based on freezing the model parameters of the feature extractor to retain the common features learned by the pre-trained sensor unit model.
[0043] Optionally, during the fine-tuning process, at the beginning of fine-tuning, the various layers of the classifier can be trained on the basis of freezing the model parameters of the feature extractor; as the training proceeds, more layers can be gradually unfrozen as needed, and the order of unfreezing is usually starting from the terminal network layer of the feature extractor and gradually moving to the input end of the network in the feature extractor. During the fine-tuning process, the network layers of the feature extractor and the classifier can use different learning rates. The learning rate of the network layer of the feature extractor is relatively small to avoid destroying the features that have been learned, while the learning rate of the network layer of the classifier is relatively large to quickly adapt to new tasks.
[0044] It is understandable that the pre-trained sensor unit model can be a model that has been pre-trained on a large-scale data set, and the parameters of the feature extractor in the pre-trained sensor unit model include model common parameters (parameters shared by different sensor unit models). For example, the model common parameters include: detection system common mode noise parameters and signal feature extraction shared parameters. The detection system common mode noise parameters can be used to remove the common mode noise introduced by the detection system in each sensor unit model, and the signal feature extraction shared parameters can be used to pre-process the disturbance signal and extract the common features between the disturbance signals of different sensor units, providing a basis for subsequent classification by the classifier.
[0045] After fine-tuning the pre-trained sensor unit model, the fine-tuned parameters in the feature extractor and the parameters in the classifier constitute the model characteristic parameters (parameters unique to a sensor unit model). For example, the model characteristic parameters include: noise suppression parameters in the environment where the sensor unit is located and specific signal feature extraction parameters in the environment where the sensor unit is located. The noise suppression parameters in the environment where the sensor unit is located can be used to remove noise interference in the environment where the sensor unit is located. The specific signal feature extraction parameters in the environment where the sensor unit is located can be used to analyze and extract specific signal features in the environment where the sensor unit is located.
[0046] Therefore, by migrating the pre-trained sensor unit model to the sensor unit model corresponding to each sensor unit, the amount of computation required for learning and training common parameters in each sensor unit model can be effectively reduced, thereby improving training efficiency. At the same time, through the above-mentioned fine-tuning, the adaptability of each sensor unit model to its respective environment scene can be improved, thereby improving the accuracy of the sensor unit model.
[0047] In one possible implementation, the pre-trained sensor unit model is obtained by the following steps: constructing a second sensing data set based on disturbance signal samples and corresponding event labels of a plurality of monitoring areas, wherein the plurality of monitoring areas do not overlap with the target monitoring area; Based on the second sensor data set, the initial sensor unit model is trained to obtain a pre-trained sensor unit model.
[0048] In a possible implementation, the unit prediction result includes the confidence of each event category. The above-mentioned unit prediction results corresponding to each sensor unit are used to make a joint decision to determine the regional prediction result, including: Based on the confidence threshold, the target unit prediction result is screened out from the unit prediction results corresponding to the multiple sensor units (the unit prediction results with low confidence are excluded to avoid the unit prediction results with low confidence affecting the accuracy of the joint decision), and the maximum confidence value of the target unit prediction result is greater than the confidence threshold; Based on the prediction results of each screened target unit, the regional prediction results are determined through joint decision-making.
[0049] Here, an example is given to illustrate the maximum confidence value of the unit prediction result. For example, the unit prediction result includes the confidence values corresponding to three event categories, where the confidence value corresponding to the first event category is 0.2, the confidence value corresponding to the second event category is 0.7, and the confidence value corresponding to the third event category is 0.1. Then the maximum confidence value of the unit prediction result is 0.7.
[0050] In a possible implementation, the joint decision-making is specifically a voting method or a weighted average method based on the confidence of the predicted event.
[0051] Optionally, when a voting method is adopted, the event categories corresponding to the maximum confidence values in the prediction results of each target unit may be counted, and the event category with the largest number of occurrences may be selected as the target prediction result.
[0052] Optionally, when adopting the weighted average method based on the confidence of the predicted event, the unit prediction result of each sensor unit is assigned a weight, and based on the weight ratio between the unit prediction results of different sensor units, the screened target unit prediction results are weighted averaged to obtain the target prediction result.
[0053] It can be understood that through joint decision-making, the uncertainty of the prediction of a single sensor unit model can be reduced, the accuracy of the regional prediction results can be improved, and the robustness of the prediction method can be improved.
[0054] The joint decision-making device based on a multi-sensor unit model provided in the present application is described below. The joint decision-making device based on a multi-sensor unit model described below and the joint decision-making method based on a multi-sensor unit model described above can be referenced to each other.
[0055] Figure 2 is a structural diagram of a joint decision-making device based on a multi-sensor unit model provided in an embodiment of the present application, such as Figure 2 As shown, the device includes: a prediction module 10 and a joint decision module 20.
[0056] The prediction module 10 is used to process the disturbance signal at each sensing unit based on the sensing unit model corresponding to each sensing unit, and obtain the unit prediction result corresponding to each sensing unit. The sensing unit is determined by dividing the sensing optical fiber of the target monitoring area into multiple sections. The unit prediction result is used to characterize the event occurring at the corresponding sensing unit. The sensing unit models corresponding to different sensing units have different model parameters. The joint decision module 20 is used to make joint decisions based on the unit prediction results corresponding to each sensor unit to determine the regional prediction results, and the regional prediction results are used to characterize the events occurring in the target monitoring area.
[0057] It can be understood that the detailed functional implementation of each of the above-mentioned units / modules can be found in the introduction of the aforementioned method embodiment, and will not be repeated here.
[0058] It should be understood that the above-mentioned device is used to execute the method in the above-mentioned embodiment. The implementation principle and technical effect of the corresponding program module in the device are similar to those described in the above-mentioned method. The working process of the device can refer to the corresponding process in the above-mentioned method, which will not be repeated here.
[0059] Based on the method in the above embodiment, an embodiment of the present application provides an electronic device, Figure 3 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application, such as Figure 3 As shown, the electronic device may include: a processor (Processor) 810, a communication interface (Communications Interface) 820, a memory (Memory) 830 and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call the logic instructions in the memory 830 to execute the method in the above embodiment.
[0060] In addition, the logic instructions in the above-mentioned memory 830 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 application, 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 a number of 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 application.
[0061] Based on the method in the above embodiment, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method in the above embodiment.
[0062] Based on the method in the above embodiment, an embodiment of the present application provides a computer program product. When the computer program product runs on a processor, the processor executes the method in the above embodiment.
[0063] It is understandable that the processor in the embodiment of the present application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.
[0064] The method steps in the embodiments of the present application can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.
[0065] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions may be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.
[0066] It should be understood that the various numerical numbers involved in the embodiments of the present application are only used for the convenience of description and are not used to limit the scope of the embodiments of the present application.
[0067] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.
Claims
1. A joint decision-making method based on a multi-sensor unit model, characterized in that: include: Based on the sensor unit model corresponding to each sensor unit, the disturbance signal at each sensor unit is processed to obtain the unit prediction result corresponding to each sensor unit, wherein the sensor unit is determined by dividing the sensing optical fiber of the target monitoring area into multiple sections, and the unit prediction result is used to characterize the event occurring at the corresponding sensor unit, and the sensor unit models corresponding to different sensor units have different model parameters; Based on the unit prediction results corresponding to each sensor unit, a joint decision is made to determine the regional prediction result, and the regional prediction result is used to characterize the events occurring in the target monitoring area.
2. The simultaneous decision-making method based on the multi-sensor unit model according to claim 1, characterized in that: The sensor unit model corresponding to the sensor unit is obtained by the following steps: By collecting disturbance signal samples of the sensing unit, a first sensing data set corresponding to the sensing unit is constructed, wherein the first sensing data set includes a plurality of disturbance signal samples and event labels corresponding to each disturbance signal sample; Model training is performed based on a first sensor data set corresponding to the sensor unit to obtain a sensor unit model corresponding to the sensor unit.
3. The simultaneous decision-making method based on the multi-sensor unit model according to claim 2 is characterized in that: The performing model training based on the first sensor data set corresponding to the sensor unit to obtain the sensor unit model corresponding to the sensor unit includes: Based on the pre-trained sensor unit model and the first sensor data set corresponding to the sensor unit, transfer learning is performed to obtain the sensor unit model corresponding to the sensor unit.
4. The simultaneous decision-making method based on the multi-sensor unit model according to claim 3 is characterized in that: The pre-trained sensor unit model is obtained by the following steps: constructing a second sensing data set based on disturbance signal samples and corresponding event labels of a plurality of monitoring areas, wherein the plurality of monitoring areas do not overlap with the target monitoring area; Based on the second sensor data set, the initial sensor unit model is trained to obtain the pre-trained sensor unit model.
5. The simultaneous decision-making method based on a multi-sensor unit model according to any one of claims 1 to 4, characterized in that: The unit prediction result includes the confidence of each event category. The unit prediction result corresponding to each sensor unit is determined by joint decision making to determine the regional prediction result, including: Based on a confidence threshold, a target unit prediction result is screened out from unit prediction results respectively corresponding to a plurality of sensing units, wherein the maximum confidence value of the target unit prediction result is greater than the confidence threshold; Based on the prediction results of each screened target unit, a joint decision is made to determine the regional prediction result.
6. The simultaneous decision-making method based on the multi-sensor unit model according to claim 5, characterized in that: The joint decision making is a weighted average method based on the confidence of the predicted events.
7. A joint decision-making device based on a multi-sensor unit model, characterized in that: include: A prediction module, for processing disturbance signals at each sensing unit based on a sensing unit model corresponding to each sensing unit, and obtaining a unit prediction result corresponding to each sensing unit, wherein the sensing unit is determined by dividing the sensing optical fiber of the target monitoring area into multiple sections, and the unit prediction result is used to characterize an event occurring at the corresponding sensing unit, and the sensing unit models corresponding to different sensing units have different model parameters; The joint decision module is used to make joint decisions based on the unit prediction results corresponding to each sensor unit to determine the regional prediction results, and the regional prediction results are used to characterize the events occurring in the target monitoring area.
8. An electronic device, characterized in that: include: at least one memory for storing a computer program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 1 to 6.
9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program runs on a processor, the processor is caused to execute the method according to any one of claims 1 to 6.
10. A computer program product, characterized in that When the computer program product runs on a processor, the processor is caused to execute the method according to any one of claims 1 to 6.
Citation Information
Patent Citations
Intelligent health monitoring self-repairing system and method for fiber grating sensor network
CN106998226A
Intelligent data prediction method and device, computer equipment and storage medium
CN110348536A
System and method for estimating interference of mechanical equipment on distributed optical fiber sensor
CN111538246A
Distributed optical fiber disturbance positioning method based on regression probability distribution
CN112161778A
Monitoring and early warning method based on distributed optical fiber sound wave sensing and related device
CN117275209A
