A method for judging geological hazards based on geological settlement data of power facilities

Abstract

The present invention discloses a method for judging geological hidden dangers based on geological settlement data of power facilities, including an AI model training step and a judging step; the AI model training step includes the following steps: A1. Establish an image feature model for geological hidden danger scenarios; A2. Establish a neural model for judging geological hidden dangers; A3. Store the established image feature model for geological hidden danger scenarios and the neural model for judging geological hidden dangers in an AI recognition library; the judging step includes a sudden shooting step, an accumulated shooting step, and a hidden danger judging step; the hidden danger judging step includes the following steps: D1. Obtain image data with only geological hidden danger features remaining; D2. Output the severity of the hidden danger. The present invention can analyze the causes of geological hidden dangers, can determine the severity of on-site geological hidden dangers, can discover geological hidden dangers found by settlement data, and reduce errors.

Description

Technical Field

[0001] The present invention relates to the technical field of geological hazard judgment, and more specifically, to a geological hazard judgment method based on geological settlement data of power facilities. Background Art

[0002] Currently, the monitoring of geological hazards of power facilities mainly relies on the analysis and judgment of geological settlement data and trends. The settlement monitoring accuracy can reach 2.5 mm (millimeters) relying on the Beidou post-differential positioning technology. However, there are many limitations when using these monitoring result data to determine geological hazards:

[0003] 1. For some sudden geological settlement situations, such as when the settlement data mutates, if only relying on the settlement data, it is impossible to see what happened at the scene. For example, it can be seen from the settlement system that the settlement data from point A to point B has increased sharply, as Figure 1 shown, but it is impossible to directly draw a conclusion about the on-site reason for the sudden settlement of the data; it is also unknown whether there is an error in the settlement data;

[0004] 2. For some situations with a cumulative subsidence trend, where the settlement is the result of geological change movement, for some situations with a cumulative subsidence trend (warning for subsidence exceeding 30 mm), relying solely on the data cannot determine the severity of the on-site situation. Even if someone goes to the site for observation, it is impossible to directly obtain the severity of some potential geological hazards, and it cannot provide a comprehensive scientific basis for taking the next step.

[0005] 3. For some geological hazards, since the settlement data has not mutated and cannot be discovered from the data in a short time, there may be an omission in the hazard judgment. Summary of the Invention

[0006] The purpose of the present invention is to provide a geological hazard judgment method based on geological settlement data of power facilities, to solve the problems mentioned in the background art, to be able to analyze the reasons for the occurrence of geological hazards, to be able to determine the severity of on-site geological hazards, to discover geological hazards found by settlement data, and to reduce errors.

[0007] To achieve the above object, a geological hazard judgment method based on geological settlement data of power facilities is provided, including an AI model training step and a judgment step;

[0008] The AI model training step includes the following steps:

[0009] A1. Establish a geological hazard scene image feature model; the input of the geological hazard scene image feature model is the image data of the geological scene, and the output is the image data with only geological hazard features remaining;

[0010] A2. Establish a neural model for geological hazard judgment; the input of the geological hazard judgment neural model is a hazard data set; the output is the severity of the hazard; the hazard data set is a set composed of image data of geological hazard characteristics and power facility geological settlement data in sequence;

[0011] A3. Store the established geological hazard scene image feature model and geological hazard judgment neural model in the AI recognition library;

[0012] The judgment steps include a sudden shooting step, an accumulated shooting step, and a hazard judgment step;

[0013] The sudden shooting step includes the following sub-steps:

[0014] B1. Set a judgment sudden time interval and a sudden settlement difference threshold; the sudden settlement difference threshold is the maximum settlement difference of the power facility geological settlement data under safe conditions within each time interval;

[0015] B2. If the power facility geological settlement data obtained in the sudden time interval exceeds the settlement difference threshold, then proceed to the next step;

[0016] B3. Trigger the video monitoring device to take a photo as the image data of the geological scene and store it in the AI recognition library, and fix it and display it on the settlement mutation time axis of the settlement system, and record the settlement data occurring at the mutation moment and store it in the AI recognition library; execute the judgment step;

[0017] The accumulated shooting step includes the following sub-steps:

[0018] C1. Set an accumulated settlement difference threshold; the accumulated settlement difference threshold is the maximum settlement difference of the power facility geological settlement data under safe conditions every day;

[0019] C2. Wake up the video monitoring device to take a photo at the same fixed position every day and store it as the image data of the geological scene in the AI recognition library, and perform image position comparison and recognition on different time axes at the same position; if the settlement difference exceeds the accumulated settlement difference threshold, then store the current settlement data occurring at the site in the AI recognition library and execute the judgment step;

[0020] The hazard judgment step includes the following steps:

[0021] D1. Input the newly stored image data of the geological scene into the geological hazard scene image feature model to obtain image data with only geological hazard characteristics remaining;

[0022] D2. Sequentially form a hazard data set with the newly stored image data of the geological scene and the power facility geological settlement data and input it into the geological hazard judgment neural model to output the severity of the hazard.

[0023] Specifically, in step A1, establishing the image feature model for geological hazard scenarios specifically includes the following steps:

[0024] (1) Obtain image data of a large number of geological hazard scenarios;

[0025] (2) Rotate, scale, enhance, and sharpen the image data obtained in step (1); and divide the processed image data into an image training set, an image validation set, and an image test set according to a certain ratio;

[0026] (3) Based on the traditional image feature extraction network, construct an instance segmentation model that can output the image features of geological hazard scenarios;

[0027] (4) Input the image training set into the instance segmentation model in step (3) for parameter training. During the training process, input the image validation set in step (1) into the trained instance segmentation model for verification;

[0028] (5) Receive the input image data into the instance segmentation model to obtain the area that only contains the image features of geological hazard scenarios, and invert and delete the image data of the areas that do not contain the image features of geological hazard scenarios;

[0029] (6) Use the image data that only remains with geological hazard features as the output.

[0030] Specifically, the image data is divided into an image training set, an image validation set, and an image test set according to the ratio of 0.8:0.1:0.1.

[0031] Specifically, establishing the neural model for geological hazard judgment specifically includes the following steps;

[0032] (1) Normalize the settlement data corresponding to the image features of a large number of geological hazard scenarios and the severity levels of hazards determined by experts from the above data, and sequentially form a data set;

[0033] (2) Based on the traditional neural network, construct a neural model for hazard severity that can output the severity level of hazards;

[0034] (3) Divide the data set into a data training set, a data validation set, and a data test set according to a certain ratio;

[0035] (4) Input the data training set in step (3) into the neural model for hazard severity in step (2) for parameter training. During the training process, input the data validation set in step (1) into the trained neural model for hazard severity for verification;

[0036] (6) Receive the input hazard data set into the neural model for hazard severity, and use the severity level of hazards as the output.

[0037] In particular, the data set is divided into a data training set, a data validation set and a data test set in a ratio of 0.8:0.1:0.1.

[0038] In particular, the severity of the hidden dangers includes no hidden dangers, slight hidden dangers, general hidden dangers, medium hidden dangers, and serious hidden dangers.

[0039] Particularly, the burst time interval is 1 hour.

[0040] Particularly, the sudden settlement difference threshold is 2.5 mm.

[0041] Particularly, the cumulative settlement difference threshold is 20 mm.

[0042] The beneficial effects of the present invention are as follows:

[0043] The present invention proposes a method for judging geological hazards based on geological settlement data of power facilities, which adopts AI technology to establish a geological hazard scene image feature model and a geological hazard judgment neural model, and can make internal judgments through geological settlement data of power facilities, and can intuitively judge the changes in the external environment of the site through images in combination with the geological hazard scene image feature model. The two are used synchronously and linked, and the geological hazard scene image feature model and the geological hazard judgment neural model are continuously trained to obtain a more accurate severity of the hazard, which is more scientific and intuitive. The present invention also adopts a sudden shooting step and a cumulative shooting step to obtain image data on site according to different time dimensions, and can respond to sudden geological hazards and discover slowly occurring and difficult-to-find geological hazards through longer time intervals, can analyze the causes of geological hazards, can determine the severity of geological hazards on site, can discover geological hazards discovered by settlement data, and reduce errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in 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 only 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.

[0045] Figure 1 A graph of geological settlement data of power facilities displayed by a settlement system which is the background technology of the present invention.

[0046] Figure 2 Schematic diagram of the method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0048] It should be noted that similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0049] It should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the inventive product is usually placed during use. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0050] In addition, the terms "horizontal", "vertical", "hanging", etc. do not mean that the components are required to be absolutely horizontal or hanging, but may be slightly inclined. For example, "horizontal" only means that its direction is more horizontal relative to "vertical", and does not mean that the structure must be completely horizontal, but may be slightly inclined.

[0051] As Figure 2 shown, a geological hazard judgment method based on geological settlement data of power facilities includes an AI model training step and a judgment step;

[0052] The AI model training step includes the following steps:

[0053] A1. Establish an image feature model for geological hazard scenarios. The input of the image feature model for geological hazard scenarios is the image data of geological scenarios, and the output is the image data with only geological hazard features remaining;

[0054] The establishment of the image feature model for geological hazard scenarios specifically includes the following steps:

[0055] (1) Obtain a large amount of image data of geological hazard scenarios;

[0056] (2) Rotate, scale, enhance, and sharpen the image data obtained in step (1); and divide the processed image data into an image training set, an image validation set, and an image test set according to the ratio of 0.8:0.1:0.1;

[0057] (3)On the basis of the traditional image feature extraction network, construct an instance segmentation model capable of outputting the image features of geological hazard scenarios;

[0058] (4)Input the image training set into the instance segmentation model in step (3) for parameter training. During the training process, input the image validation set in step (1) into the trained instance segmentation model for verification;

[0059] (5)Receive the input image data into the instance segmentation model to obtain the area containing only the image features of geological hazard scenarios, and invert and delete the image data of the area that does not contain the image features of geological hazard scenarios;

[0060] (6)Use the image data with only geological hazard features left as the output.

[0061] A2. Establish a geological hazard judgment neural model. The input of the geological hazard judgment neural model is the hazard data set; the output is the severity of the hazard; the hazard data set is a set composed of the image data of geological hazard features and the geological settlement data of power facilities in sequence; the severity of the hazard includes no hazard, minor hazard, general hazard, medium hazard, and serious hazard.

[0062] The establishment of the geological hazard judgment neural model specifically includes the following steps;

[0063] (1)Normalize the settlement data corresponding to the massive geological hazard scenario image features and the severity level of the hazard determined by experts from the above data, and form a data set in sequence;

[0064] (2)On the basis of the traditional neural network, construct a hazard severity neural model capable of outputting the severity of the hazard;

[0065] (3)Divide the data set into a data training set, a data validation set, and a data test set according to the ratio of 0.8:0.1:0.1;

[0066] (4)Input the data training set in step (3) into the hazard severity neural model in step (2) for parameter training. During the training process, input the data validation set in step (1) into the trained hazard severity neural model for verification;

[0067] (6)Receive the input hazard data set into the hazard severity neural model, and use the severity level of the hazard as the output.

[0068] A3. Store the established geological hazard scenario image feature model and the geological hazard judgment neural model in the AI recognition library;

[0069] The judgment steps include the emergency shooting step, the cumulative shooting step, and the hazard judgment step;

[0070] The emergency shooting steps include the following sub-steps:

[0071] B1. Set the judgment emergency time interval and the emergency settlement difference threshold. The emergency settlement difference threshold is the maximum settlement difference of the geological settlement data of the power facility within each time interval under safe conditions; the emergency time interval is 1 hour. The emergency settlement difference threshold is 2.5 mm.

[0072] B2. If the geological settlement data of the power facility obtained within the emergency time interval exceeds the settlement difference threshold, then proceed to the next step;

[0073] B3. Trigger the video monitoring device to take a photo, store it as image data of the geological scene in the AI recognition library, and fixedly display it on the settlement mutation time axis of the settlement system. Record the settlement data occurring at the mutation moment and store it in the AI recognition library; execute the judgment step;

[0074] The cumulative shooting steps include the following sub-steps:

[0075] C1. Set the cumulative settlement difference threshold; the cumulative settlement difference threshold is the maximum settlement difference of the geological settlement data of the power facility per day under safe conditions; the cumulative settlement difference threshold is 20 mm.

[0076] C2. Wake up the video monitoring device to take a photo at the same fixed position every day and store it as image data of the geological scene in the AI recognition library. Compare and identify the image positions on different time axes at the same position; if the settlement difference exceeds the cumulative settlement difference threshold, then store the current settlement data occurring at the site in the AI recognition library and execute the judgment step;

[0077] The hidden danger judgment steps include the following steps:

[0078] D1. Input the newly stored image data of the geological scene into the geological hidden danger scene image feature model to obtain the image data with only geological hidden danger features remaining;

[0079] D2. Sequentially form a hidden danger data set with the newly stored image data of the geological scene and the geological settlement data of the power facility, and input it into the geological hidden danger judgment neural model to output the severity of the hidden danger.

[0080] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, the patent owner can make various deformations or modifications within the scope of the appended claims. As long as it does not exceed the protection scope described by the claims of the present invention, it should be within the protection scope of the present invention.

Claims

1. A method for judging geological hazards based on geological settlement data of power facilities, characterized in that: it includes an AI model training step and a judgment step; The AI model training step includes the following steps: A1. Establish an image feature model for geological hazard scenarios; the input of the image feature model for geological hazard scenarios is the image data of geological scenarios, and the output is the image data with only geological hazard features remaining; A2. Establish a geological hazard judgment neural model; the input of the geological hazard judgment neural model is a hazard data set; the output is the severity of the hazard; the hazard data set is a set composed of the image data of geological hazard features and the geological settlement data of power facilities in sequence; A3. Store the established image feature model for geological hazard scenarios and the geological hazard judgment neural model in the AI recognition library; The judgment step includes a sudden shooting step, an accumulated shooting step, and a hazard judgment step; The sudden shooting step includes the following sub-steps: B1. Set a judgment sudden time interval and a sudden settlement difference threshold; the sudden settlement difference threshold is the maximum settlement difference of the geological settlement data of power facilities under safe conditions within each time interval; B2. If the geological settlement data of the power facilities obtained within the sudden time interval exceeds the settlement difference threshold, then proceed to the next step; B3. Trigger the video monitoring device to take pictures as the image data of the geological scenario and store them in the AI recognition library, and fixedly display them on the settlement mutation time axis of the settlement system, and record the settlement data occurring at the mutation moment and store it in the AI recognition library; execute the judgment step; The accumulated shooting step includes the following sub-steps: C1. Set an accumulated settlement difference threshold; the accumulated settlement difference threshold is the maximum settlement difference of the geological settlement data of power facilities under safe conditions every day; C2. Wake up the video monitoring device to take pictures at the same fixed position every day and store them as the image data of the geological scenario in the AI recognition library, and perform image position comparison and recognition on different time axes at the same position; if the settlement difference exceeds the accumulated settlement difference threshold, then store the current settlement data occurring at the scene in the AI recognition library and execute the judgment step; The hazard judgment step includes the following steps: D1. Input the newly stored image data of the geological scenario into the image feature model for geological hazard scenarios to obtain the image data with only geological hazard features remaining; D2. Sequentially form a hazard data set with the newly stored image data of the geological scenario and the geological settlement data of power facilities and input it into the geological hazard judgment neural model to output the severity of the hazard.

2. The method for judging geological hazards based on geological settlement data of power facilities according to claim 1, characterized in that: In the step A1, establishing the image feature model for geological hazard scenarios specifically includes the following steps: (1) Obtain a large amount of image data of geological hazard scenarios; (2) Rotate, scale, enhance the image, and sharpen the image data obtained in step (1); and divide the processed image data into an image training set, an image validation set, and an image test set according to a certain ratio; (3) On the basis of a traditional image feature extraction network, construct an instance segmentation model capable of outputting the image features of geological hazard scenarios. (4) Input the image training set into the instance segmentation model in step (3) for parameter training. During the training process, input the image validation set in step (1) into the trained instance segmentation model for validation; (5) Receive the input image data into the instance segmentation model to obtain the region containing only the image features of the geological hazard scenario, and invert and delete the image data of the region that does not contain the image features of the geological hazard scenario; (6) Use the image data with only the geological hazard features left as the output.

3. A method for judging geological hazards based on geological subsidence data of power facilities according to claim 2, characterized in that: The image data is divided into an image training set, an image validation set, and an image test set according to the ratio of 0.8:0.1:0.

1.

4. A method for judging geological hazards based on geological subsidence data of power facilities according to claim 1, characterized in that: The establishment of the geological hazard judgment neural model specifically includes the following steps; (1) Normalize the settlement data corresponding to a large number of geological hazard scenario image features and the severity levels of hazards determined by experts from the above data, and sequentially form a data set; (2) On the basis of a traditional neural network, construct a hazard severity neural model capable of outputting the severity level of hazards; (3) Divide the data set into a data training set, a data validation set, and a data test set according to a certain ratio; (4) Input the data training set in step (3) into the hazard severity neural model in step (2) for parameter training. During the training process, input the data validation set in step (1) into the trained hazard severity neural model for validation; (6) Receive the input hazard data set into the hazard severity neural model, and use the severity level of hazards as the output.

5. A method for judging geological hazards based on geological subsidence data of power facilities according to claim 4, characterized in that: The data set is divided into a data training set, a data validation set, and a data test set according to the ratio of 0.8:0.1:0.

1.

6. A method for judging geological hazards based on geological subsidence data of power facilities according to claim 1, characterized in that: The severity levels of hazards include no hazard, minor hazard, general hazard, medium hazard, and severe hazard.

7. A method for judging geological hazards based on geological subsidence data of power facilities according to claim 1, characterized in that: The sudden time interval is 1 hour.

8. A method for judging geological hazards based on geological subsidence data of power facilities according to claim 1, characterized in that: The sudden settlement difference threshold is 2.5 mm.

9. A method for judging geological hazards based on geological subsidence data of power facilities according to claim 1, characterized in that: The cumulative settlement difference threshold is 20 mm.

Images