CN117877232A - Geological disaster early warning method and system based on GIS - Google Patents
Geological disaster early warning method and system based on GIS Download PDFInfo
- Publication number
- CN117877232A CN117877232A CN202410045126.4A CN202410045126A CN117877232A CN 117877232 A CN117877232 A CN 117877232A CN 202410045126 A CN202410045126 A CN 202410045126A CN 117877232 A CN117877232 A CN 117877232A
- Authority
- CN
- China
- Prior art keywords
- model
- geological disaster
- sub
- cost function
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012545 processing Methods 0.000 claims description 31
- 230000008569 process Effects 0.000 claims description 13
- 230000002708 enhancing effect Effects 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 11
- 231100001267 hazard identification Toxicity 0.000 claims description 8
- 238000013461 design Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 51
- 230000000875 corresponding effect Effects 0.000 description 42
- 238000004364 calculation method Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000000844 transformation Methods 0.000 description 2
- 238000006424 Flood reaction Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013434 data augmentation Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 229920006395 saturated elastomer Polymers 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Landscapes
- Alarm Systems (AREA)
Abstract
The invention discloses a geological disaster early warning method and a geological disaster early warning system based on GIS, comprising the following steps: firstly, acquiring original GIS data of a target area, and extracting a target remote sensing image from the data. And then, inputting the target remote sensing images into a pre-trained geological disaster recognition model to obtain the current disaster type of the target area. And finally, generating early warning information aiming at the target area according to the recognized disaster type. By means of the design, the remote sensing image can be analyzed through the specially trained model, potential geological disasters can be identified and predicted more accurately, GIS data can be acquired and processed in real time, disaster early warning information can be generated and released rapidly, decision basis is provided for corresponding personnel, precautionary measures are taken timely, and loss caused by the disasters is reduced.
Description
Technical Field
The invention relates to the field of geological disaster monitoring, in particular to a geological disaster early warning method and system based on GIS.
Background
Geological disasters are common and strong natural disasters, and early warning of the natural disasters has important significance in preventing and reducing losses caused by the disasters. However, conventional geological disaster warning methods generally rely on manual field inspection or statistical prediction based on historical data, which are inefficient, inaccurate and unable to respond in real time.
Disclosure of Invention
The invention aims to provide a geological disaster early warning method and system based on GIS.
In a first aspect, an embodiment of the present invention provides a geological disaster early warning method based on GIS, including:
acquiring original GIS data of a target area, and acquiring a target remote sensing image from the original GIS data;
inputting the target remote sensing image into a pre-trained target geological disaster identification model to obtain a current disaster type corresponding to the target area;
and generating early warning information for the target area according to the disaster type.
In a second aspect, an embodiment of the present invention provides a server system, including a server, where the server is configured to execute the geological disaster early warning method based on GIS in the first aspect.
Compared with the prior art, the invention has the beneficial effects that: by adopting the geological disaster early warning method and system based on the GIS, disclosed by the invention, the original GIS data of the target area are obtained, and the target remote sensing image is extracted from the data. And then, inputting the target remote sensing images into a pre-trained geological disaster recognition model to obtain the current disaster type of the target area. And finally, generating early warning information aiming at the target area according to the recognized disaster type. By means of the design, the remote sensing image can be analyzed through the specially trained model, potential geological disasters can be identified and predicted more accurately, GIS data can be acquired and processed in real time, disaster early warning information can be generated and released rapidly, decision basis is provided for corresponding personnel, precautionary measures are taken timely, and loss caused by the disasters is reduced.
Drawings
In order to more clearly illustrate the technical solution of the embodiments of the present invention, the drawings that are required to be used in the embodiments will be briefly described. It is appreciated that the following drawings depict only certain embodiments of the invention and are therefore not to be considered limiting of its scope. Other relevant drawings may be made by those of ordinary skill in the art without undue burden from these drawings.
Fig. 1 is a schematic block diagram of a step flow of a geological disaster early warning method based on a GIS according to an embodiment of the present invention;
fig. 2 is a schematic block diagram of a computer device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It will be apparent that the described embodiments are some, but not all, embodiments of the invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
The following describes specific embodiments of the present invention in detail with reference to the drawings.
In order to solve the technical problems in the foregoing background art, fig. 1 is a schematic flow chart of a geological disaster early warning method based on GIS according to an embodiment of the present disclosure, and the geological disaster early warning method based on GIS is described in detail below.
Step S201, acquiring original GIS data of a target area, and acquiring a target remote sensing image from the original GIS data;
step S202, inputting the target remote sensing image into a pre-trained target geological disaster identification model to obtain a current disaster type corresponding to the target area;
step S203, generating early warning information for the target area according to the disaster type.
In the present embodiment, it is assumed, by way of example, that a potential landslide hazard area including various geographical information such as land coverage, elevation, rainfall, etc. is being studied. Raw GIS data for the area, such as satellite images, digital Elevation Models (DEMs), and land utilization data, are first collected. Then, a target remote sensing image is extracted according to the original GIS data, for example, a satellite image is converted into a required form by using a remote sensing image processing technology. A deep learning model has been trained with a large number of geological disaster image datasets to automatically identify different types of geological disasters, such as landslides, debris flows, and the like. The previously acquired target remote sensing image is now entered into the pre-trained geologic hazard recognition model. The model analyzes the characteristics in the image, compares the characteristics with known geological disaster types, and finally outputs the current disaster type corresponding to the target area. According to the current disaster type obtained in the previous step, corresponding early warning rules and strategies can be formulated according to previous research and experience. For example, if the current disaster type of the target area is identified as a landslide, it may be determined whether the current geological condition reaches a critical point of occurrence of the landslide based on the history data and the information of the monitored site. If so, pre-warning information for the area may be generated, including possible pre-warning levels, suggested emergency actions to be taken, etc., so that staff and residents can react in time.
In one possible implementation, the target geological disaster identification model is obtained in the following manner.
(1) Acquiring a sample remote sensing image and a geological disaster recognition model, wherein the sample remote sensing image is configured with a disaster type label, and the geological disaster recognition model comprises at least two geological disaster recognition sub-models;
(2) Respectively executing identification operation on the sample remote sensing image based on each geological disaster identification sub-model to obtain the prediction characteristics and the geological disaster types respectively determined by each geological disaster identification sub-model;
(3) Obtaining a target cost function according to the first cost function and the second cost function respectively corresponding to each geological disaster identification sub-model; executing a training process on each geological disaster recognition sub-model according to the target cost function to obtain a target geological disaster recognition model;
the first cost function corresponding to the first geological disaster recognition sub-model is used for measuring deviation between a geological disaster type determined by the first geological disaster recognition sub-model and the disaster type label, the second cost function corresponding to the first geological disaster recognition sub-model is used for measuring deviation between a first prediction feature determined by the first geological disaster recognition sub-model and each second prediction feature, the first geological disaster recognition sub-model is any geological disaster recognition sub-model in each geological disaster recognition sub-model, and each second prediction feature is a prediction feature determined by each geological disaster recognition sub-model except the first geological disaster recognition sub-model in each geological disaster recognition sub-model.
In the embodiment of the invention, a large number of sample remote sensing images of different types are collected in the research of a geological disaster early warning method, and the images cover various geological disaster situations. At the same time, a geologic hazard recognition model is also prepared, which is composed of a plurality of geologic hazard recognition sub-models, each of which is dedicated to recognizing a particular type of geologic hazard. In order to train and evaluate the models, manual labeling is carried out on the sample remote sensing images, and corresponding disaster type labels are added to each image. Each geological disaster identification sub-model is respectively applied to the sample remote sensing image, and by executing the identification operation, each sub-model can determine the prediction characteristics and the geological disaster type. For example, the first sub-model may determine prediction features according to features such as texture, color, shape, etc. in the image, and predict a type of geological disaster such as landslide; the second sub-model may be concerned with the water distribution and topography relief features in the image and predict the type of flood, a geological disaster. In order to improve the accuracy and robustness of the geologic hazard identification model, a pair of cost functions is defined for each geologic hazard identification sub-model. The first cost function is used for measuring deviation between the geological disaster type determined by the sub-model and the disaster type label of the sample remote sensing image. The second cost function is used for measuring deviation between the predicted features determined by the sub-model and the predicted features determined by other geological disaster recognition sub-models. According to the cost functions, a target cost function can be calculated, and training and optimizing are carried out on each geological disaster recognition sub-model by using the cost function so as to obtain a final target geological disaster recognition model.
It is noted that when the geological disaster recognition sub-model performs a recognition operation on the sample remote sensing image, a prediction feature is generated. The prediction feature refers to a numerical value or a feature vector calculated by the sub-model through an algorithm or a model according to an input remote sensing image, and is used for describing the features of geological disasters existing in the image. For example, assume that satellite images are being used for geological disaster warning, where one geological disaster identification sub-model is dedicated to identifying landslides. For this sub-model, it may determine the predicted features by analyzing features such as texture, color, and shape in the image. In particular, the sub-model may calculate the degree of texture variation of the surface of the earth in the image, such as a variegated, uneven slope region, with a greater texture variation than a flat, stable region. In addition, the sub-model may also analyze color distribution in the image, e.g., landslide areas may exhibit darker or more saturated color features. Meanwhile, the submodel may also focus on shape information of an object in the image, for example, a landslide region generally has an abnormal shape such as a depression or a crack. These calculation and analysis processes will constitute predictive features in the geologic hazard identification submodel. The dimensions and types of predictive features will vary depending on the algorithm and model used and may be a one-dimensional or multi-dimensional numerical vector. In the above example, each geologic hazard identification sub-model may generate its own unique predictive features. These features describe the basis of the sub-model for the determination of a particular geological disaster type. By comparing the prediction features generated by the different sub-models, the difference and similarity between the sub-models can be evaluated, and training and optimization can be performed according to the objective cost function so as to obtain a more accurate and reliable geological disaster recognition model.
In a possible implementation manner, the step of obtaining the target cost function according to the first cost function and the second cost function corresponding to each geological disaster identification sub-model respectively may be implemented by the following example implementation.
(1) Acquiring a diversity enhancing cost function, wherein the diversity enhancing cost function is used for enhancing diversity among the prediction features determined by each geological disaster recognition sub-model;
(2) And obtaining the target cost function according to the first cost function, the second cost function and the diversity enhancing cost function which are respectively corresponding to each geological disaster identification sub-model.
In the embodiment of the invention, the diversity of the geological disaster identification model is improved in an exemplary geological disaster early warning method. For this purpose, a diversity enhancing cost function is designed, which aims to enhance the diversity between the individual geologic hazard recognition sub-models by constraint and optimization. In particular, this diversity-enhancing cost function may be calculated based on the degree of difference in predicted features between the sub-models to facilitate more unique and diversified determination of predicted features for each sub-model. Based on the previous example, it is assumed that there are two geologic hazard recognition sub-models for the recognition of landslide and flood, respectively. For the landslide sub-model, defining a first cost function to measure the deviation between the predicted result and the label; meanwhile, a second cost function is defined to measure the difference between the landslide sub-model prediction features and the flood sub-model prediction features. In addition, a diversity enhancing cost function is introduced to enhance the diversity of the predicted features between the two sub-models. By combining these cost functions, an objective cost function can be obtained. The accuracy, the prediction characteristic difference and the diversity of each geological disaster recognition sub-model are comprehensively considered by the target cost function. By minimizing the objective cost function, each sub-model can be optimized through a training process to achieve better performance in the geological disaster identification task.
In one possible implementation manner, the foregoing manner of obtaining the second cost function corresponding to the first geological disaster identification sub-model may be implemented by the following example implementation.
(1) Obtaining a first expected output according to each second prediction characteristic;
(2) And obtaining a second cost function corresponding to the first geological disaster identification sub-model according to the first prediction characteristic and the first expected output.
In the embodiment of the present invention, it is assumed that there is a geologic hazard recognition sub-model a for recognizing floods, for example. The sub-model performs flood identification based on the remote sensing image, generates a second prediction feature and describes the condition that flood exists in the image. In order to obtain the second cost function corresponding to the sub-model a, the second prediction feature needs to be converted into the first expected output. Specifically, the second predicted feature may be mapped to a value representing the probability or confidence of the flood, which is the first expected output of sub-model a, by comparison and verification with the actual flood area. In the example of flood identification submodel a above, a first predicted feature and a first expected output have been obtained. Now, a second cost function corresponding to sub-model a needs to be calculated from this information. In particular, the difference between the first prediction feature and the first expected output may be utilized to construct the second cost function. For example, a loss function, such as a mean square error, may be defined for measuring the deviation of the first predicted characteristic from the first expected output. This difference metric will serve as a second cost function for sub-model a for training and optimizing the sub-model. By minimizing the second cost function, the prediction result of the sub-model A can be more similar to the expected output, so that the accuracy and the robustness of flood identification are improved.
In one possible implementation, the following examples are also provided by the present embodiments.
(1) Obtaining processing results corresponding to the prediction features determined by each geological disaster recognition sub-model respectively;
the step of obtaining the first expected output according to each second prediction characteristic may be implemented by the following example.
(1) Obtaining a first expected output according to the processing results respectively corresponding to the second prediction features;
the step of obtaining the second cost function corresponding to the first geological disaster identification sub-model according to the first prediction feature and the first expected output can be implemented through the following example execution.
(1) And obtaining a second cost function corresponding to the first geological disaster identification sub-model according to the processing result corresponding to the first prediction characteristic and the first expected output.
In the embodiment of the invention, in the geological disaster early warning method, each geological disaster identification sub-model generates prediction characteristics according to the input remote sensing image. To further process these predicted features, they need to be correlated and converted with other information. Specifically, the predicted features determined by each sub-model may be subjected to subsequent processing, such as normalization, dimension reduction, feature selection, or the like, to obtain corresponding processing results. These processing results will be used for calculation and analysis in the subsequent steps. And a geological disaster recognition sub-model B is assumed to be used for recognizing landslide. The sub-model generates a second prediction feature through the remote sensing image, and describes the condition that landslide exists in the image. In order to obtain the first expected output of this sub-model B, it is necessary to use the processing results corresponding to the second predicted features. The processing results may be feature vectors obtained after the dimension reduction operation. By inputting these processing results into a model or algorithm, a first expected output of sub-model B for the landslide may be obtained, for example a value representing the probability or confidence of the landslide. In the example of sub-model B above, the first predicted feature and the first expected output have been obtained. Now, a second cost function corresponding to sub-model B needs to be calculated from this information. In particular, the difference between the processing result corresponding to the first prediction feature and the first expected output may be utilized to construct the second cost function. For example, the deviation of the first predicted feature processing result from the first expected output may be measured using the mean square error as a loss function. This difference metric will serve as a second cost function for sub-model B for training and optimizing the sub-model. By minimizing the second cost function, the prediction result of the sub-model B can be more similar to the expected output, so that the accuracy and the robustness of landslide identification are improved.
In a possible embodiment, the step of obtaining the first expected output according to each of the second prediction features may be performed by the following example.
(1) Acquiring correlation factors of each second prediction feature and the first prediction feature respectively;
(2) And obtaining the first expected output according to each second predicted feature and the correlation factor between each second predicted feature and the first predicted feature.
In an exemplary embodiment of the present invention, in a geological disaster warning method, it is desirable to infer a first expected output from a second predicted characteristic. To achieve this objective, it is necessary to obtain correlation information, i.e. a correlation factor, between each second predicted feature and the first predicted feature. Specifically, the degree of correlation between the two may be measured by calculating a correlation coefficient (e.g., pearson correlation coefficient) or other correlation metric therebetween. These correlation factors will be used for calculation and inference in subsequent steps. Assume that a geologic hazard recognition sub-model C is provided for volcanic eruption recognition. The sub-model generates a second prediction feature according to the remote sensing image, describing the condition that volcanic eruption exists in the image. In order to obtain the first expected output of the sub-model C, it is necessary to use the second prediction feature and the correlation factor corresponding thereto. Specifically, the second prediction features may be multiplied by corresponding correlation factors and weighted summed to obtain the first desired output. This weighted summation process will take into account the correlation between each second predicted feature and the first predicted feature and calculate the first expected output based on its importance. The end result may be a value that represents volcanic eruption probability or confidence.
In a possible implementation manner, the step of obtaining the correlation factor of each second prediction feature and the first prediction feature may be implemented by the following example.
(1) Acquiring first weight distribution corresponding to the first prediction feature;
(2) Acquiring second weight distribution corresponding to any second prediction feature;
(3) And obtaining the correlation factor of any second predicted feature and the first predicted feature according to the first weight distribution corresponding to the first predicted feature and the second weight distribution corresponding to any second predicted feature.
In an embodiment of the present invention, for example, in a geological disaster early warning method, it is desirable to obtain a first weight distribution of a first prediction feature. This weight distribution is used to measure the importance and contribution of the first predicted feature in the identification of geological disasters. Specifically, the feature weights or feature importance ranks in the model training process may be utilized to obtain a first weight distribution corresponding to the first predicted feature. This distribution will reflect the relative importance between the different features, thereby affecting the weight calculation and the inference of the correlation factor in the subsequent steps. Assume that a geologic hazard identification submodel D is provided for use in seismic prediction. The sub-model generates a second prediction feature according to the remote sensing image, describing the existence of an earthquake in the image. In order to obtain a second weight distribution corresponding to a second predicted feature, the importance and contribution of that feature in the seismic prediction needs to be considered. In particular, a similar approach, such as feature weights or feature importance ranking during model training, may be used to obtain a second weight distribution corresponding to a second predicted feature. This distribution will reflect the relative importance between the different features, thereby affecting the weight calculation and the inference of the correlation factor in the subsequent steps. In the example of the sub-model D described above, the first weight distribution corresponding to the first predicted feature and the second weight distribution corresponding to the second predicted feature have been obtained. Now, it is necessary to calculate a correlation factor between any of the second predicted features and the first predicted features from this information. Specifically, the first weight distribution corresponding to the first prediction feature and the second weight distribution corresponding to any one of the second prediction features may be weighted and summed, and normalized to obtain the correlation factor. This correlation factor will reflect the degree of association and the importance between the first predicted feature and any of the second predicted features. In this way, the weight distribution can be utilized to infer correlation factors corresponding to the geologic hazard identification submodel.
In one possible implementation manner, each geological disaster recognition sub-model is obtained by replacing a reference processing unit in an initial geological disaster recognition model with a target task unit, the consumption resource load corresponding to the target task unit is lower than the consumption resource load corresponding to the reference processing unit, and the initial geological disaster recognition model is a basic model with only one path for carrying out geological disaster recognition.
In an exemplary embodiment of the present invention, in a geological disaster recognition system, there is a base model a for performing recognition of geological disasters. The model A consists of a plurality of processing units, each processing unit corresponds to a specific task and plays a key role in the identification process of geological disasters. These processing units may include image preprocessing, feature extraction, classification, and judgment, among others. Each processing unit consumes some computational resources and has a relatively high resource load throughout the identification process. In order to reduce the resource load of the geological disaster recognition model, it is considered to construct a geological disaster recognition sub-model B by replacing a part of the reference processing units. A target task unit is selected that is similar to the reference processing unit but has a corresponding lower load of consumed resources. For example, a lightweight neural network structure or a simplified feature extraction method may be used instead of the original complex processing unit. In this way, the sub model B has lower resource consumption when carrying out geological disaster identification, thereby improving the efficiency and performance of the system. It is assumed that a simpler feature extraction method is selected as the target task unit and replaces a reference processing unit in the original geologic hazard recognition model. In practice, this simplified feature extraction method may require less computing resources and memory space and may complete processing in a shorter time. In contrast, the original reference processing unit may be a complex convolutional neural network, requiring a significant amount of computational resources and time. Therefore, by using the target task unit, the resource load of the geological disaster recognition sub-model is reduced, so that the recognition process is more efficient.
In one possible implementation, the step of acquiring the sample remote sensing image may be performed by the following example.
(1) And carrying out data expansion on the initial sample image in the time domain and/or the space domain to obtain the sample remote sensing image.
In the embodiment of the invention, for example, in a geological disaster early warning system, a remote sensing image is required to be used as input to identify a geological disaster. To increase the diversity and number of training samples, data expansion may be performed on the initial sample image. Data augmentation is the process of generating a new sample image by performing a series of transformations and operations on the original image. In the time domain, expansion of the time sequence can be performed, namely remote sensing images at different time points are selected to form an input sample. This allows for the dynamic evolution of geological disasters. In the spatial domain, geometric transformations (e.g., rotation, scaling, flipping, etc.) or image processing (e.g., blurring, sharpening, etc.) may be performed to expand the perspective and appearance of the sample. In the geological disaster early warning system, a new set of sample remote sensing images can be obtained after the data expansion process. These images are generated from the original sample image by a transformation and manipulation in the time and/or spatial domain. These sample remote sensing images will be used to train a geologic hazard recognition model to improve its robustness and generalization ability. Each sample image has different features and perspectives, thereby increasing the model's ability to learn different geological disaster conditions.
The following provides an overall implementation of an embodiment of the present invention.
It is assumed that a town is located in a mountain area, and geological disasters such as landslide, debris flow, etc. often occur in the area. In order to improve the early warning capability of the disasters, the related departments decide a geological disaster early warning method based on GIS.
First, the original data of the town are obtained from a GIS system, and a remote sensing image is extracted from the original data. Next, a trained geologic hazard identification model is needed, which is obtained by: a collection of sample remote sensing images, such as landslide, debris flow, etc., with disaster type labels is collected. The sample remote sensing images are input into at least two geological disaster recognition sub-models to carry out recognition operation, and respectively determined prediction characteristics and geological disaster types are obtained. And obtaining a target cost function according to the first cost function (measuring deviation between the disaster type and the label) and the second cost function (measuring deviation between the predicted features) of each geological disaster identification sub-model. A diversity enhancing cost function is also added to enhance diversity between the sub-model predictive features. And finally, training each geological disaster recognition sub-model according to the target cost function to obtain a target geological disaster recognition model.
And then, inputting the collected town remote sensing image into the trained geological disaster recognition model to obtain the type of the disaster which is likely to happen currently, for example, the model predicts that landslide occurs in the area with larger probability in the future week.
Finally, according to the prediction result, generating and issuing early warning information about landslide possibly occurring in the town, wherein the early warning information comprises early warning level, predicted influence range, recommended counter measures and the like. Thus, residents and corresponding responsible personnel of the town can be prepared in advance, and the possible loss caused by disasters is reduced.
In the process, the original sample image is further subjected to data expansion in the time domain and/or the space domain so as to enrich the sample remote sensing image. In addition, the target task unit with lower consumption resource load is used for replacing the reference processing unit in the sub-model, so that the model is more efficient.
The embodiment of the invention provides a computer device 100, wherein the computer device 100 comprises a processor and a nonvolatile memory storing computer instructions, and when the computer instructions are executed by the processor, the computer device 100 executes the geological disaster early warning method based on GIS. As shown in fig. 2, fig. 2 is a block diagram of a computer device 100 according to an embodiment of the present invention. The computer device 100 comprises a memory 111, a processor 112 and a communication unit 113. For data transmission or interaction, the memory 111, the processor 112 and the communication unit 113 are electrically connected to each other directly or indirectly.
The foregoing description, for purpose of explanation, has been presented with reference to particular embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the disclosure and its practical application, to thereby enable others skilled in the art to best utilize the disclosure and various embodiments with various modifications as are suited to the particular use contemplated. The foregoing description, for purpose of explanation, has been presented with reference to particular embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the disclosure and its practical application, to thereby enable others skilled in the art to best utilize the disclosure and various embodiments with various modifications as are suited to the particular use contemplated.
Claims (10)
1. The geological disaster early warning method based on the GIS is characterized by comprising the following steps of:
acquiring original GIS data of a target area, and acquiring a target remote sensing image from the original GIS data;
inputting the target remote sensing image into a pre-trained target geological disaster identification model to obtain a current disaster type corresponding to the target area;
and generating early warning information for the target area according to the disaster type.
2. The method of claim 1, wherein the target geologic hazard identification model is obtained by:
acquiring a sample remote sensing image and a geological disaster recognition model, wherein the sample remote sensing image is configured with a disaster type label, and the geological disaster recognition model comprises at least two geological disaster recognition sub-models;
respectively executing identification operation on the sample remote sensing image based on each geological disaster identification sub-model to obtain the prediction characteristics and the geological disaster types respectively determined by each geological disaster identification sub-model;
obtaining a target cost function according to the first cost function and the second cost function respectively corresponding to each geological disaster identification sub-model; executing a training process on each geological disaster recognition sub-model according to the target cost function to obtain a target geological disaster recognition model;
the first cost function corresponding to the first geological disaster recognition sub-model is used for measuring deviation between a geological disaster type determined by the first geological disaster recognition sub-model and the disaster type label, the second cost function corresponding to the first geological disaster recognition sub-model is used for measuring deviation between a first prediction feature determined by the first geological disaster recognition sub-model and each second prediction feature, the first geological disaster recognition sub-model is any geological disaster recognition sub-model in each geological disaster recognition sub-model, and each second prediction feature is a prediction feature determined by each geological disaster recognition sub-model except the first geological disaster recognition sub-model in each geological disaster recognition sub-model.
3. The method according to claim 2, wherein obtaining the target cost function according to the first cost function and the second cost function corresponding to each geological disaster recognition sub-model respectively includes:
acquiring a diversity enhancing cost function, wherein the diversity enhancing cost function is used for enhancing diversity among the prediction features determined by each geological disaster recognition sub-model;
and obtaining the target cost function according to the first cost function, the second cost function and the diversity enhancing cost function which are respectively corresponding to each geological disaster identification sub-model.
4. The method according to claim 2, wherein the obtaining manner of the second cost function corresponding to the first geological disaster identification sub-model includes:
obtaining a first expected output according to each second prediction characteristic;
and obtaining a second cost function corresponding to the first geological disaster identification sub-model according to the first prediction characteristic and the first expected output.
5. The method according to claim 4, wherein the method further comprises:
obtaining processing results corresponding to the prediction features determined by each geological disaster recognition sub-model respectively;
and obtaining a first expected output according to each second prediction characteristic, wherein the method comprises the following steps:
obtaining a first expected output according to the processing results respectively corresponding to the second prediction features;
and obtaining a second cost function corresponding to the first geological disaster identification sub-model according to the first prediction characteristic and the first expected output, wherein the second cost function comprises the following components:
and obtaining a second cost function corresponding to the first geological disaster identification sub-model according to the processing result corresponding to the first prediction characteristic and the first expected output.
6. The method of claim 4, wherein said deriving a first expected output from said each second predicted feature comprises:
acquiring correlation factors of each second prediction feature and the first prediction feature respectively;
and obtaining the first expected output according to each second predicted feature and the correlation factor between each second predicted feature and the first predicted feature.
7. The method of claim 6, wherein said obtaining correlation factors for each of the second predicted features and the first predicted features, respectively, comprises:
acquiring first weight distribution corresponding to the first prediction feature;
acquiring second weight distribution corresponding to any second prediction feature;
and obtaining the correlation factor of any second predicted feature and the first predicted feature according to the first weight distribution corresponding to the first predicted feature and the second weight distribution corresponding to any second predicted feature.
8. The method according to claim 2, wherein each of the geological disaster recognition sub-models includes a geological disaster recognition sub-model obtained by replacing a reference processing unit in an initial geological disaster recognition model with a target task unit, the target task unit corresponding to a lower consumption resource load than the reference processing unit corresponding to a consumption resource load, and the initial geological disaster recognition model is a base model for performing geological disaster recognition in which only one path exists.
9. The method of claim 2, wherein the acquiring the sample remote sensing image comprises:
and carrying out data expansion on the initial sample image in the time domain and/or the space domain to obtain the sample remote sensing image.
10. A server system comprising a server for performing the GIS-based geological disaster warning method as claimed in any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410045126.4A CN117877232A (en) | 2024-01-12 | 2024-01-12 | Geological disaster early warning method and system based on GIS |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410045126.4A CN117877232A (en) | 2024-01-12 | 2024-01-12 | Geological disaster early warning method and system based on GIS |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117877232A true CN117877232A (en) | 2024-04-12 |
Family
ID=90594455
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410045126.4A Pending CN117877232A (en) | 2024-01-12 | 2024-01-12 | Geological disaster early warning method and system based on GIS |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117877232A (en) |
-
2024
- 2024-01-12 CN CN202410045126.4A patent/CN117877232A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110400021B (en) | Bank branch cash usage prediction method and device | |
| CN110232203B (en) | Knowledge distillation optimization RNN short-term power failure prediction method, storage medium and equipment | |
| Kankanala et al. | AdaBoost $^{+} $: An Ensemble Learning Approach for Estimating Weather-Related Outages in Distribution Systems | |
| CN113496104B (en) | Precipitation prediction correction method and system based on deep learning | |
| CN112598248B (en) | Load prediction method, load prediction device, computer equipment and storage medium | |
| CN111242351A (en) | Tropical cyclone track prediction method based on self-encoder and GRU neural network | |
| CN112668238B (en) | Rainfall processing method, rainfall processing device, rainfall processing equipment and storage medium | |
| CN112733997A (en) | Hydrological time series prediction optimization method based on WOA-LSTM-MC | |
| Li et al. | Deep spatio-temporal wind power forecasting | |
| CN116702090A (en) | Multi-mode data fusion and uncertain estimation water level prediction method and system | |
| CN116861331A (en) | Expert model decision-fused data identification method and system | |
| CN117556197A (en) | Typhoon vortex initialization method based on artificial intelligence | |
| CN117235608B (en) | Risk detection method, risk detection device, electronic equipment and storage medium | |
| CN114154622A (en) | Algorithm model for traffic operation system flow data acquisition missing completion | |
| CN109636194B (en) | Multi-source cooperative detection method and system for major change of power transmission and transformation project | |
| CN117877232A (en) | Geological disaster early warning method and system based on GIS | |
| CN116226748A (en) | Multi-label co-occurrence network discrimination-based electricity stealing type detection method and system | |
| CN112529315B (en) | Landslide prediction method, landslide prediction device, landslide prediction equipment and storage medium | |
| Zhang et al. | Regional economic prediction model using backpropagation integrated with bayesian vector neural network in big data analytics | |
| Darvishi Salookolaei et al. | Application of grey system theory in rainfall estimation | |
| Ramlan et al. | Rainfall prediction in flood prone area using deep learning approach | |
| CN117827863B (en) | Atmospheric environment monitoring and analyzing method and system based on CLDAS database | |
| Shahoud et al. | Incorporating unsupervised deep learning into meta learning for energy time series forecasting | |
| Derras et al. | Prediction of recovery time of infrastructure functionalities after an earthquake using machine learning | |
| Sumalatha et al. | Real Time Big Data Analytics for Agricultural Land Hotspot Prediction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |