A whole-process planning and management method and system for scenic areas based on GIS platform

Abstract

The present invention discloses a method and system for the full-process planning and management of scenic areas based on a GIS platform. The method includes raw data collection, preliminary data processing, identification of ecological risk areas, planning and management optimization, and scenic area planning and management. The present invention relates to the technical field of intelligent planning and management of scenic areas, and specifically refers to a method and system for the full-process planning and management of scenic areas based on a GIS platform. The present invention obtains raw data through raw data collection; employs a preliminary data processing method involving data spatial alignment, missing value repair, semantic feature extraction, data standardization, and dataset segmentation; employs a deep contrast graph convolutional network model for ecological risk area identification, improving the accuracy of ecological risk identification in complex landforms and providing risk support for subsequent planning; and employs a multi-party reinforcement learning model for scenic area planning and management optimization, capable of responding to changes in environmental conditions in real time and iterating optimization strategies based on historical data and real-time feedback.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent planning and management of scenic areas, and in particular to a whole-process planning and management method and system for scenic areas based on a GIS platform. Background Art

[0002] Scenic area planning and management based on the GIS platform refers to the use of geographic information system (GIS) technology to collect, analyze, store and display spatial data of scenic areas in order to achieve comprehensive planning and management of natural resources, cultural heritage, environmental protection, tourism facilities and other factors. It can provide scientific decision-making basis for scenic areas, help rationally plan tourist routes, facility construction and environmental protection, achieve a balance between ecological protection and economic development, and enhance the sustainability and attractiveness of scenic areas.

[0003] However, traditional scenic area planning and management methods lack a systematic independent ecological risk assessment link, resulting in planning decisions that rely too much on empirical judgment or coarse-grained environmental data, making it difficult to quantify the spatiotemporal differences in regional ecological carrying capacity, and easily leading to technical problems such as an imbalance between development and protection. Traditional scenic area planning and management methods are limited by static model architecture and linear decision-making logic, making it difficult to effectively integrate the inherent coupling relationship of dynamic environmental evolution, multiple subject demands and complex spatial associations, and unable to cope with problems such as dynamic changes in ecologically sensitive areas and nonlinear fluctuations in tourist behavior. In addition, there is a technical problem of insufficient depth in modeling the interaction effects of geographical elements and the impact on infrastructure networks. Summary of the Invention

[0004] In view of the above situation, in order to overcome the defects of the existing technology, the present invention provides a whole-process planning and management method and system for scenic areas based on the GIS platform. In view of the fact that the traditional planning and management methods of scenic areas lack a systematic independent assessment link of ecological risks, which leads to excessive reliance on empirical judgment or coarse-grained environmental data in planning decisions, making it difficult to quantify the spatiotemporal differences in regional ecological carrying capacity, and easily causing an imbalance between development and protection, this solution creatively adopts a deep contrast graph convolutional network model to identify ecological risk areas. By integrating geographic spatial data with multi-source dynamic information and combining it with an adaptively enhanced deep learning model, it can efficiently capture the characteristics of spatial heterogeneity and risk association, thereby improving the accuracy and robustness of ecological risk identification under complex landforms. At the same time, through the contrast learning machine The system reduces data noise interference and provides scientific and dynamic risk support for subsequent planning; in view of the fact that traditional scenic area planning and management methods are limited by static model architecture and linear decision-making logic, it is difficult to effectively integrate the inherent coupling relationship of dynamic environmental evolution, multiple subject demands and complex spatial associations, and cannot cope with problems such as dynamic changes in ecologically sensitive areas and nonlinear fluctuations in tourist behavior, and the technical problem of insufficient depth in modeling the interaction effects of geographical elements and the impact of infrastructure networks, this solution creatively adopts a multi-party reinforcement learning model to optimize scenic area planning and management. By introducing a multi-party collaborative reinforcement learning framework, a multi-objective decision-making model with adaptive weight allocation capabilities is constructed, which can respond to changes in environmental conditions in real time and iterate optimization strategies based on historical data and real-time feedback.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides a method for the whole process planning and management of scenic areas based on a GIS platform, the method comprising the following steps:

[0006] Step S1: raw data collection;

[0007] Step S2: preliminary data processing;

[0008] Step S3: identification of ecological risk areas;

[0009] Step S4: planning management optimization;

[0010] Step S5: Scenic area planning and management.

[0011] Furthermore, in step S1, the original data collection is used to collect the original data required for scenic area planning and management, specifically by performing data collection to obtain a scenic area planning original data set, the scenic area planning original data set specifically including a historical scenic area original data set and a current scenic area original data set, the historical scenic area original data set and the current scenic area original data set both including environmental sensor data, geospatial data and human intervention data, the historical scenic area original data set also including a historical scenic area aesthetic evaluation label, a historical scenic area tourist comment label and historical tourist flow data.

[0012] Furthermore, in step S2, the data is initially processed, which is used to perform preliminary processing on the collected raw data, and specifically includes the following steps:

[0013] Step S21: Data spatial alignment, which is used to unify the geographic reference frame. Specifically, the projection difference between the original dataset of the historical scenic area and the original dataset of the current scenic area is eliminated by affine transformation to obtain the historical registration dataset and the current registration dataset.

[0014] Step S22: Missing value repair, for repairing missing values, specifically reconstructing missing values ​​for the historical registration dataset and the current registration dataset through local statistical features to obtain a historical correction dataset and a current correction dataset;

[0015] Step S23: semantic feature extraction, for mining semantic information of scenic areas based on tourist review data of scenic areas, specifically obtaining semantic features of scenic areas through a BERT model, and incorporating them into the historical revised dataset and the current revised dataset to obtain a historical feature set and a current feature set;

[0016] Step S24: Data standardization, which is used to perform data standardization, specifically standardizing the historical feature set and the current feature set using a Z-Score standardization method to obtain a standardized historical set and a standardized current set;

[0017] Step S25: Dataset segmentation, which is used to segment the data set, specifically, to segment the standardized history set into a planning training set and a planning test set.

[0018] Furthermore, in step S3, the ecological risk area discrimination is used to discriminate ecological risk areas in scenic areas, specifically by constructing a deep contrast graph convolutional network model, and using the deep contrast graph convolutional network model to perform ecological risk area discrimination on the standardized current set, the planning training set, and the planning test set, respectively. The obtained ecological risk area discrimination features are respectively incorporated into the standardized current set, the planning training set, and the planning test set to obtain the current planning management set, the processed planning training set, and the processed planning test set;

[0019] The identification of ecological risk areas specifically includes the following steps:

[0020] Step S31: constructing a random deformation enhancement module to improve the model's robustness to terrain changes. Specifically, the module simulates natural landforms through random deformation to obtain spatially elastically augmented samples.

[0021] Step S32: constructing a spectrum random enhancement module to enhance feature invariance, specifically simulating sensor failures through band masking to enhance feature stability;

[0022] Step S33: Design an adjacency matrix for designing a graph convolutional network adjacency matrix, specifically based on infrastructure distribution, quantifying regional spatial accessibility, and combining geometric distance to construct the adjacency matrix;

[0023] Step S34: constructing a graph convolution module for performing graph convolution operations;

[0024] Step S35: design comparative learning;

[0025] Step S36: Construct a model and identify ecological risk areas, specifically through the deformation random enhancement module construction, the spectral random enhancement module construction, the design adjacency matrix, the graph convolution module construction and the design comparative learning, to construct a deep contrast graph convolution network model, train the model based on the planning training set, verify the model performance based on the planning test set, and obtain a deep contrast graph convolution network model, and use the standardized current set, the planning training set and the planning test set as inputs of the deep contrast graph convolution network model respectively, to identify ecological risk areas, and obtain the current planning management set, the processed planning training set and the processed planning test set.

[0026] Furthermore, in step S4, the planning and management optimization is used for scenic area planning and management optimization, specifically by constructing a multi-party reinforcement learning model and performing scenic area planning and management optimization based on the multi-party reinforcement learning model;

[0027] The planning management optimization specifically includes the following steps:

[0028] Step S41: Modeling a state space for modeling the planning decision-making environment state, specifically by comprehensively processing the discriminant features of ecological risk areas, semantic features of scenic areas, ecological sensitivity scores of scenic areas based on GIS overlay analysis, and environmental sensor data through a graph attention network to encode the planning decision-making environment state;

[0029] Step S42: Multi-party strategy design, used to design a strategy to balance the demands of multiple parties, specifically combining the ecological party, the tourist party and the management party to carry out multi-party strategy design;

[0030] Step S43: constructing a reward function to quantify the comprehensive benefits of the planning scheme;

[0031] Step S44: strategy optimization, for iterative strategy optimization, specifically, continuously optimizing the strategy by adopting a proximal strategy optimization method;

[0032] Step S45: construct a model and perform optimization, specifically, construct a multi-party reinforcement learning model through the modeling state space, the multi-party strategy design, the reward function construction and the strategy optimization, train the model based on the processed planning training set, verify the model performance based on the processed planning test set, obtain the multi-party reinforcement learning model, and use the current planning management set as the input of the multi-party reinforcement learning model to perform planning management optimization and obtain a scenic area planning and management plan.

[0033] Furthermore, in step S5, the scenic area planning and management is specifically to systematically carry out overall planning and implementation management of the scenic area based on the scenic area planning and management plan and in accordance with relevant policies and standards.

[0034] The present invention provides a GIS-based scenic area full-process planning and management system, which includes an original data acquisition module, a data preliminary processing module, an ecological risk area identification module, a planning and management optimization module, and a scenic area planning and management module.

[0035] The original data acquisition module is used for original data acquisition, obtains the original data set of the scenic area planning by acquiring the original data, and sends the original data set of the scenic area planning to the data preliminary processing module;

[0036] The data preliminary processing module is used for preliminary data processing, and obtains a standardized current set, a planned training set, and a planned test set through preliminary data processing, and sends the standardized current set, the planned training set, and the planned test set to the ecological risk area discrimination module;

[0037] The ecological risk area discrimination module is used to discriminate ecological risk areas by constructing a deep contrast graph convolutional network model to discriminate ecological risk areas, obtain a current planning management set, a processed planning training set, and a processed planning test set, and send the current planning management set, the processed planning training set, and the processed planning test set to the planning management optimization module;

[0038] The planning management optimization module is used for planning management optimization, which performs planning management optimization by constructing a multi-party reinforcement learning model to obtain a scenic area planning management plan, and sends the scenic area planning management plan to the scenic area planning management module;

[0039] The scenic area planning and management module is used for scenic area planning and management, and performs scenic area planning and management based on the scenic area planning and management scheme.

[0040] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0041] (1) In view of the fact that traditional scenic area planning and management methods lack a systematic independent ecological risk assessment link, which leads to excessive reliance on empirical judgment or coarse-grained environmental data in planning decisions, making it difficult to quantify the spatiotemporal differences in regional ecological carrying capacity, and easily causing technical problems such as imbalance between development and protection, this scheme creatively adopts a deep contrast graph convolutional network model to identify ecological risk areas. By integrating geographic spatial data with multi-source dynamic information and combining it with an adaptively enhanced deep learning model, it can efficiently capture spatial heterogeneity and risk association characteristics, improve the accuracy and robustness of ecological risk identification under complex landforms, and reduce data noise interference through a contrast learning mechanism, providing scientific and dynamic risk support for subsequent planning.

[0042] (2) In view of the technical problems that traditional scenic area planning and management methods are limited by static model architecture and linear decision logic, and are difficult to effectively integrate the inherent coupling relationship of dynamic environmental evolution, multiple subject demands and complex spatial associations, and are unable to cope with problems such as dynamic changes in ecologically sensitive areas and nonlinear fluctuations in tourist behavior, and the modeling depth of the interaction effects of geographical elements and the impact of infrastructure networks is insufficient, this scheme creatively adopts a multi-party reinforcement learning model to optimize scenic area planning and management. By introducing a multi-party collaborative reinforcement learning framework, a multi-objective decision-making model with adaptive weight allocation capabilities is constructed, which can respond to changes in environmental status in real time and iterate optimization strategies based on historical data and real-time feedback. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the flow chart of the whole-process planning and management method for scenic areas based on the GIS platform provided by the present invention;

[0044] Figure 2A schematic diagram of the modules of the GIS platform-based scenic area full-process planning and management system provided by the present invention;

[0045] Figure 3 This is a schematic diagram of the process of preliminary data processing in step S2;

[0046] Figure 4 This is a flow chart of ecological risk area identification in step S3;

[0047] Figure 5 Schematic diagram of the process flow for planning management optimization in step S4.

[0048] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0050] In the description of the present invention, it should be understood that terms such as "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0051] Example 1, see Figure 1 The technical solution adopted by the present invention is as follows: The present invention provides a whole-process planning and management method for scenic areas based on a GIS platform, the method comprising the following steps:

[0052] Step S1: raw data collection;

[0053] Step S2: preliminary data processing;

[0054] Step S3: identification of ecological risk areas;

[0055] Step S4: planning management optimization;

[0056] Step S5: Scenic area planning and management.

[0057] Example 2, see Figure 1 and Figure 2In step S1, the raw data collection is used to collect the raw data required for scenic area planning and management. Specifically, the raw data set for scenic area planning is obtained by performing data collection. The raw data set for scenic area planning specifically includes a raw data set for historical scenic areas and a raw data set for current scenic areas. Both the raw data set for historical scenic areas and the raw data set for current scenic areas include environmental sensor data, geospatial data, and human intervention data. The raw data set for historical scenic areas also includes aesthetic evaluation labels for historical scenic areas, tourist comment labels for historical scenic areas, and historical tourist flow data.

[0058] The geospatial data specifically includes remote sensing image data of scenic areas, LiDAR point cloud data of scenic areas, and vector boundary data of scenic areas. The remote sensing image data of scenic areas is used as a base map in the GIS platform to provide a realistic visual representation of surface coverage. The LiDAR point cloud data of scenic areas is used to generate digital elevation models and digital surface models through 3D analysis tools of the GIS platform. The vector boundary data of scenic areas is used to define planning scopes through spatial queries on the GIS platform.

[0059] The human intervention data specifically includes tourist comment data of scenic areas and infrastructure distribution layer data of scenic areas. The tourist comment data of scenic areas is used to generate semantic features of scenic areas and quantify the cultural value of the region. The infrastructure distribution layer data of scenic areas is used to quantify the spatial accessibility of the region based on the distribution of infrastructure.

[0060] Example 3, see Figure 1 、 Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the data is preliminarily processed to preliminarily process the collected raw data. Specifically, the following steps are included:

[0061] Step S21: Data spatial alignment, which is used to unify the geographic reference frame. Specifically, the projection difference between the original dataset of the historical scenic area and the original dataset of the current scenic area is eliminated by affine transformation to obtain the historical registration dataset and the current registration dataset.

[0062] Step S22: Missing value repair, for repairing missing values, specifically reconstructing missing values ​​for the historical registration dataset and the current registration dataset through local statistical features to obtain a historical correction dataset and a current correction dataset;

[0063] Step S23: semantic feature extraction, which is used to mine semantic information of scenic areas based on tourist review data of scenic areas. Specifically, semantic features of scenic areas are obtained through the BERT model and incorporated into the historical revised dataset and the current revised dataset to obtain a historical feature set and a current feature set. The formula used is as follows:

[0064] ;

[0065] Where, represents the semantic features of the scenic area, N represents the total number of tourist comments, Indicates the BERT model running function, It represents the word segmentation result of the nth tourist comment in the mth grid in the GIS platform. represents the coordinates of the landscape associated with the nth tourist comment in the mth grid in the GIS platform, represents the coordinates of the center point of the mth grid in the GIS platform, sd represents the spatial attenuation coefficient, Indicates calculation of L2 norm;

[0066] Step S24: Data standardization, which is used to perform data standardization, specifically standardizing the historical feature set and the current feature set using a Z-Score standardization method to obtain a standardized historical set and a standardized current set;

[0067] Step S25: Dataset segmentation, which is used to segment the data set, specifically, to segment the standardized history set into a planning training set and a planning test set.

[0068] Example 4, see Figure 1 、 Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the ecological risk area discrimination is used to discriminate ecological risk areas in scenic areas. Specifically, a deep contrast graph convolutional network model is constructed, and the deep contrast graph convolutional network model is used to perform ecological risk area discrimination on the standardized current set, the planning training set, and the planning test set. The obtained ecological risk area discrimination features are respectively incorporated into the standardized current set, the planning training set, and the planning test set to obtain the current planning management set, the processed planning training set, and the processed planning test set.

[0069] The identification of ecological risk areas specifically includes the following steps:

[0070] Step S31: constructing a random deformation enhancement module to improve the model's robustness to terrain changes. Specifically, random deformation is used to simulate natural landforms to obtain spatial elastic augmentation samples. The formula used is as follows:

[0071] ;

[0072] Where (x,y) represents the input coordinates, represents the horizontal random shape variable of the input coordinates, represents the vertical random shape variable of the input coordinate, A represents the number of Gaussian kernels, represents the learnable weight of the horizontal random shape variable of the a-th Gaussian kernel, represents the vertical random shape variable learnable weight of the a-th Gaussian kernel, Indicates that the center position of the Gaussian kernel is , the covariance matrix is Two-dimensional Gaussian distribution of ;

[0073] Step S32: Spectral random enhancement module is constructed to enhance feature invariance. Specifically, the module simulates sensor failure by band masking to enhance feature stability. The band masking mask matrix is ​​expressed as follows:

[0074] ;

[0075] Where Ma represents the band mask matrix, represents a random masking control parameter that is uniformly distributed within [0,1]. represents the occlusion probability;

[0076] Step S33: Design an adjacency matrix for designing a graph convolutional network adjacency matrix. Specifically, the adjacency matrix is ​​constructed based on infrastructure distribution, quantifying regional spatial accessibility, and combining geometric distance. The formula used is as follows:

[0077] ;

[0078] Where, represents the average spatial accessibility index between the center point of the mth grid and the infrastructure of the lth grid in the GIS platform, B represents the number of infrastructures in the lth grid in the GIS platform, represents the service capability score of the bth infrastructure of the lth grid in the GIS platform, represents the coordinates of the bth infrastructure in the lth grid in the GIS platform, Am represents the adjacency matrix, represents the elements of the adjacency matrix, represents the coordinates of the center point of the lth grid in the GIS platform, dis represents the empirical attenuation distance, and Sl represents the physical side length of the grid in the GIS platform;

[0079] Step S34: Build a graph convolution module for performing graph convolution operations. The formula used is as follows:

[0080] ;

[0081] Where, represents the discriminant characteristics of ecological risk areas, Represents the graph convolution operation function, Represents the enhanced features after deformation random enhancement and spectrum random enhancement;

[0082] Step S35: Design contrastive learning, the formula used is as follows:

[0083] ;

[0084] Where, represents the contrast loss value, represents the cosine similarity function, and represents the discriminant characteristics of ecological risk areas under different random deformations and random spectral disturbances, C represents the number of negative samples, represents the discriminant feature of the cth negative sample ecological risk area, represents the temperature coefficient;

[0085] Step S36: Construct a model and identify ecological risk areas, specifically through the deformation random enhancement module construction, the spectral random enhancement module construction, the design adjacency matrix, the graph convolution module construction and the design comparative learning, to construct a deep contrast graph convolution network model, train the model based on the planning training set, verify the model performance based on the planning test set, and obtain a deep contrast graph convolution network model, and use the standardized current set, the planning training set and the planning test set as inputs of the deep contrast graph convolution network model respectively, to identify ecological risk areas, and obtain the current planning management set, the processed planning training set and the processed planning test set.

[0086] By performing the above operations, we address the technical issues that traditional scenic area planning and management methods lack a systematic independent ecological risk assessment link, resulting in planning decisions that rely too much on empirical judgment or coarse-grained environmental data, making it difficult to quantify the spatiotemporal differences in regional ecological carrying capacity, and easily leading to an imbalance between development and protection. This solution creatively uses a deep contrast graph convolutional network model to identify ecological risk areas. By integrating geographic spatial data with multi-source dynamic information and combining it with an adaptively enhanced deep learning model, it can efficiently capture spatial heterogeneity and risk association characteristics, thereby improving the accuracy and robustness of ecological risk identification under complex landforms. At the same time, it reduces data noise interference through a contrastive learning mechanism, providing scientific and dynamic risk support for subsequent planning.

[0087] Example 5, see Figure 1 、 Figure 2 and Figure 5This embodiment is based on the above embodiment. In step S4, the planning and management optimization is used for optimizing the planning and management of scenic areas. Specifically, a multi-party reinforcement learning model is constructed, and the planning and management optimization of scenic areas is performed based on the multi-party reinforcement learning model.

[0088] The planning management optimization specifically includes the following steps:

[0089] Step S41: Modeling state space for modeling the planning decision-making environment state. Specifically, the graph attention network comprehensively processes the discriminant features of ecological risk areas, semantic features of scenic areas, ecological sensitivity scores of scenic areas based on GIS overlay analysis, and environmental sensor data to encode the planning decision-making environment state. The formula used is as follows:

[0090] ;

[0091] Where St represents the state of the planning decision environment, Represents the graph attention network operation function, Represents the splicing operation function, represents the ecological sensitivity score of the scenic area based on GIS overlay analysis, and Ed represents environmental sensor data;

[0092] Step S42: Multi-party strategy design, which is used to design a strategy that balances the demands of multiple parties. Specifically, the multi-party strategy design is carried out by combining the ecological party, the tourist party, and the management party. The formula used is as follows:

[0093] ;

[0094] Where, Indicates that the parameter is The policy function, represents the softmax function, Indicates that the parameter is The multi-layer perceptron running function, represents the comprehensive strategy function, at represents the action space, represents the action space of the k-th party, represents the planning decision environment state of the k-th party, , when k=1, it represents the ecological strategy, and its action space , when k=2, it represents the tourist strategy, and its action space , when k=3, it represents the management strategy, and its action space , represents the resource allocation vector, where Indicates the number of entities that require resource allocation;

[0095] Step S43: Construct a reward function to quantify the comprehensive benefits of the planning scheme. The formula used is as follows:

[0096] ;

[0097] Where, represents the ecological reward function value, M represents the total number of grids in the GIS platform, Indicates the vegetation cover change of the mth grid in the GIS platform, represents the area of ​​the mth grid in the GIS platform, represents the development impact weight, Indicates the developed area of ​​the scenic area. represents the value of the economic reward function, Tv represents the number of tourists in the scenic area, represents the average consumption per tourist. represents the value of the social benefit reward function, Represents the visual transformer operation function for the aesthetic evaluation of scenic areas, Indicates the scenic area image data, represents the reward weight of category i, represents the adjustment weight of the learnable category i, Represents the ecological reward weight, represents the economic reward weight, represents the social benefit reward weight, Represents the comprehensive reward function value;

[0098] Step S44: strategy optimization, for iterative strategy optimization, specifically, continuously optimizing the strategy by adopting a proximal strategy optimization method;

[0099] Step S45: construct a model and perform optimization, specifically, construct a multi-party reinforcement learning model through the modeling state space, the multi-party strategy design, the reward function construction and the strategy optimization, train the model based on the processed planning training set, verify the model performance based on the processed planning test set, obtain the multi-party reinforcement learning model, and use the current planning management set as the input of the multi-party reinforcement learning model to perform planning management optimization and obtain a scenic area planning and management plan.

[0100] By performing the above operations, this solution creatively adopts a multi-party reinforcement learning model to optimize the planning and management of scenic areas. By introducing a multi-party collaborative reinforcement learning framework, it constructs a multi-objective decision-making model with adaptive weight allocation capabilities, which can respond to changes in environmental conditions in real time and iterate optimization strategies based on historical data and real-time feedback.

[0101] Example 6, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S5, the scenic area planning and management is specifically based on the scenic area planning and management plan, in accordance with relevant policies and standards, to systematically carry out overall planning and implementation management of the scenic area.

[0102] Example 7, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. The present invention provides a GIS platform-based scenic area full-process planning and management system, including an original data acquisition module, a data preliminary processing module, an ecological risk area identification module, a planning management optimization module, and a scenic area planning management module;

[0103] The original data acquisition module is used for original data acquisition, obtains the original data set of the scenic area planning by acquiring the original data, and sends the original data set of the scenic area planning to the data preliminary processing module;

[0104] The data preliminary processing module is used for preliminary data processing, and obtains a standardized current set, a planned training set, and a planned test set through preliminary data processing, and sends the standardized current set, the planned training set, and the planned test set to the ecological risk area discrimination module;

[0105] The ecological risk area discrimination module is used to discriminate ecological risk areas by constructing a deep contrast graph convolutional network model to discriminate ecological risk areas, obtain a current planning management set, a processed planning training set, and a processed planning test set, and send the current planning management set, the processed planning training set, and the processed planning test set to the planning management optimization module;

[0106] The planning management optimization module is used for planning management optimization, which performs planning management optimization by constructing a multi-party reinforcement learning model to obtain a scenic area planning management plan, and sends the scenic area planning management plan to the scenic area planning management module;

[0107] The scenic area planning and management module is used for scenic area planning and management, and performs scenic area planning and management based on the scenic area planning and management scheme.

[0108] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0109] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

[0110] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A whole-process planning and management method for scenic areas based on a GIS platform, characterized by: The method comprises the following steps: S1: Original data collection: obtaining an original data set for scenic area planning by performing data collection. The original data set for scenic area planning specifically includes an original data set for historical scenic areas and an original data set for current scenic areas; S2: Preliminary data processing: preliminary processing of the collected raw data to obtain the standardized current set, planning training set and planning test set; S3: Ecological risk area discrimination, used to discriminate ecological risk areas in scenic areas, specifically by constructing a deep contrast graph convolutional network model, and using the deep contrast graph convolutional network model to perform ecological risk area discrimination on the standardized current set, the planning training set, and the planning test set, respectively. The obtained ecological risk area discrimination features are incorporated into the standardized current set, the planning training set, and the planning test set, respectively, to obtain the current planning management set, the processed planning training set, and the processed planning test set; The deep contrast graph convolutional network model specifically includes a deformation random enhancement module, a spectral random enhancement module, and a graph convolution module. The deformation random enhancement module is used to improve the model's robustness to terrain changes, specifically by simulating natural landforms through random deformation to obtain spatial elastic augmented samples. The spectral random enhancement module and the graph convolution module are used to enhance feature invariance, specifically by simulating sensor failures through band masking to enhance feature stability. S4: Planning and management optimization, which is used for optimizing the planning and management of scenic areas. Specifically, a multi-party reinforcement learning model is constructed, and the planning and management of scenic areas is optimized based on the multi-party reinforcement learning model to obtain a planning and management plan for the scenic areas. The planning management optimization specifically includes the following steps: S41: Modeling state space, used to model the environmental state of planning and decision-making. Specifically, the graph attention network comprehensively processes the discriminant features of ecological risk areas, semantic features of scenic areas, ecological sensitivity scores of scenic areas based on GIS overlay analysis, and environmental sensor data to encode the environmental state of planning and decision-making. S42: Multi-party strategy design, used to design strategies that balance the demands of multiple parties, specifically combining the ecological side, the tourist side, and the management side to conduct multi-party strategy design; S43: Reward function construction to quantify the comprehensive benefits of planning schemes; S44: Strategy optimization, used for iterative optimization of strategies, specifically, continuously optimizing strategies by adopting proximal strategy optimization methods; S45: Constructing a model and performing optimization, specifically, constructing a multi-party reinforcement learning model through the modeling state space, the multi-party strategy design, the reward function construction, and the strategy optimization, training the model based on the processed planning training set, and verifying the model performance based on the processed planning test set to obtain a multi-party reinforcement learning model, and using the current planning management set as input to the multi-party reinforcement learning model to perform planning management optimization to obtain a scenic area planning management plan; S5: Scenic area planning and management, specifically, systematically carrying out overall planning and implementation management of the scenic area based on the scenic area planning and management plan.

2. The method for whole-process planning and management of scenic areas based on a GIS platform according to claim 1 is characterized in that: The identification of ecological risk areas specifically includes the following steps: Step S31: constructing a deformation random enhancement module; Step S32: constructing a spectrum random enhancement module; Step S33: Design an adjacency matrix for designing a graph convolutional network adjacency matrix, specifically based on infrastructure distribution, quantifying regional spatial accessibility, and combining geometric distance to construct the adjacency matrix; Step S34: constructing a graph convolution module for performing graph convolution operations; Step S35: design comparative learning; Step S36: Construct a model and identify ecological risk areas, specifically through the deformation random enhancement module construction, the spectral random enhancement module construction, the design adjacency matrix, the graph convolution module construction and the design comparative learning, to construct a deep contrast graph convolution network model, train the model based on the planning training set, verify the model performance based on the planning test set, and obtain a deep contrast graph convolution network model, and use the standardized current set, the planning training set and the planning test set as inputs of the deep contrast graph convolution network model respectively, to identify ecological risk areas, and obtain the current planning management set, the processed planning training set and the processed planning test set.

3. The method for overall planning and management of scenic areas based on a GIS platform according to claim 1 is characterized in that: The original dataset of the historical scenic area and the original dataset of the current scenic area both include environmental sensor data, geospatial data and human intervention data. The original dataset of the historical scenic area also includes aesthetic evaluation labels of the historical scenic area, tourist comment labels of the historical scenic area and historical tourist flow data.

4. The method for overall planning and management of scenic areas based on a GIS platform according to claim 1 is characterized in that: The data preliminary processing is used to perform preliminary processing on the collected raw data, and specifically includes the following steps: Step S21: Data spatial alignment, which is used to unify the geographic reference frame. Specifically, the projection difference between the original dataset of the historical scenic area and the original dataset of the current scenic area is eliminated by affine transformation to obtain the historical registration dataset and the current registration dataset. Step S22: Missing value repair, for repairing missing values, specifically reconstructing missing values ​​for the historical registration dataset and the current registration dataset through local statistical features to obtain a historical correction dataset and a current correction dataset; Step S23: semantic feature extraction, for mining semantic information of scenic areas based on tourist review data of scenic areas, specifically obtaining semantic features of scenic areas through a BERT model, and incorporating them into the historical revised dataset and the current revised dataset to obtain a historical feature set and a current feature set; Step S24: Data standardization, which is used to perform data standardization, specifically standardizing the historical feature set and the current feature set using a Z-Score standardization method to obtain a standardized historical set and a standardized current set; Step S25: Dataset segmentation, which is used to segment the data set, specifically, to segment the standardized history set into a planning training set and a planning test set.

5. A GIS-based scenic area full-process planning and management system, for implementing the GIS-based scenic area full-process planning and management method according to any one of claims 1 to 4, characterized in that: It includes original data collection module, preliminary data processing module, ecological risk area identification module, planning management optimization module and scenic area planning management module.

6. The GIS-based scenic area full-process planning and management system according to claim 5 is characterized by: The original data acquisition module is used for original data acquisition, obtains the original data set of the scenic area planning by acquiring the original data, and sends the original data set of the scenic area planning to the data preliminary processing module; The data preliminary processing module is used for preliminary data processing, and obtains a standardized current set, a planned training set, and a planned test set through preliminary data processing, and sends the standardized current set, the planned training set, and the planned test set to the ecological risk area discrimination module; The ecological risk area discrimination module is used to discriminate ecological risk areas by constructing a deep contrast graph convolutional network model to discriminate ecological risk areas, obtain a current planning management set, a processed planning training set, and a processed planning test set, and send the current planning management set, the processed planning training set, and the processed planning test set to the planning management optimization module; The planning management optimization module is used for planning management optimization, which performs planning management optimization by constructing a multi-party reinforcement learning model to obtain a scenic area planning management plan, and sends the scenic area planning management plan to the scenic area planning management module; The scenic area planning and management module is used for scenic area planning and management, and performs scenic area planning and management based on the scenic area planning and management scheme.

Images