Mine Ecological Restoration Monitoring and Assessment System
By using a hybrid cloud platform and a multi-scale assessment model, the problems of data security, real-time performance, and fragmentation in mine ecological restoration monitoring have been solved, enabling full-cycle, intelligent monitoring and assessment, which is suitable for mine ecological restoration in complex terrain.
Patent Information
- Application Number
- CN202611133210.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-29
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2046-07-29
AI Technical Summary
Traditional mine ecological restoration supervision models suffer from problems such as data fragmentation, discontinuous monitoring, subjective assessment, and difficulty in balancing data security and real-time performance. In particular, they lack effective monitoring and assessment methods in the land-sea interaction zone.
A hybrid cloud platform architecture is adopted, which combines public and intranet data acquisition subsystems with business subsystems. Data security is achieved through physically isolated data exchange modules, and data is collected in the absence of network environment using offline application interaction modules. Data fusion and quantitative evaluation are carried out by combining spatiotemporal data governance engine and multi-scale evaluation model.
It has enabled full-cycle, real-time, and intelligent monitoring and evaluation of mine ecological restoration projects while ensuring data security, solving the problems of data fragmentation and evaluation subjectivity, and improving the continuity of monitoring and the objectivity of evaluation.
Smart Images

Figure CN122675284B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of ecological restoration monitoring and assessment technology, and in particular to a mine ecological restoration monitoring and assessment system. Background Technology
[0002] Mine ecological restoration is a core task in the field of natural resource management. In particular, coastal areas, where terrestrial and marine ecosystems are intertwined and the geological environment is complex, ecological restoration projects face severe challenges due to special and complex geological and ecological environments such as steep slopes, deep and large mining pits, salinization, and seawater intrusion.
[0003] Currently, traditional mine ecological restoration supervision mainly relies on manual on-site inspections, periodic surveys, and paper-based record management. This model has significant drawbacks: First, monitoring data is fragmented, and data from different sources and periods are difficult to effectively integrate and manage uniformly, resulting in a lagging regulatory perspective and an inability to achieve full life-cycle, multi-dimensional, and efficient tracking of a large number of scattered and diverse restoration sites. Second, the assessment of restoration effectiveness often relies on periodic reports and expert judgment, which is highly subjective and lacks continuous, automated quantitative assessment from micro-level sites to macro-level engineering. The assessment results cannot objectively and dynamically reflect the true restoration effect, especially lacking specific assessments for ecological problems unique to the land-sea interaction zone, such as seawater intrusion and soil salinization.
[0004] Furthermore, the traditional model presents a significant contradiction between the feasibility and security of data transfer. On the one hand, field data collection operations often face network signal blind spots or weak network environments, resulting in the inability to transmit critical monitoring data in real time, making it difficult to guarantee the continuity and authenticity of the data. On the other hand, geographic information data related to core businesses such as mining area spatial planning and ecological protection red lines must be deployed in physically isolated intranets due to security requirements. This creates a significant gap between convenient mobile data collection and secure intranet data analysis. Public network mobile terminals cannot directly and securely access intranet data, and intranet analysis services cannot immediately obtain real-time data collected from the field, severely restricting the real-time performance and intelligence level of monitoring.
[0005] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention
[0006] The main purpose of this application is to provide a monitoring and evaluation system for mine ecological restoration, which aims to solve the technical problem that the traditional regulatory model for mine ecological restoration cannot achieve full-cycle, real-time and intelligent monitoring and evaluation of mine restoration projects while ensuring data security.
[0007] To achieve the above objectives, this application proposes a mine ecological restoration monitoring and assessment system, which includes: The hybrid cloud platform includes a data acquisition subsystem deployed on the public network and a business subsystem deployed on the intranet. The data acquisition subsystem and the business subsystem perform targeted data synchronization through physically isolated data exchange modules. The data acquisition subsystem includes an offline application interaction module, which is used to generate an offline map package of the mining task area, for mobile terminals to load and collect data locally in an offline environment, and to perform incremental synchronization to the data acquisition subsystem after the network is restored. The business subsystem includes: The spatiotemporal data governance engine is used to access and fuse multi-source heterogeneous spatiotemporal data of the task area, and to perform coordinate unification and boundary processing on the monitoring data of the land-sea interaction zone in the spatiotemporal data. The ecological restoration assessment module is used to quantitatively assess the ecological restoration effectiveness of the task area based on processed spatiotemporal data and a nested multi-scale assessment model, thereby obtaining a multi-scale comprehensive assessment index.
[0008] In one embodiment, the ecological restoration assessment module includes: The patch assessment module is used to input the first feature data from the processed spatiotemporal data into a pre-trained first machine learning model for evaluation calculation at the patch scale, so as to obtain the ecological quality assessment results of each patch in the task area; the first feature data includes at least marine features and terrestrial features, the terrestrial features include soil salinity content, and the marine features include groundwater chloride ion concentration; The sub-project evaluation module is used to aggregate the ecological quality assessment results of the map patches under the jurisdiction of the sub-project and the project management indicators of the sub-project at the sub-project scale using the analytic hierarchy process (AHP) to obtain the effectiveness level assessment results of the sub-project. The project evaluation module is used to calculate the comprehensive effectiveness index of the project at the project scale, based on the effectiveness level evaluation results of each sub-project under the project and the socio-economic benefit indicators of the project, using a mechanism evaluation model.
[0009] In one embodiment, the ecological restoration assessment module further includes: The calibration module is used to trigger a data quality review process when the calculation results of the project are inconsistent with the evaluation results of the lower-level scale under the preset proportion; and to dynamically adjust the evaluation weight of at least one of the patch evaluation module, the sub-project evaluation module and the project evaluation module when it is determined that there are no abnormalities in the data quality.
[0010] In one embodiment, the data exchange module includes a manual data packet transfer mode and a one-way transmission gateway automatic synchronization mode; In the one-way transmission gateway automatic synchronization mode, the public network data acquisition subsystem packages and encrypts the collected data, and then transmits it to the internal network business subsystem through the physically isolated one-way transmission gateway. The business subsystem is also used to decrypt, parse, and verify the encrypted data transmitted by the data acquisition subsystem before storing it in the database.
[0011] In one embodiment, the offline application interaction module is specifically used for: During the process of generating offline map packages of the mining task area, the texture features and color complexity of the remote sensing images of the task area are used to distinguish sub-regions of different landforms within the task area. For sub-regions with color complexity below a preset threshold, a preset first compression algorithm is used for processing; for sub-regions with color complexity higher than or equal to the preset threshold, a preset second compression algorithm is used for processing, generating an offline map package for the task area; wherein, the compression ratio of the first compression algorithm is higher than that of the second compression algorithm.
[0012] In one embodiment, the spatiotemporal data governance engine is specifically used for: By fusing tidal data and remote sensing imagery from the spatiotemporal data, instantaneous waterline is identified, and a coordinate transformation algorithm is used to unify nautical and land map data from the spatiotemporal data into a pre-constructed standard coordinate system.
[0013] In one embodiment, the business subsystem further includes: The spatiotemporal collaborative early warning engine is used to provide real-time early warning of ecological risk indicators unique to the land-sea interaction zone in the task area based on preset rules, and to use a time-series prediction model combined with the spatial information of the monitoring points in the land-sea interaction zone to predict the changing trend of the ecological risk indicators and the range of spatial impact. The input features of the time-series prediction model include at least tidal data, groundwater level, precipitation data, groundwater chloride ion concentration, and soil salinity.
[0014] In one embodiment, the spatiotemporal collaborative early warning engine is further used to couple the monitoring data of the monitoring points with the spatial information to construct a risk transmission map, so as to deduce the spread speed and affected area of the early warning event; the spatial information includes geological structure, topography and groundwater flow direction; the early warning event includes at least seawater intrusion event.
[0015] In one embodiment, the business subsystem further includes an evaluation and optimization module, used for: Based on the periodic assessment results output by the ecological restoration assessment module and the early warning information output by the spatiotemporal collaborative early warning engine, when an early warning is detected in the periodic assessment results, ecological restoration optimization suggestions are generated. The ecological restoration optimization suggestions include process adjustment measures and / or monitoring task adjustment suggestions. The ecological restoration optimization suggestions are then converted into monitoring tasks and sent to the data acquisition subsystem.
[0016] In one embodiment, the data acquisition subsystem further includes a task recommendation module, used to: dynamically generate an inspection route using a clustering algorithm based on the evaluation results of the ecological restoration assessment module, the early warning information of the spatiotemporal collaborative early warning engine, and the real-time location of the mobile terminal, and identify key monitoring points within the task area; and push the inspection route and the key monitoring points to the mobile terminal.
[0017] One or more technical solutions proposed in this application have at least the following technical effects: First, by constructing a hybrid cloud platform comprising a data acquisition subsystem deployed on the public network and a business subsystem deployed on the intranet, and establishing a physically isolated data exchange module between the two, the system achieves physical separation of the network environment at the architectural level. This ensures data security within the intranet and guarantees physical isolation between the intranet business subsystem carrying core business data and models and the external network attack surface, fundamentally meeting the security management requirements of sensitive areas for core geographic information data. Simultaneously, the targeted data synchronization mechanism, while ensuring security, provides a foundation for controllable data exchange between isolated environments, thus resolving the contradiction between convenient mobile data acquisition and intranet data security at the system architecture level.
[0018] Secondly, the offline application interaction module in the data acquisition subsystem addresses the monitoring interruption issue caused by network signal blind spots in the field. By generating offline map packages of the mining task area, it enables mobile terminals to load and collect local data in environments without network coverage. Incremental synchronization is achieved after network recovery, ensuring the continuity of data acquisition and allowing monitoring tasks to continue uninterrupted regardless of network conditions. This provides a foundation for achieving full-cycle monitoring and evaluation. The offline acquisition and post-event synchronization mode enables real-time data collection.
[0019] Furthermore, the spatiotemporal data governance engine in the business subsystem is responsible for accessing and integrating multi-source heterogeneous spatiotemporal data, and for unifying coordinates and processing boundaries of land-sea interaction zone data. Addressing the shortcomings of data fragmentation and difficulty in integration in the traditional model, unified data governance integrates monitoring data from different sources and in different formats into a standardized data base with consistent spatial location, providing a high-quality and computable data foundation for subsequent assessment and analysis, which is a prerequisite for achieving intelligent assessment.
[0020] Finally, the ecological restoration assessment module, based on the spatiotemporal data after restoration, employs a nested multi-scale assessment model for quantitative evaluation, replacing the traditional assessment method which is highly subjective and lacks automated quantification. Through the pre-set, structured multi-scale assessment model, the system can automatically analyze and calculate the processed spatiotemporal data, outputting a quantitative comprehensive assessment index. This significantly reduces the subjectivity of manual judgment, improves the efficiency and objectivity of the assessment, and realizes intelligent assessment.
[0021] The mine ecological restoration monitoring and assessment system provided in this application resolves the conflict between data security and access through a physically isolated hybrid cloud platform, ensures continuous monitoring throughout the entire lifecycle through offline application interaction, achieves the fusion of multi-source heterogeneous data through a data governance engine, and realizes automated quantitative assessment through a multi-scale evaluation model. The synergistic effect of all system components overcomes the shortcomings of traditional regulatory models, enabling more complete, continuous, and objective intelligent monitoring and assessment of mine ecological restoration projects while ensuring data security. Attached Figure Description
[0022] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0024] Figure 1 A schematic diagram of the system structure provided for the mine ecological restoration monitoring and assessment system in this application; Figure 2 This is a schematic diagram of the data transmission process provided in an embodiment of this application; Figure 3 This is one of the schematic diagrams illustrating the generation and application process of the offline map package provided in the embodiments of this application; Figure 4 The second schematic diagram illustrating the generation and application process of the offline map package provided in this application embodiment; Figure 5 A schematic diagram of the evaluation process of the multi-scale evaluation model provided in the embodiments of this application; Figure 6 and Figure 7 This is a schematic diagram of the workflow of the spatiotemporal collaborative early warning engine provided in the embodiments of this application; Figure 8 This is a schematic diagram of the monitoring and assessment process for mine ecological restoration provided in the embodiments of this application; Figure 9 This is a schematic diagram of the system architecture provided by the mine ecological restoration monitoring and assessment system in this application.
[0025] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0026] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.
[0027] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.
[0028] Existing technologies have the following shortcomings in the monitoring and assessment of mine ecological restoration, especially in mine areas with complex topography such as coastal zones: At the data level, traditional manual inspection and phased surveying methods lead to data fragmentation and discontinuous collection (especially affected by network blind spots in the field), and it is difficult to integrate multi-source heterogeneous data from "sky-air-ground-sea". The physical isolation between the government intranet and the public network exacerbates the difficulty of secure data transfer.
[0029] At the assessment level, manual assessment methods that rely on expert experience are subjective and outdated, lacking multi-scale, automated, and quantitative assessments from micro-level plots to macro-level engineering, and especially lacking specialized assessments for special environments in the land-sea interaction zone (such as salinization and seawater intrusion).
[0030] At the application level, there is a lack of mobile solutions that can ensure the continuity of high-precision map services and data collection in environments without network coverage, as well as a data-based intelligent early warning and engineering optimization closed-loop mechanism.
[0031] To address the shortcomings of existing solutions, this application provides a monitoring and assessment system for mine ecological restoration. Targeting mine areas requiring integrated land-sea management, it employs an intelligent monitoring and assessment approach that integrates cloud-edge-device collaboration, land-sea coordination, and assessment and early warning systems to monitor and evaluate the effectiveness of ecological restoration. The core system architecture is a hybrid cloud platform combining cloud-edge-device collaboration with physical isolation. It connects a data acquisition subsystem and mobile terminals deployed on the public network, as well as a business subsystem deployed on the intranet. It processes and integrates multi-source, heterogeneous spatiotemporal data through a unified governance engine; it utilizes a three-scale nested assessment model ("map patch-sub-project-engineering") to achieve automated quantitative effectiveness assessment; it employs a spatiotemporal collaborative early warning engine combining "GIS spatial analysis + AI time-series prediction" for intelligent risk early warning; and it utilizes offline data processing... Figure 1Integrated preparation and mobile application modules ensure continuous operation in un-networked environments in the field. Ultimately, through secure data transfer and an "assessment-optimization" closed loop, dynamic optimization of data-driven ecological restoration projects is achieved.
[0032] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, such as a mine ecological restoration monitoring and assessment server. The following description uses a mine ecological restoration monitoring and assessment server as an example to illustrate this embodiment and the subsequent embodiments.
[0033] This application provides a mine ecological restoration monitoring and assessment system. Through a hybrid cloud-edge-device architecture, it achieves intelligent monitoring and assessment of the entire lifecycle of mine ecological restoration projects while maintaining data security isolation. It is particularly suitable for ecological restoration monitoring and assessment in mine areas requiring integrated land-sea management. Specifically, refer to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the mine ecological restoration monitoring and assessment system of this application.
[0034] like Figure 1 As shown, in this embodiment, the mine ecological restoration monitoring and assessment system (hereinafter referred to as the system) includes: The hybrid cloud platform includes a data acquisition subsystem deployed on the public network and a business subsystem deployed on the intranet. The data acquisition subsystem and the business subsystem perform targeted data synchronization through physically isolated data exchange modules. The data acquisition subsystem includes an offline application interaction module, which is used to generate an offline map package of the mining task area, for mobile terminals to load and collect data locally in an offline environment, and to perform incremental synchronization to the data acquisition subsystem after the network is restored. The business subsystem includes: The spatiotemporal data governance engine is used to access and fuse multi-source heterogeneous spatiotemporal data of the task area, and to perform coordinate unification and boundary processing on the monitoring data of the land-sea interaction zone in the spatiotemporal data. The ecological restoration assessment module is used to quantitatively assess the ecological restoration effectiveness of the task area based on processed spatiotemporal data and a nested multi-scale assessment model, thereby obtaining a multi-scale comprehensive assessment index.
[0035] Understandably, the system comprises terminals, a data interaction module, and a hybrid cloud platform. The hybrid cloud platform includes a data acquisition subsystem and a business subsystem. The data acquisition subsystem operates on the edge cloud, while the business subsystem operates on the central cloud. In other words, the business subsystem, data acquisition subsystem, and terminals within the hybrid cloud platform constitute a cloud-edge-device collaborative system architecture. Furthermore, the business subsystem deployed on the intranet and the data acquisition subsystem deployed on the public network achieve targeted data synchronization through the data exchange module.
[0036] In one embodiment, the system employs a physically isolated hybrid cloud platform architecture to achieve data collection and secure transmission, as well as monitoring and assessment of ecological restoration in mining areas. Specifically, the system deploys the data collection subsystem on the public network to handle high-frequency mobile interactions and data aggregation in the field; the core business subsystem is deployed on the government intranet to store authoritative ecological red line data and run a high-performance analysis engine. Directed data synchronization between the two is achieved through a physically isolated data exchange module. The data collection subsystem has a built-in offline application interaction module that can generate and load offline map packages for the task area in field environments with no or weak network coverage. This supports local data collection and encrypted storage by field personnel. Once the network is restored, the collected incremental data is automatically synchronized to the data collection subsystem deployed on the public network, and then to the business subsystem on the intranet.
[0037] Meanwhile, the business subsystem incorporates a spatiotemporal data governance engine and an ecological restoration assessment module. The spatiotemporal data governance engine is used to access and integrate multi-source heterogeneous spatiotemporal data from the mining task area, and to perform coordinate unification and boundary processing on the monitoring data of the land-sea interaction zone within this multi-source heterogeneous spatiotemporal data. Specifically targeting the complex environment of the land-sea interaction zone, it performs coordinate system normalization, spatial topology restoration, and dynamic boundary processing on the multi-source heterogeneous monitoring data to obtain a standardized and clean spatiotemporal data base. The ecological restoration assessment module, based on this processed clean data base, uses a nested multi-scale assessment model to quantitatively evaluate the restoration effectiveness of the task area at multiple scales, obtaining a multi-scale comprehensive assessment index.
[0038] It should be noted that the mining task area is the area within the mining area that requires ecological restoration monitoring and assessment. This task area can be part or all of the mining area, and there is no specific limitation on this.
[0039] As an example, multi-scale quantitative assessment includes, but is not limited to, bottom-up quantitative scoring of different spatial scales within the task area. Multi-source heterogeneous spatiotemporal data includes remote sensing data, IoT sensor data (such as real-time telemetry data from sensors deployed in mining areas for surface displacement, soil moisture, and groundwater quality), manual survey data (such as on-site inspection records and sample testing reports reported via mobile terminals), and basic geographic data (such as digital elevation models (DEMs) and land use maps).
[0040] The spatiotemporal data governance engine utilizes NiFi's processor queues and other methods to perform format parsing, coordinate transformation, noise reduction, cleaning, and standardization on these multi-source, heterogeneous spatiotemporal data. Specifically, it unifies the coordinates and handles boundaries for monitoring data from the land-sea interaction zone. For example, the coordinate system of nearshore monitoring data may differ from that used in land-based planning data. The spatiotemporal data governance engine can unify all spatial data to a specified reference coordinate system by calling an object library or a custom seven-parameter coordinate transformation algorithm. For dynamic boundaries such as the intertidal zone, it can generate a dynamic land-sea boundary mask by integrating tidal models with instantaneous waterline data from satellite imagery for subsequent spatial analysis, thus solving the challenge of fusing land and sea data into a single map.
[0041] Through a hybrid cloud architecture of "cloud-edge-device collaboration + physical isolation", ubiquitous access and convenient operation of mobile data collection in the field are achieved while ensuring the security of ecological data, solving the pain point of easy data loss in network blind spots of traditional operations. At the same time, through the spatiotemporal data governance engine and multi-scale evaluation module, the problems of severe data heterogeneity, subjective evaluation and fragmentation in the land-sea integrated scenario are specifically tackled, realizing objective, quantitative and full-cycle intelligent supervision of mine ecological restoration.
[0042] In this embodiment, firstly, a hybrid cloud platform is constructed, comprising a data acquisition subsystem deployed on the public network and a business subsystem deployed on the intranet. A physically isolated data exchange module is established between the two. This system achieves physical separation of the network environment at the architectural level, ensuring intranet data security and guaranteeing physical isolation between the intranet business subsystem carrying core business data and models and the external network attack surface. This fundamentally meets the security management requirements of sensitive areas for core geographic information data. Simultaneously, the targeted data synchronization mechanism provides a foundation for controllable data exchange between isolated environments while ensuring security, thus resolving the contradiction between convenient mobile data acquisition and intranet data security at the system architecture level.
[0043] Secondly, the offline application interaction module in the data acquisition subsystem addresses the monitoring interruption issue caused by network signal blind spots in the field. By generating offline map packages of the mining task area, it enables mobile terminals to load and collect local data in environments without network coverage. Incremental synchronization is achieved after network recovery, ensuring the continuity of data acquisition and allowing monitoring tasks to continue uninterrupted regardless of network conditions. This provides a foundation for achieving full-cycle monitoring and evaluation. The offline acquisition and post-event synchronization mode enables real-time data collection.
[0044] Furthermore, the spatiotemporal data governance engine in the business subsystem is responsible for accessing and integrating multi-source heterogeneous spatiotemporal data, and for unifying coordinates and processing boundaries of land-sea interaction zone data. Addressing the shortcomings of data fragmentation and difficulty in integration in the traditional model, unified data governance integrates monitoring data from different sources and in different formats into a standardized data base with consistent spatial location, providing a high-quality and computable data foundation for subsequent assessment and analysis, which is a prerequisite for achieving intelligent assessment.
[0045] Finally, the ecological restoration assessment module, based on the spatiotemporal data after restoration, employs a nested multi-scale assessment model for quantitative evaluation, replacing the traditional assessment method which is highly subjective and lacks automated quantification. Through the pre-set, structured multi-scale assessment model, the system can automatically analyze and calculate the processed spatiotemporal data, outputting a quantitative comprehensive assessment index. This significantly reduces the subjectivity of manual judgment, improves the efficiency and objectivity of the assessment, and realizes intelligent assessment.
[0046] The mine ecological restoration monitoring and assessment system provided in this application resolves the conflict between data security and access through a physically isolated hybrid cloud platform, ensures continuous monitoring throughout the entire lifecycle through offline application interaction, achieves the fusion of multi-source heterogeneous data through a data governance engine, and realizes automated quantitative assessment through a multi-scale evaluation model. The synergistic effect of all system components overcomes the shortcomings of traditional regulatory models, enabling more complete, continuous, and objective intelligent monitoring and assessment of mine ecological restoration projects while ensuring data security.
[0047] In one feasible implementation, the data exchange module includes a manual data packet transfer mode and a one-way transmission gateway automatic synchronization mode. In the one-way transmission gateway automatic synchronization mode, the public network data acquisition subsystem packages and encrypts the collected data, and then transmits it to the internal network business subsystem through the physically isolated one-way transmission gateway. The business subsystem is also used to decrypt, parse, and verify the encrypted data transmitted by the data acquisition subsystem before storing it in the database.
[0048] In this embodiment, the data exchange module is designed with a data transmission mechanism that combines manual and automatic methods. It includes a manual data packet transfer mode and a one-way transmission gateway automatic synchronization mode. The two modes can be switched freely according to the security level of the actual scenario.
[0049] In the one-way transmission gateway automatic synchronization mode, the public network data acquisition subsystem automatically packages and compresses the collected business data using encryption algorithms to generate standard encrypted data packets. These encrypted data packets are then pushed unidirectionally into the internal network via a physically isolated one-way transmission gateway. The internal network business subsystem has a dedicated data receiving interface that automatically receives the encrypted data packets, decrypts, decompresses, parses, and performs hash verification. After successful verification, the received encrypted data packets are updated to the internal network spatiotemporal database.
[0050] As an example, refer to Figure 2 The data transmission process shown is as follows: Figure 2 In this system, both the mobile terminal and the data acquisition subsystem are deployed on the public network, or the mobile terminal communicates with the data acquisition subsystem deployed on the public network. The data acquisition subsystem encrypts and compresses the collected business data, packaging it into corresponding encrypted data packets. Then, through the data exchange module, it sends the encrypted data packets to the internal network using either a manual data packet transfer mode or a one-way transmission gateway automatic synchronization mode. The business subsystem deployed on the internal network receives the encrypted data packets from the data acquisition subsystem through a pre-set data receiving interface, decrypts and verifies them, and stores the data in a spatiotemporal database, which can also be called the core business database.
[0051] It is understandable that, such as Figure 2 As shown, in the encrypted data packet manual transfer mode, the data exchange module physically exports the encrypted data packets from the data acquisition subsystem to an external storage device or external storage medium (such as an authorized administrator CD or USB flash drive), and then physically imports them through the data receiving interface of the business subsystem on the intranet. In the one-way transmission gateway automatic synchronization mode, the data acquisition subsystem uses a one-way transmission method, sending the encrypted data packets to the business subsystem on the intranet through a physically isolated one-way data transmission gateway. The business subsystem on the intranet unidirectionally receives the encrypted data packets from the data acquisition subsystem through the data receiving interface.
[0052] The encrypted data packet manual transfer mode achieves physical isolation between the internal and external networks, providing a high level of security. The gateway automatic synchronization mode is a hardware-level unidirectional isolation mode, suitable for high-frequency services. Building upon the manual transfer mode, the automatic synchronization mode of the gateway, through unidirectional transmission, maintains the original level of physical isolation security while allowing for switching to higher-frequency automated data synchronization based on actual scenario requirements. This eliminates manual intervention, significantly improves the timeliness of data collection, and helps meet the business demands of high-frequency monitoring.
[0053] As a feasible implementation method, the multi-source heterogeneous spatiotemporal data packet contains remote sensing images of the mission area. Based on this, the offline application interaction module of the data acquisition subsystem is specifically used for: During the process of generating offline map packages of the mining task area, the texture features and color complexity of the remote sensing images of the task area are used to distinguish sub-regions of different landforms within the task area. For sub-regions with color complexity below a preset threshold, a preset first compression algorithm is used for processing; for sub-regions with color complexity higher than or equal to the preset threshold, a preset second compression algorithm is used for processing, generating an offline map package for the task area; wherein, the compression ratio of the first compression algorithm is higher than that of the second compression algorithm.
[0054] To address the challenges of map applications in offline environments, the offline application interaction module employs an adaptive raster tiling tool to generate offline base maps for mobile terminals to collect data in offline conditions. Specifically, during the generation of raster tiles for the offline map package of the task area, image feature recognition algorithms are used to extract texture features and color complexity from remote sensing images, automatically distinguishing sub-regions of different landforms such as bare soil, forests, and intertidal zones. Adaptive compression is then performed: for areas with color complexity below a preset threshold (such as large areas of monotonous bare soil), since these areas inherently have limited color and detail information, a high compression ratio introduces almost no perceptible distortion visually while significantly reducing data volume; therefore, a first compression algorithm with a high compression ratio is used. For areas with color complexity higher than or equal to the preset threshold (such as forests, water bodies, or intertidal zones with complex textures), a second compression algorithm, which preserves better details but has a slightly lower compression ratio than the first algorithm, is used to generate the offline map package. Although the second compression algorithm has a relatively lower compression ratio, it perfectly preserves key visual information such as subtle color gradations of vegetation and clear edges of features, ensuring the readability and analytical value of the generated map.
[0055] For texture features, image processing algorithms can be used to perform convolution calculations on remote sensing images to extract texture responses at different scales and directions. Regions with coarse textures and obvious directionality usually correspond to forests, grasslands, or strip-shaped engineering facilities; regions with smooth and uniform textures may correspond to water bodies, hardened ground, or large areas of bare soil.
[0056] Regarding color complexity, the color complexity of local regions in a remote sensing image can be calculated simultaneously. In one feasible implementation, the remote sensing image is converted from the RGB color space to the Lab color space, which is more suitable for visual perception. Then, the information entropy of the color histogram of the local region is calculated. The higher the entropy value, the more dispersed the color distribution, the richer the colors, and the higher the color complexity. Conversely, the lower the entropy value, the more monotonous the colors, and the lower the color complexity, such as azure water, gray bare rock, and yellow bare soil.
[0057] Furthermore, the extracted texture features and the entropy value corresponding to the calculated color complexity are used as input. A pre-trained lightweight decision tree or support vector machine decision classifier is then used to classify pixels or pixel blocks in the remote sensing image into sub-regions corresponding to different landform types. Examples include "forest / high vegetation area," "water area," "bare soil / rock area," and "construction area." The preset threshold for color complexity represents the classifier's decision boundary in the color complexity dimension, used to initially distinguish between monochromatic and colorful areas.
[0058] After completing the region classification, different compression strategies are applied to different categories of sub-regions when generating standard map tiles. Specifically, sub-regions with low color complexity are processed using lossy compression algorithms with high compression ratios, while sub-regions with high color complexity are processed using lossless or near-lossless compression algorithms with higher fidelity.
[0059] In one embodiment, reference is made to Figure 3 The illustrated offline map package generation and application process includes an offline application interaction module comprising a multi-source map downloader and a rendering engine deployed on the mobile terminal. The multi-source map downloader downloads remote sensing images of the task area and performs image feature recognition on these images, including texture features and color complexity. Based on the identified impact features, different sub-regions of different terrain types within the task area are distinguished, and a compression algorithm is adaptively selected based on color complexity to compress different sub-regions. For the compressed remote sensing image data, an intelligent recommendation process is used to recommend optimal tile parameters for each sub-region. Based on the recommended optimal tile parameters, a tile selection tool is adaptively selected to perform parallel tile processing on different sub-regions. Tiles using different compression algorithms are then uniformly packaged and encapsulated into a standard offline map package format and output as the offline data package.
[0060] Furthermore, the offline data packet is distributed and loaded over the network when available, incrementally synchronized to the application terminal, and encrypted and stored on the mobile terminal. When the mobile terminal collects data offline, it only needs to render the encrypted offline data packet using the built-in rendering engine to smoothly load and display the offline collection interface in a network-free environment for data collection tasks. This achieves the goal of carrying high-precision offline maps within limited storage space, effectively supporting long-term continuous field survey operations.
[0061] That is, such as Figure 3 As shown, in some embodiments, the offline application interaction module is specifically used for: During the process of generating offline map packages of the mining task area, the texture features and color complexity of the remote sensing images of the task area are used to distinguish sub-regions of different landforms within the task area. For sub-regions with color complexity below a preset threshold, a preset first compression algorithm is used for processing. For sub-regions with color complexity higher than or equal to the preset threshold, a preset second compression algorithm is used for processing to obtain compressed data packets. The compression ratio of the first compression algorithm is higher than that of the second compression algorithm.
[0062] Furthermore, using a slicing tool, the compressed data package is divided into tiles, and the tiles are uniformly packaged and encapsulated into a standard offline map package format and output to obtain the offline map package of the task area.
[0063] By using incremental synchronization, the offline map package is distributed to the mobile terminal when there is a network connection, and stored encrypted in the mobile terminal. The map rendering engine (such as MapLidre) in the mobile terminal loads and renders the offline map package when the mobile terminal is offline, resulting in an offline collection interface for performing data collection tasks.
[0064] As an example, refer to Figure 4 , Figure 4 This diagram illustrates another workflow for generating and applying offline map packages. The tiling tool is adaptively selected based on the data source type. Specifically, remote sensing images are tiled using a raster tiling tool to obtain raster tiles, while vector data is tiled using a vector tiling tool to obtain vector tiles. The resulting tiles are then published through a server for use by front-end devices (such as mobile terminals).
[0065] The tiles published by the server serve as an online base map. Mobile terminals can use a map downloader to capture the online map published by the server when there is a network connection, and use it as an offline data package when the mobile terminal is offline. The mobile terminal conducts field data collection through offline mapping and stores the data in the mobile terminal database. After the network is restored, an incremental synchronization mechanism is used to synchronize the data. The collected data is synchronized to the data collection subsystem through the data exchange module, and then to the core business database for use by the front end (such as the business subsystem).
[0066] It is understandable that the mobile terminal uses the offline data collection interface and a mobile database (such as SQLite) to collect and encrypt data locally, and then uploads the collected data to the public network data collection subsystem through an incremental synchronization mechanism after the network is restored.
[0067] In one feasible implementation, the multi-source heterogeneous spatiotemporal data of the mining task area specifically includes nautical chart data and land map data. Based on this, the spatiotemporal data governance engine is specifically used to: fuse the tidal data and remote sensing images in the spatiotemporal data to identify the instantaneous waterline, and unify the nautical chart data and land map data in the spatiotemporal data to a pre-constructed standard coordinate system through a coordinate transformation algorithm.
[0068] In some embodiments, the spatiotemporal data governance engine includes: The fusion recognition module is used to fuse tidal data and remote sensing images in the spatiotemporal data to identify instantaneous waterline. The coordinate transformation module is used to unify the nautical chart data and land chart data in the spatiotemporal data to a pre-constructed standard coordinate system through a coordinate transformation algorithm.
[0069] To address the challenges of data fusion for unique coastal geographic units, the spatiotemporal data governance engine incorporates specialized processing logic for land-sea boundaries and coordinate systems. Specifically, when processing multi-source heterogeneous spatiotemporal data, the engine utilizes tidal data and remote sensing image fusion technology for spatial analysis, automatically identifying the dynamically changing instantaneous waterline of the intertidal zone. Addressing the data discrepancies caused by nautical charts typically using different coordinate systems than land maps, the engine seamlessly stitches nautical and land map data together and unifies them to a pre-built standard coordinate system by invoking coordinate transformation algorithms (such as the seven-parameter coordinate transformation algorithm), eliminating spatial deviations caused by coordinate inconsistencies.
[0070] In coastal areas, the coastline (waterline) is dynamically changing due to tidal fluctuations. Using static coastline data at a specific moment for fusion analysis will introduce boundary errors, affecting the accuracy of monitoring and assessment. Therefore, it is necessary to extract the instantaneous waterline that matches the acquisition time of the remote sensing image.
[0071] As a feasible implementation method, when identifying the waterline, firstly, a threshold segmentation method or edge detection algorithm (such as the Canny operator) is used to preliminarily process the remote sensing image (especially the near-infrared band, which strongly absorbs water bodies) to extract a preliminary, coarse water-non-water boundary. Then, by calling a digital elevation model (DEM) that covers the coastal zone and has vertical reference information, and using the instantaneous tidal height value obtained from tidal data as the reference surface, the DEM is used to simulate the submergence of all areas with elevation values lower than the tidal height value. By calculating the intersection of the DEM and the instantaneous tidal height surface, a theoretical instantaneous waterline can be generated.
[0072] Finally, the theoretical waterline calculated from the DEM is spatially overlaid and its consistency verified with the boundary initially extracted from the remote sensing image. Inconsistent areas (such as those affected by waves or cloud shadows) are smoothed and corrected, ultimately outputting a high-precision vector data of the instantaneous waterline that matches the remote sensing image. This waterline dynamically distinguishes between land and water at a specific moment, serving as the accurate boundary basis for subsequent land-sea data fusion.
[0073] The spatiotemporal data governance engine can effectively overcome the inconsistencies between land and sea data in terms of spatiotemporal benchmarks. The standardized and purified spatiotemporal data base generated ensures that subsequent tasks, such as simulating land-based pollution diffusion, assessing seawater intrusion risks, or calculating the scope of intertidal ecological restoration, have accurate and unified spatial references, laying a reliable data foundation for the precise monitoring and assessment of "land-sea integration".
[0074] Based on this, the ecological restoration assessment module employs a nested multi-scale assessment model to quantitatively evaluate the ecological restoration effectiveness of the task area, resulting in a comprehensive assessment index across multiple scales. Specifically, it adopts a three-scale nested quantitative assessment mechanism of "map feature - sub-project - project." Based on this, the ecological restoration assessment module includes: The patch assessment module is used to input the first feature data from the processed spatiotemporal data into a pre-trained first machine learning model for evaluation calculation at the patch scale, so as to obtain the ecological quality assessment results of each patch in the task area; the first feature data includes at least marine features and terrestrial features, the terrestrial features include soil salinity content, and the marine features include groundwater chloride ion concentration; The sub-project evaluation module is used to aggregate the ecological quality assessment results of the map patches under the jurisdiction of the sub-project and the project management indicators of the sub-project at the sub-project scale using the analytic hierarchy process (AHP) to obtain the effectiveness level assessment results of the sub-project. The project evaluation module is used to calculate the comprehensive effectiveness index of the project at the project scale, based on the effectiveness level evaluation results of each sub-project under the project and the socio-economic benefit indicators of the project, using a mechanism evaluation model.
[0075] It should be understood that a map patch is the smallest task unit for performing monitoring and assessment tasks within the task area, corresponding to the smallest assessment scale within the task area; a sub-project is the superior task unit of a map patch, and one sub-project can include multiple map patches, meaning that the area corresponding to one sub-project can include multiple map patches; similarly, an engineering project is the superior task unit of a sub-project, and one engineering project can include multiple sub-projects, meaning that the area corresponding to one engineering project can include multiple sub-projects. Based on this, the ecological restoration assessment module's quantitative assessment of the ecological restoration effectiveness of the task area is specifically conducted at multiple different levels of scale, including map patches, sub-projects, and engineering projects.
[0076] It is understood that, in this embodiment, the nested multi-scale evaluation model includes at least a first machine learning model at the patch scale, a hierarchical analysis model at the sub-project scale, and a mechanism evaluation model at the engineering scale.
[0077] For the multi-scale assessment of the ecological restoration effectiveness in mining areas, specifically at the patch scale, the patch assessment module uses a first machine learning model (e.g., a random forest model) pre-trained based on a large number of coastal zone samples. The first feature data, including terrestrial features such as soil salinity and terrestrial features such as groundwater chloride concentration, which are specific indicators of the land-sea interaction zone, are input into the model for feature extraction and classification regression. The model outputs the ecological quality assessment results of each patch with accurate diagnosis. These ecological quality assessment results can be characterized by indicators such as the restoration degree index.
[0078] As one embodiment, the patch assessment module employs an ensemble learning algorithm, such as a random forest model, as the first machine learning model. In the initial stage of system deployment, a large amount of sample data from historical coastal mine restoration patches is used as training samples. The restoration effectiveness level of the patches, determined by expert experience or long-term observation, is used as a label to construct a sample dataset for model training. The first feature data of each sample includes multiple monitoring indicators for the corresponding patch, including but not limited to soil salinity and groundwater chloride ion concentration, which are sensitive indicators of ecological risk in the land-sea interaction zone. Further indicators may include vegetation cover, soil organic matter content, and slope stability coefficient. Vegetation cover can be obtained through remote sensing image inversion.
[0079] In the first stage of machine learning application, for the task area to be monitored and evaluated, the latest processed monitoring data for each patch is obtained from the spatiotemporal data governance engine. First feature data is extracted from this data and input into a trained random forest model. The model uses a decision tree within the model to make a voting decision, ultimately obtaining a quantitative assessment result of the patch's ecological quality. In one embodiment, the ecological quality assessment result of a patch can be a patch restoration index between 0 and 100.
[0080] At the sub-project scale, the sub-project evaluation module uses methods such as the Analytic Hierarchy Process (AHP) to aggregate and weight the ecological quality assessment results of each map patch under its jurisdiction with project management indicators such as project progress and investment completion rate to obtain the sub-project's effectiveness level, such as excellent, good, or qualified.
[0081] As a feasible implementation method, each sub-project typically governs multiple adjacent or functionally related governance patches on the control panel. The sub-project evaluation module adopts a comprehensive decision-making method based on weight determination, namely the analytic hierarchy process, to evaluate the sub-projects.
[0082] Specifically, firstly, a hierarchical model is constructed for each sub-project. This model includes a target layer, an indicator layer, and a criterion layer. The target layer represents the sub-project's effectiveness level. The indicator layer includes at least two core elements: one is the aggregated value of the ecological quality assessment results of all the assigned map patches, which can be an average or a weighted average; the other is project management indicators, such as the completion rate of engineering work and the investment completion rate. Secondly, the criterion layer assigns appropriate weights to each element of the indicator layer using a built-in expert scoring template or based on historical management rules. For example, the weight of ecological effectiveness can be set to 0.6, and the weight of management indicators can be 0.4. The specific implementation of the analytic hierarchy process includes: constructing a judgment matrix, calculating the weight vector, and performing a consistency check (e.g., requiring a consistency ratio CR < 0.1) to ensure the rationality of the weight allocation. Finally, the aggregated map patch assessment results and the standard scores of the project management indicators are linearly weighted and comprehensively calculated according to the determined weights to obtain the effectiveness level assessment results of the corresponding sub-project, for example, divided into four levels: "Excellent," "Good," "Pass," and "Fail."
[0083] At the engineering scale, the engineering assessment module calls upon a mechanism assessment model (such as the InVEST model) to calculate the value of ecosystem services such as water conservation and carbon storage based on the effectiveness level of the sub-projects under its jurisdiction, as well as underlying data such as land use, meteorology, and soil. This value is then weighted with socio-economic benefit indicators to ultimately calculate the comprehensive effectiveness index of the project.
[0084] As an example, the engineering assessment module conducts a top-level comprehensive assessment at the entire scale of the mine ecological restoration project, reflecting the overall ecological and socio-economic benefits of the project. Specifically, it adopts a combination of mechanistic assessment models and comprehensive index methods. By calling mechanistic assessment models that can quantify ecosystem service functions, such as the InVEST model, this model is based on spatial input parameters such as land use / cover data, soil type data, digital elevation models, and meteorological data of the project area. It runs its internal hydrological, biogeochemical, and other process modules to calculate the quantity and value of key ecosystem services such as water conservation, soil retention, carbon storage, and biodiversity maintenance, reflecting the pure ecological benefits brought by the mine ecological restoration project.
[0085] In one embodiment, the ecosystem service value calculated by the mechanistic assessment model is standardized. Simultaneously, socio-economic benefit indicators, such as the number of local jobs created by the project, its promotion of surrounding industrial development, and the satisfaction scores of residents in surrounding communities, are incorporated and standardized in the same way. Finally, the two standardized values are weighted and synthesized according to preset weights (e.g., ecosystem service value accounts for 0.7, and socio-economic benefit accounts for 0.3) to generate a single, horizontally comparable Project Effectiveness Index (PEI), used to comprehensively characterize the overall restoration effectiveness of the project.
[0086] The business subsystem may also include: a calibration module, used to trigger a data quality review process when the calculation results of the project are inconsistent with the evaluation results of the lower-level scale under the predefined proportion; and to dynamically adjust the evaluation weight of at least one of the patch evaluation module, the sub-project evaluation module and the project evaluation module when it is determined that there are no abnormalities in the data quality.
[0087] In one embodiment, the calibration module operates as an independent monitoring and scheduling subroutine. Its workflow is triggered based on preset rules, and it performs data verification and parameter adjustment tasks. The calibration module continuously monitors the output results of the patch assessment module, sub-project assessment module, and project assessment module. Its built-in triggering logic is based on the principle of logical consistency of assessment results across scales. The module presets inconsistency judgment thresholds. For example, a preset proportion is specified as follows: when there is a significant contradiction between the rating of the comprehensive effectiveness index (PEI) at the project scale (e.g., good) and the ecological quality assessment results (e.g., the patch restoration index RMI) of more than 80% of the patches under the project's jurisdiction (this proportion is a configurable parameter) (e.g., unqualified), this large-scale divergence between top-down and bottom-up assessment conclusions indicates that there may be a systematic bias within the assessment system. Once the triggering conditions are detected, the calibration module automatically triggers the data quality verification process.
[0088] The data quality review process specifically includes: Verification of the authenticity of the monitoring data of the patches: The calibration module first retrieves the original monitoring data (such as soil salinity and chloride ion concentration) of patches with inconsistent trends. By comparing the spatiotemporal watermarked field images uploaded by the mobile terminal, the sensor ID, and the continuity of the collection time series, it verifies whether the data is real collection and not anomalies or incorrect input. At the same time, it checks whether the data is within the preset reasonable value range.
[0089] Mechanism Assessment Model Input and Parameter Verification: Next, the quality of the input data and parameter settings of the mechanism assessment model (such as the InVEST model) used by the engineering assessment module are checked. This includes verifying the timeliness of the input land use classification map, the accuracy of the digital elevation model, and whether soil parameters and climate data match the assessment area and time period. The calibration module may call a parameter sensitivity analysis subroutine to quickly examine the impact of changes in key parameters on the output results (PEI).
[0090] Evaluation process retrospective: The calibration module will retrospectively examine whether the weight settings for aggregating patch results and management indicators in the analytic hierarchy process (AHP) of the sub-project evaluation module are reasonable, and whether the weights currently used match the phased priorities of the project.
[0091] If data quality issues are detected, such as specific sensor malfunctions or data entry errors, the calibration module will mark the problematic data, remove it from the current assessment, and notify the operations and maintenance system to generate a repair work order. Subsequently, the system will rerun the assessment calculations at the patch, sub-project, and project levels using the cleaned data.
[0092] If the data verification confirms no anomalies: This indicates that the inconsistency in the evaluation results is likely due to the current evaluation model or weight allocation failing to accurately reflect the actual situation. In this case, the calibration module initiates a dynamic weight adjustment procedure. For example: Given the significant conflict between extensive field data (RMI) at the map scale and the results of the engineering-scale model (PEI), the module may determine that the current engineering-scale assessment places excessive weight on indicators such as socio-economic benefits, while failing to adequately reflect the basic ecological conditions. Therefore, it will automatically increase the influence of the map-scale ecological quality assessment results in the upward aggregation process. Specifically, for example, in the hierarchical analysis model of the sub-project assessment module, the weight of the aggregated value of the map-scale ecological quality assessment results can be dynamically increased from 0.6 to 0.7 or higher, while correspondingly reducing the weight of other management indicators. Alternatively, when synthesizing the comprehensive effectiveness index (PEI) in the engineering assessment module, the weight of socio-economic benefit indicators can be reduced while the weight of ecosystem service value indicators can be increased.
[0093] The weight adjustments can be based on a predefined library of adjustment strategies or calculated using a lightweight optimization algorithm (such as trying several different weight combinations to see which one best alleviates the inconsistencies between scales). After the weight adjustments are complete, the system uses the new weight configuration to recalculate the multi-scale quantitative evaluation results until the multi-scale results reach an acceptable level of consistency.
[0094] Through the calibration module, the system overcomes the rigidity and misjudgment that may be caused by static models, enabling it to adapt to the assessment needs of different repair stages and different regional characteristics. It forms an intelligent closed loop of "assessment-verification-optimization-reassessment", which significantly improves the reliability, adaptability and decision support value of the assessment results, and is the key to realizing intelligent assessment.
[0095] Furthermore, the ecological restoration assessment module also includes a calibration module to ensure the logical consistency of multi-scale assessment results. In actual operation, the system compares the assessment conclusions of upper and lower scales in real time. When a discrepancy is detected between the macroscopic calculation results of the engineering scale (e.g., the overall engineering effectiveness is "good") and the assessment results of the lower-level scale (e.g., the patch scale) with a preset proportion (e.g., 80%) (e.g., a large number of patches showing "unqualified"), the calibration module automatically triggers a data quality review process to check the authenticity of the underlying monitoring data and the parameter settings of the mechanism assessment model. If the data is normal, the calibration module dynamically adjusts the assessment weights between scales, for example, increasing the weight of the lower-level patch assessment results in the overall engineering assessment, and recalculates the output results.
[0096] In one embodiment, reference is made to Figure 5 The multi-scale assessment model illustrated outlines an assessment process that includes patch-scale assessment, sub-project-scale assessment, and engineering-scale assessment. Patch-scale assessment, based on fundamental feature mining, is the cornerstone of mine ecological restoration assessment, performing ecological quality diagnosis for each smallest governance unit (patch). Specifically, the latest multi-source data for each patch is obtained from the spatiotemporal data governance engine, and its land and marine features are input into a pre-trained first machine learning model to obtain the patch-scale ecological quality assessment results. The ecological quality assessment results of each patch constitute an ecological quality feature set.
[0097] The terrestrial features mainly include soil physicochemical indicators such as pH value, organic matter content, soil salinity, vegetation indices (e.g., NDVI), and slope stability monitoring data. Marine features mainly include groundwater chloride ion concentration, nearshore water turbidity, and shoreline type and stability, which are key indicators for assessing seawater intrusion and intertidal ecology. Internally, the model automatically calculates the importance weight of each input feature (such as salinity and vegetation indices) to the classification result based on the reduction in Gini impurity or entropy, serving as the evaluation weight at the patch scale. This process requires no manual intervention, ensuring the objectivity of the weights.
[0098] Sub-project scale assessment conducts a comprehensive evaluation within the sub-project scope. The data input for this assessment primarily includes two parts: the ecological quality assessment results of all sub-plots, and the sub-project's project management indicators, such as project progress, investment execution rate, and maintenance measure implementation rate. The sub-project scale assessment process employs the analytic hierarchy process (AHP) for comprehensive calculation, constructing a judgment matrix based on preset rules or project objectives, considering criteria such as ecological effectiveness and management performance. The determination of assessment weights can be combined with AI-predicted matrices; for example, it can be revised based on historical data to learn the importance patterns of each indicator and expert scores, forming the final sub-project scale assessment weights. Furthermore, the standardized scores of the sub-plot ecological quality feature set and management indicators are weighted according to the assessment weights to obtain the sub-project's comprehensive effectiveness index.
[0099] Engineering-scale assessment is a macro-level effectiveness calculation used to conduct a comprehensive top-level ecological benefit assessment of the entire ecological restoration project. Its data input consists of the effectiveness assessment results of each sub-project under its jurisdiction, and macro-level ecosystem data of the project area, such as land use / cover change data, biomass data, habitat quality data, etc., which can be obtained through remote sensing inversion and model simulation.
[0100] The engineering-scale assessment is achieved by using the InVEST (Integrated Valuation of Ecosystem Services and Tradeoffs) mechanistic assessment model. Based on input data, the model quantifies the quality of key ecosystem services brought about by the project implementation, such as water conservation, soil retention, carbon storage, and biodiversity conservation value, to obtain a comprehensive effectiveness index. The assessment weights of the mechanistic assessment model can be obtained through model training.
[0101] Finally, the calibration module performs deviation warnings and parameter optimizations on the first machine learning model of the patch evaluation module, the hierarchical analysis model of the sub-project evaluation module, and the mechanism evaluation model of the engineering evaluation module, forming a closed-loop feedback.
[0102] Specifically, deviation warnings are used to monitor the consistency of assessment results at different scales, thereby determining whether the triggering conditions for the data quality review process are met. For example, if the overall project effectiveness index shows "good," but more than 70% of the map features are assessed as "poor," a deviation warning will be triggered immediately.
[0103] The optimization process begins after an early warning is triggered, initiating a data quality review. First, the quality of the underlying map data is checked. If the data is correct, the problem is determined to be a mismatch between the evaluation model parameters or weight settings and the actual stage of the current project. Subsequently, system parameter optimization is automatically performed, adjusting the evaluation weights of at least one of the map-scale evaluation module, sub-project evaluation module, and project evaluation module. For example, the weights of soil salinity and chloride ion concentration—two crucial indicators for coastal engineering—in the map-scale model may be dynamically increased; or the weight ratio of ecological effectiveness and management indicators in the sub-project scale AHP model may be adjusted. After parameter optimization, a new round of multi-scale evaluation calculations is automatically initiated until the consistency of the multi-scale evaluation results meets the requirements, ensuring the adaptability and reliability of the evaluation system.
[0104] In one feasible implementation, the business subsystem may further include: a spatiotemporal collaborative early warning engine, used to provide real-time early warning of ecological risk indicators specific to the land-sea interaction zone in the task area based on preset rules, and to use a time-series prediction model combined with the spatial information of the monitoring points in the land-sea interaction zone to predict the changing trend of the ecological risk indicators and the range evolution of their spatial impact; wherein, the input features of the time-series prediction model include at least tidal data, groundwater level, precipitation data, groundwater chloride ion concentration, and soil salinity.
[0105] In terms of early warning and prevention, the business subsystem has constructed a spatiotemporal collaborative early warning engine. On the one hand, it sets static rule thresholds to conduct real-time analysis of exceeding limits in monitoring data; on the other hand, it focuses on the unique risks of the land-sea interaction zone and uses AI-driven time-series prediction models (such as multi-layer LSTM network models) to perform pre-event trend extrapolation. As an example, the time-series prediction model is trained using historical time-series data from at least the past 24 months. Its input features include conventional meteorological and hydrological characteristics, and may further include tidal data (tidal height / tidal time), groundwater level, precipitation data, groundwater chloride ion concentration, and soil salinity. Based on the real-time input monitoring data, the model can accurately predict the changing trends of ecological risk indicators within a preset time period (such as 72 hours).
[0106] Furthermore, the spatiotemporal collaborative early warning engine is also used to couple the monitoring data of the monitoring points with the spatial information to construct a risk transmission map, so as to deduce the spread speed and affected area of the early warning event; the spatial information includes geological structure, topography and groundwater flow direction; the early warning event includes at least seawater intrusion event.
[0107] In a specific implementation, the spatiotemporal collaborative early warning engine operates two analysis channels in parallel: a real-time early warning channel based on static thresholds and a predictive early warning channel based on time-series models. The impact range is assessed through a unified spatial analysis module. By coordinating real-time threshold early warning with spatiotemporal trend prediction, and coupling spatial analysis to construct a risk transmission map, an intelligent upgrade is achieved for the unique ecological risks of the land-sea interaction zone, especially seawater intrusion events, moving from post-event alarms to pre-event warnings, and from point-based alerts to spatial extrapolation.
[0108] First, the real-time early warning system based on preset rules is responsible for instantly assessing the real-time monitoring data, thus functioning as a risk sentinel. Through a built-in configurable rule base, specific static early warning thresholds are preset for ecological risk indicators unique to the land-sea interaction zone. These indicators and thresholds can be set based on historical data, environmental standards, and expert experience.
[0109] For example, soil salinity is a key indicator of whether soil is affected by seawater or saline-alkali land. When the monitored value of soil salinity is ≥3‰, a mild salinization warning (yellow warning) is triggered; when the monitored value of soil salinity is ≥5‰, a severe salinization warning (red warning) is triggered.
[0110] Chloride ion concentration in groundwater is a characteristic ion of seawater, and an increase in its concentration is direct evidence of seawater intrusion into freshwater aquifers. When the monitored value of chloride ion concentration in groundwater is ≥200 mg / L, a seawater intrusion warning (orange warning) is triggered.
[0111] The spatiotemporal collaborative early warning engine receives monitoring data in real time after it has been standardized and processed by the spatiotemporal data governance engine. Once a certain indicator value at a monitoring point exceeds its corresponding preset threshold, the system immediately generates a real-time early warning event. The event record of this early warning event includes the warning point location, indicator, exceeding value, threshold, and occurrence time, and is automatically sent to the relevant responsible persons through a message push interface.
[0112] Trend early warning based on time-series prediction models aims to prevent problems before they occur by predicting future risk trends and providing early warnings. The time-series prediction model used by the spatiotemporal collaborative early warning engine can be a long short-term memory network model, which is a recurrent neural network specifically designed for processing time-series data and can effectively capture long-term dependencies of data over time.
[0113] The input to the time series prediction model is a multivariate time series feature set, containing key factors that comprehensively reflect the water-salt dynamic processes in the land-sea interaction zone, such as tidal data (tidal height, tidal time): reflecting the periodic external force driving the sea level and being a key boundary condition affecting changes in seawater intrusion pressure; groundwater level: reflecting the pressure state of the aquifer, and its relative height to the sea level directly controls the direction and intensity of seawater intrusion; precipitation data: precipitation infiltration is an important positive process for replenishing groundwater and diluting salinity; groundwater chloride ion concentration and soil salinity: serving as the historical sequence of the target risk indicators themselves, and forming the basis for the model to learn their evolution patterns.
[0114] The time-series prediction model is trained using at least 24 months of continuous historical monitoring data of the task area as samples to fully learn the coupled variation patterns of various indicators under different seasons, tides, and rainfall scenarios. The trained model can use current and past monitoring data (e.g., the past 7 days) as input to continuously predict the changes in groundwater chloride ion concentration and soil salinity at specific monitoring points over the next 72 hours. The system not only monitors whether the predicted values exceed static thresholds but also sets dynamic baselines (e.g., ±20% confidence interval of the predicted values). When the latest measured data continuously deviates from the predicted dynamic baseline, or when the predicted value itself is approaching the risk threshold, the spatiotemporal collaborative early warning engine triggers a higher-level predictive warning, indicating that a risk is accumulating or about to occur.
[0115] The spatiotemporal collaborative early warning engine not only provides dual-channel early warning for monitoring points, but also predicts and assesses the spread and evolution of risk range based on spatial information through spatial analysis. Specifically, when early warning points appear in the monitoring points of the land-sea interaction zone, the spatiotemporal collaborative early warning engine couples the early warning information of all discrete monitoring points with the spatial information layers of the task area, such as geological structure maps, topographic slope maps, land use maps, and groundwater flow field simulation results, to construct a risk transmission map. For seawater intrusion early warning, the spatiotemporal collaborative early warning engine takes the early warning point as the source point, combines the hydrogeological conditions of the land-sea interaction zone in the task area (such as aquifer permeability coefficient) and the current groundwater flow direction, and uses simplified models such as particle tracking or groundwater solute transport simulation to deduce the possible movement speed and direction of the intrusion front.
[0116] Furthermore, the spatiotemporal collaborative early warning engine can not only predict "when the risk will occur," but also predict "how the risk will spread." Specifically, through spatial overlay analysis, it can dynamically plot the potential spatial impact range of the warning event on a map in the current and future period, and assess whether it will threaten nearby sensitive targets, such as freshwater sources, important vegetation restoration areas, or residential areas. This risk impact range projection map will be included in the early warning information, providing intuitive spatial decision support for taking precise emergency control measures (such as deploying emergency pumping wells in specific locations).
[0117] When an early warning event is triggered by rule-based or predictive alerts, spatial simulation is initiated to couple spatial information and construct a risk transmission map. Specifically, the spatiotemporal collaborative early warning engine retrieves multi-layered spatial information surrounding the early warning point from the spatiotemporal data base. This layered spatial information includes: Geological structures, such as faults and fissure zones, may be advantageous channels for seawater intrusion.
[0118] Topography: Digital elevation models (DEMs) are used to determine surface water flow direction and potential low-lying catchment areas.
[0119] Groundwater flow direction: The regional groundwater flow field map obtained by interpolation of hydrogeological model simulation or monitoring data is the core basis for determining the direction of pollutant or saline water transport.
[0120] The spatial information layers are spatially overlaid and coupled with monitoring data (such as chloride ion concentration exceeding standards) from warning points to construct a risk transmission map. For example, using warning points as the source, and based on information such as formation permeability coefficient and hydraulic gradient (from groundwater flow direction), cost distance or particle tracking algorithms in a geographic information system are used to simulate the possible migration paths and velocities of saline water bodies in aquifers, dynamically generating a seawater intrusion risk transmission map. This risk transmission map visually displays the possible diffusion directions and relative risk intensity of seawater intrusion in the form of a heat map or arrow clusters.
[0121] Specifically, the derivation of diffusion velocity, combined with the formula for calculating actual groundwater flow velocity (Darcy's Law) and the spatial distribution of aquifer parameters, allows for the estimation of the advance velocity of the saline water front in different directions. For assessing the impact area, the derivation of the diffusion velocity's path and speed can be used to further predict the spatial extent that seawater intrusion may affect within a specified future timeframe. This spatial extent can be output as a polygonal impact zone, which can be overlaid with sensitive targets within this area (such as freshwater wells, ecological protection zones, and construction projects) to assess potential impacts and provide precise spatial decision support for developing emergency control measures (such as closing water wells and initiating artificial recharge).
[0122] The spatiotemporal collaborative early warning engine, through dual-channel early warning monitoring, not only achieves a fundamental shift in ecological risk management from post-event alarms to pre-event early warnings, but also upgrades the traditional passive threshold alarms into a proactive, three-dimensional early warning system that combines real-time monitoring, trend prediction, and spatial extrapolation. It also particularly strengthens the special early warning capabilities for core risks in the land-sea interaction zone, such as seawater intrusion and soil salinization, significantly improving the foresight, accuracy, and scientific nature of risk prevention and control in mine ecological restoration projects. It is a key guarantee for achieving safe and controllable operation throughout the entire project lifecycle.
[0123] In one feasible implementation, the system not only focuses on monitoring and evaluating mine ecological restoration but also achieves a closed-loop management system at the operational level through an evaluation and optimization module. Specifically, the operational subsystem may also include an evaluation and optimization module, used to: generate ecological restoration optimization suggestions based on the periodic evaluation results output by the ecological restoration evaluation module and the early warning information output by the spatiotemporal collaborative early warning engine; generate ecological restoration optimization suggestions when an early warning is detected in the periodic evaluation results; the ecological restoration optimization suggestions include process adjustment measures and / or monitoring task adjustment suggestions; and convert the ecological restoration optimization suggestions into monitoring tasks and distribute them to the data acquisition subsystem.
[0124] In some embodiments, the spatiotemporal collaborative early warning engine includes: The rule-based early warning module is used to provide real-time early warnings of ecological risk indicators specific to the land-sea interaction zone in the task area based on preset rules. The prediction and early warning module is used to predict the changing trend of the ecological risk indicators and the evolution of the spatial impact range by using a time-series prediction model combined with the spatial information of the monitoring points in the land-sea interaction zone.
[0125] The prediction and early warning module is specifically used to: when an early warning is generated at a monitoring point, couple the monitoring data of the early warning point with spatial information to construct a risk transmission map, so as to deduce the diffusion speed and impact area of ecological risk indicators in the early warning event; the spatial information includes geological structure, topography and groundwater flow direction; the early warning event includes at least seawater intrusion event.
[0126] In one feasible implementation, the business subsystem further includes an evaluation and optimization module, configured to: generate ecological restoration optimization suggestions based on the periodic evaluation results output by the ecological restoration evaluation module and the early warning information output by the spatiotemporal collaborative early warning engine; generate ecological restoration optimization suggestions when an early warning is detected in the periodic evaluation results; the ecological restoration optimization suggestions include process adjustment measures and / or monitoring task adjustment suggestions; and convert the ecological restoration optimization suggestions into monitoring tasks and send them to the data acquisition subsystem.
[0127] Accordingly, the data acquisition subsystem also includes a task recommendation module, which is used to: dynamically generate inspection routes through clustering algorithms based on the evaluation results of the ecological restoration assessment module, the early warning information of the spatiotemporal collaborative early warning engine, and the real-time location of the mobile terminal, and identify key monitoring points within the task area; and push the inspection routes and the key monitoring points to the mobile terminal.
[0128] The assessment and optimization module runs as a background service in the system, starting periodically (e.g., monthly or quarterly) or triggered by specific events (e.g., high-level alerts). Specifically, the assessment and optimization module synchronously acquires the latest periodic assessment results (e.g., sub-project effectiveness levels, patch restoration index RMI, etc.) output by the ecological restoration assessment module and real-time alert information generated by the spatiotemporal collaborative early warning engine.
[0129] The built-in judgment logic of the evaluation and optimization module is used to determine whether a warning has occurred in the periodic evaluation results. One specific implementation method is to determine whether a warning has occurred in the periodic evaluation results output by the ecological restoration evaluation module by setting association rules. For example, if the sub-project's effectiveness level is unsatisfactory, and / or, the patch restoration index of more than a preset proportion of patches within the sub-project's jurisdiction declines for two consecutive periods, and / or, a seawater intrusion warning has recently been triggered within the task area, then ecological restoration optimization suggestions are generated when a warning has been determined in the periodic evaluation results.
[0130] As another feasible implementation method, the early warning information is correlated with the assessment results. Based on the early warning information of the spatiotemporal collaborative early warning engine, it is determined whether the periodic assessment results output by the ecological restoration assessment module have an early warning. If the early warning information of the spatiotemporal collaborative early warning engine determines that the periodic assessment results output by the ecological restoration assessment module have an early warning, ecological restoration optimization suggestions are generated.
[0131] As a feasible implementation method, when generating ecological restoration optimization suggestions, structured ecological restoration optimization suggestions are generated by matching and combining them from a predefined restoration strategy knowledge base based on assessment results and early warning data. These ecological restoration optimization suggestions include process adjustment measures and monitoring task adjustment suggestions.
[0132] Among them, the process adjustment measures address specific ecological problems and propose optimized solutions for restoration projects. For example, if the triggering condition is a soil salinity warning + low RMI, the process adjustment measures in the generated ecological restoration optimization suggestions might be: for the area numbered 001, it is recommended to adjust the current ordinary soil covering and greening process to a process of soil replacement combined with salt-tolerant vegetation planting, and to add drainage and salt removal ditches.
[0133] The monitoring task adjustment recommendations target risk areas or assessment blind spots, proposing enhanced monitoring schemes. For example, if the triggering condition is a predictive early warning of seawater intrusion, the monitoring task adjustment recommendations in the generated ecological restoration optimization suggestions might be: within 500 meters downstream of the early warning point P003, add two groundwater monitoring wells, increasing the monitoring frequency from once a week to once a day, focusing on chloride ion concentration and groundwater level changes.
[0134] The generated ecological restoration optimization suggestions do not remain merely at the report level. The evaluation and optimization module automatically transforms each monitoring task adjustment suggestion into a standard monitoring task that can be executed by the data acquisition subsystem, and distributes it to the data acquisition subsystem via work orders or other means. For example, suggestions to add new monitoring wells are transformed into task items containing specific coordinates, monitoring indicators, and monitoring frequencies; suggestions to increase monitoring frequency are updated into the configuration of existing tasks. These newly generated or adjusted tasks are securely distributed to the public network-side data acquisition subsystem through the data exchange module, awaiting assignment to field personnel for execution, thus completing a closed loop from analyzing and identifying problems to instructing them to be resolved.
[0135] The task recommendation module, deployed within the data acquisition subsystem, serves as an intelligent scheduling center to improve the efficiency and focus of individual field inspection operations. When field personnel are preparing to go out to perform a batch of monitoring tasks, the task recommendation module is activated to plan the optimal on-site work path and priorities for them.
[0136] The task recommendation module makes decisions based on three types of real-time dynamic information: the evaluation results of the ecological restoration assessment module, the early warning information from the spatiotemporal collaborative early warning engine, and the real-time location of the mobile terminal.
[0137] The assessment results of the ecological restoration assessment module further include the RMI of the map patch assessment module and the sub-project effectiveness level of the sub-project assessment module, which are used to locate areas with poor effectiveness and those that need to be given special attention.
[0138] The early warning information provided by the spatiotemporal collaborative early warning engine includes real-time early warning information based on rules, as well as predictive early warning information based on time-series prediction models, such as seawater intrusion early warning points, which are used to locate high-risk areas that require urgent verification.
[0139] The real-time location of the mobile terminal, i.e. the real-time GPS location of the field personnel's handheld smart terminal, can be used as the starting point for route planning.
[0140] In one embodiment, spatial analysis and path optimization are performed using algorithms such as K-means clustering to dynamically generate inspection routes. All task points to be inspected and the real-time locations of personnel are used as input. Based on the spatial distribution density of the task points, they are dynamically divided into K clusters as inspection areas, and the center of each cluster is calculated. Task points include regular inspection points and key points added based on early warning and evaluation results. Subsequently, the task recommendation module uses an approximate solution (such as the nearest neighbor method) to plan the shortest access sequence for the points within each cluster, thereby forming several efficient sub-routes for area inspections to avoid long-distance, inefficient round trips.
[0141] During the clustering process, clustering is assisted by the attribute weights of task points. Each point is assigned a comprehensive weight, which is calculated based on factors such as the basic task weight, the RMI of the patch it belongs to (the lower the weight, the higher the weight), and whether it is in an alert state (if so, the weight increases). Points with high weights are given priority for inclusion in the current inspection route and are marked as key monitoring points, which are highlighted with prominent icons on mobile terminals to prompt field personnel to conduct more detailed inspections and sampling.
[0142] The final generated inspection route (displayed as lines and sequence on a map) and the identified key monitoring points are sent to the field personnel's mobile terminals in real time via push notifications and / or application interface updates. Field personnel can perform inspection and data collection tasks according to this intelligently recommended inspection route, ensuring priority coverage of the most critical areas within a limited time, greatly improving the targeting and efficiency of field data collection.
[0143] In one embodiment, reference is made to Figure 6 and Figure 7 The workflow of the spatiotemporal collaborative early warning engine shown includes multiple stages such as data acquisition, data preprocessing, spatiotemporal collaborative early warning analysis, early warning event generation, early warning simulation, and early warning response.
[0144] Specifically, the spatiotemporal collaborative early warning engine begins with high-quality, multi-source spatiotemporal data input. During the data acquisition phase, the data sources include: Real-time IoT streams: Real-time access to sensor network data deployed in the ecological restoration task area via the MQTT protocol, including second-level / minute-level updates of key indicators such as groundwater chloride ion concentration, groundwater level, and soil salinity. Manually collected batch data: Regularly receiving groundwater sampling and testing results and on-site inspection records reported by field personnel using mobile terminals via API interfaces; this data is highly accurate but has a timeliness of up to days. Remote sensing detection results: Acquiring the latest satellite remote sensing inversion data, such as surface temperature, vegetation index, and surface deformation, through scheduled tasks, providing macroscopic change information across the region.
[0145] In the data preprocessing stage, the raw spatiotemporal input data undergoes unified preprocessing, including data cleaning: identifying and interpolating or removing instantaneous spikes (outliers) and continuous missing values caused by sensor disconnections in the IoT stream data. Data standardization: unifying the units of data from different sources (e.g., standardizing chloride ion concentration to mg / L) and performing normalization processing to eliminate the influence of dimensions. Feature engineering index calculation: calculating derived indices based on the original indices, such as calculating the "daily change rate of chloride ion concentration" and "salinity-water level correlation coefficient," etc. These derived features can more sensitively indicate anomalies.
[0146] In the spatiotemporal collaborative early warning analysis stage, the processed data flows into the core ecological restoration assessment module. This module adopts a parallel architecture of rule channels and time-series prediction channels to achieve collaboration between real-time judgment and trend prediction.
[0147] Among them, the rule channel provides real-time early warnings based on static thresholds and logical rules, including land and sea-specific early warning indicator threshold settings. For example, there are specific threshold rules for seawater intrusion. For instance, a yellow warning is triggered when the chloride ion concentration is ≥250mg / L.
[0148] Building upon static thresholds, for more complex scenarios, the Drools rule engine and logical rules (such as AND, OR, and NOT) execute complex logic to achieve spatiotemporal correlation rules. Using multiple indicators (such as chloride ions, water level, and salinity) as features, it determines in real-time whether the indicator status falls under "normal," "risk," or "intrusion," providing a more comprehensive and immediate diagnosis. For example, "IF chloride ion concentration at monitoring point A exceeds the standard AND water level at point B within 500 meters downstream (according to groundwater flow direction) continues to drop within 24 hours THEN triggers a 'suspected intrusion vanguard advance' orange alert." This spatiotemporal correlation based on AND / OR / NOT logical combinations can capture complex risk patterns that a single indicator cannot reflect.
[0149] The time-series prediction channel provides trend prediction and early warning based on time-series data. It features specialized prediction models specifically designed for seawater intrusion, such as the Long Short-Term Memory (LSTM) network model, which is specifically trained to predict changes in chloride ion concentration and soil salinity. Its input features include not only historical sequences of indicators calculated through feature engineering but also external driving factors such as tidal data and precipitation forecasts. This enables the detection of time-series anomalies in risk indicators, spatial cluster analysis, and classification-based early warning.
[0150] In particular, for time-series anomaly detection, during model operation, on the one hand, the indicator value is predicted for a future period of time (such as the next 72 hours). If the predicted value will exceed the threshold, a predictive warning is issued in advance. On the other hand, by comparing the deviation between the predicted value and the actual inflow value, time-series anomaly detection is achieved, which can discover sudden deterioration that the model did not anticipate.
[0151] Spatial clustering analysis employs density-based clustering algorithms such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise) to perform spatial clustering analysis on monitoring points that simultaneously trigger warnings. If multiple points cluster together spatially, it indicates a potential regional intrusion event rather than a single-point anomaly, thus triggering a higher-level regional warning.
[0152] Categorized early warning generates different types or levels of early warning information based on the risk type or event type, risk level, etc. of the early warning event.
[0153] During the early warning event generation phase, the analysis results of parallel dual-channel early warnings are aggregated into the early warning event generator to form structured and actionable early warning information. Based on this information, early warnings are categorized, aggregated, and deduplicated to generate early warning events. Specifically, early warning categorization is based on the severity of rule violations or the magnitude of predicted exceedances, marking early warning events into different levels such as "blue" (alert), "yellow" (moderate), and "red" (severe). Early warning aggregation and deduplication are performed to avoid information overload. Multiple early warnings of the same type and from the same region within a short period are aggregated (merged into a single regional early warning) and deduplicated (removing duplicate warnings for anti-shaking) to ensure that the pushed early warning information is concise and effective.
[0154] During the early warning simulation phase, the scope of the early warning is simulated. For high-level early warnings, the spatiotemporal collaborative early warning engine automatically initiates the spatial simulation process. For example, by combining geological structure maps and groundwater flow data, the range of seawater that may spread in the next 24-48 hours is simulated, and a layer corresponding to the potential impact area is generated on the map as a risk transmission map.
[0155] During the early warning and response phase, the early warning and response process is triggered based on the early warning event or information. This allows structured early warning events (including level, type, location, scope of simulation, and recommended measures) to be automatically pushed to the mobile terminals or command terminals of relevant personnel. It may also automatically generate investigation work orders and issue them to field personnel, or trigger the activation of emergency plans, thereby completing the complete business loop of "data perception - analysis and early warning - response and feedback".
[0156] In one embodiment, reference is made to Figure 8 The illustrated mine ecological restoration monitoring and assessment process involves collecting mine geology and ecological environment data through mobile terminals during fieldwork. This data is then uploaded to the data acquisition subsystem via an incremental synchronization mechanism. Simultaneously, the real-time location of the mobile terminals is uploaded, creating trajectory data for the field personnel. The data acquisition subsystem can distribute task data, including but not limited to field data acquisition tasks, to the mobile terminals.
[0157] Furthermore, in this embodiment, the mine ecological restoration monitoring and assessment system also includes an evaluation index and evaluation system database, used for users to configure the required evaluation indicators and provide the evaluation indicators to the business subsystem. The data acquisition subsystem transmits ecological restoration monitoring data to the business subsystem via the data exchange module. The business subsystem calculates indicators based on the evaluation indicators provided by the evaluation index and evaluation system database, and performs ecological restoration assessment and spatiotemporal coordinated early warning analysis based on the calculated indicators, sending the assessment results and early warning information to the user terminal. In this embodiment, the geographic base map and remote sensing images required for the business subsystem's assessment and analysis are sourced from the user terminal.
[0158] In another embodiment, reference Figure 9 The system architecture shown in this application embodiment of the mine ecological restoration monitoring and assessment system includes a physical layer, a data layer, a service layer, a business layer, and a user layer. The physical layer is mainly used for the physical deployment of the hardware equipment required by the system; the data layer is equipped with a database, mainly used for data storage; the service layer is equipped with a variety of services for the business layer to call; the business layer calls the corresponding service in the corresponding service layer according to the user's request to respond to the user's request.
[0159] Specifically, such as Figure 9 As shown, the system adopts a typical layered, domain-based microservice architecture, logically divided from bottom to top into a physical layer, data layer, service layer, business layer, and user layer. It is further divided into an internal private network zone and an internet zone on the physical network to ensure a balance between core data security and business flexibility. Specifically, the internal private network zone (internal network zone) deploys core business subsystems, carrying all confidential or sensitive business data, analytical models, and geographic information servers, serving system administrators and business department users. The internet zone (public network zone) deploys data acquisition subsystems and mobile application services, responsible for interacting with field mobile terminals, serving task managers and field data acquisition users (operation personnel).
[0160] The two areas are logically isolated and exchange data in a controlled manner through a data exchange module, achieving the security requirements of "network isolation and data controllability". The data exchange module is configured with a secure VxLAN (Virtual Extended LAN) tunnel and strictly configured firewall policies.
[0161] Specifically, the physical layer is used for deploying network security architecture and is the hardware and network foundation for system operation. Its core is physical partition deployment based on security considerations.
[0162] Internal private network zone: Deploys application servers, data servers, and map servers to carry out the system's core business processing and sensitive data storage functions. It is physically isolated from the public Internet and strictly follows the security management regulations for geographic information and government data to prevent external network attacks and ensure the security of core business data.
[0163] Internet Zone: Connected to the public network via security devices such as firewalls, primarily carrying interactive services for field users and mobile applications. Controlled, targeted data transfer occurs between the internal and external network zones via a data exchange module, including data encryption and synchronization, achieving a balance between business connectivity and security isolation.
[0164] The data layer is primarily used for data storage and management, responsible for the persistent storage of all system data. It employs a hybrid database strategy to adapt to different data requirements. The data layer is configured with a core business database (PostgreSQL + PostGIS), deployed within the intranet. The PostgreSQL relational database stores structured business data such as project information, user permissions, monitoring indicators, and evaluation reports. Its spatial extension, PostGIS, enables efficient storage, querying, and analysis of massive amounts of spatial data (such as vector boundaries and monitoring points), forming the foundation for complex GIS operations such as spatial overlay analysis and buffer analysis.
[0165] High-speed caching database (Redis): Deployed at the front end of the service layer, it is used to cache hot data (such as frequently queried alert rules, user sessions, etc.), which can improve the system's concurrent response speed.
[0166] Map Tile Database (MBTiles): A lightweight format based on SQLite used to store pyramid map tiles made from satellite remote sensing imagery, electronic maps, etc. It is accessed by the GeoServer map server to provide efficient online map services.
[0167] Mobile local database (SQLite): Deployed on the mobile devices of field personnel. Used in environments without network access, it stores offline map packages and data collection tasks issued from the internet zone, and temporarily caches track, photo, and form data collected on-site. Once the network is restored, it is synchronized to the internet zone server via an encrypted channel.
[0168] The service layer consists of a series of reusable microservices or technical components, providing general technical capabilities to upper-layer businesses in the form of APIs, mainly including: Spatial data services: Geoserver provides standard OGC map services (WMS, WMTS), and MapLibre (GLJS / Native) provides the front-end map rendering engine, which together support the visualization of the system.
[0169] Core analysis services: encapsulates the system's key algorithm models for threshold judgment; performs real-time data matching with early warning rules to trigger real-time early warnings; trend analysis: runs sequential prediction models (such as LSTM) to generate predictive early warnings; performance evaluation: performs evaluation using a multi-scale nested evaluation model of "map patch-sub-project-project"; early warning analysis: performs spatial aggregation and extrapolation analysis on early warning events.
[0170] Data and business services: Provide general services such as information retrieval, historical data review, indicator calculation, statistical analysis, and report generation to meet the data query and processing needs of various business functions.
[0171] At the business level, the technical capabilities of the service layer are assembled into complete business function modules for system users, tailored to specific application scenarios. These modules include: comprehensive viewing: providing a global, unified view that integrates and displays the distribution of repair projects, real-time monitoring data, early warning events, and assessment results; monitoring and early warning: providing a closed-loop management interface for monitoring, tracking, confirming, and providing feedback on early warning events; analysis and assessment: providing functions for configuring, triggering, viewing, and comparing assessment tasks; mobile data acquisition: corresponding to fieldwork, covering the entire process management including task allocation, data synchronization, trajectory recording, encrypted data transmission, and format verification, ensuring standardized field operations and data quality; and data aggregation: responsible for cleaning, transforming, standardizing, and aggregating heterogeneous data from multiple sources to provide a high-quality data lake for upper-level analysis.
[0172] The user layer defines the main users of the system and their access channels. User roles include management decision-making users, namely system administrators and business department users. These users mainly access the system through the intranet PC terminal and use the business layer's functions such as "comprehensive viewing", "monitoring and early warning", and "analysis and evaluation" to conduct macro-level supervision and decision-making.
[0173] User roles also include field execution users, namely task managers and field data collection users. They primarily receive tasks, conduct mobile data collection using offline maps, and synchronize data with the system via encrypted channels through mobile smart terminals (in the internet zone). Task managers are also responsible for task assignment and approval.
[0174] The system architecture provided in this embodiment, through clear physical isolation and logical division, not only meets the management requirements for geographic information security and confidentiality but also ensures the flexibility and real-time nature of mobile data collection in the field. The front-end and back-end separation combined with the microservice architecture design guarantees the system's scalability, high availability, and maintainability, forming the foundation for the stable operation of complex monitoring and assessment services for mine ecological restoration.
[0175] In this embodiment, firstly, at the system architecture and data security level, a physically isolated hybrid cloud platform architecture is adopted. The system is clearly divided into a data acquisition subsystem deployed on the public network and a core business subsystem deployed on the intranet. The two subsystems synchronize data via encrypted data packet transfer or a one-way transmission gateway. This fundamentally resolves the contradiction between the security of core data on the intranet and the convenience of mobile data collection in the field. While meeting data security requirements, it provides a feasible basis for ubiquitous access and elastic resource scheduling, ensuring the stable and reliable operation of the entire system.
[0176] Secondly, at the data fusion and intelligent assessment level, the system, through a spatiotemporal data governance engine, specifically addresses the challenge of unified aggregation of multi-source heterogeneous data (such as remote sensing, sensor data, nautical charts, and land maps) in the land-sea interaction zone. In particular, through tidal data fusion and coordinate transformation algorithms, it achieves precise unification of dynamic waterline identification and land-sea coordinate systems, providing a solid and clean spatiotemporal data foundation for upper-level analysis. Building upon this, a multi-scale nested assessment model of map features, sub-projects, and engineering projects is adopted, integrating machine learning, the analytic hierarchy process (AHP), and ecosystem service mechanism models. It also incorporates land-sea interaction zone-specific indicators such as soil salinity and groundwater chloride ion concentration, achieving a full-chain, automated, and quantitative assessment from micro-ecological parameters to macro-ecological benefits. The assessment results are objective and traceable, and the calibration module ensures the inherent consistency of multi-scale assessment conclusions, significantly improving the scientific rigor and efficiency of ecological restoration assessment results.
[0177] Furthermore, at the level of risk warning and proactive prevention, the spatiotemporal collaborative early warning engine constructs a dual-channel early warning analysis model combining spatiotemporal collaborative analysis and time-series prediction. It not only provides real-time threshold warnings for specific risks such as seawater intrusion and salinization based on preset rules, but also achieves trend prediction and pre-emptive warning of risks through a time-series prediction model combined with driving factors such as tides and precipitation. Moreover, the spatiotemporal collaborative early warning engine can couple discrete early warning points with spatial information such as geological structures and groundwater flow directions to construct risk transmission maps, dynamically extrapolating the spread speed and impact range of early warning events. This elevates single-point alarms to regional, forward-looking spatial situational awareness, greatly enhancing the initiative and accuracy of risk prevention and control.
[0178] Finally, at the business closed-loop and engineering optimization level, the system adopts an intelligent "assessment-early warning-optimization-execution" closed-loop mechanism. The assessment and optimization module can combine periodic effectiveness assessment results with real-time early warning information to automatically generate suggestions for adjusting repair processes or optimizing monitoring tasks, and then convert them into executable tasks for distribution. Simultaneously, the task recommendation module of the data acquisition subsystem can dynamically plan the optimal inspection route and identify key monitoring points based on assessment results, early warning information, and personnel locations using clustering algorithms, and push these to mobile terminals, greatly improving the targeting and efficiency of field data collection. This closed-loop optimization mechanism makes the system not only a monitoring and assessment tool, but also an intelligent management platform capable of automatically feeding back data and continuously optimizing repair projects and monitoring strategies.
[0179] This application's embodiments, through the organic synergy of various modules, systematically solve the inherent defects of traditional mine ecological restoration supervision models in terms of data security, subjective assessment, delayed early warning, and intermittent data collection. It achieves full-cycle, intelligent, safe, and controllable monitoring, assessment, and risk management of ecological restoration projects in complex land-sea interaction environments.
[0180] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the mine ecological restoration monitoring and assessment system of this application. Any simple modifications based on this technical concept are within the protection scope of this application.
[0181] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
[0182] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.
Claims
1. A monitoring and assessment system for mine ecological restoration, characterized in that, include: Mobile terminal; The hybrid cloud platform includes a data acquisition subsystem deployed on the public network and a business subsystem deployed on the intranet. The data acquisition subsystem and the business subsystem perform targeted data synchronization through physically isolated data exchange modules. The data acquisition subsystem is an edge cloud, and the business subsystem is a central cloud. The data acquisition subsystem, the business subsystem, and the mobile terminal constitute a cloud-edge-device architecture. The data acquisition subsystem includes an offline application interaction module, which is used to generate an offline map package of the mining task area, for the mobile terminal to load and collect data locally in an offline environment, and to perform incremental synchronization to the data acquisition subsystem after the network is restored. The business subsystem includes: The spatiotemporal data governance engine is used to access and fuse multi-source heterogeneous spatiotemporal data of the task area, and to perform coordinate unification and boundary processing on the monitoring data of the land-sea interaction zone in the spatiotemporal data. The ecological restoration assessment module is used to quantitatively assess the ecological restoration effectiveness of the task area based on the processed spatiotemporal data and a nested multi-scale assessment model, thereby obtaining a multi-scale comprehensive assessment index. The business subsystem also includes: The spatiotemporal collaborative early warning engine is used to provide real-time early warning of ecological risk indicators unique to the land-sea interaction zone in the task area based on preset rules, and to use a time-series prediction model combined with the spatial information of the monitoring points in the land-sea interaction zone to predict the changing trend of the ecological risk indicators and the evolution of the spatial impact range. The input features of the time-series prediction model include at least tidal data, groundwater level, precipitation data, groundwater chloride ion concentration, and soil salinity. The spatiotemporal collaborative early warning engine is also used to couple the monitoring data of the monitoring points with the spatial information to construct a risk transmission map, so as to deduce the spread speed and affected area of the early warning event; the spatial information includes geological structure, topography and groundwater flow direction; the early warning event includes at least seawater intrusion event; The data acquisition subsystem also includes a task recommendation module, which is used to: dynamically generate inspection routes through clustering algorithms based on the evaluation results of the ecological restoration assessment module, the early warning information of the spatiotemporal collaborative early warning engine, and the real-time location of the mobile terminal, and identify key monitoring points within the task area; and push the inspection routes and the key monitoring points to the mobile terminal.
2. The mine ecological restoration monitoring and assessment system as described in claim 1, characterized in that, The ecological restoration assessment module includes: The patch assessment module is used to input the first feature data from the processed spatiotemporal data into a pre-trained first machine learning model for evaluation calculation at the patch scale, so as to obtain the ecological quality assessment results of each patch in the task area; the first feature data includes at least marine features and terrestrial features, the terrestrial features include soil salinity content, and the marine features include groundwater chloride ion concentration; The sub-project evaluation module is used to aggregate the ecological quality assessment results of the map patches under the jurisdiction of the sub-project and the project management indicators of the sub-project at the sub-project scale using the analytic hierarchy process (AHP) to obtain the effectiveness level assessment results of the sub-project. The project evaluation module is used to calculate the comprehensive effectiveness index of the project at the project scale, based on the effectiveness level evaluation results of each sub-project under the project and the socio-economic benefit indicators of the project, using a mechanism evaluation model.
3. The mine ecological restoration monitoring and assessment system as described in claim 2, characterized in that, The ecological restoration assessment module also includes: The calibration module is used to trigger a data quality review process when the calculation results of the project are inconsistent with the evaluation results of the lower-level scale under the preset proportion; and to dynamically adjust the evaluation weight of at least one of the patch evaluation module, the sub-project evaluation module and the project evaluation module when it is determined that there are no abnormalities in the data quality.
4. The mine ecological restoration monitoring and assessment system as described in claim 1, characterized in that: The data exchange module includes a manual data packet transfer mode and a one-way transmission gateway automatic synchronization mode. In the one-way transmission gateway automatic synchronization mode, the public network data acquisition subsystem packages and encrypts the collected data, and then transmits it to the internal network business subsystem through the physically isolated one-way transmission gateway. The business subsystem is also used to decrypt, parse, and verify the encrypted data transmitted by the data acquisition subsystem before storing it in the database.
5. The mine ecological restoration monitoring and assessment system as described in claim 1, characterized in that, The offline application interaction module is specifically used for: During the process of generating offline map packages of the mining task area, the texture features and color complexity of the remote sensing images of the task area are used to distinguish sub-regions of different landforms within the task area. For sub-regions with color complexity below a preset threshold, a preset first compression algorithm is used for processing; for sub-regions with color complexity higher than or equal to the preset threshold, a preset second compression algorithm is used for processing, generating an offline map package for the task area; wherein, the compression ratio of the first compression algorithm is higher than that of the second compression algorithm.
6. The mine ecological restoration monitoring and assessment system as described in claim 1, characterized in that, The spatiotemporal data governance engine is specifically used for: By fusing tidal data and remote sensing imagery from the spatiotemporal data, instantaneous waterline is identified, and a coordinate transformation algorithm is used to unify nautical and land map data from the spatiotemporal data into a pre-constructed standard coordinate system.
7. The mine ecological restoration monitoring and assessment system according to claim 1, characterized in that, The business subsystem also includes an evaluation and optimization module, used for: Based on the periodic assessment results output by the ecological restoration assessment module and the early warning information output by the spatiotemporal collaborative early warning engine, when it is determined that an early warning has occurred in the periodic assessment results, ecological restoration optimization suggestions are generated. The ecological restoration optimization recommendations include recommendations for process adjustment measures and / or monitoring task adjustment; The ecological restoration and optimization suggestions are transformed into monitoring tasks and sent to the data acquisition subsystem.
Citation Information
Patent Citations
Emergency escape guide method and emergency escape guide system based on multi-data infusion
CN103954278A
