Generation method and evaluation method of target monitoring data

Abstract

The invention discloses a target monitoring data generation method and an evaluation method. Through space grid division and time alignment processing, the problem of space-time heterogeneity of multi-source monitoring data is effectively solved, and a unified space-time reference is established; further, the conditional probability relation between the data is mined through an ecological factor association model, the limitation of traditional simple weighted fusion is broken through, and data fusion based on ecological logic association is achieved. According to the method, the quality and reliability of the target monitoring data are remarkably improved, and a solid data basis is provided for accurate evaluation of the subsequent vegetation recovery progress, so that the key technical problem of low fusion quality of a traditional data fusion method in a specific scene of temporary land occupation of a power transmission line is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vegetation recovery monitoring, and in particular to a target monitoring data generation method and an evaluation method. BACKGROUND

[0002] Transmission line temporary land occupation vegetation recovery monitoring can be represented as, after the construction of the transmission line temporarily occupies the land ends, through manual field investigation, remote sensing image interpretation or sensor layout, etc. Way, the re-growth and coverage of the ground vegetation are observed and recorded periodically or irregularly.

[0003] In the related art, in the scene of transmission line temporary land occupation vegetation recovery monitoring, the method for data fusion has the defects of insufficient processing of spatio-temporal heterogeneity and ecological logical correlation, resulting in the technical problem of low data fusion quality. SUMMARY

[0004] The technical problem to be solved by the present application is that in the scene of transmission line temporary land occupation vegetation recovery monitoring, the method for data fusion has the defects of insufficient processing of spatio-temporal heterogeneity and ecological logical correlation, resulting in the technical problem of low data fusion quality. The purpose is to provide a target monitoring data generation method and an evaluation method, which solves the technical problem of low data fusion quality.

[0005] The present application is realized by the following technical scheme:

[0006] In a first aspect, the present application provides a target monitoring data generation method, comprising:

[0007] The transmission line temporary land occupation is spatially gridded according to a preset scale to obtain a plurality of spatial grid units;

[0008] The collected multi-source data is spatially classified according to the spatial grid unit to which it belongs, and the multi-source data includes remote sensing data, sensor data and measured data;

[0009] For the multi-source data in each spatial grid unit, time alignment processing is performed to align the data of different sources to a unified time node in the time dimension;

[0010] Based on a preset ecological factor correlation model, the multi-source data after time alignment is subjected to ecological logical correlation analysis to identify the conditional probability relationship between the basic factors, state indicators and verification true values;

[0011] According to the conditional probability relationship, the multi-source data is weighted and fused to output the fused target monitoring data.

[0012] Further, the step of performing time alignment processing on the multi-source data in each spatial grid cell to align data of different sources to a uniform time node in the time dimension comprises:

[0013] The multi-source data of different acquisition frequencies is divided into time windows using a preset sliding window method.

[0014] The multi-source data divided into the same time window is aligned to a representative time node of the time window.

[0015] Further, the step of performing ecological logical correlation analysis on the time-aligned multi-source data based on a preset ecological factor correlation model to identify the conditional probability relationship between the basic factors, state indicators, and validation true values comprises:

[0016] The time-aligned multi-source data is divided into basic factors, state indicators, and validation true values based on a preset data classification rule.

[0017] The divided data is input into a preset Bayesian network model, and the conditional probability relationship between the basic factors, state indicators, and validation true values is learned and output by the Bayesian network model.

[0018] Further, the step of inputting the divided data into a preset Bayesian network model and learning and outputting the conditional probability relationship between the basic factors, state indicators, and validation true values by the Bayesian network model comprises:

[0019] Based on the ecological function zone type to which the spatial grid cell belongs, a function zone weight is assigned to the basic factors, state indicators, and validation true values.

[0020] Based on the function zone weight and the mutual information between nodes in the Bayesian network model, a priority score of each node is calculated.

[0021] According to the priority score, the nodes are sorted to establish a weighted Bayesian network structure model.

[0022] Based on the weighted Bayesian network structure model, the conditional probability relationship between the basic factors, state indicators, and validation true values is determined.

[0023] Further, the step of determining the conditional probability relationship between the basic factors, state indicators, and validation true values based on the weighted Bayesian network structure model comprises:

[0024] Based on the weighted Bayesian network structure model, an initial prediction conditional probability is obtained.

[0025] calculating a prediction error between the initial predicted conditional probability and an actual probability value;

[0026] determining an error threshold based on a statistical value of the prediction errors calculated continuously for multiple times;

[0027] determining whether the prediction error is greater than the error threshold;

[0028] if yes, re-distributing the functional zone weight and returning to the step of calculating the node priority score to update the weighted Bayesian network structure model until the prediction error is not greater than the error threshold and outputting a conditional probability relationship based on a final model;

[0029] if no, outputting a conditional probability relationship based on the current weighted Bayesian network structure model.

[0030] Further, the step of performing weighted fusion on the multi-source data according to the conditional probability relationship to output the fused target monitoring data comprises:

[0031] determining a fusion priority weight of the base factor, the state index and the verification true value based on the ecological function zone type to which the spatial grid cell belongs;

[0032] selecting a target probability relationship greater than a preset probability threshold from the conditional probability relationship;

[0033] performing weighted calculation on the multi-source data corresponding to the target probability relationship based on the fusion priority weight to output the fused target monitoring data.

[0034] In a second aspect, the present application provides a method for evaluating vegetation restoration progress of temporary land occupation of a power transmission line, comprising:

[0035] determining an ecological function zone type of the temporary land occupation of the power transmission line and configuring a corresponding evaluation index set for the ecological function zone type;

[0036] distributing an index weight to each evaluation index in the evaluation index set based on a preset weight distribution rule;

[0037] constructing a restoration progress quantitative evaluation model according to the evaluation index set and the weight of each evaluation index;

[0038] inputting target monitoring data into the restoration progress quantitative evaluation model to obtain a vegetation restoration progress evaluation result; wherein the target monitoring data is obtained by using a target monitoring data generation method as described above.

[0039] In a third aspect, the present application provides a method for determining an ecological management decision of vegetation restoration of temporary land occupation of a power transmission line, comprising the following steps:

[0040] determine a current recovery state level based on the vegetation recovery progress evaluation result, wherein the vegetation recovery progress evaluation result is obtained by using the vegetation recovery progress evaluation method for temporary land occupation of a power transmission line as described above;

[0041] match one or more candidate decision schemes corresponding to the current recovery state level according to a preset decision rule library;

[0042] determine an optimal decision scheme from the one or more candidate decision schemes based on a preset optimization target.

[0043] In a fifth aspect, the present application provides a target monitoring data generation device, comprising:

[0044] a division module for performing spatial grid division on the temporary land occupation of the power transmission line according to a preset scale to obtain a plurality of spatial grid units;

[0045] a classification module for classifying the collected multi-source data according to the spatial grid units, wherein the multi-source data includes remote sensing data, sensing data and measured data;

[0046] an alignment module for performing time alignment processing on the multi-source data in each spatial grid unit, so that the data of different sources are aligned to a unified time node in the time dimension;

[0047] an association analysis module for performing ecological logical association analysis on the time-aligned multi-source data based on a preset ecological factor association model, and identifying the conditional probability relationship between the basic factors, state indicators and verification true values;

[0048] a weighted fusion module for performing weighted fusion on the multi-source data according to the conditional probability relationship, and outputting the fused target monitoring data.

[0049] In a sixth aspect, the present application provides an electronic device, comprising a memory and one or more processors in communication connection with the memory; the memory stores instructions executable by the one or more processors, and the instructions are executed by the one or more processors to enable the one or more processors to implement the above-mentioned method.

[0050] In a seventh aspect, the present application provides a computer-readable storage medium, wherein the readable storage medium stores a computer program, and the computer program is executed by a processor to implement the above-mentioned method.

[0051] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0052] The application effectively solves the spatio-temporal heterogeneity problem of multi-source monitoring data by spatial grid division and time alignment processing, and establishes a unified spatio-temporal reference; further, through an ecological factor correlation model, the conditional probability relationship between data is mined, breaking through the limitation of traditional simple weighted fusion, and realizing data fusion based on ecological logical correlation. This method significantly improves the quality and reliability of the target monitoring data, provides a solid data foundation for subsequent accurate evaluation of vegetation restoration progress, and solves the key technical problem of low fusion quality of traditional data fusion methods in the specific scene of temporary occupation of power transmission lines. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the example embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor. In the drawings:

[0054] Figure 1 An architecture diagram of a target monitoring data generation method provided for the embodiments in the specification;

[0055] Figure 2 A flowchart of a target monitoring data generation method provided for the embodiments in the specification;

[0056] Figure 3 A block diagram of an electronic device provided for the embodiments in the specification. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the present application more clear and obvious, the following will further describe the present application in combination with embodiments and drawings. The illustrative embodiments of the present application and their descriptions are only used to explain the present application, and are not regarded as a limitation on the present application.

[0058] In the related art, with the continuous advancement of power grid construction in China, power transmission line projects inevitably need to temporarily occupy land, especially in ecologically sensitive areas (such as the Qinghai-Tibet Plateau, hilly areas, etc.). The vegetation in these areas needs to be artificially intervened and naturally restored after the construction is completed, and the restoration process is directly related to the ecological safety and soil and water conservation of the region. Therefore, effective monitoring and evaluation of the vegetation restoration progress of the temporary occupation of the power transmission line has become a key link for the coordinated development of power infrastructure construction and ecological protection.

[0059] In the current related technical practice, monitoring mainly relies on the collection of multi-source data. Specifically, satellite or unmanned aerial vehicle platforms can be used to obtain large-scale vegetation index (e.g., NDVI) images, which have the advantage of wide coverage but limited spatial and temporal resolution and are susceptible to weather conditions. Internet of Things devices deployed at key points can be used to collect real-time soil temperature, humidity, and electrical conductivity, and other micro-ecological parameters, which have high data frequency but are sparse in point locations. Field surveys using the sampling method can be used to obtain accurate data on vegetation coverage and species richness, which are accurate but time-consuming, labor-intensive, and difficult to fully cover.

[0060] The above multi-source data forms the information basis for vegetation restoration monitoring. However, the related art has a fundamental technical defect in the fusion processing of these multi-source data, resulting in low-quality final data fusion.

[0061] First, the related art is insufficient in handling the spatio-temporal heterogeneity of data. Remote sensing, sensing, and field measurement data are unevenly distributed in space (point and area) and have different collection frequencies and times in time (real-time and periodic). The related art usually uses simple data stacking or interpolation methods without establishing a unified spatial and temporal reference. For example, data from different sources are not anchored to a unified spatial grid unit, and effective time alignment algorithms (such as sliding windows) are not used to coordinate data streams of different frequencies. This rough processing method results in misalignment and fragmentation of data in the spatio-temporal dimension, making it impossible to form a complete and consistent picture of the ecological conditions of the same spatio-temporal unit.

[0062] Second, the related art lacks the mining and utilization of ecological logical correlation. Vegetation restoration is a complex ecological process, and there is an inherent and uncertain causal and response relationship between soil conditions (basic factors), vegetation growth (state indicators), and final ecological functions (validation truth values). The data fusion of the related art often stops at simple weighted averaging at the physical level or uses general mathematical models unrelated to ecological mechanisms, failing to build a model that can quantitatively express this ecological logical relationship. By ignoring the deep ecological meaning behind the data, the fused data is integrated in form but fragmented in logic, making it impossible to truly reflect the internal driving mechanism of the ecological process.

[0063] In summary, due to the serious defects in the handling of spatio-temporal heterogeneity and ecological logical correlation in the data fusion process of the related art, the final generated monitoring data is of low quality and insufficient reliability. Any subsequent evaluation and decision-making based on this low-quality data cannot guarantee accuracy and scientificity, thereby restricting the level of fine and intelligent management of temporary land occupation for power transmission line ecological restoration.

[0064] The core idea of the present application is to solve the problem of spatio-temporal heterogeneity by introducing spatial grid division and time alignment processing, and further to mine and quantify the ecological logical correlation between data by constructing an ecological factor correlation model based on a Bayesian network, thereby fundamentally improving the fusion quality of multi-source data and providing a reliable data basis for accurate ecological assessment and decision-making.

[0065] As shown in Figure 1 and Figure 2 The present embodiment provides a method for generating target monitoring data. The execution subject of the method can be a server, which can be a cloud server cluster, for example, a data processing center, or an edge computing gateway, for example, a near-end processing device deployed on a power transmission line site, or an industrial-grade embedded computer or a special edge server, etc.

[0066] The method can include:

[0067] Step S12: performing spatial grid division on the temporary land occupation of the power transmission line according to a preset scale to obtain a plurality of spatial grid units.

[0068] In the present embodiment, the temporary land occupation of the power transmission line refers to the temporarily occupied land area during the construction, maintenance or expansion of the power transmission line.

[0069] In the present embodiment, the preset scale is used to define the size and fineness of the generated spatial grid units. The specific value can be pre-set when the execution subject is deployed, and can be configured according to the specific monitoring requirements, topographic features or computing resources. Specifically, it can be a fixed scale, for example, dividing the entire occupied area into square grid with a side length of 50m x 50m. It can also be a dynamic scale, for example, using a smaller scale (e.g., 30m x 30m) in areas with large topographic relief and obvious ecological gradient to obtain more detailed information; using a larger scale (e.g., 100m x 100m) in areas with flat terrain and uniform ecology to reduce computational complexity. It can also be a scale based on topographic features or a scale based on ecological function zones. That is, the preset scale can be associated with the pre-defined ecological function zones (e.g., water and soil conservation zone, biodiversity core zone, general restoration zone) in the temporary land occupation of the power transmission line, and different grid division scales are set for different function zones to achieve precise management in different zones.

[0070] In this embodiment, the spatial grid cell is the basic spatial management unit obtained from the above-described partitioning operation. Specifically, each spatial grid cell can be assigned a unique identifier (ID) within the executing entity and associated with its spatial geometric information (e.g., boundary coordinates, center point coordinates) and attribute information (e.g., the type of ecological functional zone it belongs to, its altitude, etc.). This spatial grid cell can be a square grid, a rectangular grid, or an irregular grid.

[0071] Step S14: Spatial classification of the collected multi-source data according to the spatial grid unit to which they belong. The multi-source data includes remote sensing data, sensor data, and measured data.

[0072] In this embodiment, the multi-source data refers to various data sets acquired by the executing entity from different sensing platforms, which differ in format, resolution, acquisition cycle, and physical meaning, but collectively serve the purpose of monitoring vegetation restoration status.

[0073] Specifically, remote sensing data can be surface observation data acquired in a non-contact manner through sensors mounted on aerial platforms such as satellites and drones. It can be optical remote sensing data, hyperspectral remote sensing data, or radar remote sensing data, etc.

[0074] The sensor data can be local environmental parameters collected in real time or near real time by IoT sensor nodes deployed at the temporary site of power transmission lines. These parameters can include soil parameters, meteorological parameters, or vegetation physiological parameters, etc.

[0075] Measured data can be obtained from vegetation and soil samples through manual field surveys, measurements, and recording. For example, vegetation cover, plant species and quantity (species richness), plant height, biomass, etc., can be obtained through quadrat surveys. Soil organic matter content, pH value, erosion resistance indicators, etc., can be obtained through laboratory analysis of soil samples collected on-site.

[0076] In this embodiment, spatial classification means that the executing entity, based on the geographical location information of each data point or data area, uses a spatial query and matching algorithm to classify it into a specific spatial grid cell with a unique identifier established in step S12.

[0077] In a specific implementation plan, the server can perform spatial relationship determination by comparing the geographic coordinates of the data with the geometric boundaries of each spatial grid cell. First, for sensor data and measured data, it can determine which polygon (grid cell) a point falls into. For remote sensing data, it can calculate the intersection relationship between the remote sensing data coverage area and the grid cell.

[0078] Step S16: Perform time alignment processing on the multi-source data within each spatial grid cell to align data from different sources to a unified time node in the time dimension.

[0079] In this embodiment, the time alignment process can be represented as the execution subject mapping multiple data sequences belonging to that unit but with different timestamps to a series of preset, unified time nodes for each spatial grid cell through a preset time series calculation and interpolation method, thereby generating a regular multivariate dataset aligned in the time dimension.

[0080] Specifically, the executing entity can use a sliding window method to perform time alignment processing. For high-frequency data (sensor data), the values ​​of all raw data points within the window can be aggregated (calculating the average, median, maximum, and minimum values), and the aggregated value can be used as the value of the center time node (i.e., the unified time node) of the window. For low-frequency data (measured data), if its collection timestamp falls within a certain time window, the measured data is associated with the unified time node corresponding to that window.

[0081] The executing entity can also use nearest neighbor interpolation to align execution time.

[0082] During the alignment process, if it is found that a certain data type is completely missing at a certain time point, the execution entity can perform a preset data missing handling operation. For example, it can keep the position as empty and wait for subsequent steps to perform special data completion, or within the allowable time deviation range, it can use the aggregated value of adjacent time windows to fill it.

[0083] In this embodiment, the unified time node can be represented as a common time point or time interval referenced by the aligned data. This unified time node can be a fixed-frequency sequence, for example, specifying alignment to the 1st of each month, Monday of each week, or 00:00 UTC time each day. Alternatively, the original acquisition time of a certain type of data source (which could be the most comprehensive and periodically stable data, such as remote sensing data) can be used as the target node for alignment. For example, the date of satellite transit and successful image acquisition can be used as the unified time node. Important ecological or management events (e.g., the 30th day after sowing, the 7th day after fertilization, and the 1st and 3rd days after a heavy rain) can also be used as key alignment nodes.

[0084] Step S18: Based on the preset ecological factor association model, perform ecological logical association analysis on the time-aligned multi-source data to identify the conditional probability relationship between basic factors, state indicators and verification true values.

[0085] In this embodiment, the ecological factor association model can be a predefined computational model that can be trained with data, which can formally express and calculate the interdependence, causality, or influence relationships between different types of ecological data (factors).

[0086] Specifically, the ecological factor association model can be a Bayesian network, a structural equation model, a causal discovery algorithm model, or a custom rule network, such as an if-then rule set built based on expert knowledge, etc.

[0087] In this embodiment, the ecological logical correlation analysis refers to using a pre-defined ecological factor correlation model to analyze the ecological patterns contained in the processed multi-source data. That is, it involves using computational methods to uncover hidden ecological patterns in the data, including driving relationships, response mechanisms, and synergistic effects between factors.

[0088] Specifically, the fundamental factors can be represented as environmental background data that drives or restricts the vegetation restoration process, and can be the causes or conditions of the restoration state. These can include soil temperature, soil moisture content, soil electrical conductivity, soil pH, topographic slope, light intensity, etc. State indicators can be represented as characterizing data that directly reflect vegetation growth vitality and community structure dynamics. These can include the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Leaf Area Index (LAI), vegetation cover, etc. Validation true values ​​can be represented as data obtained through field measurements, used to calibrate and validate data from other sources and model inferences, serving as a benchmark for assessing accuracy. These can include measured vegetation cover, plant species list and quantity (species richness), aboveground biomass (dry weight), etc.

[0089] In this embodiment, the conditional probability relationship can be expressed as the probability distribution of a variable (child node) in a Bayesian network equal probability model, given the value of its parent node variable.

[0090] Specifically, the basic factors, state indicators, and validation ground truth can be defined as nodes in the network. Time-aligned data is used as the training dataset input into the model. The connections between nodes (i.e., directed edges) can be determined based on data and / or prior expert knowledge. For example, node priority scores can be calculated based on mutual information and functional area weights to guide the construction of the network structure.

[0091] Once the network structure is determined, the parameters in the conditional probability table for each node can be learned based on the input data. This step ultimately calculates all the necessary conditional probability relationships.

[0092] The prediction error of the initially obtained conditional probability relationship can be calculated. If the error exceeds a preset threshold, an optimization mechanism is triggered, for example, by readjusting the prior weights based on ecological functional zones and iteratively performing structure learning and parameter learning until the model accuracy meets the requirements.

[0093] Step S110: Based on the conditional probability relationship, perform weighted fusion of the multi-source data and output the fused target monitoring data.

[0094] In this embodiment, the executing entity can directly use conditional probability relationships as weights to perform weighted fusion of multi-source data and output the fused target monitoring data. Specifically, for a certain state indicator to be estimated (e.g., vegetation cover), it has conditional probability relationships with multiple basic factors (e.g., soil moisture content, temperature). The magnitude of this conditional probability value can be used directly or after normalization as the weight of the corresponding basic factor data in supporting the estimation of the state indicator. The higher the correlation strength (the greater the conditional probability) of a factor, the heavier its data is in the fusion.

[0095] A probability threshold can also be pre-set in the execution entity. Only when the conditional probability between data groups is greater than this threshold is it considered that there is a sufficiently strong ecological logical connection between them, and they are allowed to participate in the fusion.

[0096] In a specific implementation plan, firstly, the implementing entity can perform validation on the multi-source data to ensure that the multi-source data has been time-aligned, spatially classified, and formatted uniformly. Then, it reads the probability relationship table generated in step S18 from the model storage, for example, the probability matrix of P (vegetation cover | soil factor).

[0097] Then, the conditional probability values ​​are mapped to fusion weights. For example, for each grid cell, a weight is calculated based on P (state index | base factor), where a higher probability value corresponds to a higher weight. The weights of the multi-source data are then normalized to ensure that the sum is 1.

[0098] Next, grid-by-grid processing is performed, independently weighting the multi-source data within each spatial grid cell. For example, a weighted average is calculated for the remote sensing data (weight 0.6), sensor data (weight 0.3), and measured data (weight 0.1) in cell A. If a source data differs significantly from other data, robust statistical methods (such as truncated mean) can be used to reduce the impact of outliers.

[0099] Finally, the fusion results can be converted to a standard format and output (e.g., HDF5), and metadata (e.g., fusion algorithm version, weight source, etc.) can be added.

[0100] This embodiment effectively addresses the spatiotemporal heterogeneity challenge of multi-source monitoring data by employing spatial grid partitioning and time alignment, establishing a unified spatiotemporal benchmark. Furthermore, it leverages an ecological factor correlation model to mine conditional probabilistic relationships between data points, overcoming the limitations of traditional simple weighted fusion and achieving data fusion based on ecological logical associations. This method significantly improves the quality and reliability of target monitoring data, providing a solid data foundation for accurate assessment of subsequent vegetation restoration progress, thus solving the key technical problem of low fusion quality in the specific scenario of temporary land occupation by transmission lines using traditional data fusion methods.

[0101] In some implementations, the step of performing time alignment processing on multi-source data within each spatial grid cell to align data from different sources to a unified time node in the time dimension includes:

[0102] Step S162: Using a preset sliding window method, divide the multi-source data with different acquisition frequencies into time windows.

[0103] In this embodiment, the parameters of the sliding window method are determined before the algorithm is executed. Specifically, these parameters may include the window length, which is the time span covered by a single window, and the sliding step size, which is the time interval between each window movement. The sliding step size can be equal to the window length, in which case the windows are connected end-to-end and do not overlap; the sliding step size can also be less than the window length, in which case there is partial time overlap between adjacent windows.

[0104] In this embodiment, the executing entity can generate a series of time windows on the time axis based on the aforementioned preset parameters. For each time window, the executing entity can traverse all multi-source data belonging to the current spatial grid cell and determine whether it falls within the start and end time range of the window based on the timestamp of each data point. A data point can belong to one or more time windows (in the case of overlapping windows).

[0105] Step S164: Align the multi-source data that are divided into the same time window to the representative time node of that time window.

[0106] In this embodiment, this step is used to assign a unified time identifier to all data within each time window, completing the final normalization in the time dimension. The representative time node is represented as a discrete absolute time point that represents the entire time window. It can be a midpoint within the window, such as the midpoint between the window's start and end times. It can also be the window's start or end point, etc.

[0107] In this embodiment, for high-frequency data (such as sensor data), all raw readings within the window (soil temperature data one per hour) can be statistically aggregated. Aggregation methods include, but are not limited to, calculating the arithmetic mean, median, maximum / minimum value, or integral value, etc.

[0108] In some implementations, the step of performing ecological logical correlation analysis on time-aligned multi-source data based on a preset ecological factor correlation model to identify the conditional probability relationship between basic factors, state indicators, and verification truth values ​​includes:

[0109] Step S182: Based on preset data classification rules, the time-aligned multi-source data is divided into basic factors, state indicators, and verification truth values.

[0110] In this embodiment, the preset data classification rule can be a pre-set set of logical judgment criteria or mapping relationships, used to divide the input data items into three predefined functional categories.

[0111] Specifically, this can be a mapping rule based on data sources. The implementing entity can maintain a classification mapping table, where the following definitions apply: data from soil sensors that measure soil physicochemical properties (e.g., temperature, moisture, electrical conductivity) are classified as basic factors. Vegetation spectral indices (e.g., NDVI, EVI) obtained through inversion from remote sensing platforms, or data from specific canopy sensors, are classified as state indicators. Vegetation community structure parameters or detailed soil physicochemical properties (e.g., species number, biomass, organic matter content) obtained from manual field surveys and sampling and subsequently analyzed in laboratories are classified as verification true values.

[0112] It can also be a classification rule based on metadata. The metadata accompanying each data point (e.g., measurement method, sensor model, data accuracy) can be used as a classification criterion. For example, data labeled as measured by quadrat surveys are classified as true values.

[0113] Step S184: Input the partitioned data into a preset Bayesian network model, and learn and output the conditional probability relationship between the basic factors, state indicators and verification truth values ​​through the Bayesian network model.

[0114] In this embodiment, the preset Bayesian network model can be a probabilistic graphical model with a predefined basic framework (e.g., the definition of node variables, possible discrete states). The model consists of two parts: structure (directed acyclic graph) and parameters (conditional probability table). Initially, the parameters are empty or have initial values ​​and need to be learned through data.

[0115] Specifically, each node in the network can correspond to a basic factor, state index, or a variable that verifies the truth value (e.g., soil moisture content node, NDVI node, vegetation cover_measured node, etc.). Each node variable can also be discretized into several ecologically meaningful state intervals. For example, the soil moisture content node can be defined as three states: {low, medium, high}.

[0116] In this implementation, the executing entity can automatically discover and quantify the probabilistic relationships between variables from the data using a preset algorithm. That is, it establishes a mapping function from input data to probability parameters, thereby determining the directed connections between nodes in the network (i.e., who is whose parent node).

[0117] Specifically, the preset algorithm can be a score-based search algorithm, for example, searching for a structure in the possible network structure space that best fits the input data. It can also be a constraint-based algorithm, for example, using statistical tests (such as the chi-square test) to determine the conditional independence between variables to derive the structure.

[0118] In this implementation, after the network structure is fixed, the specific values ​​in the conditional probability table of each node can be calculated based on the input dataset. For any node, its conditional probability table defines the probability of the node itself taking each state given a certain combination of states of all its parent nodes. For example, P(NDVI=high | soil moisture content=high, sunshine hours=high) = 0.92. This probability value is learned from the data. After the learning process is complete, the output of the Bayesian network model is a complete set of conditional probability relationships. These relationships are stored in the model in the form of conditional probability tables, quantifying the strength and uncertainty of all ecological logical associations identified by the model.

[0119] In some implementations, the step of inputting the partitioned data into a preset Bayesian network model, and learning and outputting the conditional probability relationship between the basic factors, state indicators, and verification truth values ​​through the Bayesian network model, includes:

[0120] Step S1842: Based on the ecological functional zone type to which the spatial grid unit belongs, assign functional zone weights to the basic factors, state indicators, and verification truth values.

[0121] In this implementation, prior knowledge of ecological functional zones can be introduced, thereby enabling the subsequent Bayesian network structure learning process to have a clear ecological goal orientation, thus overcoming the limitation of traditional algorithms that rely solely on pure mathematical relationships and ignore actual ecological functional needs.

[0122] Specifically, the ecological functional zone type refers to the spatial zoning of temporary land occupied by power transmission lines based on the dominant ecological functions, protection objectives, and management needs of the region. Each functional zone has a differentiated focus on vegetation restoration and evaluation standards. For example, there are soil and water conservation zones (areas whose core function is to prevent soil erosion and reduce soil loss), biodiversity conservation zones (areas whose core function is to protect and restore plant species diversity and maintain community structure stability), and general restoration zones (areas with relatively low ecological sensitivity, whose main goal is routine vegetation restoration and greening), etc.

[0123] In this implementation, the executing entity can pre-store or receive a functional area-weight configuration table via an interface. This table defines the relative importance weights assigned to the three major categories of data—basic factors, state indicators, and verification truth values—under different ecological functional area types.

[0124] Specifically, in soil and water conservation areas, implementing entities can assign higher weights (e.g., 0.6) to basic factors (soil moisture content, erosion resistance), medium weights (e.g., 0.3) to state indicators (vegetation cover), and lower weights (e.g., 0.1) to verification values ​​(species richness). This reflects the region's high level of attention to soil baseline conditions and land cover status.

[0125] In biodiversity conservation areas, implementing entities can assign higher weights (e.g., 0.4 and 0.4, respectively) to validation ground truths (species richness, community structure) and state indicators (indices characterizing habitat complexity), while assigning relatively lower weights (e.g., 0.2) to basic factors. This reflects the region's high reliance on direct observational data of biological communities.

[0126] Step S1844: Calculate the priority score of each node based on the functional area weights and the mutual information between nodes in the Bayesian network model.

[0127] In this implementation, mutual information is represented as an information-theoretic metric used to quantify the degree of statistical dependence between two random variables (i.e., two Bayesian network nodes). A higher mutual information value indicates that the two variables share more information and have a closer relationship.

[0128] Specifically, for each node in the Bayesian network, firstly, the functional area weight assigned to it in step S1842 is obtained; then, the mutual information value between the node and all other nodes in the network is calculated; next, a priority score can be calculated by multiplying the functional area weight by the sum of the mutual information between the node and all other nodes. This priority score characterizes the ecological importance of the node in the current functional area, as well as the breadth and density of its connections throughout the data network (reflected by the sum of mutual information).

[0129] In a specific implementation, the priority score of a node in a Bayesian network model can be calculated in the following way;

[0130]

[0131] In the formula, This represents the priority score of the i-th node in the Bayesian network model. This represents the functional area weight of the i-th node in the Bayesian network model. This represents the mutual information between the i-th node and the j-th node in a Bayesian network model.

[0132] Step S1846: Sort the nodes according to the priority scores to establish a weighted Bayesian network structure model.

[0133] In this implementation, the executing entity can sort all nodes in descending order according to their calculated priority scores. The node with the highest score is placed at the top of the list. This sorting defines the processing order of nodes in the subsequent network structure learning algorithm. That is, high-priority nodes will be processed first, allowing them to have a greater probability of becoming the parent node (i.e., the cause node) of other nodes. Then, a preset Bayesian network structure learning algorithm (such as the K2 algorithm) is used, with this sorted list of nodes as input, to perform network structure search. The final learned weighted Bayesian network structure model is a directed acyclic graph whose structure reflects the ecological functional zone preferences. In this structure, variables that can be assigned high importance in certain functional zones tend to be located upstream in the causal chain.

[0134] Step S1848: Based on the weighted Bayesian network structure model, determine the conditional probability relationship between the basic factors, state indices, and verification truth values.

[0135] In this embodiment, with the network structure (i.e., directed edges) determined in step S1846 remaining fixed, the executing entity can use time-aligned, partitioned multi-source data as the training dataset. Then, a parameter learning algorithm (e.g., maximum likelihood estimation or Bayesian estimation) is used to calculate the conditional probability table for each node in the network. Each value in the conditional probability table quantifies the probability that a variable will take a certain state given the state of its parent variable. For example, in a model for soil and water conservation areas, a conditional probability relationship of P(vegetation cover = high | soil moisture content = high) = 0.90 can be learned.

[0136] The set of all the values ​​learned and populated into the conditional probability table constitutes the conditional probability relationship between the base factors, state indices, and verification truth values ​​in the final output. This relationship serves as the direct basis for subsequent weighted data fusion.

[0137] In some implementations, the step of determining the conditional probability relationship between the basic factors, state indices, and verification truth values ​​based on the weighted Bayesian network structure model includes:

[0138] Step S18482: Analyze the weighted Bayesian network structure model to obtain the initial prediction conditional probability.

[0139] In this embodiment, probabilistic inference can be performed using a pre-constructed weighted Bayesian network structure model, that is, given the known state (evidence) of some variables (nodes), the probability distribution of the states of other variables (nodes) in the network is calculated.

[0140] Specifically, standard inference algorithms for Bayesian networks (e.g., variable elimination, connection tree algorithms) can be used to input the observed values ​​of some variables from time-aligned multi-source data as evidence into the model. The goal of inference computation is to obtain the model's probabilistic predictions of the states of nodes that validate the true values. For example, by inputting the observed states of soil moisture content (basic factor) and NDVI (state index) into the model, the model is asked to output the probabilities of vegetation cover_measured (validating the true value) being high, medium, or low. The model can calculate the probability values ​​based on the current structure and parameters (conditional probability table), which are the initial predicted conditional probabilities. This represents the model's preliminary quantification of the uncertainty of ecological relationships under the current configuration.

[0141] Step S18484: Calculate the prediction error between the initial predicted conditional probability and the actual probability value.

[0142] In this embodiment, the actual probability value can be expressed as the probability value used in the evaluation and regarded as the benchmark.

[0143] Specifically, with sufficient historical data or test datasets, empirical probabilities can be directly derived statistically from the data. For example, historical data shows that when soil moisture content and NDVI are high, the measured vegetation cover is frequently high. When sufficient data is lacking, ecological experts can use their expertise and experience to establish an acceptable probability range or expected value for the relationships between relevant variables as a benchmark.

[0144] In this embodiment, the executing entity can compare the initial predicted conditional probability output by the model with the actual probability values ​​mentioned above. For multiple prediction points, statistical error measurement methods, such as calculating the root mean square error (RMSE) or mean absolute error (MAE), can be used to obtain a comprehensive, scalar prediction error value.

[0145] Step S18485: Determine the error threshold based on the statistical value of the prediction error calculated in multiple consecutive calculations.

[0146] In this embodiment, the prediction error calculated in multiple consecutive iterations can be represented as a series of prediction error values ​​calculated in multiple iterations or on different subsets of data during the historical optimization process.

[0147] Specifically, the mean and standard deviation of the above-mentioned consecutive prediction errors can be calculated, and the error threshold can be determined by the mean and standard deviation.

[0148] Step S18486: Determine whether the prediction error is greater than the error threshold. If yes, reallocate the functional area weights and return to the step of calculating node priority scores to update the weighted Bayesian network structure model until the prediction error is not greater than the error threshold, and output the conditional probability relationship based on the final model; if no, output the conditional probability relationship based on the current weighted Bayesian network structure model.

[0149] In this implementation, the executing entity can compare the currently calculated prediction error with an error threshold. If the prediction error is greater than the error threshold, it means that the prediction accuracy of the current model has not reached the preset standard. In this case, the weight allocation of the three categories—basic factors, state indicators, and verification ground truth—under the current ecological functional zone type can be adjusted based on the error analysis results. For example, if the model's prediction of verification ground truth consistently deviates significantly, the functional zone weights for verification ground truth can be appropriately increased to enhance their influence in subsequent network structure learning. The executing entity can then return to step S1844 (calculate node priority scores) with the newly allocated functional zone weights. Subsequent steps S1846 (establish a weighted model) and S1848 (determine conditional probabilities) will be re-executed based on the new weights, thereby generating an updated weighted Bayesian network structure model whose structure may have been optimized. This process continues iteratively until the prediction error of the new model is less than the error threshold.

[0150] In some implementations, the step of weightedly fusing multi-source data according to the conditional probability relationship and outputting fused target monitoring data includes:

[0151] Step S1102: Based on the ecological functional zone type to which the spatial grid unit belongs, determine the fusion priority weight of the basic factors, state indicators and verification truth values.

[0152] In this embodiment, the execution entity can pre-store an ecological functional zone-fusion weight mapping table. This mapping table defines the pre-set weight values ​​for three types of data—basic factors, state indicators, and verification truth values—under different ecological functional zone types.

[0153] The executing entity can read the corresponding weight combination from the mapping table according to the functional area to which the current spatial grid cell belongs.

[0154] Step S1104: Select a target probability relationship that is greater than a preset probability threshold from the conditional probability relationships.

[0155] In this embodiment, the preset probability threshold can be a pre-set threshold value used to filter conditional probability relationships. This threshold defines the strength of ecological associations considered valid in the fusion process. For example, the preset probability threshold can be set to 0.7, meaning that only when the conditional probability between two variables reaches 0.7 or higher is a sufficiently strong dependency considered to exist, which can then be used to guide the fusion of these two types of data.

[0156] In this implementation, all conditional probability relationships learned by the Bayesian network model can be traversed. For each conditional probability relationship, its probability value can be compared with a preset probability threshold. All conditional probability relationships greater than the threshold are filtered out to form the target probability relationship set.

[0157] Step S1106: Based on the fusion priority weight, perform weighted calculation on the multi-source data corresponding to the target probability relationship, and output the fused target monitoring data.

[0158] In this implementation, for the current spatial grid cell and the unified time node, the executing entity can select only the data associated with the target probability relationship set to participate in the fusion. For example, if soil moisture content → vegetation cover is a target probability relationship, then the soil moisture content data and vegetation cover data of the current cell can be selected.

[0159] In this embodiment, weighting can be performed directly using a linear weighted fusion method.

[0160] This embodiment provides a method for assessing the progress of vegetation restoration on temporary land occupied by power transmission lines. The execution entity of the method can be a server, including:

[0161] Step S22: Determine the ecological functional zone type of the temporary land occupied by the transmission line, and configure the corresponding evaluation index set for the ecological functional zone type.

[0162] In this embodiment, the executing entity can access a pre-stored digital map (vector data) of ecological functional zoning. Through spatial querying, based on the geographic coordinates of the area, it is located within a specific functional zone polygon, thereby automatically determining its ecological functional zone type.

[0163] In this implementation, the implementing entity can activate or invoke a pre-defined set of evaluation parameters for measuring vegetation restoration progress, based on a pre-defined ecological functional zone type. Specifically, for soil and water conservation zones, the evaluation indicator set can focus on soil stability and land cover. Specific indicators may include: vegetation cover, soil erosion modulus, root depth, and surface roughness. For biodiversity conservation areas, the evaluation indicator set can focus on community structure and species composition. Specific indicators may include: species richness, Simpson diversity index, proportion of native species, and community hierarchical structure. For general restoration zones, the evaluation indicator set can focus more on basic vegetation growth conditions, such as vegetation cover, vegetation height, and biomass. The above indicator sets and their correspondence with functional zones can be preset and stored in the implementing entity's knowledge base or configuration file.

[0164] Step S24: Based on the preset weight allocation rules, assign indicator weights to each evaluation indicator in the evaluation indicator set.

[0165] In this embodiment, weights can be assigned to each evaluation indicator in the evaluation indicator set based on an expert scoring method (e.g., the Analytic Hierarchy Process, AHP). Specifically, the implementing entity can store or receive the weight results obtained by a group of experts, including ecologists and electrical engineering experts, who perform pairwise comparisons and scoring on the evaluation indicator set for each functional area, and then process the results.

[0166] In this embodiment, the entropy weight method can also be used to assign weights to each evaluation indicator in the evaluation indicator set.

[0167] In this embodiment, the fixed weight method can also be used to assign weights to each evaluation indicator in the evaluation indicator set.

[0168] Step S26: Construct a quantitative assessment model for recovery progress based on the set of assessment indicators and the weights of each assessment indicator.

[0169] In this embodiment, the quantitative assessment model for vegetation restoration progress can be a mathematical function or calculation rule. Its inputs are the specific values ​​and weights of each assessment indicator, and its output is a comprehensive quantitative value of vegetation restoration progress (which can be a score or an index). This quantitative assessment model for vegetation restoration progress can be a linear weighted comprehensive model or a nonlinear model, etc.

[0170] Step S28: Input the target monitoring data into the restoration progress quantitative assessment model to obtain the vegetation restoration progress assessment result; wherein, the target monitoring data is target monitoring data obtained by one of the target monitoring data generation methods described above.

[0171] In this embodiment, the executing entity can extract data values ​​corresponding to each indicator defined in the evaluation indicator set from the target monitoring data. For example, the field value of integrated vegetation coverage in the target monitoring data can be used as the input value of the evaluation indicator vegetation coverage. Then, the extracted indicator values ​​and their assigned weights from step S24 are input into the quantitative assessment model of restoration progress (linear weighted model) constructed in step S26. Next, the model performs calculations and outputs a quantitative assessment result of vegetation restoration progress. This result can be a numerical score (e.g., 85 out of 100) or a grade (e.g., excellent, good, average, poor).

[0172] This embodiment provides a method for determining ecological management decisions regarding vegetation restoration for temporary land occupied by power transmission lines. The method can be implemented by a server and includes the following steps:

[0173] Step S32: Determine the current restoration status level based on the vegetation restoration progress assessment results; wherein, the vegetation restoration progress assessment results are obtained by using the vegetation restoration progress assessment method for temporary land occupation by power transmission lines as described above.

[0174] In this embodiment, the executing entity can map continuous vegetation restoration progress assessment scores to discrete state categories with clearly defined upper and lower limits based on a preset grading standard.

[0175] In one specific implementation scheme, the implementing entity can receive the vegetation restoration progress assessment result obtained through the aforementioned assessment method. This result can be a numerical score (e.g., 85 points) or a preliminary grade label. The implementing entity can then access a pre-defined status grade classification table. This table defines the correspondence between different score ranges (or initial grades) and the final restoration status grade. The restoration status grade can include four levels: Excellent, Good, Average, and Poor. The corresponding score ranges can be pre-defined as: [90, 100] for Excellent, [75, 90) for Good, [60, 75) for Average, and [0, 60) for Poor.

[0176] The implementing entity can compare the obtained assessment score with the grading criteria to determine its corresponding range, thereby outputting the final current recovery status level. For example, an assessment score of 85 will be mapped and classified as a "good" level.

[0177] Step S34: Match one or more candidate decision schemes corresponding to the current recovery status level according to the preset decision rule library.

[0178] In this embodiment, the preset decision rule base can be a structured knowledge base that stores a set of rules specifying "what measures should be taken under what conditions". The rules can exist in the form of "IF-THEN".

[0179] In this embodiment, the executing entity can use the current recovery status level determined in step S32, along with the optional ecological functional zone type and key indicator deficiency information, as query conditions to search the decision rule base. All rules whose conditions match the current status will have their corresponding candidate decision schemes extracted to form a set of candidate decision schemes.

[0180] Step S36: Based on the preset optimization objective, determine the optimal decision scheme from the one or more candidate decision schemes.

[0181] In this embodiment, the preset optimization objective can be a multi-objective system, which may include: Expected ecological benefits: the expected improvement in recovery progress after the implementation of the plan; Implementation costs: the economic costs of manpower, materials, machinery, etc., required by the plan; Implementation timeliness: the time period required for the plan to produce significant effects from initiation; Operational feasibility: the difficulty of implementation under current site conditions, technical capabilities, and policies and regulations; Environmental risks: the potential negative impacts of the plan, such as soil compaction, species invasion risks, etc. In this embodiment, the implementing entity can quantify and score each candidate decision plan against the above optimization objectives. The scoring data can come from historical implementation effect statistics or expert experience assignments, etc. In this embodiment, the implementing entity can also use a multi-objective decision-making algorithm to comprehensively compare the scores of each plan. Specifically, a weight is assigned to each optimization objective, the weighted average score of each plan is calculated, and the plan with the highest score is selected.

[0182] Based on the above analysis, the executing entity ultimately outputs a decision scheme that is determined to be optimal under the current constraints.

[0183] In one specific implementation plan, a visualization platform, method, and system for monitoring the vegetation restoration progress of temporary land occupied by power transmission lines are provided.

[0184] Specifically, the accelerated construction of my country's new power system is pushing the transmission network to ecologically sensitive areas. In regions like the Qinghai-Tibet Plateau, the temporary land occupation for line construction has led to significant problems such as soil disturbance and water erosion caused by the destruction of surface vegetation. Vegetation restoration in these areas is characterized by long cycles, low ecological thresholds, and varying functional zone standards. Traditional management models have fundamental shortcomings in data integration and measure determination.

[0185] The lack of a systematic data fusion mechanism prevents the release of the value of multi-source data. Specifically, existing monitoring relies on a multi-source model of manual measurement, remote sensing interpretation, and sensor acquisition, but lacks a unified fusion framework. Current fusion methods mostly use simple weighting, without considering the special characteristics of ecological monitoring or establishing functional zone adaptation models.

[0186] The measures lack data support, resulting in a serious lack of precise policy implementation capabilities. Specifically, the formulation of measures relies on empiricism, lacks integrated data-driven approaches, and fails to establish a linkage mechanism between monitoring data, recovery effects, and measure optimization. As a result, measure adjustments become a passive response. The recovery targets of different ecological functional zones vary significantly, but the existing measures have not utilized integrated data to construct quantitative models.

[0187] To address the aforementioned issues, there is an urgent need to construct a visualization platform, method, and system for the vegetation restoration progress of temporary land occupied by power transmission lines, integrating multi-source data fusion, quantitative assessment, dynamic visualization, and intelligent decision-making. This would improve the efficiency and accuracy of ecological restoration management and ensure the coordinated development of power transmission line construction and ecological protection.

[0188] Specific implementation plan 1:

[0189] This implementation plan provides a visualization platform for the vegetation restoration progress of temporary land occupied by power transmission lines, comprising: a data layer for collecting multi-source data related to vegetation restoration of temporary land occupied by power transmission lines; a processing layer for preprocessing and fusing the multi-source data to obtain target monitoring data; an evaluation layer for constructing a quantitative evaluation model of restoration progress in different ecological functional zones, and obtaining a restoration progress score by combining the target monitoring data; and a decision layer for generating maintenance and adjustment strategies for the vegetation on temporary land occupied by power transmission lines based on the score difference between the restoration progress score and a preset restoration score. In this embodiment, the multi-source data includes remote sensing data, IoT sensor data, and field measurement data. In this embodiment, the different ecological functional zones include soil and water conservation zones and biodiversity conservation zones. In this embodiment, the maintenance and adjustment strategies include the selection of planting vegetation types and the application parameters of water-retaining agents.

[0190] Specific Implementation Plan 2:

[0191] Based on implementation scheme 1, the implementation scheme of the present invention provides a visualization platform for the vegetation restoration progress of temporary land occupied by transmission lines, and further includes a visualization layer for dynamically visually displaying the vegetation restoration progress based on the restoration progress score.

[0192] The beneficial effects of the above design scheme are: by dynamically visualizing the vegetation restoration progress based on the restoration progress score, it breaks through the limitations of data abstraction, realizes intuitive perception of restoration progress, and provides a basis for accurately locating the temporary land occupation location.

[0193] Specific implementation plan 3:

[0194] Based on specific implementation scheme 1, this specific implementation scheme provides a visualization platform for the vegetation restoration progress of temporary land occupied by power transmission lines. The data layer includes: a remote sensing data acquisition unit for acquiring remote sensing images of vegetation restoration in temporary land occupied by power transmission lines and extracting vegetation growth index data from the vegetation restoration remote sensing images; a sensor data acquisition unit for acquiring ecological sensor data at key points in the temporary land occupied by power transmission lines; a measured data acquisition unit for collecting vegetation data using a sampling method on a quarterly basis in the temporary land occupied by power transmission lines to obtain measured data; and a transmission unit for transmitting the acquired vegetation growth index data, ecological sensor data, and measured data as a multi-source data to the processing layer. In this embodiment, vegetation growth indicators include normalized difference vegetation index, enhanced vegetation index, vegetation cover, and net primary productivity. In this embodiment, ecological sensor data includes soil temperature and humidity, moisture content, electrical conductivity, plant height, and canopy width. The beneficial effects of the above design scheme are as follows: by extracting vegetation growth index data through remote sensing data, large-scale dynamic monitoring can be achieved; by collecting ecological sensing data at key points in the temporary occupation of power transmission lines, micro-ecological data at key points can be collected in real time, avoiding the lag of manual measurement; by collecting vegetation data in the temporary occupation of power transmission lines using the quadrat method on a quarterly basis, measured data can be obtained, which can provide accurate vegetation parameters, calibrate the errors of remote sensing and sensor data, and solve the problems of inefficiency, fragmentation and limited coverage of traditional monitoring data.

[0195] Specific implementation plan 4:

[0196] Based on specific implementation scheme 1, this invention provides a visualization platform for the vegetation restoration progress of temporary land occupied by power transmission lines. The processing layer includes: a data processing unit for cleaning and standardizing multi-source data to obtain standard multi-source data; and a data fusion unit for fusing the standard multi-source data to obtain target monitoring data. The beneficial effects of the above design scheme are: by preprocessing and fusing multi-source data to obtain target monitoring data, data quality is improved, and the problems of fragmented multi-source data that cannot be directly used for evaluation are solved.

[0197] Specific implementation plan 5:

[0198] Based on specific implementation scheme 4, this specific implementation scheme provides a visualization platform for the vegetation restoration progress of temporary land occupied by transmission lines. The data fusion unit includes: a spatial fusion unit, used to divide the temporary land occupied by transmission lines into spatial grids to obtain several spatial grids, and to divide multi-source data of the same spatial grid standard into the same spatial data to obtain several sets of spatial data; a temporal fusion unit, used to perform temporal alignment on each set of spatial data based on sliding window alignment to obtain initial aligned data; and a data completion unit, used to determine data missing data in the initial aligned data at the same time to obtain the data missing type, and to determine the data missing type based on the initial aligned data corresponding to other data types and the initial aligned data corresponding to other grids under the same data missing type. The data completion process yields target-aligned data. A data association unit divides the target-aligned data into three types: basic factors, state indicators, and verification truth values. These basic factors, state indicators, and verification truth values ​​are input into a Bayesian network model for identification. The conditional probabilities between the basic factors, state indicators, and verification truth values ​​are determined, and based on these conditional probabilities, the basic factors, state indicators, and verification truth values ​​are associated to obtain associated data groups. A weighted fusion unit sets fusion priority weights for the basic factors, state indicators, and verification truth values ​​based on the functional area type of the spatial grid. It selects target associated data from the associated data groups with a probability greater than a preset conditional probability and fuses the target associated data based on the fusion priority weights to obtain target monitoring data. In this embodiment, the specific steps for selecting target associated data from the associated data groups with a probability greater than a preset conditional probability and fusing the target associated data based on the fusion priority weights to obtain target monitoring data are as follows:

[0199] Calculate the total priority weight of the current target associated data based on the fusion priority weight;

[0200]

[0201] in, This indicates the total priority weight of the data associated with the current target. This indicates the fusion priority weight of the basic factors in the current target associated data. This indicates the fusion priority weight of status indicators in the current target associated data. This indicates the fusion priority weight for verifying truth values ​​in the current target associated data;

[0202] All target-related data are sequentially merged according to their overall priority weight to obtain target monitoring data.

[0203] In this specific implementation plan, all target-related data are sequentially fused according to the overall priority weight to obtain target monitoring data, ensuring the orderly fusion of data and prioritizing the fusion of important data to ensure fusion quality.

[0204] In this specific implementation, the time dimension adopts a sliding window pair: if the remote sensing data is monthly (such as January 15, February 12), the high-frequency sensor data (once per hour) within that month is averaged within a window of ±15 days from the remote sensing imaging time. At the same time, 1-2 measured data within that month (such as January 20, February 5) are associated with the corresponding window to avoid the problem of no data on fixed dates.

[0205] In this specific implementation plan, if the normalized vegetation index (NVI) data for a certain grid is missing, it is supplemented by a random forest model based on the NVI of adjacent grids and the soil temperature, humidity and precipitation data of that grid (e.g., the NVI should show an increasing trend when soil moisture content increases, to avoid contradictions between the supplemented value and ecological logic). For example, if sensor data is missing, it is inferred by back-calculation based on the rate of change of the NVI of the same grid (vegetation growth index) and the measured soil erosion resistance (e.g., measured value in January) (e.g., when the NVI increases by 5%, the soil moisture content should increase by 3%-5%, which is consistent with the correlation mechanism between vegetation water consumption and soil moisture).

[0206] In this specific implementation plan, the basic factors include soil moisture content, soil temperature, soil electrical conductivity, etc., the state indicators include vegetation growth index data, macro growth vitality, high coverage area precision optimization, etc., and the verification true values ​​include vegetation coverage, species richness, soil erosion resistance, etc.

[0207] In this specific implementation plan, the conditional probabilities are, for example, that when the soil moisture content is ≥18% and the soil temperature is ≥10℃, the probability of macroscopic growth vitality ≥0.6 is 85%, and for example, that when the macroscopic growth vitality is ≥0.6, the probability of vegetation coverage is 90%.

[0208] In this specific implementation plan, for biodiversity conservation areas, the focus is on integrating vegetation accuracy and soil data in high-coverage areas, and for soil and water conservation areas, the focus is on integrating vegetation cover and soil data. In this specific implementation plan, the basic factors, state indicators, and validation ground truths are input into a Bayesian network model for identification, determining the conditional probabilities between the basic factors, state indicators, and validation ground truths, including: assigning functional area weights to the basic factors, state indicators, and validation ground truths based on functional area characteristics.

[0209] Based on the basic factors, state indicators, and verification true values, and their corresponding functional area weights, establish the correspondence between the nodes of the Bayesian network model and the basic factors, state indicators, and verification true values, and determine the indicators corresponding to the nodes.

[0210] Calculate the priority score of nodes in a Bayesian network model;

[0211]

[0212] in, This represents the priority score of the i-th node in the Bayesian network model. This represents the functional area weight of the i-th node in the Bayesian network model. This represents the mutual information between the i-th node and the j-th node in a Bayesian network model.

[0213] Establish a weighted Bayesian network structure model by arranging nodes in descending order of their scores.

[0214] Based on the weighted Bayesian network structure model, the initial prediction conditional probability is determined by analyzing the basic factors, state indices, and verification truth values.

[0215] Calculate the prediction error of the initial prediction conditional probability. ;

[0216]

[0217]

[0218] Where n represents the number of initial prediction condition probabilities, Let represent the value of the conditional probability of the 'a'th initial prediction. This represents the a-th actual probability value. Indicates the error threshold. This represents the mean of three consecutive prediction errors. This represents the confidence level, with a value of 0.933.

[0219] Determine if the prediction error is less than the error threshold. If so, keep the weighted Bayesian network structure unchanged and continue with the subsequent conditional probability calculation and analysis.

[0220] Otherwise, the weights of the basic factors, state indicators, and functional areas for verifying the truth value are reallocated to adjust the weighted Bayesian network structure model, and the final conditional probability is obtained after the adjustment.

[0221] In this specific implementation plan,

[0222]

[0223] in express The probability of them happening simultaneously express The probability of occurrence express The probability of occurrence.

[0224] In this specific implementation scheme, the traditional K2 algorithm in the Bayesian model requires manual specification of the node order and does not consider the differences in functional areas. This scheme automatically learns and adapts the network structure to different functional areas to ensure prediction accuracy.

[0225] In this specific implementation plan, for example in a soil and water conservation area, the functional zone weight of basic factors is 0.6, the functional zone weight of state indicators is 0.3, and the functional zone weight of verification truth value is 0.1, which is used to adjust node priority. For another example, in a biodiversity conservation area, the functional zone weight of basic factors is 0.3, the functional zone weight of state indicators is 0.4, and the functional zone weight of verification truth value is 0.3.

[0226] In this specific implementation plan, the model is iterated and updated by calculating the difference between the predicted conditional probability and the actual value, so as to ensure the timeliness and accuracy of the model.

[0227] The beneficial effects of the above scheme are as follows: by configuring functional area weights based on functional area characteristics, the problem of model lacking functional area differentiation adaptation is solved; by calculating node priority scores, the problem of fixed node order is solved, making the node priority ranking more in line with functional area goals and ecological laws, providing a scientific basis for the node order in subsequent network structure construction; by calculating the prediction error of the initial prediction conditional probability, a measurable judgment standard is provided for subsequent model adjustment; by judging whether the error is less than the threshold, it is decided whether to maintain or adjust the model, ensuring the timeliness and accuracy of the model, and ultimately improving the efficiency and accuracy of ecological restoration management, ensuring the coordinated development of power transmission construction and ecological protection.

[0228] The beneficial effects of the above design scheme are as follows: By dividing multi-source data into the same spatial grid standard into several sets of spatial data, all multi-source data are anchored to the same spatial cell, solving the problem of spatial misalignment of data. Initial aligned data is obtained by temporally aligning each set of spatial data based on sliding window alignment, preserving key temporal features. By combining correlation of similar data types and cross-grid references of the same missing type, error accumulation from a single completion method is avoided, providing a reliable data foundation for subsequent evaluation. Furthermore, by dividing the target aligned data into three types—basic factors, state indicators, and verification ground truth—the system clearly defines... The role of data in the ecosystem is defined by determining the conditional probabilities between basic factors, state indicators, and verification truth values ​​through a Bayesian network model. This upgrades data from scattered indicators into interpretable causal chains, providing a basis for subsequent deviation tracing. By setting the fusion priority weights for basic factors, state indicators, and verification truth values ​​based on the functional area type of the spatial grid, the fusion results are ensured to be highly matched with functional objectives. By selecting target associated data with a probability greater than the preset conditional probability from the associated data group, redundant information interference is avoided, improving fusion efficiency and evaluation accuracy. This perfectly adapts to the refined management needs of vegetation restoration for temporary land occupation by transmission lines.

[0229] Specific Implementation Plan 6:

[0230] Based on specific implementation scheme 1, this specific implementation scheme provides a visualization platform for the vegetation restoration progress of temporary land occupied by transmission lines. The evaluation layer includes:

[0231] The system comprises four sub-units: an indicator configuration unit, used to identify the ecological functional zones temporarily occupied by power transmission lines and to configure corresponding evaluation indicators for these zones; a weight determination unit, used to score the importance of the evaluation indicators for each ecological functional zone based on assessments by ecologists and power engineering experts, and to determine the weights of the evaluation indicators according to the scoring results; a model construction unit, used to construct a quantitative evaluation model for restoration progress based on the evaluation indicators and their weights; and a score determination unit, used to input the target monitoring data into the quantitative evaluation model for restoration progress to obtain a score for the restored scenic area. The beneficial effects of this design scheme are: by constructing quantitative evaluation models for restoration progress in different ecological functional zones and combining them with target monitoring data, restoration progress scores are obtained, meeting the differentiated needs of ecological functions, achieving accurate assessment by zoning and classification, and providing objectivity in the assessment through quantitative progress scores, facilitating progress comparison and traceability.

[0232] Specific implementation plan 7:

[0233] Based on specific implementation scheme 1, this specific implementation scheme provides a visualization platform for the vegetation restoration progress of temporary land occupied by transmission lines, wherein the decision-making layer includes:

[0234] The system comprises four main components: a grade determination unit, a relationship determination unit, and a scope determination unit. The first component determines the maintenance grade based on the difference between the recovery progress score and the preset recovery score, using a difference-dimensional grade matching table. The second component determines the contribution of each evaluation indicator to the score difference and establishes a correspondence between the evaluation indicators and maintenance measures in the basic maintenance strategy. The third component calculates the weighted maintenance strategy by weighting the maintenance measures based on their contribution and the correspondence. The fourth component calculates the scope of each measure's parameter values ​​in the weighted maintenance strategy, obtains future environmental parameters for temporary land occupation of transmission lines based on an environmental prediction model, and determines the future environmental parameters for each measure's parameter values ​​based on feasibility and compatibility. The system employs a multi-stage approach: an implementation effectiveness score is given, and parameter values ​​for measures with implementation effectiveness scores lower than preset scores are adjusted to obtain a target range of parameter values. A strategy adjustment unit adjusts the weighted maintenance strategy based on this target range of parameter values ​​to obtain a range-based maintenance strategy. A strategy determination unit uses maintenance benefits, maintenance costs, and maintenance timeliness as strategy evaluation indicators, determines the maintenance weights for these indicators based on current needs, establishes a strategy evaluation system, selects different maintenance parameter groups from the range-based maintenance strategies using a sampling method, comprehensively evaluates these groups based on the strategy evaluation system, obtains a maintenance evaluation value, and selects the maintenance parameter group with the highest evaluation value as the maintenance adjustment strategy for temporarily occupied vegetation. In this specific implementation plan, the contribution of each evaluation indicator to the score difference is determined using the Sharpe ratio method. Specifically, adjusting parameter values ​​for measures with implementation effectiveness scores lower than preset scores involves determining the adjustment range based on the score difference.

[0235] In this specific implementation plan, based on contribution and corresponding relationships, the maintenance measures in the basic maintenance strategy are weighted to obtain a weighted maintenance strategy, including:

[0236] Based on the contribution level, determine the adjustment range of maintenance measures in the basic maintenance strategy;

[0237]

[0238] in, Indicates the extent of the adjustment of the measures. Indicates the degree of contribution;

[0239] The maintenance measures in the basic maintenance strategy are weighted according to the adjustment range of the measures to obtain the weighted maintenance strategy.

[0240] In this specific implementation plan, based on feasibility and suitability, the implementation effect score of each measure parameter value under future environmental parameters is determined, including:

[0241]

[0242] in, This indicates the score for the implementation effect of current measures under future environmental parameters. This indicates the feasibility of the current measure's parameter value under future environmental parameters. This indicates the matching value between the current measure parameter value and the future environmental parameter value. Indicates the feasibility weight. This represents the matching weight.

[0243] In this specific implementation plan, the feasibility weight and the matching weight are determined based on actual needs and sum to 1.

[0244] The beneficial effects of the above design scheme are as follows: By determining the maintenance level from the difference between the recovery progress score and the preset recovery score in the difference-dimensional level matching table, the basic maintenance strategy under the maintenance level is obtained, realizing the hierarchical response to maintenance needs. By determining the contribution of each evaluation indicator to the score difference, the correspondence between the evaluation indicators and the maintenance measures in the basic maintenance strategy is established. Based on the contribution and the correspondence, the maintenance measures in the basic maintenance strategy are weighted to obtain the weighted maintenance strategy, realizing targeted optimization of different measures. By obtaining the parameter values ​​of the measures in the weighted maintenance strategy, the future environmental parameters of the temporary occupation of the transmission line are obtained based on the environmental prediction model. Based on feasibility and matching, the implementation effectiveness of each measure parameter value under the future environmental parameters is determined. The evaluation process involves scoring the implementation effect and adjusting the parameter values ​​of measures that are lower than the preset scores to obtain the target range of parameter values. This adapts to future environmental changes, prevents measures from failing due to sudden environmental changes, and improves the adaptability of measure parameters to environmental changes. A strategy evaluation system is established by using maintenance benefits, maintenance costs, and maintenance timeliness as strategy evaluation indicators and determining the maintenance weights for these factors based on current needs. Different maintenance parameter groups are selected from the range of maintenance strategies using a sampling method. The maintenance parameter groups are then comprehensively evaluated based on the strategy evaluation system to obtain maintenance evaluation values. The maintenance parameter group with the highest maintenance evaluation value is selected as the maintenance adjustment strategy for temporarily occupied vegetation, achieving optimal decision-making with multi-objective balance, maximizing the satisfaction of current needs, and improving maintenance efficiency.

[0245] Specific implementation plan 8:

[0246] Based on specific implementation scheme 7, this specific implementation scheme provides a visualization platform for the vegetation restoration progress of temporary land occupied by transmission lines. The weighting unit includes: a determination unit, used to determine the weighting adjustment amplitude based on contribution; and an adjustment unit, used to adjust the maintenance measures based on the correspondence relationship to obtain the target maintenance measures, and to integrate the target maintenance measures to obtain a weighted maintenance strategy. The beneficial effect of the above design scheme is that, based on contribution and combined with the correspondence relationship, the maintenance measures in the basic maintenance strategy are weighted to obtain a weighted maintenance strategy, thereby achieving targeted optimization of different measures.

[0247] Specific Implementation Plan 9:

[0248] A specific embodiment of the present invention provides a method for the restoration of vegetation on temporary land occupied by power transmission lines, including:

[0249] S1: Collect multi-source data related to vegetation restoration for temporary land occupied by power transmission lines;

[0250] S2: Preprocess and fuse multi-source data to obtain target monitoring data;

[0251] S3: Construct a quantitative assessment model for the restoration progress of different ecological functional zones, and obtain a restoration progress score by combining target monitoring data;

[0252] S4: Based on the score difference between the recovery progress score and the preset recovery score, generate a maintenance and adjustment strategy for the vegetation temporarily occupying the land.

[0253] In this specific implementation plan, multi-source data includes remote sensing data, IoT sensor data, and field measurement data. Different ecological functional zones include soil and water conservation zones and biodiversity conservation zones. Maintenance and adjustment strategies include the selection of planting vegetation types and the application parameters of water-retaining agents. The beneficial effects of the above design scheme are: by collecting multi-source data, it solves the problems of inefficiency, fragmentation, and limited coverage of traditional monitoring data; by preprocessing and fusing multi-source data, it obtains target monitoring data, improving data quality and solving the problems of fragmented multi-source data that cannot be directly used for assessment; by constructing a quantitative assessment model of restoration progress for different ecological functional zones, combined with target monitoring data, it obtains restoration progress scores, meeting the differentiated needs of ecological functions and achieving accurate assessment by zoning and classification; by quantifying progress scores, it provides objectivity for assessment, facilitating progress comparison and traceability; by generating maintenance and adjustment strategies for temporarily occupied vegetation based on the score difference between the restoration progress score and the preset restoration score, it solves the problems of simply displaying data without targeted adjustment plans and poor restoration effects in traditional management, accurately locating restoration deviations, avoiding blind measures, and improving maintenance efficiency.

[0254] Specific implementation plan 10:

[0255] This invention provides a system for monitoring the vegetation restoration progress of temporary land occupied by power transmission lines, comprising: a data acquisition module for collecting multi-source data related to vegetation restoration of temporary land occupied by power transmission lines; a data processing module for preprocessing and fusing the multi-source data to obtain target monitoring data; a data evaluation module for constructing a quantitative evaluation model of restoration progress in different ecological functional zones and obtaining a restoration progress score by combining the target monitoring data; and a decision-making module for generating maintenance and adjustment strategies for the vegetation on temporary land occupied by power transmission lines based on the score difference between the restoration progress score and a preset restoration score. In this specific embodiment, the multi-source data includes remote sensing data, IoT sensor data, and field measurement data. In this specific embodiment, the different ecological functional zones include soil and water conservation zones and biodiversity conservation zones. In this specific embodiment, the maintenance and adjustment strategies include the selection of planting vegetation types and the application parameters of water-retaining agents. The beneficial effects of the above design scheme are as follows: By collecting multi-source data, it solves the problems of inefficiency, fragmentation, and limited coverage of traditional monitoring data; by preprocessing and fusing multi-source data, it obtains target monitoring data, improves data quality, and solves the problems of fragmented multi-source data that cannot be directly used for assessment; by constructing a quantitative assessment model for the restoration progress of different ecological functional zones, combined with target monitoring data, it obtains restoration progress scores, meets the differentiated needs of ecological functions, and achieves accurate assessment of zoning and classification; by quantifying the progress scores, it provides objectivity for assessment, facilitates progress comparison and traceability; by generating maintenance and adjustment strategies for temporarily occupied vegetation based on the score difference between the restoration progress score and the preset restoration score, it solves the problems of simply displaying data, lacking targeted adjustment plans, and having poor restoration effects in traditional management, accurately locates restoration deviations, avoids blind implementation of policies, and improves maintenance efficiency.

[0256] According to an embodiment of the present invention, a device for generating target monitoring data is provided, comprising: a partitioning module, used to partition the temporary land occupation of power transmission lines into spatial grids according to a preset scale to obtain a plurality of spatial grid units; a classification module, used to spatially classify the collected multi-source data according to the spatial grid units to which they belong, wherein the multi-source data includes remote sensing data, sensor data, and measured data; an alignment module, used to perform time alignment processing on the multi-source data within each spatial grid unit, so that data from different sources are aligned to a unified time node in the time dimension; a correlation analysis module, used to perform ecological logical correlation analysis on the time-aligned multi-source data based on a preset ecological factor correlation model, and identify the conditional probability relationship between basic factors, state indicators, and verification truth values; and a weighted fusion module, used to perform weighted fusion of the multi-source data according to the conditional probability relationship, and output the fused target monitoring data.

[0257] According to an embodiment of the present invention, an electronic device is provided; please refer to... Figure 3The electronic device in this embodiment may include one or more of the following components: a processor, a network interface, memory, non-volatile memory, and one or more application programs, wherein the one or more application programs may be stored in non-volatile memory and configured to be executed by one or more processors, and the one or more programs are configured to perform the methods as described in the foregoing method embodiments.

[0258] According to embodiments of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a computer, causes the computer to perform the method described in any of the above embodiments.

[0259] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for generating target monitoring data, characterized in that, include: The temporary land occupied by power transmission lines is divided into spatial grids according to a preset scale to obtain several spatial grid units; The collected multi-source data are spatially categorized according to their respective spatial grid units. The multi-source data includes remote sensing data, sensor data, and measured data. For multi-source data within each spatial grid cell, time alignment processing is performed to align data from different sources to a unified time node in the time dimension. Based on a pre-defined ecological factor association model, ecological logical association analysis is performed on time-aligned multi-source data to identify the conditional probability relationship between basic factors, state indicators, and verification truth values. Based on the conditional probability relationship, the multi-source data is weighted and fused to output the fused target monitoring data.

2. The generation method according to claim 1, characterized in that, The step of performing time alignment processing on multi-source data within each spatial grid cell to align data from different sources to a unified time node in the time dimension includes: Using a preset sliding window method, time windows are divided for the multi-source data with different acquisition frequencies; Multi-source data that are divided into the same time window are aligned to the representative time nodes of that time window.

3. The generation method according to claim 2, characterized in that, The steps of performing ecological logical correlation analysis on time-aligned multi-source data based on the preset ecological factor correlation model, and identifying the conditional probability relationships between basic factors, state indicators, and verification truth values, include: Based on preset data classification rules, the time-aligned multi-source data is divided into basic factors, state indicators, and verification truth values. The segmented data is input into a preset Bayesian network model, which learns and outputs the conditional probability relationship between the basic factors, state indicators, and verification truth values.

4. The generation method according to claim 3, characterized in that, The step of inputting the segmented data into a preset Bayesian network model, and learning and outputting the conditional probability relationship between the basic factors, state indicators, and verification truth values ​​through the Bayesian network model includes: Based on the ecological functional zone type to which the spatial grid unit belongs, functional zone weights are assigned to the basic factors, state indicators, and verification truth values. Based on the functional area weights and the mutual information between nodes in the Bayesian network model, the priority score of each node is calculated. The nodes are sorted according to the priority scores to establish a weighted Bayesian network structure model. Based on the weighted Bayesian network structure model, the conditional probability relationship between the basic factors, state indices, and verification truth values ​​is determined.

5. The generation method according to claim 4, characterized in that, The step of determining the conditional probability relationship between the basic factors, state indices, and verification truth values ​​based on the weighted Bayesian network structure model includes: Based on the weighted Bayesian network structure model, the initial prediction conditional probability is obtained through analysis. Calculate the prediction error between the initial predicted conditional probability and the actual probability value; The error threshold is determined based on the statistical value of the prediction error calculated in multiple consecutive calculations. Determine whether the prediction error is greater than the error threshold; If so, the functional area weights are reallocated, and the step of calculating node priority scores is returned to update the weighted Bayesian network structure model until the prediction error is no greater than the error threshold, and the conditional probability relationship is output based on the final model. If not, then output the conditional probability relationship based on the current weighted Bayesian network structure model.

6. The generation method according to claim 2, characterized in that, The step of weighted fusing of multi-source data according to the conditional probability relationship and outputting the fused target monitoring data includes: Based on the ecological functional zone type to which the spatial grid unit belongs, the fusion priority weights of the basic factors, state indicators, and verification truth values ​​are determined; Select a target probability relationship that is greater than a preset probability threshold from the conditional probability relationships; Based on the fusion priority weight, the multi-source data corresponding to the target probability relationship are weighted and calculated to output the fused target monitoring data.

7. A method for assessing the progress of vegetation restoration on land temporarily occupied by power transmission lines, characterized in that, include: Determine the ecological functional zone type for the temporary land occupation of the transmission line, and configure a corresponding set of evaluation indicators for the ecological functional zone type; Based on the preset weight allocation rules, an indicator weight is assigned to each evaluation indicator in the set of evaluation indicators; Based on the set of evaluation indicators and the weights of each indicator, a quantitative evaluation model for recovery progress is constructed. The target monitoring data is input into the restoration progress quantitative assessment model to obtain the vegetation restoration progress assessment result; wherein, the target monitoring data is target monitoring data obtained by a target monitoring data generation method as described in any one of claims 1-6.

8. A method for determining ecological management decisions regarding vegetation restoration for temporary land occupied by power transmission lines, characterized in that, Includes the following steps: Based on the vegetation restoration progress assessment results, the current restoration status level is determined; wherein, the vegetation restoration progress assessment results are obtained by using the vegetation restoration progress assessment method for temporary land occupation by transmission lines as described in claim 7. According to a preset decision rule base, match one or more candidate decision schemes corresponding to the current recovery status level; Based on a preset optimization objective, the optimal decision scheme is determined from one or more candidate decision schemes.

9. A device for generating target monitoring data, characterized in that, include: The partitioning module is used to divide the temporary land occupied by transmission lines into spatial grids according to a preset scale, resulting in several spatial grid units. The classification module is used to spatially classify the collected multi-source data according to the spatial grid unit to which they belong. The multi-source data includes remote sensing data, sensor data, and measured data. The alignment module is used to perform time alignment processing on multi-source data within each spatial grid cell, so that data from different sources are aligned to a unified time node in the time dimension. The correlation analysis module is used to perform ecological logical correlation analysis on time-aligned multi-source data based on a preset ecological factor correlation model, and to identify the conditional probability relationship between basic factors, state indicators and verification truth values. The weighted fusion module is used to perform weighted fusion of multi-source data according to the conditional probability relationship and output the fused target monitoring data.

10. An electronic device, characterized in that, include: A memory, and one or more processors communicatively connected to the memory; The memory stores instructions that can be executed by the one or more processors to cause the one or more processors to implement the method as described in any one of claims 1 to 6.

11. A computer-readable storage medium, characterized in that, The readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 6.

Images