Restoration strategy evaluation method and equipment based on marine ecological data and medium

Abstract

The invention discloses a restoration strategy assessment method and device based on marine ecological data and a medium, and relates to the technical field of marine ecology, and the method comprises the steps: collecting multi-source marine ecological data and historical restoration strategy records, carrying out the time alignment and space mapping, generating an aligned continuous ecological state sequence on a unified time axis and a sea area grid, and carrying out the restoration strategy assessment. Performing multi-dimensional time sequence feature extraction on the continuous ecological state sequence, and outputting an ecological response trajectory set; based on the strategy look-ahead prediction sequence and the partition level evolution sequence, a look-ahead comparison relation reflecting ecological change differences before and after the action of the restoration measures is constructed, and a future restoration difference sequence is output; and converting the future repair difference sequence into a repair strategy evaluation conclusion, and outputting the repair strategy evaluation conclusion for decision selection of the management end. According to the trajectory cluster of the sea area partition, the spatial representation precision of an ecological evolution sequence is enhanced, the local characteristics of the ecological process are highlighted, and the analysis capability of an overall analysis framework is improved.

Description

Technical Field

[0001] This invention relates to the field of marine ecological technology, and in particular to a method, equipment and medium for evaluating restoration strategies based on marine ecological data. Background Technology

[0002] In recent years, marine ecological monitoring technology has continued to evolve. Multi-source data acquisition systems (such as remote sensing, buoy networks, and in-situ sensors) can acquire ecological parameters with high spatiotemporal resolution and achieve long-term dynamic tracking of marine environmental elements with the support of big data analysis frameworks. At the same time, time series alignment algorithms, spatial grid mapping technology, and machine learning methods such as clustering and prediction have been gradually introduced into ecological research, providing a technical foundation for continuous modeling of ecological states and identification of complex ecological patterns.

[0003] However, existing technologies typically lack a unified spatiotemporal correlation modeling method for policy records and ecological observation sequences in the context of restoration policy assessment. This makes it difficult to establish a reusable mapping relationship between policy implementation information and ecological evolution response under the same time benchmark and spatial scope, thus hindering stable quantitative characterization and forward-looking assessment of ecological changes caused by restoration measures. Summary of the Invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, this invention provides a method for evaluating restoration strategies based on marine ecological data to address the problem that it is difficult to stably correlate strategy implementation information and ecological evolution response under a unified spatiotemporal reference system, which makes it difficult to quantitatively characterize and support forward-looking assessment and analysis.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for evaluating restoration strategies based on marine ecological data, which includes: collecting multi-source marine ecological data and historical restoration strategy records, performing time alignment and spatial mapping, generating an aligned continuous ecological state sequence on a unified time axis and marine grid, performing multi-dimensional temporal feature extraction on the continuous ecological state sequence, and outputting an ecological response trajectory set; The ecological response trajectory set is clustered using a time-series trajectory clustering method to generate trajectory clusters that include marine area partitions; Using trajectory clusters, we construct zone-level ecological evolution within the corresponding marine areas of each cluster and output the zone-level evolution sequence. The partition-level evolution sequence is combined with historical repair strategy records to form a prediction input sequence. Lead prediction is performed on the prediction input sequence to generate a strategy forward prediction sequence in multiple future time windows. Based on the strategic forward prediction sequence and the regional evolution sequence, a forward comparative relationship reflecting the differences in ecological changes before and after the restoration measures are implemented is constructed, and the future restoration difference sequence is output. The future repair difference sequence is transformed into repair strategy evaluation conclusions, and the repair strategy evaluation conclusions are output for management decision-making selection.

[0007] As a preferred embodiment of the marine ecological data-based restoration strategy evaluation method of the present invention, the steps include: collecting multi-source marine ecological data and historical restoration strategy records, performing time alignment and spatial mapping, and generating an aligned continuous ecological state sequence on a unified time axis and marine grid. The specific steps are as follows. Collect multi-source marine ecological data and historical restoration strategy records, and sort them according to the original timestamps to generate a basic time-series dataset; Based on the basic time-series dataset, a unified time axis is used to align multi-source marine ecological data with historical restoration strategy records at each observation time position, generating a time-aligned data stream; Based on the spatial location of the marine grid, the time-aligned data stream is mapped item by item to the corresponding spatial location to generate a spatially aligned dataset; Based on the ecological variables recorded by the spatial location of the marine grid in the spatially aligned dataset, an aligned continuous ecological state sequence is generated by continuous arrangement on a unified time axis.

[0008] As a preferred embodiment of the marine ecological data-based restoration strategy evaluation method of the present invention, the specific steps of performing multi-dimensional temporal feature extraction on the continuous ecological state sequence and outputting an ecological response trajectory set are as follows. For continuous ecological state sequences, a unified time axis is used to analyze the magnitude, direction, and trend of changes in ecological variables at different time locations in the marine grid, and extract temporal variation characteristics. Spatial variation features are generated by comparing the temporal variation characteristics of adjacent spatial locations within a sea area grid on a unified time axis. Based on temporal and spatial variation characteristics, ecological variable change trajectories are constructed for each ecological variable. The ecological variable change trajectories are aggregated according to their spatial location in the marine grid to generate an ecological response trajectory set.

[0009] As a preferred embodiment of the marine ecological data-based restoration strategy evaluation method of the present invention, the specific steps for generating trajectory clusters containing marine area partitions are as follows: Based on the ecological response trajectory set, the trajectory difference measurement values ​​of any two ecological response trajectories in terms of change magnitude, change direction and change trend are calculated on a unified time axis, and a trajectory difference measurement set is generated. Using a trajectory difference metric set, a time-series trajectory clustering method is employed to cluster the ecological response trajectories in the ecological response trajectory set, assigning group identifiers to each cluster group, and generating a clustering record set containing the group identifiers. Based on the clustering of record sets, a correspondence between group identifiers and spatial locations of the sea area grid is established on the sea area grid. Trajectory clustering structures are generated and mapped to the sea area grid, and trajectory clusters containing sea area partitions are output.

[0010] As a preferred embodiment of the marine ecological data-based restoration strategy evaluation method of the present invention, the specific steps of constructing a partition-level ecological evolution within the marine area partition corresponding to each cluster using trajectory clustering and outputting the partition-level evolution sequence are as follows. Within the marine area corresponding to the trajectory cluster, the trajectory difference measurement set is used to compare the changes in all ecological variables within the trajectory cluster one by one in pairs, generating a set of paired difference measurement results. The set of pairwise difference measurement results is merged according to the ecological variable change trajectory. A summary operation is performed on the pairwise difference measurement results corresponding to each ecological variable change trajectory, and the difference measurement within the cluster is output. Based on the intra-cluster variance measure, the ecological variable change trajectory with the smallest intra-cluster variance measure is selected as the representative ecological variable change trajectory. Based on the change trajectory of representative ecological variables, the magnitude, direction and trend of change at each time point are connected in chronological order according to a unified time axis to generate a continuous ecological change process. Within the spatial range of marine zoning, zoning-level ecological evolution expressions are generated based on continuous ecological change processes, and zoning-level ecological evolution expressions are connected according to the temporal order of a unified time axis to generate zoning-level evolution sequences.

[0011] As a preferred embodiment of the marine ecological data-based restoration strategy evaluation method of the present invention, the specific steps for generating a forward-looking prediction sequence of strategies within multiple future time windows are as follows: The partition-level evolution sequence is synchronized point by point with the historical repair strategy record on a unified time axis to generate a joint time series sequence, and the joint time series sequence is organized into a prediction input sequence according to the unified time axis. Based on the historical time position of the predicted input sequence on a unified time axis, the exponential smoothing method is used to calculate the lead time of the changing trend of each future time window, and generate a forward trend model. Based on forward-looking trend patterns, strategic forward-looking predictions are made for multiple future time windows one by one, generating a strategic forward-looking prediction sequence.

[0012] As a preferred embodiment of the marine ecological data-based restoration strategy evaluation method of the present invention, the specific steps for outputting the future restoration difference sequence are as follows: The future time position of the strategy forward prediction sequence on a unified time axis is mapped point by point to the partition-level evolution sequence, and a forward control sequence is output. Based on prospective control sequences, the difference between the strategic prospective prediction sequence and the regional evolution sequence is calculated in the future time window to identify the differences in ecological changes in the future time window and generate difference candidates. The candidate differences are categorized and structured according to future time windows to generate a set of structured difference information. Based on the future time positions of the structured difference information set on a unified time axis, a forward-looking comparison relationship is gradually formed, and a future repair difference sequence is output.

[0013] As a preferred embodiment of the marine ecological data-based restoration strategy evaluation method of the present invention, the specific steps for transforming the future restoration difference sequence into restoration strategy evaluation conclusions and outputting these conclusions for management decision-making are as follows. Based on the analysis of future restoration difference sequences, the ecological change differences in future time windows are analyzed. The cumulative degree and frequency of future restoration difference sequences are statistically analyzed according to future time windows, and an evaluation element set is generated to summarize the evaluation elements of restoration strategies. Based on the chronological order of future time windows on a unified timeline, the elements for evaluating the remediation strategy are summarized and organized to generate a remediation strategy evaluation conclusion, which is then output to the management end as a basis for decision-making.

[0014] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the marine ecological data-based restoration strategy evaluation method as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the marine ecological data-based restoration strategy evaluation method as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: by using a time-series trajectory clustering method to cluster the ecological response trajectory set, a trajectory cluster containing marine zoning is generated, which enhances the spatial representation accuracy of the ecological evolution sequence, highlights the local characteristics of the ecological process, and improves the analytical capability of the overall analysis framework; it enables the historical restoration strategy records to obtain the zoning effect attribution caliber, establishes a consistent correlation between the restoration measure type and restoration implementation intensity parameters and the zoning-level ecological evolution expression, forms a quantifiable basis for comparing the differences in the effects of restoration measures, and supports the interpretable output of restoration strategy evaluation conclusions. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method for evaluating restoration strategies based on marine ecological data.

[0019] Figure 2 This is a flowchart for time-series feature extraction.

[0020] Figure 3 This is a flowchart for trajectory clustering.

[0021] Figure 4 This is a flowchart for forward forecasting. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0025] Reference Figures 1-4This is one embodiment of the present invention, which provides a method for evaluating restoration strategies based on marine ecological data, including the following steps: S1: Collect multi-source marine ecological data and historical restoration strategy records, perform time alignment and spatial mapping, generate aligned continuous ecological state sequences on a unified time axis and marine grid, perform multi-dimensional temporal feature extraction on the continuous ecological state sequences, and output ecological response trajectory sets.

[0026] S1.1: Collect multi-source marine ecological data and historical restoration strategy records, sort them according to the original timestamps, and generate a basic time-series dataset.

[0027] Collect multi-source marine ecological data, including observation data from marine observation equipment, water quality monitoring records, and biological observation records. Record the original timestamp and observation spatial location identifier corresponding to each observation data, water quality monitoring record, and biological observation record from marine observation equipment. The observation spatial location identifier is used to indicate the target spatial location in the marine grid.

[0028] Collect historical restoration strategy records, including restoration measure type, restoration start and end time and location, and restoration intensity parameters. Record the original timestamp and target sea area scope identifier for each historical restoration strategy record. The target sea area scope identifier is used to indicate the set of spatial locations of the restoration measures on the sea area grid.

[0029] The observation data from marine observation equipment, water quality monitoring records, biological observation records, and historical restoration strategy records are merged and sorted according to all the corresponding original timestamps to generate a basic time-series dataset.

[0030] S1.2: Based on the basic time-series dataset, a unified time axis is used to align multi-source marine ecological data with historical restoration strategy records at each observation time position, generating a time-aligned data stream.

[0031] Based on the basic time series dataset, all original timestamps are read, the earliest and latest time positions in the original timestamps are determined, and a time interval is formed.

[0032] Based on the basic time series dataset, the sampling period of various observation data is extracted, and the minimum sampling period is selected as the candidate time interval; the candidate time interval set is aligned one by one within the historical data period.

[0033] The time alignment error is defined as the arithmetic mean of the absolute values ​​of the minimum time difference of all successfully mapped records within the historical data period. The candidate time interval with the smallest time alignment error is selected as the time interval for the unified time axis.

[0034] Multiple observation time positions are set sequentially at set time intervals within the time interval to generate a unified time axis covering the time interval.

[0035] The time alignment tolerance threshold is determined based on the time interval of the unified time axis, and is set to half of the time interval of the unified time axis.

[0036] Based on the basic time series dataset, the original timestamps corresponding to the observation data of marine observation equipment, water quality monitoring records and biological observation records are read. The time difference between the original timestamps and the observation time positions on a unified time axis is calculated, and the observation time position with the smallest time difference is selected as the mapping target.

[0037] When there are multiple observation time locations with the same and minimum time difference, the earliest observation time location is selected as the mapping target. When the minimum time difference is greater than the time alignment tolerance threshold, a time-aligned data record is generated for the corresponding observation time position, and the numerical fields that failed to be mapped are written into the time-aligned data record as missing markers.

[0038] Based on the basic time series dataset, the original timestamps of historical restoration strategy records are read, the time difference between the original timestamps and the observation time positions on the unified time axis is calculated, and the historical restoration strategy records are mapped to the observation time positions with the smallest time difference.

[0039] At each observation time point on a unified timeline, observation data, water quality monitoring records, biological observation records, and historical restoration strategy records from marine observation equipment mapped to that observation time point are collected.

[0040] When multiple numerical records of the same type exist at the same observation time and location, a single representative value is generated by averaging the records of the same type.

[0041] When records of the same type are categorical fields, a single representative category is generated by majority voting, and a time-aligned data record for the current observation time position is generated.

[0042] Based on the time order of a unified timeline, all time-aligned data records are sequentially concatenated to generate a time-aligned data stream.

[0043] S1.3: Based on the spatial location of the sea area grid, map the time-aligned data stream item by item to the corresponding spatial location to generate a spatially aligned dataset.

[0044] Based on the spatial extent of the target sea area, the target sea area is divided into multiple sea area grids according to the candidate grid resolution set. Spatial mapping is performed on each candidate grid resolution, and statistics are collected within the historical data period. Spatial alignment coverage: The proportion of marine grid spatial locations mapped to by at least one observation record out of all marine grid spatial locations.

[0045] Temporal coverage stability: The candidate grid resolutions are sorted from coarse to fine based on the number of effective observation time locations of each sea area grid. The sequence of coverage changes with resolution is output. Among the inflection point resolution and its adjacent candidate resolutions, the candidate grid resolution with the best temporal coverage stability is selected as the target grid resolution to determine the spatial location of the sea area grid.

[0046] Based on time-aligned data streams, the spatial location identifiers corresponding to the observation data, water quality monitoring records, and biological observation records of marine observation equipment are read.

[0047] The spatial location markers of the observations and the coordinates of the center of the sea area grid are unified into latitude and longitude coordinates. The distance between the spatial location markers of the observations and the coordinates of the center of each sea area grid is calculated using the great circle distance, and then mapped to the spatial location of the sea area grid with the smallest distance.

[0048] Based on time-aligned data streams, the target sea area scope identifier recorded in historical repair strategies is read, and it is determined whether the center coordinates of the sea area grid are located inside or on the boundary of the spatial scope polygon, which serves as the criterion for spatial coverage determination.

[0049] A mapping relationship is established between the spatial locations of intersecting marine grids and the corresponding historical restoration strategy records, and the historical restoration strategy records are written into the spatial locations of the intersecting marine grids.

[0050] At each marine grid location, observation data, water quality monitoring records, biological observation records, and historical restoration strategy records of marine observation equipment mapped to the spatial location and belonging to each observation time location are collected according to the observation time location order of a unified time axis.

[0051] S1.4: Generate an aligned continuous ecological state sequence by arranging the ecological variables recorded in the spatial location of the marine grid in the spatial alignment dataset on a unified time axis.

[0052] At each marine grid location, for each ecological variable, the ecological variables corresponding to each observation time location are read in chronological order according to the observation time location on a unified time axis.

[0053] When there are no ecological variables at the observation time position or the ecological variables are marked as missing, linear interpolation is performed on the ecological variables corresponding to the previous most recent valid observation time position and the next most recent valid observation time position based on a unified time axis to generate interpolated ecological variables. When the current most recent valid observation time location or the next most recent valid observation time location does not exist, the ecological variable corresponding to the most recent valid observation time location is used as the imputed ecological variable, and the missing marker is retained.

[0054] By arranging ecological variables sequentially over time, a continuous ecological state sequence aligned with the spatial location of the marine grid is generated.

[0055] S1.5: For continuous ecological state sequences, use a unified time axis to analyze the magnitude, direction and trend of changes in ecological variables at different time locations in the marine grid, and extract time change characteristics.

[0056] At each sea area grid spatial location, for each ecological variable in the continuous ecological state sequence, the ecological variables at the next observation time position and the previous observation time position are read at two adjacent observation time positions.

[0057] When an ecological variable at any observation time location is marked as missing, for the corresponding pair of adjacent observation time locations, the magnitude of change, direction of change, trend of change, and rate of change are marked as missing, and the calculation of the difference between the adjacent pair of observation time locations is skipped.

[0058] When no missing marker is written for the ecological variables at two adjacent observation time positions, calculate the difference between the ecological variables at the later observation time position and the earlier observation time position, and use the difference in ecological variables as the increase or decrease value. Extract the sign of the increase / decrease value as a change direction marker; perform consistency judgment on the change direction markers of two adjacent differences and output the change trend marker; calculate the change rate value based on the ratio of the increase / decrease value to the time interval between adjacent observation positions.

[0059] For each ecological variable, the increase / decrease, direction of change, and trend of change between observation time points are arranged in chronological order according to a unified time axis, generating time change characteristics that include the overall rate of change, change direction markers, and change trend markers.

[0060] S1.6: Based on the difference comparison of the temporal variation characteristics of adjacent spatial locations within the sea area grid on a unified time axis, spatial variation characteristics are generated.

[0061] Based on the temporal variation characteristics, for ecological variables belonging to adjacent sea area grid spatial locations, the change rate value, change direction marker, and change trend marker in the temporal variation characteristics corresponding to the adjacent sea area grid spatial locations are read.

[0062] When any field being compared is marked as missing, the spatial variation characteristics of the corresponding observation time and location and the corresponding ecological variable are marked as missing, and the difference comparison is skipped. When none of the fields being compared have missing markers written, perform comparisons of the rate of change, the direction of change, and the trend of change, and output the comparison results.

[0063] The comparison results of the spatial positions of adjacent sea area grids are arranged in chronological order according to a unified time axis to generate spatial variation characteristics that reflect the differences in the spatial positions of adjacent sea area grids.

[0064] S1.7: Based on the temporal and spatial variation characteristics, construct the ecological variable change trajectory for each ecological variable.

[0065] Temporal and spatial variation features are concatenated in field order to generate a trajectory vector sequence indexed by observation time and location. This trajectory vector sequence serves as the trajectory of ecological variable changes.

[0066] It should be noted that the fields are ordered as follows: change magnitude value, change direction marker, change trend marker, change rate value, change rate difference value, change direction difference value, and change trend difference value.

[0067] S1.8: Aggregate the ecological variable change trajectories according to their spatial location in the marine grid to generate an ecological response trajectory set.

[0068] At each marine grid spatial location, all ecological variable change trajectories belonging to that marine grid spatial location are collected according to ecological variable identifiers and formed into ecological response trajectories.

[0069] All ecological response trajectories are arranged and organized according to the spatial index order of the marine grid spatial location, generating a set of ecological response trajectories covering the target marine area and a unified time axis.

[0070] S2: Use the time-series trajectory clustering method to cluster the ecological response trajectory set and generate trajectory clusters that include marine area partitions.

[0071] S2.1: Based on the ecological response trajectory set, calculate the trajectory difference measurement values ​​of any two ecological response trajectories in terms of change magnitude, change direction and change trend on a unified time axis, and generate a trajectory difference measurement set.

[0072] Based on the ecological response trajectory set, for any two ecological response trajectories, at each observation time position, the ecological variable change trajectory corresponding to the same ecological variable identifier in the two ecological response trajectories is read, and the two ecological variable change trajectories are compared field by field in the three fields of change amplitude, change direction and change trend, and the difference values ​​of change amplitude, change direction and change trend are output. The arithmetic mean of the differences between all ecological variable identifiers at the same observation time location is calculated as the comprehensive difference value at that time location.

[0073] The cumulative sum of the comprehensive difference values ​​is calculated based on the order of observation time and location, and the cumulative sum is used as the trajectory difference measure of the two ecological response trajectories.

[0074] Using a pairwise combination method of ecological response trajectories, the trajectory difference metric values ​​of all trajectory pairs in the ecological response trajectory set are calculated, and the trajectory difference metric values ​​are organized to generate a trajectory difference metric set.

[0075] It should be noted that the smaller the value of the trajectory difference measure, the more similar the two ecological response trajectories are. The difference in direction of change and the difference in trend of change adopt a numerical mapping rule of 0 for consistency and 1 for inconsistency.

[0076] S2.2: Using the trajectory difference metric set, the ecological response trajectories in the ecological response trajectory set are clustered using the time-series trajectory clustering method. Each cluster group is assigned a group identifier, and a clustering record set containing the group identifier is generated.

[0077] Based on the trajectory difference measurement set, the order of the ecological response trajectory set is used as the row and column index of the difference measurement matrix, and the trajectory difference measurement values ​​of the corresponding trajectory pairs are used as matrix elements to construct the difference measurement matrix.

[0078] The trajectory difference measurement values ​​are sorted within the historical data period. The range of extreme value removal is determined based on the overall distribution characteristics of the trajectory difference measurement values. Extreme difference measurement values ​​that significantly deviate from the main distribution are removed.

[0079] For example, after sorting the trajectory difference measurement values ​​from smallest to largest, the difference measurement values ​​at both ends of the sorted sequence that have a difference exceeding the multiple relationship threshold with the adjacent values ​​of the main distribution interval are removed. The remaining continuous value interval is used as the main distribution interval. The multiple relationship threshold is determined based on the statistical distribution of the adjacent differences of the trajectory difference measurement values, and is taken as 1.5 to 3.0 times the arithmetic mean of the adjacent differences, which is used to identify the extreme difference measurement values ​​at both ends of the sorted sequence.

[0080] After removing the extreme values, the range of the remaining difference measures is considered (e.g., 0.50–2.00).

[0081] Within the range of candidate merging threshold values, select multiple sets of candidate merging thresholds and perform clustering partitioning one by one: When the trajectory difference metric value corresponding to any two ecological response trajectories is not greater than the candidate merging threshold, a similar connection relationship is established between the trajectory pairs.

[0082] Based on similar connectivity, ecological response trajectories with similar connectivity links are grouped into the same cluster group, and a candidate cluster partitioning record set is output.

[0083] Perform partitioned spatial connectivity verification on the candidate clustering record set. Use four-adjacency relation to calculate the number of connected components in the spatial location set of the sea area grid corresponding to the cluster group. If the number of connected components is equal to 1, it is determined that the partitioned spatial connectivity is satisfied.

[0084] Perform minimum partition area constraint and spatial resolution requirement verification on the candidate cluster division record set: count the number of spatial locations of the sea area grid for each cluster group and calculate the partition area in combination with the area of ​​the sea area grid unit. Summarize the partition areas of all candidate partitions to generate a partition area set.

[0085] Sort the partition area sets from smallest to largest. The number of end values ​​to be removed is determined by the total number of partition area sets. The number of end values ​​to be removed is the number of values ​​that correspond to one value for every ten values ​​in the total number. After removing the number of end values ​​to be removed from the smallest and largest ends of the partition area sets, take the partition area interval corresponding to the remaining middle segment as the allowed partition area interval.

[0086] If the partition area falls within the allowed partition area range, it is determined that both the minimum partition area constraint and the spatial resolution requirement are met, and a candidate set that passes the verification is output.

[0087] Based on the candidate set that has passed the constraint verification, the average trajectory difference metric within the partition and the average trajectory difference metric between the intervals are calculated. The candidate clustering record set with the smallest average trajectory difference metric within the partition and the largest average trajectory difference metric between the intervals is selected as the clustering record set of the ecological response trajectory.

[0088] Assign a unique group identifier to each cluster group, write the group identifier into the ecological response trajectory record in the cluster partitioning record set, and output the cluster partitioning record set containing the group identifier.

[0089] S2.3: Based on the clustering of the record set, establish the correspondence between the group identifier and the spatial location of the sea area grid on the sea area grid, generate the trajectory clustering structure and map it to the sea area grid, and output the trajectory cluster body containing the sea area partition.

[0090] Based on the clustering record set containing group identifiers, the spatial locations of all sea area grids with the same group identifier are collected according to the group identifier and arranged according to the spatial index order to generate a trajectory clustering structure for expressing the spatial range of the group.

[0091] Based on the trajectory clustering structure, the trajectory clustering structure corresponding to each group identifier is mapped to the marine grid.

[0092] Establish a unique correspondence between the spatial location of each sea area grid and the corresponding group identifier, and output the trajectory cluster containing the sea area partitions according to the arrangement order of the sea area grid spatial index.

[0093] S3: Construct zone-level ecological evolution within the marine area corresponding to each cluster using trajectory clusters, and output zone-level evolution sequences.

[0094] S3.1: Within the marine area corresponding to the trajectory cluster, use the trajectory difference measure set to compare the changes in all ecological variables within the trajectory cluster one by one in pairs, and generate a set of paired difference measure results.

[0095] Based on trajectory clustering, within a marine area partition, ecological variable change trajectories are collected according to ecological variable identifiers, and each pair is enumerated in each collection result to generate a set of ecological variable change trajectory pairs.

[0096] Based on the set of ecological variable change trajectories, and according to the spatial location identifiers of the marine grid corresponding to the two trajectories, ecological response trajectories are located one by one in the corresponding marine area partitions and ecological response trajectory pairs are formed.

[0097] In the trajectory difference measurement set, entries are matched and recorded according to the ecological response trajectory pairs, the correspondence between observation time and location and ecological variables. The difference values ​​of change magnitude, change direction and change trend are extracted to generate paired difference measurement records.

[0098] Organize all pairwise difference measurement records within the sea area partition and output the set of pairwise difference measurement results corresponding to the sea area partition.

[0099] S3.2: Merge the set of paired difference measures according to the ecological variable change trajectory, perform a summary operation on the paired difference measures corresponding to each ecological variable change trajectory, and output the difference measure within the cluster.

[0100] Based on all pairwise difference measurement records in the pairwise difference measurement result set, the records are merged according to the ecological variable change trajectory pairs, and the number of effective overlapping observation time and location pairs for each ecological variable change trajectory pair is calculated. The number of effective overlapping observation time locations is counted according to the number of observation time locations where neither of the two ecological variable change trajectories has a missing marker written at the same observation time location.

[0101] In the aggregated results corresponding to each ecological variable change trajectory, the ecological variable change trajectory pairs are sorted from largest to smallest according to the number of effective overlapping observation time and location, and the ecological variable change trajectory pairs corresponding to the highest number of targets in the sorted order are selected.

[0102] For the selected pairs of ecological variable change trajectories, the arithmetic mean of the paired difference measures is calculated and used as the intra-cluster difference measure of the ecological variable change trajectories.

[0103] S3.3: Based on the intra-cluster variance measure, select the ecological variable change trajectory with the smallest intra-cluster variance measure as the representative ecological variable change trajectory.

[0104] Based on the list of intra-cluster differential measures, the intra-cluster differential measures of all ecological variable change trajectories within each marine area are read, the numerical values ​​of the intra-cluster differential measures are compared, and the value with the smallest intra-cluster differential measure is identified.

[0105] By locating the ecological variable change trajectory marker corresponding to the minimum value of the difference metric within a cluster, the corresponding ecological variable change trajectory is determined as the representative ecological variable change trajectory of the marine area.

[0106] When there are multiple ecological variable change trajectories with the same minimum value within a cluster, the ecological variable change trajectory with the smallest marine grid spatial index is selected as the representative ecological variable change trajectory according to the order of the marine grid spatial index.

[0107] S3.4: Based on the change trajectory of representative ecological variables, the magnitude, direction and trend of change at each time position are connected in a time sequence according to a unified time axis to generate a continuous ecological change process.

[0108] Based on the change trajectory of representative ecological variables, at each observation time location within each marine area, the change direction, change magnitude, and change trend of the representative ecological variable change trajectory are read.

[0109] The time and location of observation, the direction of change, the magnitude of change, and the trend of change are combined into time change record entries, which are then arranged in chronological order according to the time and location of observation to form a record sequence of representative ecological variables evolving over time, serving as a continuous ecological change process for marine zoning.

[0110] S3.5: Within the spatial range of marine zoning, generate zoning-level ecological evolution expressions based on continuous ecological change processes, and connect the zoning-level ecological evolution expressions according to the time sequence of a unified time axis to generate a zoning-level evolution sequence.

[0111] Based on the continuous ecological change process, the spatial location of the corresponding marine grid is read within each marine area partition.

[0112] By establishing a correspondence between continuous ecological change processes and the spatial location of marine grids, a correspondence is established between each observation time location in the continuous ecological change process and the spatial range of the marine sub-region, generating a sub-regional ecological evolution expression that can characterize the overall ecological change trend of the marine sub-region.

[0113] Based on the temporal order of the observation time and location, the regional-level ecological evolution expressions of all sea areas are arranged sequentially according to the spatial index order of the sea areas, forming a regional-level evolution sequence covering the target sea area.

[0114] S4: Combine the partition-level evolution sequence with historical repair strategy records into a prediction input sequence, perform lead prediction on the prediction input sequence, and generate a strategy forward prediction sequence in multiple future time windows.

[0115] S4.1: Synchronize the partition-level evolution sequence with the historical repair strategy record point by point on a unified time axis to generate a joint time series sequence, and organize the joint time series sequence into a prediction input sequence according to the unified time axis.

[0116] Based on a unified timeline, the observation time positions are listed one by one in chronological order. For each group identifier and each observation time position, the corresponding regional ecological evolution expression of the regional evolution sequence at the observation time position is read and recorded as an ecological evolution entry.

[0117] For each group identifier and each observation time location, iterate through the historical restoration strategy records and filter historical restoration strategy records that simultaneously meet the following conditions: Repair the implementation of start and end time locations that cover observation time locations; The set of spatial locations corresponding to the target sea area's scope of action covers the spatial location of the sea area grid corresponding to the grouping identifier. The coverage determination criteria are based on the fact that the center coordinates of the sea area grid are located inside the polygon of the spatial scope of action corresponding to the set of spatial locations of action.

[0118] When the screening results contain multiple historical repair strategy records, the historical repair strategy records are merged according to the repair measure type, and the repair implementation intensity parameter is aggregated. The group identifier and the repair measure type and repair implementation intensity parameter corresponding to the observation time and location are output.

[0119] The aggregation of the repair intensity parameters is expressed as follows: ; In the formula, To implement strength parameters for the repair after polymerization, For group identification, For the observation time and location, This is an identifier for the type of remediation measure (valued during the traversal of remediation measure types). To meet both temporal and spatial coverage conditions, and the type of remediation measure is equal to... A set of historical restoration strategy record identifiers, Identify and record historical restoration strategies Repair implementation strength parameters, This is a single record identifier for historical restoration strategies.

[0120] When no historical remediation strategy record exists in the screening results, the remediation measure type record is empty, and the remediation implementation intensity parameter record is zero.

[0121] The group identifier, observation time and location, regional ecological evolution expression, restoration measure type and restoration implementation intensity parameters are concatenated into a joint time series data record, and then connected according to the time order of observation time and location and the order of group identifier to output the joint time series.

[0122] Based on the joint time series, the observation time positions are traversed item by item according to a unified time axis. At each observation time position, the group identifiers are traversed item by item and the corresponding joint time series data records are located.

[0123] The joint time series data records obtained from the location are appended sequentially according to the traversal order of the observation time position to generate a record sequence, thus obtaining the prediction input sequence.

[0124] S4.2: Based on the historical time position of the predicted input sequence on a unified time axis, the exponential smoothing method is used to calculate the lead time of the changing trend of each future time window, and generate a forward trend model.

[0125] Based on the predicted input sequence, the regional ecological evolution expression is read in the order of observation time and location for each group identifier to generate the historical evolution sequence.

[0126] Based on the historical evolution sequence, multiple future time windows are set on a unified time axis. These multiple future time windows are represented by consecutive future observation time position intervals and are numbered sequentially according to their chronological order.

[0127] For each group identifier, the smoothed evolution level and smoothed evolution trend are calculated using the exponential smoothing method based on the historical evolution sequence, and the initial values ​​of exponential smoothing are set: the initial value of evolution level is taken as the value of the partition-level ecological evolution expression at the first historical observation time position, and the initial value of evolution trend is taken as the difference between the values ​​of the partition-level ecological evolution expression at the first two historical observation time positions.

[0128] The smoothed evolution level and smoothed evolution trend are calculated using the exponential smoothing method and are expressed as follows: ; In the formula, This represents the evolution level after exponential smoothing. Group identifier At the time of observation The values ​​of the partition-level ecological evolution expression. Evolutionary level smoothing coefficient (satisfying) ), For the observation time and location The level of evolution after exponential smoothing. For the observation time and location The evolution trend after exponential smoothing.

[0129] ; In the formula, This represents the evolutionary trend after exponential smoothing. Evolutionary trend smoothing coefficient (satisfying) ).

[0130] Based on each group identifier and each future time window, for each future observation time location covered by the future time window, the corresponding lead is used to output the evolutionary look-ahead value, and the values ​​are connected in the order of the future observation time locations to form an evolutionary look-ahead value sequence.

[0131] For each future observation time location covered by the future time window, the evolutionary look-ahead value is output using the corresponding lead time, expressed as follows: ; In the formula, Group identifier At the time of observation The evolutionary look-ahead value of the lead prediction output. This is the lead time, corresponding to the number of steps at the future observation time and location.

[0132] It should be noted that, The evolution level smoothing coefficient and satisfies , The evolution trend smoothing coefficient is satisfied with ; and By selecting multiple groups of candidates at equal intervals within the range (0 to 1) Values ​​and multiple candidate groups The values ​​are selected and paired to form candidate combinations.

[0133] A validation interval is set at the end of the historical data. The validation interval consists of the continuous observation time positions traced back from the end.

[0134] For each candidate combination, for each group identifier, before the verification interval, the exponentially smoothed evolution level and the exponentially smoothed evolution trend are recursively calculated at each observation time position according to the recursive relationship of the exponentially smoothed evolution level and the recursive relationship of the exponentially smoothed evolution trend. Then, one-step back substitution prediction and absolute error calculation are performed at each observation time position within the verification interval.

[0135] The average absolute error between all observation time locations and all group identifiers within the verification interval is taken as the comprehensive prediction error. The candidate combination with the smallest comprehensive prediction error is selected for determination. and .

[0136] Numbering multiple future time windows sequentially means dividing multiple adjacent future observation time locations into several continuous intervals, with each continuous interval corresponding to a future time window. The continuous interval that occurs first corresponds to the future time window with the earlier number, and the continuous interval that occurs later corresponds to the future time window with the later number.

[0137] S4.3: Based on the forward-looking trend pattern, make forward-looking predictions for multiple future time windows one by one to generate a forward-looking prediction sequence for strategies.

[0138] Based on historical restoration strategy records, the types of restoration measures that have appeared in each group are categorized according to the coverage relationship between the spatial location of the sea area grid corresponding to the group identifier and the set of spatial locations corresponding to the target sea area scope identifier.

[0139] For each group identifier and each remediation measure type, read the remediation implementation intensity parameter from the historical remediation strategy record.

[0140] The repair implementation intensity parameters are sorted in ascending order of value and the sorted sequence is output. The median of the sorted sequence is defined as the representative repair implementation intensity parameter.

[0141] The representative repair implementation intensity parameter is calculated and expressed as follows: ; In the formula, As representative of the repair implementation strength parameters, To meet the coverage requirements, the number of historical repair strategy records is required. After sorting all repair implementation intensity parameters that meet the coverage conditions in ascending order of their values, the sequence number in the sorted sequence is... The repair implementation strength parameter value, After sorting all repair implementation intensity parameters that meet the coverage conditions in ascending order of their values, the sequence number in the sorted sequence is... The repair implementation strength parameter value, After sorting all repair implementation intensity parameters that meet the coverage conditions in ascending order of their values, the sequence number in the sorted sequence is... The numerical values ​​of the strength parameters for the repair implementation.

[0142] It should be noted that when When it is an odd number, take As a representative parameter for the intensity of repair implementation; when When it is even, take and arithmetic mean This serves as a representative parameter for the intensity of repair implementation.

[0143] Based on the type of restoration measures, historical restoration strategy records and corresponding ecological variable observation records are collected. The average value of the change in target ecological variables for each type of restoration measure during the evaluation period is calculated. When the average change value is greater than zero, the direction of influence is determined to be enhancing; when the average change value is less than zero, the direction of influence is determined to be weakening; and when the average change value is equal to zero, the direction of influence is determined to be neutral.

[0144] The average absolute value of change corresponding to each type of remediation measure is normalized across all types of remediation measures to determine the influence intensity coefficient within the interval (0-1), and to establish the correspondence between the type of remediation measure, the direction of influence, and the influence intensity coefficient.

[0145] The intensity parameters of representative restorations are normalized and mapped to intensity level values ​​within the range of 0 to 1.

[0146] The magnitude of the strategy's impact is determined by multiplying the intensity level value by the influence intensity coefficient, and by taking into account the time offset of the future observation time location relative to the starting time location of the future time window.

[0147] The time offset is mapped to a time decay weight within the interval (0~1) that monotonically decreases as the time offset increases, so that the magnitude of the policy's influence gradually decreases as the time offset increases under the action of the time decay weight.

[0148] At each future observation time location, the influence direction, the magnitude of the policy influence, and the time decay weight are combined to output the policy influence coefficient corresponding to the future observation time location; At each future observation time position, the policy influence coefficient is applied to the value of the evolution prospective value sequence at that future observation time position, generating the evolution prospective value sequence under the policy.

[0149] The types of remediation measures, representative remediation implementation intensity parameters, and evolutionary prospective value sequences under the influence of the strategy are concatenated to form prospective prediction entries for the strategy.

[0150] Connect all forward-looking policy prediction entries according to the order of future time windows, group identifiers, and remediation measure types, and output the forward-looking policy prediction sequence.

[0151] S5: Based on the strategy-based forward prediction sequence and the regional evolution sequence, construct a forward comparison relationship that reflects the differences in ecological changes before and after the restoration measures take effect, and output the future restoration difference sequence.

[0152] S5.1: Match the future time position of the strategy look-ahead prediction sequence on a unified time axis with the partition-level evolution sequence point by point, and output the look-ahead control sequence.

[0153] Based on a unified timeline, the last historical observation time position is read, subsequent observation time positions are used as future time positions, and the correspondence between group identifiers and spatial positions of the sea area grid is read.

[0154] For each group identifier, read the partition-level ecological evolution expression corresponding to the last historical observation time position of the partition-level evolution sequence, and read the partition-level ecological evolution expression corresponding to the second-to-last historical observation time position of the partition-level evolution sequence to determine the recent change magnitude of the partition-level evolution sequence.

[0155] The magnitude of the most recent change in the partition-level evolutionary sequence is represented as: ; In the formula, For partition-level evolutionary sequences in grouping identifiers In the corresponding marine zoning, the variation of zoning-level ecological evolution between the last historical observation time and the penultimate historical observation time is shown. This refers to the location of the last historical observation. This is the second to last historical observation time position. Group identifier At the location of the last historical observation time The values ​​of the partition-level ecological evolution expression. Group identifier At the penultimate historical observation time position The values ​​for the regional-level ecological evolution expression.

[0156] Based on the strategy forward prediction sequence, the group identifier, the type of remediation measure and the evolutionary forward value corresponding to the future time position are generated item by item, and the representative remediation implementation intensity parameters are recorded.

[0157] Based on the partition-level evolutionary sequence, the partition-level ecological evolution expression is extrapolated to the same future time position, and the extrapolated value of the partition-level ecological evolution expression is output.

[0158] The extrapolated values ​​of the output partition-level ecological evolution expression are represented as follows: ; In the formula, Group identifier Future Time Location Extrapolated values ​​of regional-level ecological evolution expression. For future time location, For future time location relative to the last historical observation time position The number of steps to the future time position.

[0159] The future time location is mapped to its corresponding future time window identifier, and the group identifier, future time window identifier, future time location, type of restoration measure, representative restoration implementation intensity parameter, evolution prospective value and regional ecological evolution expression extrapolation value are arranged in chronological order and group identifier order to output a prospective control sequence.

[0160] S5.2: Based on prospective control sequences, calculate the difference measure between the strategic prospective prediction sequence and the regional evolution sequence in the future time window, identify the differences in ecological changes in the future time window, and generate difference candidates.

[0161] Based on prospective control sequences, the records in the prospective control sequences are traversed in chronological order on a unified time axis according to their future time positions.

[0162] Read the future time window identifier and future time position from each prospective control sequence record.

[0163] The prospective control sequence records are segmented according to the future time window identifier, and the future time location range covered by each future time window is output. Within each future time window, the prospective control sequence records are aggregated according to the group identifier and the type of remediation measure.

[0164] Based on the representative remediation intensity parameters recorded in the aggregated prospective control sequence records, and by traversing the representative remediation intensity parameters of all remediation measure types within the same group identifier, the maximum representative remediation intensity parameter is determined.

[0165] The intensity weighting coefficient is calculated based on the ratio of the representative restoration implementation intensity parameter to the maximum representative restoration implementation intensity parameter, and a correspondence is established between the intensity weighting coefficient and the group identifier and restoration measure type.

[0166] The intensity weighting coefficient is calculated as follows: ; ; In the formula, This is the intensity weighting coefficient (example values ​​are: 0 to 1). Group identifier The maximum value of the representative repair implementation intensity parameter corresponding to each type of repair measure.

[0167] It should be noted that the intensity weighting coefficient is dimensionless; the evolution prospective value and the extrapolated value of regional ecological evolution expression have the same physical dimension; the physical dimension of the intensity weighting difference is consistent with the dimension of the ecological index, thus maintaining dimensional consistency.

[0168] For each future time location covered by the future time window, the difference between the evolutionary prospective value and the extrapolated value of the regional ecological evolution expression is calculated. The difference is used as the ecological change difference of the future time location, and the intensity weighting coefficient is used to perform intensity weighting on the ecological change difference to obtain the intensity weighted difference corresponding to the future time location.

[0169] The intensity-weighted differences of each future time location are connected according to their temporal order to form an intensity-weighted difference sequence.

[0170] The intensity-weighted difference corresponding to future time locations is represented as: ; In the formula, For future time location Intensity-weighted differences For strategy-oriented prediction of sequences in grouping identifiers Repair measure type identifier Future time location Evolutionary predictive value.

[0171] The difference measure is calculated for the intensity-weighted difference series in order of future time position, and is expressed as: ; In the formula, For future time window Difference measurement For future time window Number of future time locations covered For future time window The range of future time locations covered This serves as a marker for a future time window.

[0172] Examine intensity-weighted differences point-by-point according to the future time locations covered by the future time window: When the intensity-weighted difference of all future time locations covered by the future time window is zero, the difference metric is determined to be zero and no difference candidate record is generated. When the difference measure is not zero, the future time window identifier, group identifier, remediation measure type, and difference measure are recorded as difference candidate records.

[0173] Where the difference measure is determined to be zero, it is represented as follows: ; S5.3: Classify and structure the candidate differences according to future time windows to generate a structured set of difference information.

[0174] The candidate differences are processed one by one according to the time sequence of the future time window, and the candidate differences are collected by grouping identifier within each future time window.

[0175] When there are candidate differences within a future time window, the candidate difference with the largest difference measure is selected as the main difference entry. If no difference candidate exists within a future time window, the main difference entry is marked as empty and the generation of structured records for the future time window is skipped.

[0176] Among them, the candidate with the largest difference measure is selected as the principal difference item, represented as follows: ; In the formula, For future time window With group identifier The type of remediation action corresponding to the main difference item selected within.

[0177] The future time window identifier, group identifier, the type of remediation measure corresponding to the main difference item, the difference measure corresponding to the main difference item, the intensity weighting coefficient corresponding to the main difference item, and the future time location range covered by the future time window are concatenated into a structured record, and all structured records are connected to output a set of structured difference information.

[0178] S5.4: Based on the future time positions of the structured difference information set on a unified time axis, a forward-looking comparison relationship is gradually formed, and a future repair difference sequence is output.

[0179] Based on the structured differential information set, for each future time location covered by each structured record, the corresponding prospective control sequence record is located in the prospective control sequence according to the group identifier, the type of restoration measure and the future time location, and the evolutionary prospective value and the regional ecological evolution expression extrapolation value in the record are obtained.

[0180] Based on the intensity weighting coefficients recorded in the structured records, intensity weighting is applied to the difference between the evolution prospective value and the extrapolated value of the regional ecological evolution expression, and the intensity weighted difference of the future time position is output.

[0181] The group identifier, future time location, and weighted differences in the type and intensity of remediation measures are arranged sequentially according to the time order of the future time location and the order of the group identifier to form a forward-looking comparison relationship.

[0182] Based on the forward comparison relationship, intensity-weighted differences are extracted according to the time order of future time positions, and sorted according to the group identifier order. Each group identifier is then connected with the intensity-weighted differences corresponding to each future time position to output the future repair difference sequence.

[0183] S6: Transform the future repair difference sequence into repair strategy evaluation conclusions, and output the repair strategy evaluation conclusions for management decision-making.

[0184] S6.1: Analyze the differences in ecological changes in future time windows based on the future restoration difference sequence, statistically analyze the cumulative degree and frequency of occurrence of the future restoration difference sequence according to the future time window, generate a set of evaluation elements, and summarize the evaluation elements of restoration strategies.

[0185] Based on the future repair difference sequence, each future time window is processed sequentially according to the order of the future time windows on a unified time axis. Within each future time window, the future repair difference values ​​corresponding to all future time positions covered by the future time window are collected according to the grouping identifier.

[0186] When the future restoration difference is greater than zero, it is determined to be an ecological improvement item; when the future restoration difference is less than zero, it is determined to be an ecological risk item; when the future restoration difference is equal to zero, it is determined to be an item with no change.

[0187] For each group identifier and each future time window, iterate through all future time locations covered by the future time window to find the corresponding future repair difference values: When the future restoration difference value is greater than zero, the future restoration difference value is added to the cumulative level of ecological improvement items, and the occurrence count of ecological improvement items is incremented by one; When the future restoration difference is less than zero, the absolute value of the future restoration difference is added to the cumulative level of ecological risk items, and the occurrence count of ecological risk items is incremented by one. When the future repair difference value is zero, the number of occurrences of the unchanged item will be incremented by one.

[0188] After completing the traversal, the number of future time locations covered by the future time window is taken as the window length. The ratio of the number of occurrences of ecological improvement items to the window length is taken as the frequency of occurrence of ecological improvement items, the ratio of the number of occurrences of ecological risk items to the window length is taken as the frequency of occurrence of ecological risk items, and the ratio of the number of occurrences of unchanged items to the window length is taken as the frequency of occurrence of unchanged items.

[0189] By compiling information on ecological improvements and changes in ecological risks, a set of assessment elements for restoration strategies is generated, and these elements are then summarized.

[0190] S6.2: Summarize and organize the elements of the repair strategy evaluation according to the order of the future time windows on a unified timeline, generate the repair strategy evaluation conclusion, and output it to the management end as a basis for decision-making.

[0191] Based on the set of remediation strategy evaluation elements, the remediation strategy evaluation elements corresponding to each future time window are read one by one in chronological order on a unified time axis, and the remediation strategy evaluation elements are grouped by group identifier within the same future time window.

[0192] Based on trajectory clustering, the number of sea area grid spatial locations corresponding to each group identifier is read, and the number of sea area grid spatial locations of each group identifier is mapped to the group weight. The group weight is defined as the proportion of the number of sea area grid spatial locations corresponding to the group identifier in the total number of sea area grid spatial locations of all group identifiers, and satisfies that the sum of the group weights of all group identifiers is 1.

[0193] Based on group weights, the cumulative degree of ecological improvement items, the frequency of occurrence of ecological improvement items, the cumulative degree of occurrence of ecological risk items, and the frequency of occurrence of ecological risk items corresponding to each group identifier within the same future time window are summarized using a unified standard, and the summary results of ecological improvement and ecological risk at the future time window level are output.

[0194] Based on the summary results of ecological improvement and ecological risk at the future time window level, the cumulative summary value of ecological improvement items, the summary value of the frequency of occurrence of ecological improvement items, the cumulative summary value of the frequency of occurrence of ecological risk items, and the summary value of the frequency of occurrence of ecological risk items corresponding to the future time window are read one by one, and the number of future time locations covered by the future time window is read as the window length.

[0195] The cumulative sum of ecological improvement items and the cumulative sum of ecological risk items are normalized using the window length, and the window-level improvement intensity index and window-level risk intensity index are output.

[0196] The window-level net benefit index is calculated based on the window-level improvement intensity index and the window-level risk intensity index, and the weight sequence of future time windows is set according to the order of future time windows on a unified time axis.

[0197] The window-level net benefit index is weighted and aggregated based on the future time window weight sequence, and the repair strategy score is output.

[0198] In scenarios where the set of remediation strategy evaluation elements carries remediation measure type identifiers, a remediation strategy score is calculated for each remediation measure type identifier, and the remediation measure type identifiers are sorted according to the remediation strategy score, and the remediation strategy ranking result is output.

[0199] Based on the summary results of ecological improvement and ecological risk at the future time window level, ecological improvement trend description and ecological risk trend description are generated. The future time window identifier, ecological improvement trend description, ecological risk trend description and window-level net benefit index are then concatenated to generate future time window-level assessment items.

[0200] All future time window-level assessment items are connected in chronological order to form a repair strategy assessment conclusion. The repair strategy score and the repair strategy ranking result are then concatenated into the repair strategy assessment conclusion and output to the management end as a basis for decision-making.

[0201] This embodiment also provides a computer device applicable to the marine ecological data-based restoration strategy evaluation method, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the marine ecological data-based restoration strategy evaluation method proposed in the above embodiment.

[0202] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0203] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for evaluating restoration strategies based on marine ecological data as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0204] In summary, this invention enhances the spatial representation accuracy of ecological evolution sequences by using a time-series trajectory clustering method to cluster ecological response trajectory sets, generating trajectory clusters that include marine zoning. This highlights the local characteristics of ecological processes and improves the analytical capabilities of the overall framework. Furthermore, it enables historical restoration strategy records to obtain zoning attribution, establishes a consistent correlation between restoration measure types and restoration implementation intensity parameters and zoning-level ecological evolution expressions, forms a quantifiable basis for comparing the differences in the effects of restoration measures, and supports the interpretable output of restoration strategy evaluation conclusions.

[0205] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating restoration strategies based on marine ecological data, characterized in that: include, Collect multi-source marine ecological data and historical restoration strategy records, perform time alignment and spatial mapping, generate aligned continuous ecological state sequences on a unified time axis and marine grid, perform multi-dimensional temporal feature extraction on the continuous ecological state sequences, and output ecological response trajectory sets; The ecological response trajectory set is clustered using a time-series trajectory clustering method to generate trajectory clusters that include marine area partitions; Using trajectory clusters, we construct zone-level ecological evolution within the corresponding marine areas of each cluster and output the zone-level evolution sequence. The partition-level evolution sequence is combined with historical repair strategy records to form a prediction input sequence. Lead prediction is performed on the prediction input sequence to generate a strategy forward prediction sequence in multiple future time windows. Based on the strategic forward prediction sequence and the regional evolution sequence, a forward comparative relationship reflecting the differences in ecological changes before and after the restoration measures are implemented is constructed, and the future restoration difference sequence is output. The future repair difference sequence is transformed into repair strategy evaluation conclusions, and the repair strategy evaluation conclusions are output for management decision-making selection.

2. The method for evaluating restoration strategies based on marine ecological data as described in claim 1, characterized in that: The process involves collecting multi-source marine ecological data and historical restoration strategy records, performing time alignment and spatial mapping, and generating an aligned continuous ecological state sequence on a unified timeline and marine grid. The specific steps are as follows. Collect multi-source marine ecological data and historical restoration strategy records, and sort them according to the original timestamps to generate a basic time-series dataset; Based on the basic time-series dataset, a unified time axis is used to align multi-source marine ecological data with historical restoration strategy records at each observation time position, generating a time-aligned data stream; Based on the spatial location of the marine grid, the time-aligned data stream is mapped item by item to the corresponding spatial location to generate a spatially aligned dataset; Based on the ecological variables recorded by the spatial location of the marine grid in the spatially aligned dataset, an aligned continuous ecological state sequence is generated by continuous arrangement on a unified time axis.

3. The method for evaluating restoration strategies based on marine ecological data as described in claim 1, characterized in that: The specific steps for performing multi-dimensional temporal feature extraction on continuous ecological state sequences and outputting an ecological response trajectory set are as follows. For continuous ecological state sequences, a unified time axis is used to analyze the magnitude, direction, and trend of changes in ecological variables at different time locations in the marine grid, and extract temporal variation characteristics. Spatial variation features are generated by comparing the temporal variation characteristics of adjacent spatial locations within a sea area grid on a unified time axis. Based on temporal and spatial variation characteristics, ecological variable change trajectories are constructed for each ecological variable. The ecological variable change trajectories are aggregated according to their spatial location in the marine grid to generate an ecological response trajectory set.

4. The method for evaluating restoration strategies based on marine ecological data as described in claim 1, characterized in that: The specific steps for generating trajectory clusters that include sea area partitions are as follows. Based on the ecological response trajectory set, the trajectory difference measurement values ​​of any two ecological response trajectories in terms of change magnitude, change direction and change trend are calculated on a unified time axis, and a trajectory difference measurement set is generated. Using a trajectory difference metric set, a time-series trajectory clustering method is employed to cluster the ecological response trajectories in the ecological response trajectory set, assigning group identifiers to each cluster group, and generating a clustering record set containing the group identifiers. Based on the clustering of record sets, a correspondence between group identifiers and spatial locations of the sea area grid is established on the sea area grid. Trajectory clustering structures are generated and mapped to the sea area grid, and trajectory clusters containing sea area partitions are output.

5. The method for evaluating restoration strategies based on marine ecological data as described in claim 4, characterized in that: The specific steps are as follows: Using trajectory clustering to construct regional-level ecological evolution within the corresponding marine areas of each cluster, and outputting regional-level evolution sequences. Within the marine area corresponding to the trajectory cluster, the trajectory difference measurement set is used to compare the changes in all ecological variables within the trajectory cluster one by one in pairs, generating a set of paired difference measurement results. The set of pairwise difference measurement results is merged according to the ecological variable change trajectory. A summary operation is performed on the pairwise difference measurement results corresponding to each ecological variable change trajectory, and the difference measurement within the cluster is output. Based on the intra-cluster variance measure, the ecological variable change trajectory with the smallest intra-cluster variance measure is selected as the representative ecological variable change trajectory. Based on the change trajectory of representative ecological variables, the magnitude, direction and trend of change at each time point are connected in chronological order according to a unified time axis to generate a continuous ecological change process. Within the spatial range of marine zoning, zoning-level ecological evolution expressions are generated based on continuous ecological change processes, and zoning-level ecological evolution expressions are connected according to the temporal order of a unified time axis to generate zoning-level evolution sequences.

6. The method for evaluating restoration strategies based on marine ecological data as described in claim 1, characterized in that: The specific steps for generating the strategy forward prediction sequence within multiple future time windows are as follows. The partition-level evolution sequence is synchronized point by point with the historical repair strategy record on a unified time axis to generate a joint time series sequence, and the joint time series sequence is organized into a prediction input sequence according to the unified time axis. Based on the historical time position of the predicted input sequence on a unified time axis, the exponential smoothing method is used to calculate the lead time of the changing trend of each future time window, and generate a forward trend model. Based on forward-looking trend patterns, strategic forward-looking predictions are made for multiple future time windows one by one, generating a strategic forward-looking prediction sequence.

7. The method for evaluating restoration strategies based on marine ecological data as described in claim 1, characterized in that: The specific steps for outputting the future repaired differential sequence are as follows. The future time position of the strategy forward prediction sequence on a unified time axis is mapped point by point to the partition-level evolution sequence, and a forward control sequence is output. Based on prospective control sequences, the difference between the strategic prospective prediction sequence and the regional evolution sequence is calculated in the future time window to identify the differences in ecological changes in the future time window and generate difference candidates. The candidate differences are categorized and structured according to future time windows to generate a set of structured difference information. Based on the future time positions of the structured difference information set on a unified time axis, a forward-looking comparison relationship is gradually formed, and a future repair difference sequence is output.

8. The method for evaluating restoration strategies based on marine ecological data as described in claim 1, characterized in that: The specific steps for transforming future repair discrepancy sequences into repair strategy evaluation conclusions and outputting these conclusions for management decision-making are as follows. Based on the analysis of future restoration difference sequences, the ecological change differences in future time windows are analyzed. The cumulative degree and frequency of future restoration difference sequences are statistically analyzed according to future time windows, and an evaluation element set is generated to summarize the evaluation elements of restoration strategies. Based on the chronological order of future time windows on a unified timeline, the elements for evaluating the remediation strategy are summarized and organized to generate a remediation strategy evaluation conclusion, which is then output to the management end as a basis for decision-making.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the marine ecological data-based restoration strategy evaluation method as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the marine ecological data-based restoration strategy evaluation method as described in any one of claims 1 to 8.

Images