A species invasion detection method and system based on a correlation relationship

Abstract

The application provides a species invasion detection method and system based on a correlation relationship, relates to the cross field of ecology and information technology, and the method comprises the following steps: based on the species invasion correlation rule, performing rule matching analysis on the multi-source data acquired in real time to obtain a matching degree result; calculating a species invasion risk score according to the matching degree result; based on the comparison result of the species invasion risk score and a preset early warning threshold, performing early warning judgment to obtain an early warning judgment result; when the early warning judgment result is that early warning needs to be triggered, early warning information containing an early warning position and key data is obtained. Through multi-source data acquisition, fusion and feature extraction, adaptive correlation rules are constructed, real-time risk assessment is performed, and finally the complete process of accurate early warning is realized.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of ecology and information technology, and in particular to a method and system for detecting species invasion based on correlation relationships. Background Technology

[0002] Invasive alien species pose a serious threat to agricultural biodiversity and ecosystem stability. Establishing an effective integrated prevention and control system has become an important issue in ecological and environmental protection. Existing methods for detecting invasive species mostly rely on single data sources or simple data overlay analysis, which have obvious limitations at the data processing level, resulting in low accuracy of early warning and delayed prevention and control response.

[0003] Existing methods for processing spatial data sometimes remain at the level of simple visualization or statistics, lacking in-depth exploration of the characteristics of ecological gradient changes; they fail to construct adaptive monitoring networks based on the spatial heterogeneity of environmental parameters and species distribution, which may lead to a lack of basis for the placement of sampling points and the potential neglect of invasion risks in key ecological transition areas.

[0004] Existing methods sometimes rely on apparent correlations for risk assessment, failing to effectively distinguish between causal relationships and spurious associations. For example, they may misjudge the simultaneous occurrence of environmental factors and species with no biological connection as invasion signals, leading to reduced credibility of early warnings and thus wasting resources for prevention and control. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a species invasion detection method and system based on correlation, which achieves a complete process from multi-source data collection, fusion and feature extraction to constructing adaptive correlation rules and conducting real-time risk assessment, and finally realizes accurate early warning.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0007] Firstly, a species invasion detection method based on association relationships, the method comprising:

[0008] Collect species observation data, environmental monitoring data, and human activity data to obtain an initial multi-source dataset. Perform spatial processing and spatiotemporal alignment on the initial multi-source dataset to obtain a spatiotemporally aligned multi-source dataset.

[0009] Data fusion processing is performed on spatiotemporally aligned multi-source datasets to obtain a standardized fusion dataset containing ecological feature elements;

[0010] Based on the standardized fusion dataset, multiple spatial reference points are deployed within the monitoring area to construct a monitoring network; the monitoring network is divided into spatial units to obtain each spatial unit; rule correction parameters are obtained based on the ecological characteristics of each spatial unit; and species invasion association rules are constructed using the rule correction parameters.

[0011] Based on the aforementioned species invasion association rules, rule matching analysis is performed on real-time acquired multi-source data to obtain matching degree results; a species invasion risk score is calculated based on the matching degree results.

[0012] Based on the comparison between the species invasion risk score and the preset warning threshold, a warning judgment is performed to obtain a warning judgment result; when the warning judgment result indicates that a warning needs to be triggered, warning information containing the warning location and key data is obtained.

[0013] Secondly, a species invasion detection system based on association relationships includes:

[0014] The data acquisition module is used to collect species observation data, environmental monitoring data, and human activity data to obtain an initial multi-source dataset. The initial multi-source dataset is then spatially processed and spatiotemporally aligned to obtain a spatiotemporally aligned multi-source dataset.

[0015] The fusion module is used to perform data fusion processing on spatiotemporally aligned multi-source datasets to obtain a standardized fusion dataset containing ecological feature elements;

[0016] The module is used to construct a monitoring network by deploying multiple spatial reference points within the monitoring area based on the standardized fusion dataset; to divide the monitoring network into spatial units to obtain each spatial unit; to obtain rule correction parameters based on the ecological characteristics of each spatial unit; and to construct species invasion association rules using the rule correction parameters.

[0017] The matching module is used to perform rule matching analysis on real-time acquired multi-source data based on the species invasion association rules to obtain matching degree results; and to calculate the species invasion risk score based on the matching degree results.

[0018] The early warning module is used to perform an early warning judgment based on the comparison result between the species invasion risk score and the preset early warning threshold, and obtain an early warning judgment result; when the early warning judgment result is that an early warning needs to be triggered, early warning information containing the early warning location and key data is obtained.

[0019] Thirdly, a computing device includes:

[0020] One or more processors;

[0021] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0022] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0023] The above-described solution of the present invention has at least the following beneficial effects:

[0024] Multi-source data is collected and integrated into an initial set. Through spatial processing and spatiotemporal alignment, the spatial benchmark and temporal dimension of the data are unified, enabling scattered data to form a spatiotemporally related whole, providing a consistent data foundation for subsequent fusion. The spatiotemporally aligned data is then fused to achieve organic integration and standardization of multi-source data, extracting ecological characteristic elements to ensure both data integrity and standardization, providing high-quality data support for subsequent analysis. A monitoring network is deployed based on standardized data, dividing spatial units to adapt the network layout to ecological characteristics. Association rules are constructed by modifying parameters through rules, ensuring the rules align with the actual ecological characteristics of each spatial unit and improving their relevance. Real-time data is matched and analyzed based on association rules, and risk scores are calculated using matching degrees, providing a clear rule basis for risk assessment and quantifiable scoring results. Early warning judgments are executed by comparing risk scores with preset thresholds, generating information containing warning locations and key data, making early warning judgment standards clear and enabling the direct transformation of data processing results into information with practical operational guidance. Attached Figure Description

[0025] Figure 1 This is a schematic flowchart of a species invasion detection method based on association relationships provided by an embodiment of the present invention.

[0026] Figure 2 This is a schematic diagram of a species invasion detection system based on association relationships provided in an embodiment of the present invention. Detailed Implementation

[0027] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0028] like Figure 1 As shown, embodiments of the present invention propose a species invasion detection method based on association relationships, the method comprising the following steps:

[0029] Step 100: Collect species observation data, environmental monitoring data, and human activity data to obtain an initial multi-source dataset. Perform spatial processing and spatiotemporal alignment on the initial multi-source dataset to obtain a spatiotemporally aligned multi-source dataset.

[0030] Step 200: Perform data fusion processing on the spatiotemporally aligned multi-source dataset to obtain a standardized fusion dataset containing ecological feature elements;

[0031] Step 300: Based on the standardized fusion dataset, multiple spatial reference points are deployed within the monitoring area to construct a monitoring network; the monitoring network is divided into spatial units to obtain each spatial unit; rule correction parameters are obtained based on the ecological characteristics of each spatial unit; and species invasion association rules are constructed using the rule correction parameters.

[0032] Step 400: Based on the species invasion association rules, perform rule matching analysis on the real-time acquired multi-source data to obtain the matching degree results; calculate the species invasion risk score based on the matching degree results.

[0033] Step 500: Based on the comparison result between the species invasion risk score and the preset warning threshold, a warning judgment is performed to obtain a warning judgment result; when the warning judgment result indicates that a warning needs to be triggered, warning information containing the warning location and key data is obtained.

[0034] In this embodiment of the invention, by collecting species observation data, environmental monitoring data, and human activity data, and performing spatial processing and spatiotemporal alignment on the initial multi-source dataset, data from different sources can be kept consistent in spatial coordinates and time dimensions, eliminating correlation barriers caused by inconsistent spatiotemporal attributes of the data. Data fusion processing of the spatiotemporally aligned multi-source dataset generates a standardized fusion dataset containing ecological feature elements, transforming data of different formats and types into a unified format, avoiding additional work caused by format differences in subsequent processing. Constructing a monitoring network based on spatial reference points deployed from the standardized fusion dataset allows for more reasonable monitoring coverage and clear spatial references. By dividing spatial units and obtaining rule correction parameters based on the ecological characteristics of each unit, the subsequently constructed species invasion correlation rules can remain consistent with the actual ecological conditions of each spatial unit. Constructing correlation rules using rule correction parameters enables the rules to have spatial specificity, avoiding insufficient adaptation of unified rules to regions with different ecological characteristics, and improving the fit between correlation rules and the actual situation of the monitoring area.

[0035] By performing rule-matching analysis on real-time multi-source data based on species invasion association rules, the analysis process of real-time data becomes rule-based. Calculating a species invasion risk score based on the matching results directly links the risk assessment process with the rule-matching results, providing a traceable computational basis for the risk score and improving the standardization of the risk assessment process. Executing early warning judgments based on the comparison between the species invasion risk score and preset early warning thresholds provides clear criteria for triggering early warnings. When an early warning needs to be triggered, generating early warning information containing the warning location and key data directly transforms the data processing results into information with practical operational guidance. This allows prevention and control personnel to obtain key verification evidence without additional data screening, enhancing the practical value of early warning information and the efficiency of subsequent prevention and control operations.

[0036] In a preferred embodiment of the present invention, step 100 above involves collecting species observation data, environmental monitoring data, and human activity data to obtain an initial multi-source data set. Spatial processing and spatiotemporal alignment are then performed on the initial multi-source data set to obtain a spatiotemporally aligned multi-source data set, including:

[0037] Step 101 involves collecting species observation data through on-site patrols and recordings to obtain structured observation records containing information on species type, quantity, and geographical location. Specifically, this includes: First, for the collection of species observation data, a pre-planned on-site patrol route is developed, taking into account the ecological characteristics of the monitoring area and the potential distribution areas of invasive alien species. This route must cover the core ecological zone, the marginal transition zone, and areas with frequent human activity within the monitoring area to ensure comprehensiveness of the observation scope. Then, professionally trained patrol personnel are arranged to conduct periodic patrols along the pre-set route. The patrol cycle can be set according to the ecological sensitivity of the monitoring area and the reproductive cycle of invasive species, for example, conducting 2 to 3 comprehensive patrols per month. During the patrols, patrol personnel use portable data recording devices to record the specific information of the species discovered in real time. Species type must be labeled according to a unified species classification standard, quantity must be quantified based on the actual number of individuals observed or the population size, and geographical location information is obtained through the device's built-in positioning module, ensuring that each species observation record contains these three core information categories: species type, quantity, and geographical location. Finally, the information recorded in real time is organized according to a pre-set data structure to form structured observation records.

[0038] Step 102 involves collecting environmental monitoring data through a deployed environmental sensor network to obtain a sensor dataset containing coordinate locations and corresponding environmental parameters. Specifically, for environmental monitoring data collection, the environmental sensor network is first rationally deployed within the monitoring area based on its area, topography, and ecological zoning. The sensor placement must consider both the representativeness and coverage of environmental parameters. For example, sensors are deployed in wetland ecological zones, green ecological zones, and areas with concentrated human activity, and the spacing between adjacent sensors must be controlled to ensure the continuity of monitoring data. The deployed environmental sensors must be capable of collecting multiple key environmental parameters, including but not limited to temperature, humidity, soil moisture content, and light intensity, all of which are closely related to the survival and reproduction of invasive alien species. After sensor deployment, the sensor network is activated for continuous data collection. The collection frequency can be set according to the changing characteristics of environmental parameters, for example, collecting data every 30 minutes. During each collection, the sensor not only records the corresponding environmental parameter values ​​but also simultaneously records its own coordinate location information, ensuring that each set of environmental parameter data is bound to a specific spatial location. Finally, the data collected by all sensors, including coordinate locations and corresponding environmental parameters, are summarized to form a sensor dataset.

[0039] Step 103 involves collecting human activity data through entrance gates and visitor registration data to obtain an activity log including visitor numbers, origins, types of items carried, and areas visited within the park. Specifically, this includes: firstly, establishing a data collection system based on the entrance management system and visitor service system of the monitoring area; secondly, deploying gate devices with data recording capabilities at the entrance to count the number of visitors entering the monitoring area in real time. These gate devices must be able to identify visitor entry and exit status to ensure the accuracy of visitor count statistics; and thirdly, during visitor registration, obtaining visitor origin information through the registration system. This origin information must be recorded down to the provincial administrative level for subsequent analysis. The study analyzes the potential risks of tourists carrying invasive species in different areas. Furthermore, it records the types of items tourists bring through manual questioning or intelligent detection equipment, focusing on plant, soil, and live organism items, as these are the main carriers of invasive species spread through human activities. After tourists enter the park, video surveillance equipment and tourist positioning systems deployed within the monitoring area, combined with tourist registration information, track and record the specific areas where tourists stay in the park. These records must correspond to pre-defined spatial units within the monitoring area. Finally, the collected information on the number of tourists, their origins, types of items brought, and areas where they stayed in the park is organized chronologically to form an activity log.

[0040] Step 104: Integrate structured observation records, sensor datasets, and activity logs to obtain an initial multi-source dataset. This initial multi-source dataset includes species observation data, environmental monitoring data, and human activity data. Specifically, after collecting the species observation data, environmental monitoring data, and human activity data separately, the initial multi-source dataset is constructed. First, a unified data integration standard is established. This standard needs to define the format requirements, field definitions, and timestamp formats for various types of data to ensure that data from different sources can be integrated under the same standard. Subsequently, using timestamps as the core linking element, the structured observation records, sensor datasets, and activity logs are aggregated. In the aggregation process, preliminary integrity checks are performed on various types of data, removing data entries that are obviously missing core information, such as geographical locations missing from species observation records or environmental parameters missing from sensor datasets. Simultaneously, the aggregated data is categorized and stored according to data type; for example, species-related data, environment-related data, and human activity-related data are stored in different data modules, and each module is linked through a unique identifier field to ensure rapid retrieval of various types of data within a specific time period or spatial region. Through this process, an initial multi-source data set containing species observation data, environmental monitoring data, and human activity data is ultimately obtained.

[0041] Step 105: For the species observation data in the initial multi-source data set, by calling predefined coordinate transformation parameters, the geographical location information based on different geodetic coordinate systems is uniformly transformed to the target spatial coordinate system to obtain species observation data with unified spatial reference. Specifically, this includes: first, determining the target spatial coordinate system corresponding to the monitoring area. This coordinate system needs to be selected according to the geographical location of the monitoring area and the subsequent data processing and application requirements, such as using the National Geodetic Coordinate System 2000 or the local plane rectangular coordinate system of the monitoring area.

[0042] After determining the target spatial coordinate system, the work of predefining coordinate transformation parameters begins. First, the types of source geodetic coordinate systems that may be involved in the species observation data in the initial multi-source dataset are identified, and the correspondence between each source coordinate system and the target spatial coordinate system is determined. Then, coordinate transformation parameters are acquired and predefined through two methods: one is to apply to the surveying and mapping department for officially published standard transformation parameters between corresponding coordinate systems, ensuring that the parameters meet national or industry surveying accuracy standards; the other is to obtain parameters through on-site measurement and calibration if the monitoring area has special terrain or insufficient coverage by official parameters. Specifically, multiple control points with known precise coordinates (based on the target spatial coordinate system) can be selected within the monitoring area, and the coordinate data of these control points in each source coordinate system can be collected. The transformation parameters between the source and target coordinate systems are obtained through coordinate inversion. After obtaining the parameters, their accuracy needs to be verified. The parameters are substituted into some species observation data with known coordinates for transformation testing. The transformation results are compared with the standard coordinates in the target coordinate system to ensure that the error is controlled within a preset range, such as no more than 5 meters. After verification, these transformation parameters are stored according to the source coordinate system type, forming a predefined set of coordinate transformation parameters, thus completing the predefinition of the coordinate transformation parameters.

[0043] Subsequently, the predefined coordinate transformation parameter set is invoked, and the corresponding predefined transformation parameters are matched according to the source geodetic coordinate system to which each species observation data belongs in the initial multi-source data set. Through the preset coordinate transformation program, the geographical location information of each species observation data is transformed to the target spatial coordinate system one by one. During the transformation process, the accuracy of the transformation result of each species observation data needs to be verified to ensure that the error of the transformed geographical location information is controlled within a preset range, for example, the error does not exceed 10 meters. Through the above transformation and verification process, species observation data with unified spatial reference is finally obtained.

[0044] Step 106: Based on the species observation data with unified spatial reference and the environmental monitoring data in the initial multi-source dataset, an environmental parameter spatial field covering the entire monitoring area is obtained. Specifically, this includes: First, analyzing the species observation data with unified spatial reference to extract spatial characteristics of species distribution, such as concentrated species distribution areas and ecological boundaries of species distribution. These characteristics serve as important references for constructing the environmental parameter spatial field, ensuring that the spatial field accurately reflects the actual ecological conditions of the monitoring area. Next, preprocessing the sensor dataset in the initial multi-source dataset to remove abnormal environmental parameter data, such as temperature and humidity data exceeding reasonable ranges. Then, spatially mapping the preprocessed environmental parameter data in the target spatial coordinate system based on the sensor coordinates. Next, using spatial interpolation, based on the mapped sensor environmental parameter data and combined with the spatial characteristics of species distribution, supplementary calculations are performed for environmental parameters in areas where sensors are not deployed. During interpolation, the variation patterns of environmental parameters in different ecological regions must be fully considered, such as the difference in humidity gradient between wetland and green areas, to ensure that the supplemented environmental parameter data accurately reflects the ecological gradient changes within the monitoring area. Through the above process, the environmental parameter spatial field covering the entire monitoring area is finally obtained.

[0045] Step 107 involves performing spatiotemporal correlation matching between the environmental parameter spatial field and the human activity data in the initial multi-source dataset to establish a correspondence between environmental parameters, species observations, and human activities. Based on this correspondence, a spatiotemporally aligned multi-source dataset is obtained. Specifically, this includes: first, determining a unified time granularity and spatial unit division standard. The time granularity can be set according to the data acquisition frequency, for example, one hour as a time unit. The spatial units can be divided into several equally sized or adjustable polygonal units based on the area and ecological characteristics of the monitoring area. Subsequently, based on the established time and spatial units, the environmental parameter spatial field is spatiotemporally discretized to obtain the environmental parameters corresponding to each spatial unit under each time unit. The data is processed using the same spatiotemporal discretization method. Based on the areas and durations tourists spend in the park, the number of tourists, their origin distribution, and the proportion of different types of items they carry are statistically analyzed within each spatial unit of each time period. Next, the discretized environmental parameter data, human activity data, and species observation data with a unified spatial benchmark are correlated and matched. Specifically, the environmental parameters and human activity statistics within a specific spatial unit of a given time period are bound to the species observation data recorded within that unit, establishing a one-to-one correspondence between environmental parameters, species observations, and human activities. Through this spatiotemporal correlation and matching process, a spatiotemporally aligned multi-source dataset is finally obtained.

[0046] In this embodiment of the invention, by collecting species observation data through on-site inspections and forming structured records containing species type, quantity, and geographical location, scattered observation information can be organized into standardized data, clearly presenting the correspondence between species attributes and spatial location, and providing well-organized basic species data for subsequent multi-source data integration and association; by collecting environmental monitoring data through sensor networks and generating datasets containing coordinates and environmental parameters, continuous and stable environmental data collection can be achieved, and environmental parameters are directly bound to coordinates, giving environmental data clear spatial attributes and laying the foundation for subsequent matching with other spatial data; by collecting human activity data through turnstiles and visitor registration and forming activity logs containing visitor numbers, origins, carried items, and areas of stay, human activity related to species invasion can be comprehensively captured. This system provides key information on various activities, covering multiple dimensions. It integrates three types of data to form an initial multi-source dataset, unifying scattered species, environmental, and human activity data from different channels, eliminating physical dispersion, and providing a centralized data carrier for subsequent spatial processing and spatiotemporal alignment. By calling parameters to unify the different coordinate systems of species observation data into a target coordinate system, it eliminates spatial benchmark inconsistencies, ensuring that the location information of species observation data is based on the same standard. This guarantees no positional misalignment when matching with other spatial data, ensuring consistent spatial attributes. Based on species and environmental data with unified coordinates, it constructs an environmental parameter spatial field, overcoming the limitations of sensor point data and expanding discrete environmental data into continuously distributed data covering the entire monitoring area, filling data gaps in areas where sensors are not deployed.

[0047] By matching the spatial field of environmental parameters with human activity data in a spatiotemporal manner, establishing the correspondence between the three types of data and forming a spatiotemporally aligned multi-source data set, the three types of data can form a clear correlation in the spatiotemporal dimension, ensuring that environmental, species, and human activity information within a specific spatiotemporal space can be matched.

[0048] In a preferred embodiment of the present invention, step 200 above involves performing data fusion processing on a spatiotemporally aligned multi-source dataset to obtain a standardized fusion dataset containing ecological feature elements, including:

[0049] Step 201 involves cleaning the spatiotemporally aligned multi-source dataset. This is achieved by filling in missing values ​​and smoothing outliers, resulting in a cleaned multi-source dataset. Specifically, this includes: firstly, screening the species observation data, environmental monitoring data, and human activity data in the multi-source dataset for completeness and validity to identify missing and outliers. Specifically, for species observation data, the focus is on checking for missing entries in species type, quantity, and geographical location information. For example, if a record lacks geographical location information, it is considered a missing value. For environmental monitoring data, by comparing it to a preset reasonable range of environmental parameters, such as temperatures typically between -10°C and 40°C and humidity between 0% and 100%, values ​​exceeding this range are considered outliers. For instance, an instantaneous humidity of 120% collected by a sensor is an outlier. For human activity data, the number of tourists, their origins, types of items carried, and areas of stay are checked for unrecorded items. If only the number of tourists is recorded for a certain period without recording the area of ​​stay, it is considered a missing value.

[0050] After identifying missing and outlier values, corresponding processing methods are adopted based on the characteristics of different data types. For missing values, if the species quantity is missing in the species observation data, the average species quantity of the three adjacent valid observation periods in the area is used to fill in the missing values, ensuring that the filled values ​​match the actual species distribution in the area. If the humidity is missing in the environmental monitoring data, the humidity values ​​of the two valid collection times before and after the same sensor are used for linear interpolation to fill in the missing values, ensuring the continuity of environmental parameters. If the area of ​​stay is missing in the human activity data, the area of ​​stay is determined by combining the visitor's entry time and gate exit time with the visitor's activity trajectory recorded by video surveillance during the same period. For outliers, if the instantaneous abnormal temperature is in the environmental monitoring data, the temperature values ​​of the sensor at 10 consecutive collection times on the same day are used for moving average processing to smooth abnormal fluctuations. If the number of visitors is abnormal in the human activity data, such as a negative number of visitors recorded at a certain time period, the average number of visitors in the two hours before and after that time period is used for correction. Through the above operations of filling missing values ​​and smoothing outliers, the cleaned multi-source data set is finally obtained.

[0051] Step 202 involves standardizing the cleaned multi-source dataset to obtain a numerically standardized multi-source dataset. This includes: firstly, establishing unified standardization rules based on the numerical characteristics of different types of data in the multi-source dataset; specifically, for the number of tourists in human activity data, temperature in environmental monitoring data, and the number of species in species observation data, where the number of tourists ranges from 0 to 500 people, the temperature in environmental monitoring data ranges from -10℃ to 40℃, and the number of species in species observation data ranges from 0 to 100 individuals, determining the standardization target range for each type of data. Typically, these data are uniformly converted to the range of 0 to 1 to eliminate the impact of differences in the original value scales of different data.

[0052] After determining the transformation rules, standardization transformation operations were performed on each type of data in the cleaned multi-source dataset. For data such as the number of tourists, which are non-negative integers, the transformation was achieved by calculating the ratio of the data value to the maximum number of tourists within the monitoring period. For example, if the number of tourists at a certain time period was 200 and the maximum number of tourists within the monitoring period was 500, the standardized value after transformation would be the ratio of 200 to 500. For data such as temperature, which may contain negative values, the transformation was achieved by first calculating the difference between the temperature value and the lowest temperature value, and then dividing by the difference between the highest and lowest temperature values. For example, if the temperature at a certain moment was 15℃, the lowest temperature was -10℃, and the highest temperature was 40℃, the standardized value was obtained by first calculating the difference between 15℃ and -10℃ and then dividing by the difference between 40℃ and -10℃. For data such as the number of species, which are related to population distribution, the transformation was performed by using the ratio of the number of species to the historical maximum observed number of the species in the monitoring area to ensure that the transformed data can reflect the relative scale of the species numbers. Through the above standardization transformation operations, a numerically standardized multi-source dataset was finally obtained.

[0053] Step 203 involves performing feature fusion processing on the numerically standardized multi-source dataset to obtain a fused feature set. Specifically, this includes: firstly, extracting key features with ecological relevance from the three types of data in the numerically standardized multi-source dataset; secondly, extracting spatiotemporal trend features of species type and species quantity from species observation data, such as the increase in the number of invasive alien species in a spatial unit over the past week; thirdly, extracting the mean and rate of change features of temperature, humidity, and soil moisture content from environmental monitoring data, such as the daily average humidity and daily humidity variation in a spatial unit; and fourthly, extracting features of tourist origin risk level, the proportion of tourists carrying plant-based items, and the length of stay of tourists in highly ecologically sensitive areas from human activity data. For example, the tourist origin risk level can be defined as a high-risk origin area if provinces with a high incidence of invasive alien species are classified as high-risk origin areas, such as the proportion of tourists from high-risk origin areas in a spatial unit and the average length of stay of tourists in wetland ecological areas.

[0054] After extracting key features, a feature fusion logic is constructed based on the survival, reproduction, and spread patterns of invasive alien species. Key features from different sources are organically combined to form fused features. For example, the growth rate of invasive alien species in a spatial unit is combined with the average humidity and temperature of the day to form a fused feature reflecting the species' reproduction and environmental adaptability. The proportion of tourists from high-risk sources is combined with the proportion of tourists carrying plant-based items and the growth rate of species numbers to form a fused feature reflecting the correlation between the risk of human activity spread and species invasion. The length of time tourists stay in wetland ecosystems is combined with the soil moisture content and species numbers in wetland ecosystems to form a fused feature reflecting the coupling between human activities and species distribution in a specific ecological area. Through the above feature extraction and combination operations, a fused feature set is finally obtained.

[0055] Step 204: Integrate the fused feature set with the structured observation records in the numerically standardized multi-source data set to obtain a standardized fused dataset containing ecological feature elements. Specifically, this includes: First, determining the core content of the structured observation records, which includes basic and key structured data such as species type, species quantity, and geographical location information. These data are the basis for subsequent monitoring network deployment and species invasion association rule construction, and must be completely retained in the final dataset.

[0056] While preserving the core content of structured observation records, and following the principle of spatiotemporal matching, various fusion features in the fusion feature set are associated and bound with the structured observation records. Specifically, using the geographical location information and timestamp in the structured observation records as matching keywords, fusion features within the same spatial unit and the same time period are associated with the structured observation records under that spatiotemporal unit. Among them, fusion features within the same time period include fusion features of species reproduction and environmental adaptability, and fusion features of the correlation between human activity spread risk and species invasion. For example, if a structured observation record records the species type, quantity, and geographical location of spatial unit A from 9:00 to 10:00 on a certain day, then all fusion features of spatial unit A during that time period are associated with that record.

[0057] After the association and binding are completed, the integrated data is formatted and sorted according to the preset data field order. The preset data field order is as follows: first the basic fields of the structured observation records, and then the fusion feature fields. This ensures that each data point contains basic species information, environmental parameter information, human activity information, and fusion feature information, and that the data format is uniform and the field definitions are clear. Through the above integration and standardization operations, a standardized fusion dataset containing ecological feature elements is finally obtained.

[0058] In this embodiment of the invention, spatiotemporally aligned multi-source data is cleaned to fill in missing values, smooth outliers, correct information gaps and numerical deviations, ensure data integrity and accuracy, avoid subsequent fusion biases, and provide a reliable foundation for standardization transformation and feature fusion. The cleaned data is then standardized to unify the numerical scales of different data types, eliminate weight imbalances caused by different units and value ranges, ensure the participation of various data types in analysis, provide conditions for deep fusion of multi-source data, and meet the requirements of multi-dimensional collaborative analysis. The standardized data feature fusion integrates key information from species, environment, and human activity data, uncovers potential correlations, breaks through the limitations of single data dimensions, comprehensively reflects the ecological state, and provides core support for constructing datasets containing ecological features. Finally, the fusion feature set is integrated with structured observation records in the standardized data to obtain a standardized fusion dataset containing ecological features, achieving a combination of original basic data and deep fusion features, providing comprehensive and standardized data support for deploying monitoring networks and constructing intrusion association rules.

[0059] In a preferred embodiment of the present invention, step 300 involves: deploying multiple spatial reference points within the monitoring area to construct a monitoring network based on the standardized fusion dataset; dividing the monitoring network into spatial units to obtain the divided spatial units; obtaining rule correction parameters based on the ecological characteristics of each divided spatial unit; and constructing species invasion association rules using the rule correction parameters, including:

[0060] Step 301 involves performing spatial heterogeneity analysis on the environmental parameters and species distribution data in the standardized fusion dataset to identify the ecological gradient change characteristics within the monitoring area. Specifically, this includes: first, retrieving the standardized fusion dataset containing ecological features, and then selecting environmental parameters and species distribution data closely related to the survival and reproduction of invasive alien species. Environmental parameters may include temperature, humidity, and soil moisture content, while species distribution data may include the observation location and population size of different species. Subsequently, based on the selected data, a spatial block statistical approach is used to analyze the monitoring area layer by layer. Specifically, the monitoring area is divided into several basic analysis blocks according to a preset grid scale. The mean, variance, and density differences of environmental parameters and species distribution within each basic analysis block are calculated. By comparing the degree of parameter differences between different basic analysis blocks, the spatial variation patterns of environmental parameters and species distribution are identified. Further, based on these variation patterns, the ecological gradient change characteristics within the monitoring area are identified, such as the humidity gradient change from wetland ecological zones to green ecological zones, and the species distribution density gradient change from densely populated human activity areas to core ecological zones.

[0061] Step 302: Based on the characteristics of ecological gradient changes, a differentiated spatial sampling strategy is derived. Specifically, this includes: for areas with significant ecological gradient changes, such as transitional areas where humidity drops sharply from 60% to 30%, or areas where species density increases from 10 individuals / m² to 50 individuals / m², these areas are designated as key sampling areas. Within these areas, the spacing between sampling points is appropriately reduced to increase sampling density, ensuring accurate capture of data details during gradient changes. For areas with gentle ecological gradient changes, such as areas where temperature remains stable between 25 and 28°C and species density fluctuations are less than 5 individuals / m², these areas are designated as regular sampling areas. Within these areas, the spacing between sampling points can be appropriately increased to control sampling costs. Simultaneously, the sampling strategy ensures that the direction of sampling point placement aligns with the direction of ecological gradient change. For example, sampling points are placed along the direction of humidity gradient from high to low to fully cover the entire range of gradient changes. This strategy allows the sampling plan to closely match the actual ecological differences of the monitoring area, avoiding blind sampling and laying the foundation for obtaining high-quality sampling data.

[0062] Step 303: Based on the differentiated spatial sampling strategy, and according to the preset spatial reference point quantity constraint, adjust the sampling point distribution to obtain a spatial reference point coordinate set that meets the requirements of ecological gradient coverage and spatial distribution balance. Specifically, this includes: First, based on the total area of ​​the monitoring area, topographic complexity, and ecological zoning results, such as the division into core ecological areas, edge transition areas, and densely populated human activity areas, determine the basic requirements for sampling point density for different ecological zones. For example, if the core ecological area requires higher monitoring accuracy, then the sampling point density in this area needs to be higher than in other areas. Second, referring to the available hardware resources (such as the total number of environmental sensors and data acquisition terminals) and the subsequent maintenance manpower configuration, determine the hardware setup. The upper limit of sampling points that can be supported is determined; then, combined with the effective coverage efficiency of sampling points in different areas in historical monitoring data, such as the past data showing that 8 sampling points can achieve effective monitoring of 1 square kilometer core ecological area and 12 sampling points can achieve effective monitoring of 1 square kilometer conventional area, the total number of preliminary reference points is calculated based on the area of ​​each ecological zone and the corresponding coverage efficiency; finally, the preliminary total number is matched with the equipment procurement cost and data transmission and storage cost. If the preliminary total number exceeds the cost tolerance range, the sampling point quota of the core ecological area is reserved first, and the quota of the conventional area is appropriately reduced until the total number is within the range of resources and costs that can be tolerated, and finally the spatial reference point number constraint is determined and preset.

[0063] After fulfilling the preset spatial reference point quantity constraints, sampling points are deployed in key sampling areas at a preset high-density spacing (e.g., 50 meters per point). Key sampling areas include core ecological zones and ecological transition zones. In regular sampling areas, sampling points are deployed at a preset low-density spacing (e.g., 100 meters per point). Regular sampling areas include areas surrounding human activity zones. The total number of sampling points after the initial deployment is counted. If the number of initially deployed sampling points does not exceed the preset quantity constraints, the spatial distribution of sampling points is directly checked for balance. The electronic map of the monitoring area is used to check for any instances of over-concentration or omission of sampling points in local areas. Over-concentration in local areas, such as a core ecological sub-zone, is indicated by three or more sampling points deployed within a 20-meter radius. Omission is indicated by no sampling points deployed in a certain edge transition zone. For overly concentrated sampling points, the number should be appropriately reduced according to the principle of retaining key location points and eliminating duplicate coverage points. For missed areas, sampling points should be added according to the direction of ecological gradient change. If the number of sampling points initially deployed exceeds the preset quantity constraint, sampling points in key sampling areas should be retained first, and sampling points in regular sampling areas should be reduced by interval elimination, such as retaining 2 out of every 3 sampling points. The retained points must cover the key ecological nodes in the area. At the same time, the ecological gradient change map should be used to ensure that the sampling points in the regular sampling areas after reduction can still cover the complete ecological gradient change from high humidity to low humidity and from high vegetation cover to low vegetation cover. Through the above preset quantity constraint, initial deployment and adjustment operations, the spatial reference point coordinate set that meets both the ecological gradient coverage requirements and the spatial distribution balance is finally obtained.

[0064] Step 304: Based on the spatial reference point coordinate set, construct a monitoring network covering the entire monitoring area. Specifically, this includes: First, according to the coordinate position of each spatial reference point, mark the specific location of all reference points on the electronic map of the monitoring area. When marking, it is necessary to associate the ecological zoning information corresponding to each reference point in step 303. For example, reference points in the core ecological area are marked in red, and reference points in the regular area are marked in blue. Ensure that the marked position of each reference point corresponds completely with the actual geographic coordinates, providing a clear visualization basis for subsequent spatial topology construction.

[0065] After accurately labeling the spatial reference points, a triangulation algorithm is introduced to construct the spatial topology of the reference points to build a stable and efficient basic network framework. First, the constraints of the triangulation are determined, combining the terrain features of the monitoring area with the effective communication distance of the data transmission equipment. Terrain features include avoiding signal transmission obstacles such as rivers and steep slopes, and the effective communication distance of the data transmission equipment is set to a maximum transmission radius of 500 meters, with a preset distance threshold for link construction. Then, the Delaunay triangulation algorithm is used to traverse all spatial reference points, calculating the straight-line distance between any two points and excluding point pairs with a distance exceeding 500 meters or crossing signal obstacle areas. The qualified reference points are used as vertices to construct a non-overlapping triangular network, ensuring that each reference point belongs to at least one vertex of a triangle, and that the circumcircle of each triangle does not contain other reference points. This triangulation method avoids the formation of elongated triangles, ensuring uniform link lengths between reference points and reducing data transmission latency. Finally, the sides of the triangles formed by the triangulation are used as virtual links between spatial reference points, replacing the fuzzy rules of traditional interconnection of adjacent reference points, making the link layout of the basic network framework more spatially logical and structurally stable.

[0066] After completing the basic link construction, spatial reference points are treated as nodes, and virtual links formed by triangulation are treated as edges to construct the topology of the monitoring network. Using the connected component detection method in graph theory, all nodes in the topology are traversed, and the number of reference points and link transmission parameters, such as estimated transmission delay and signal strength, are counted for each connected component. If an isolated connected component with only a single reference point is detected (i.e., the reference point has no effective link connection), the nearest reference point with unobstructed signal is selected from the neighboring triangles, and a virtual link is added to connect it to the main network. The transmission delay of the added link is verified to be within a preset threshold (e.g., 1 second). If a link transmission delay exceeding 1 second is detected within a connected component (e.g., although the distance between reference points does not exceed the threshold, there is tall vegetation obstruction), a relay reference point is added at the midpoint of the link, splitting the original link into two short links. The transmission delay is recalculated to ensure that the delay of both split links is below the threshold. Through the above connectivity analysis and optimization, it is ensured that all reference points in the monitoring network are within the same connected component, with no data transmission blind spots.

[0067] Furthermore, following the principle that each aggregation node covers 3 to 5 adjacent triangular sub-regions, reference points with the best geographical location and signal transmission conditions are selected as data aggregation nodes within each coverage area. For example, aggregation nodes in the core ecological area are preferentially selected at the centroid of the triangular sub-region to ensure coverage efficiency. Each aggregation node needs to preset its data receiving range and specify the list of spatial reference points it is responsible for receiving. For example, an aggregation node may correspond to 8 reference points, all of which come from the 3 triangular sub-regions it covers. At the same time, a dedicated data transmission channel is established between the data aggregation nodes and the monitoring center. When constructing the channel, the transmission path needs to be verified again through connectivity analysis. The path with the fewest links and the lowest transmission delay between the aggregation node and the monitoring center is selected. If there is a risk of path interruption, a backup relay node is added to the path to ensure that the monitoring data can be uploaded to the monitoring center in real time and stably.

[0068] Through the above operations of precise positioning, triangulation to establish links, connectivity analysis and optimization, aggregation node deployment, and transmission channel construction, a monitoring network covering the entire monitoring area is finally constructed.

[0069] Step 305 involves dividing the monitoring network into adjacent areas, dividing the monitoring area into multiple spatial units centered on spatial reference points. Specifically, this includes: first, determining the influence range of each spatial reference point. The radius of this influence range can be determined based on the distance between adjacent spatial reference points, typically half the distance between adjacent reference points. For example, if the distance between adjacent reference points is 100 meters, the radius of the influence range of each reference point is set to 50 meters. Then, a circular area is drawn with each spatial reference point as the center and the determined radius of the influence range as the radius. If adjacent circular areas overlap, the overlapping area is divided using a perpendicular bisector method. That is, the perpendicular bisector of the line connecting two adjacent spatial reference points is drawn, and the overlapping area is evenly distributed to the circular areas corresponding to the two spatial reference points. Through the above division method, the entire monitoring area is divided into multiple non-overlapping spatial units centered on spatial reference points, with each spatial unit corresponding to only one spatial reference point.

[0070] Step 306: Based on the standardized fusion dataset, calculate the ecological characteristic parameters within each spatial unit. These parameters include environmental parameter statistics and species distribution density. Specifically, this involves: First, for each spatial unit, extracting all environmental monitoring data and species observation data from the standardized fusion dataset. For example, extracting temperature and humidity data and species observation records for the most recent monitoring period within a spatial unit. Then, performing statistical calculations on the extracted environmental monitoring data to obtain the environmental parameter statistics within the spatial unit. This may include the mean, maximum, minimum, and variation range of the environmental parameters. For example, calculating the average temperature within a spatial unit as 26℃ and the temperature variation range as 5℃. Simultaneously, performing density calculations on the extracted species observation data, i.e., counting the total number of individuals of the target species within the spatial unit and dividing it by the actual area of ​​the spatial unit to obtain the species distribution density. For example, if the total number of individuals of the target species within a spatial unit is 200 and the unit area is 1000 square meters, the species distribution density is 0.2 individuals / square meter. Through the above calculations, the ecological characteristic parameters corresponding to each spatial unit are obtained.

[0071] Step 307: Integrate all spatial units and their corresponding ecological characteristic parameters to obtain the divided spatial units with ecological characteristic annotations. This includes: First, establishing a mapping table between spatial units and ecological characteristic parameters, recording the unique identifier, spatial range coordinates, and corresponding ecological characteristic parameters of each spatial unit in the mapping table. The unique identifier is such as the unit number, and the corresponding ecological characteristic parameters include environmental parameter statistics and species distribution density. Then, label each spatial unit with ecological characteristics according to the numerical range of the ecological characteristic parameters. For example, spatial units with an average humidity greater than 60% and a species distribution density greater than 0.3 individuals / m² are labeled as high humidity, high species density units, and spatial units with an average temperature between 25 and 30℃ and a species distribution density between 0.1 and 0.3 individuals / m² are labeled as suitable for survival - medium species density units. Further, overlay the above mapping table with the electronic map of the monitoring area, and label the corresponding ecological characteristic type of each spatial unit on the electronic map to form a visualized spatial unit ecological characteristic distribution map. Through the above integration operation, the divided spatial units with ecological characteristic annotations are obtained.

[0072] Step 308: Extract the ecological characteristic parameters of each spatial unit after division, and standardize the ecological characteristic parameters to eliminate dimensional differences, obtaining a standardized set of characteristic parameters. Specifically, this includes: First, determining the original value range of different parameters for all types of ecological characteristic parameters to be processed, such as temperature ranging from -10℃ to 40℃, humidity ranging from 0% to 100%, and species distribution density ranging from 0 to 1 individuals / square meter; Then, for each type of ecological characteristic parameter, a unified standardization method is used to eliminate dimensional differences. Specifically, for parameters with a value range, the difference between the actual value and the minimum value of the parameter is calculated, and then divided by the difference between the maximum value and the minimum value of the parameter to convert the parameter to the range of 0 to 1; for example, if the actual temperature of a spatial unit is 25℃, the minimum temperature is -10℃, and the maximum temperature is 40℃, then the standardized temperature value is 25℃ to -10℃ or 40℃ to -10℃. This method unifies parameters with different dimensions to the same numerical range; through the above processing, a standardized set of characteristic parameters is obtained.

[0073] Step 309: Based on the standardized feature parameter set, calculate its feature covariance matrix and decompose it to obtain the principal component eigenvectors of the feature covariance matrix. Specifically, this includes: First, determining the matrix dimension according to the parameter types in the standardized feature parameter set. For example, if the parameter types include temperature, humidity, and species distribution density, then a 3×3 feature covariance matrix is ​​constructed. Then, calculate the covariance value between any two types of parameters. The calculation of the covariance value is based on the parameter data of all spatial units. By statistically analyzing the degree of deviation of the two types of parameters in different spatial units, the linear correlation between the parameters is determined, and the calculated covariance values ​​are filled into the covariance matrix according to their corresponding positions. After completing the construction of the feature covariance matrix, perform eigenvalue decomposition on the matrix. By solving the matrix characteristic equation, several sets of eigenvalues ​​and corresponding eigenvectors are obtained. The magnitude of the eigenvalue reflects the degree of information contribution represented by the corresponding eigenvector, and the eigenvector reflects the combination relationship between different parameters. The principal component eigenvectors of the covariance matrix are obtained through the above operations. This process can extract the most representative core features from multi-dimensional ecological feature parameters, reducing data redundancy.

[0074] Step 310: Select the principal component feature vectors corresponding to the top k largest eigenvalues ​​as the principal component directions, and project the standardized feature parameter set onto the principal component directions to obtain the environmental feature vector of each spatial unit. Specifically, this includes: First, taking the principal component feature vectors, sorting all feature values ​​in descending order. During the sorting process, the corresponding principal component feature vectors must be associated synchronously to ensure that the correspondence between feature values ​​and vectors is not misaligned. Then, determine the number of principal components to be selected, k, based on the preset information retention threshold (e.g., 85%). The specific judgment method is as follows: calculate the cumulative information contribution rate of the first 1, the first 2, ..., the first n feature values ​​in sequence. The cumulative contribution rate = the sum of the first t feature values ​​ / the sum of all feature values. When the cumulative contribution rate reaches 85% or above for the first time, the corresponding t value is k. For example, if the cumulative contribution rate of the first 2 feature values ​​is 88%, then k = 2 is determined. This ensures that the selected k principal components can retain the core ecological information in the standardized feature parameter set to the greatest extent.

[0075] After determining the number of principal components k and the corresponding k principal component eigenvectors, a low-dimensional subspace is first constructed: using these k principal component eigenvectors as the basis vectors of the subspace, a k-dimensional subspace is spanned. This subspace is a low-dimensional subspace of the original feature space. The original feature space has the same dimension as the dimension of the standardized feature parameters. For example, if it contains m dimensions such as temperature, humidity, and species distribution density, then the original space is m-dimensional. As can be seen from the eigenvalue decomposition in step 309, this subspace can reflect the variance changes of the original feature data to the greatest extent, that is, the core ecological difference information.

[0076] Furthermore, for each spatial unit's standardized feature parameters (an m-dimensional vector, denoted as X, containing m-dimensional data such as standardized temperature, standardized humidity, and standardized species distribution density), projection calculations are performed with k basis vectors, i.e., principal component feature vectors, each of which is m-dimensional and denoted as V1, V2, ..., Vk. The specific calculation method is as follows: for the first basis vector V1, each dimension value of X is multiplied by the element of V1 of the same dimension one by one, and then all the multiplication results are summed to obtain the result. The projection value p1 of the spatial unit in the V1 direction; similarly, the projection values ​​p2 to pk of the spatial unit in each basis vector direction from V2 to Vk are calculated by multiplying and accumulating the corresponding elements; for example, if X is [0.6, 0.7, 0.5] (m=3) and V1 is [0.8, 0.1, 0.1], then p1=0.6×0.8+0.7×0.1+0.5×0.1. This calculation ensures that the projection value can accurately reflect the correlation between the spatial unit characteristics and the principal component directions.

[0077] After completing the vector projection of a single spatial unit, the k projection values ​​(p1, p2, ..., pk) of that unit are combined in the order of the basis vectors to obtain the coordinate vector of that unit in the k-dimensional subspace. This process is called subspace mapping, which maps the original m-dimensional standardized feature parameters to the k-dimensional subspace to form a low-dimensional coordinate vector. To verify the effectiveness of the mapping, the mapping results of all spatial units are checked as a whole: the cumulative variance of all coordinate vectors in the k-dimensional subspace is calculated and compared with the total variance of the original m-dimensional standardized feature parameter set to ensure that the cumulative variance ratio is not less than the preset 85% information retention threshold. If the ratio does not meet the standard, it means that the current k value has not fully retained the core information and the k value needs to be readjusted, such as increasing k from 2 to 3, and repeating the above steps of feature value sorting, principal component selection, subspace construction and vector projection until the mapping result meets the information retention requirements. Through the above subspace construction, vector projection and mapping verification operations, the standardized feature parameters of each spatial unit are finally converted into a k-dimensional coordinate vector, which is the environmental feature vector of each spatial unit.

[0078] Step 311: Based on the correlation analysis between the environmental feature vector and historical invasion record data, calculate the rule correction weight parameter for each spatial unit. Specifically, this includes: First, reviewing the historical invasion record data and extracting key information such as the invasion occurrence time, invasive species type, and invasion scale corresponding to the spatial unit, establishing the association between historical invasion records and spatial units. For example, determining if a spatial unit has experienced two invasive species invasions in the past three years and the corresponding environmental state at the time of the invasion. Then, perform correlation calculations between the environmental feature vector of each spatial unit and the historical invasion records of that unit. Specifically, analyze the correlation between each projection value in the environmental feature vector and the invasion frequency and scale. If the correlation between the environmental feature vector of a spatial unit and the historical invasion records is high, it indicates that the ecological characteristics of that unit have a more significant impact on the invasion risk, and a higher rule correction weight parameter is assigned to that unit. If the correlation is low, a lower rule correction weight parameter is assigned. Through the above correlation analysis and weight calculation, obtain the rule correction weight parameter specific to each spatial unit.

[0079] Step 312: Adjust the weight parameters according to the rules, and weight the environmental thresholds in the predefined basic association rules to generate environment-species association sub-rules. The basic association rules include unverified environment-species association thresholds, specifically: First, review the biological characteristics of invasive alien species and existing ecological research findings to identify key environmental factors affecting the survival and reproduction of target invasive species; Second, collect historical species invasion records and concurrent environmental monitoring data for the past 5 years in the monitoring area. The historical species invasion records include the specific time, spatial location, and corresponding environmental parameters of the invasion. Statistical analysis is performed on the historical data to calculate the value range of each environmental factor in each invasion event. For example, in 300 invasion events, the minimum temperature is 22℃, the maximum is 35℃, and the median is 25℃; the minimum humidity is 55%, the maximum is 80%, and the median is 60%. Based on the statistical results, combined with the species' biological characteristics, the suitable growth temperature range of 20-30℃ and suitable humidity range are determined. Based on research findings ranging from 50% to 70%, preliminary environmental thresholds for basic association rules were established, such as temperature greater than or equal to 25℃ (median value to balance sensitivity and accuracy), humidity greater than or equal to 60% (median value), soil moisture content greater than or equal to 20%, and light duration greater than or equal to 8 hours. These thresholds were then combined logically and relationally to form basic association rules. For example, a temperature greater than or equal to 25℃, humidity greater than or equal to 60%, and soil moisture content greater than or equal to 20% indicates that invasive species are likely to invade. Finally, the preliminary basic association rules were reviewed, and adjustments were made based on feedback regarding whether the thresholds covered key invasion scenarios and whether there were any redundant conditions. For instance, light duration had a more significant impact on invasion events in spring and autumn, so the condition of light duration greater than or equal to 9 hours was added in spring and autumn. The rules were then backtested using historical data from three typical sub-regions within the monitoring area, and the recognition rate of the rules for historical invasion events was required to be greater than or equal to 80%. Finally, the predefined basic association rules were determined and stored, providing a benchmark framework for the generation of subsequent sub-rules.

[0080] The predefined basic association rules are retrieved, such as a temperature greater than or equal to 25℃, humidity greater than or equal to 60%, and soil moisture content greater than or equal to 20%, indicating a combination of conditions conducive to the invasion of alien species. Subsequently, for each spatial unit, the environmental thresholds in the basic association rules are adjusted using a weighted average based on the corresponding rule correction weight parameter. The rule correction weight parameter reflects the ecological sensitivity of the spatial unit; a higher value indicates stronger sensitivity, ranging from 0 to 1. Specifically, the adjustment formula is set as: Sub-rule threshold = Basic threshold × (1 - Rule correction weight parameter × 0.1), meaning that for every 0.1 increase in the rule correction weight parameter, the sub-rule threshold decreases by 10% from the basic threshold. For example, if the rule correction weight parameter for a spatial unit is 0.8 (high sensitivity...), the sub-rule threshold is... If the sensitivity of a spatial unit is 0.2 (low sensitivity), then the temperature threshold is adjusted to 25×(1-0.8×0.1)=23℃, the humidity threshold is adjusted to 60%×(1-0.8×0.1)=55.2%, and the soil moisture content threshold is adjusted to 20%×(1-0.8×0.1)=18.4%, thereby reducing the thresholds to improve the timeliness of early warning for this sensitive unit. If the rule correction weight parameter of another spatial unit is 0.2 (low sensitivity), then the temperature threshold is adjusted to 25×(1-0.2×0.1)=24.5℃, the humidity threshold is adjusted to 60%×(1-0.2×0.1)=58.8%, and the soil moisture content threshold is adjusted to 20%×(1-0.2×0.1)=19.6%, thereby reducing false alarms in low-sensitivity areas by appropriately increasing the thresholds.

[0081] By first predefining basic association rules and then adjusting the threshold by modifying the weight parameters in conjunction with spatial unit rules, an environment-species association sub-rule adapted to its ecological characteristics is generated for each spatial unit.

[0082] Step 313: Based on the adjusted environment-species association sub-rules and combined with the predefined basic association rules between species and human activities, construct species invasion association rules. Specifically, this includes: First, determining the core content of the predefined basic association rules between species and human activities. These rules may include the association logic related to human activities, such as the proportion of tourists from high-risk source areas being no less than 20% and the proportion of tourists carrying plant-based items being no less than 15%.

[0083] Subsequently, the environment-species association sub-rules for different spatial units will be logically integrated with the aforementioned basic association rules between species and human activities. Specifically, for each spatial unit, the environment-species association sub-rules of that unit will be used as ecological suitability conditions, and the basic association rules between species and human activities will be used as propagation risk conditions. A complete logical chain will be constructed in which ecological suitability conditions and propagation risk conditions jointly point to the level of species invasion risk. For example, the final association rule for a certain spatial unit can be set as follows: the temperature is not lower than 24℃ and the humidity is not lower than 58% (ecological suitability condition), while the proportion of tourists from high-risk sources is not lower than 20% and the proportion of tourists carrying plant-based items is not lower than 15% (propagation risk condition), which indicates a high risk of species invasion. Through the above integration operation, a species invasion association rule covering two key dimensions, environment-species and species-human activities, will be constructed.

[0084] In this embodiment of the invention, spatial heterogeneity analysis of environmental parameters and species distribution data in the standardized fusion dataset can uncover hidden spatial differences in the data and accurately identify changes in the ecological gradient within the monitoring area. Based on these ecological gradient change characteristics, a differentiated spatial sampling strategy is derived, ensuring that the sampling plan aligns with the actual ecological differences in the monitoring area, guaranteeing that sampling focuses on key nodes of the ecological gradient, and improving the targeting and effectiveness of the sampling. Adjusting the distribution of sampling points based on the differentiated sampling strategy and the preset constraint on the number of reference points yields a spatial reference point coordinate set, which balances ecological gradient coverage with spatial distribution balance, avoiding data deviations caused by concentrated or omitted sampling points. Constructing a monitoring network based on the spatial reference point coordinate set ensures complete network coverage of the entire monitoring area, and the network layout relies on the prior spatial heterogeneity analysis and sampling strategy. This approach enables precise matching of the monitoring scope with key ecological areas, enhancing the practicality of the monitoring network. By dividing the monitoring network into adjacent areas and splitting the monitoring region into multiple spatial units centered on spatial reference points, a large-scale monitoring area can be refined into small-scale analysis units, laying the foundation for accurate extraction of the ecological characteristics of each unit. Based on standardized fusion datasets, the statistical values ​​of environmental parameters and ecological characteristic parameters such as species distribution density within each spatial unit are calculated, making the ecological state of each spatial unit concrete and data-driven, providing specific data basis for subsequent ecological feature annotation and rule correction. Integrating all spatial units and their corresponding ecological characteristic parameters yields spatial units with ecological feature annotations, clarifying the unique ecological attributes of each unit, allowing subsequent data processing and rule construction to focus on the unit dimension, improving the accuracy of the analysis.

[0085] Standardizing ecological characteristic parameters to eliminate dimensional differences results in a standardized parameter set. This avoids analytical biases caused by variations in parameter units and value ranges, ensuring that different types of ecological characteristic parameters can be directly used in subsequent calculations and comparisons. Calculating the characteristic covariance matrix based on the standardized parameter set and obtaining principal component eigenvectors through eigenvalue decomposition allows for the extraction of core features from multi-dimensional ecological parameters, reducing data redundancy and focusing on key ecological information regarding the impact of species invasion. Selecting the principal component eigenvectors corresponding to the k largest eigenvalues ​​as principal component directions and projecting the standardized parameter set yields the environmental characteristic vector for each spatial unit. This simplifies the feature dimensions while retaining key ecological information, improving the efficiency and accuracy of subsequent correlation analysis. Accuracy: Based on the correlation analysis between environmental feature vectors and historical invasion records, the rule correction weight parameters for each spatial unit are calculated, which enables the weight parameters to be closely related to the actual ecological status and invasion history of the unit, providing a data-driven basis for subsequent rule adjustments. The environmental thresholds of predefined basic association rules are weighted and adjusted according to the rule correction weight parameters to generate environment-species association sub-rules, which can make the association rules adapt to the ecological characteristics of each spatial unit. Based on the adjusted environment-species association sub-rules and the predefined species and human activity basic association rules, species invasion association rules are constructed, which can integrate the two key dimensions of environment-species and species-human activity, so that the rules cover multi-source data association relationships and improve the comprehensiveness and applicability of the rules.

[0086] In a preferred embodiment of the present invention, step 400 above, based on the species invasion association rule, performs rule matching analysis on the multi-source data acquired in real time to obtain a matching degree result; and calculates a species invasion risk score based on the matching degree result, including:

[0087] Step 401: Acquire real-time multi-source data. Through a data preprocessing process, convert the real-time multi-source data into a format consistent with the standardized fusion dataset to obtain a standardized real-time dataset. Specifically, this includes: first, determining the specific composition of the real-time multi-source data, ensuring consistency with the data source type of the aforementioned standardized fusion dataset. This includes species observation data transmitted in real-time by on-site inspection equipment, environmental monitoring data collected in real-time by the environmental sensor network, and human activity data generated in real-time by the entrance gate and visitor positioning system, ensuring data dimensionality completeness; further, initiating a preset data preprocessing process. This process first verifies the format of the real-time multi-source data, for example, unifying the species type field names in the species observation data to a description consistent with the standardized fusion dataset, and converting the temperature data type in the environmental monitoring data from string to numeric, ensuring complete matching of field names and data types with the standardized fusion dataset; then, supplementing missing fields in the real-time data. For example, if a piece of real-time human activity data lacks information about the area of ​​stay, it can be reasonably supplemented based on historical trajectory data from the visitor positioning system, avoiding interruptions in subsequent rule matching due to data format differences. Through the above preprocessing operations, a standardized real-time dataset is finally obtained. This dataset can form a unified format with the standardized fusion dataset on which the species invasion association rules are constructed, laying a data foundation for the subsequent accurate execution of rule matching.

[0088] Step 402: Input the standardized real-time dataset into the species invasion association rules, and perform rule-by-rule matching through the rule engine to obtain the matching degree result of each rule. Specifically, this includes: first, decomposing the standardized real-time dataset into species observation subsets, environmental monitoring subsets, and human activity subsets according to data type, so as to form corresponding matching relationships with different types of sub-rules in the species invasion association rules; then, starting the rule engine and loading the aforementioned constructed species invasion association rules. The rule engine needs to perform the matching operation one by one in the order of environment-species association sub-rules first, followed by species-human activity basic association rules: for environment-species association sub-rules, the rule engine will extract the standardized real-time dataset. The system compares environmental parameters (such as temperature and humidity) and species observation data (such as species numbers) with the environmental thresholds and species association conditions set in the sub-rules. For example, it determines whether the real-time temperature reaches the suitable invasion temperature threshold in the sub-rule and whether the real-time species number meets the potential invasion scale conditions, and gives a matching degree result between 0 and 1 based on the degree of compliance. For species-human activity basic association rules, the rule engine extracts information such as the proportion of tourists from high-risk source areas and the proportion of tourists carrying plant-based items from the real-time human activity data, compares it with the transmission risk conditions in the rules, and generates corresponding matching degree results. Through the system matching operation of the rule engine, the matching degree result corresponding to each association rule is finally obtained.

[0089] Step 403: Obtain the matching degree result, read the predefined rule weight coefficients in the species invasion association rules, multiply the matching degree result of each rule by its corresponding weight coefficient, sum them up, and calculate the preliminary risk assessment value. Specifically, this includes: First, sorting out all the sub-rule types contained in the species invasion association rules, and determining the specific entries of the environment-species association sub-rules and the species-human activity association rules. For example, the sub-classes of the environment-species association sub-rules are divided according to the ecological zoning of the monitoring area, where the ecological zoning of the monitoring area includes: core ecological zone, marginal transition zone, and densely populated human activity zone. The sub-classes of the species-human activity association rules are divided according to the intensity of human activity, where the intensity of human activity includes: high, medium, and low.

[0090] Secondly, historical invasion event data and trigger records of various rules in the monitoring area over the past five years were collected to establish a rule-invasion event association database. The actual contribution of different rules when invasion events occurred was statistically analyzed. Specifically, the frequency percentage of each rule triggered in historical invasion events and the accuracy rate of actual invasions after triggering were calculated. The frequency percentage of triggering is the number of times a rule was triggered / the total number of invasion events. The accuracy rate of actual invasions after triggering is the number of times an invasion actually occurred after the rule was triggered / the total number of times the rule was triggered. For example, the environment-species association sub-rule in the core ecological area had a trigger frequency percentage of 60% and a trigger accuracy rate of 85% in historical invasion events, while the species-human activity association rule in the densely populated human activity area had a trigger frequency percentage of 55% and a trigger accuracy rate of 80%.

[0091] Based on the above statistical results, the research conclusions on the impact weight of different rules on invasion risk are as follows: environmental adaptability is the basis of species colonization, and its weight should be higher than that of transmission risk in ecologically sensitive areas. The analytic hierarchy process (AHP) is used to initially set the weight coefficients: the trigger frequency ratio and trigger accuracy of the rule are weighted at a ratio of 4:6 to calculate the basic weight score, and then adjusted according to the characteristics of the ecological region. The environment-species association sub-rules in the core ecological area are increased by 20% on the basic weight score, the species-human activity association rules in the densely populated human activity area are increased by 15% on the basic weight score, and the weights of the two types of rules in the marginal transition area remain at the basic score to ensure that the sum of the weight coefficients is 1. For example, the weight of a certain environment-species sub-rule in the core ecological area is 0.6, and the corresponding species-human activity rule weight is 0.4.

[0092] Subsequently, the initially set weight coefficients are submitted for review. Adjustment suggestions are made based on dimensions such as the effectiveness of the rules under extreme climates and the differences in the impact of rules in different seasons. For example, if the environmental factors have a stronger impact during the rainy season, the weight of the corresponding environment-species rule needs to be increased by 5%. Then, backtesting is conducted using historical data from three typical high-incidence periods of invasion in the monitoring area to verify the accuracy of risk assessment when using the weight coefficients. The accuracy needs to reach more than 80%. If the standard is not met, the weight allocation ratio is readjusted until the verification requirements are met. Finally, the predefined weight coefficients of each rule are determined and stored to provide a benchmark for subsequent risk calculations.

[0093] After completing the rule-by-rule matching in step 402, the predefined weight coefficients of each rule are retrieved from the rule storage module. For example, the environment-species association sub-rule for ecologically sensitive areas is set with a higher weight coefficient because it has a high trigger frequency and high accuracy in historical invasion events and meets the strong dependence of ecologically sensitive areas on environmental adaptability. On the other hand, the species-human activity association rule for areas with less human activity impact is set with a relatively lower weight coefficient because its trigger frequency and accuracy are relatively low.

[0094] Subsequently, the matching degree result of each rule is associated with its corresponding weight coefficient one by one. For example, the matching degree result of the environment-species association sub-rule of a certain core ecological area is 0.8, and its predefined weight coefficient is 0.6, so the product value of the two is calculated to be 0.48; the matching degree result of the species-human activity association rule of the corresponding area is 0.7, and its predefined weight coefficient is 0.4, so the product value of the two is calculated to be 0.28.

[0095] Finally, all the product values ​​obtained from the correlation calculation are summed. For example, by adding the two product values ​​above, a preliminary risk assessment value of 0.76 is obtained.

[0096] Step 404: Normalize the preliminary risk assessment value to obtain a normalized numerical sequence; linearly map the normalized numerical sequence to a preset standard score range to obtain a species invasion risk score. Specifically, this includes: First, sorting out the historical preliminary risk assessment values ​​and corresponding actual invasion event severity data of the monitoring area over the past 5 years, establishing a preliminary risk assessment value-actual invasion severity correlation database, where the actual invasion severity is divided into 4 levels: no invasion, mild invasion, moderate invasion, and severe invasion. Mild invasion is the appearance of a single population, moderate invasion is the spread of the population to 3 or more spatial units, and severe invasion is the population covering 10% or more of the monitoring area. Statistically calculate the distribution range of the preliminary risk assessment values ​​corresponding to each level. For example, no invasion corresponds to a preliminary risk assessment value of 0 to 2, mild invasion corresponds to 2 to 5, moderate invasion corresponds to 5 to 8, and severe invasion corresponds to 8 to 10.

[0097] Secondly, considering the need for intuitive risk level identification in prevention and control work, such as the need for prevention and control personnel to quickly determine whether to activate an emergency response based on the score, and referring to common risk scoring practices in the industry, such as the 0 to 100-point system which is widely used due to its detailed and easy-to-understand interval division, the upper and lower limits of the standard score range are initially determined to be 0 to 100 points. At the same time, to ensure the correspondence between the score and the actual risk level, the preliminary risk assessment value range corresponding to each intrusion severity in historical data is mapped to the 0 to 100-point range. For example, no intrusion corresponds to 0 to 20 points, mild intrusion corresponds to 20 to 50 points, moderate intrusion corresponds to 50 to 80 points, and severe intrusion corresponds to 80 to 100 points, so that changes in the score can intuitively reflect the escalation of risk.

[0098] Based on this, optimization suggestions are proposed from the perspective of whether the score range covers extreme risk scenarios (such as whether a sudden large-scale intrusion can be covered by 100 points). This feedback helps determine whether the score division facilitates rapid decision-making, such as whether key decision nodes need to be set at 50 and 80 points. The range division is then fine-tuned based on the feedback, such as adjusting the score corresponding to moderate intrusion to 50 to 70 points and heavy intrusion to 70 to 100 points, thereby enhancing the differentiation of high-risk ranges.

[0099] Finally, the historical preliminary risk assessment values ​​are mapped to standard scores according to preset intervals, and the matching rate between different score intervals and the actual intrusion severity is calculated. The matching rate is required to be no less than 85%. If the standard is not met, the upper and lower limits of the intervals or the internal division ratios are readjusted until the verification requirements are met. Finally, the preset standard score intervals, from 0 to 100, are determined and stored to provide a benchmark for subsequent risk scoring mapping.

[0100] Based on the preliminary risk assessment value, a normalization process is first performed. This process requires first calling the range of preliminary risk assessment values ​​obtained from historical data statistics, such as a minimum value of 0 and a maximum value of 10. Then, the normalization formula is used to convert each preliminary risk assessment value to the range of 0 to 1. For example, if a preliminary risk assessment value is 6, then the normalized value is (6-0) / (10-0)=0.6. This eliminates the comparison obstacles caused by the difference in the range of values ​​of the preliminary risk assessment values ​​calculated from different batches and different spatial units.

[0101] After normalization, a normalized numerical sequence is obtained. This sequence is then linearly mapped to the preset standard score range of 0 to 100. The mapping process follows a linear correspondence rule: a normalized value of 0 corresponds to a standard score of 0, a normalized value of 1 corresponds to a standard score of 100, and any normalized value x corresponds to a standard score of x × 100. For example, a normalized value of 0.6 corresponds to a standard score of 60, ensuring that the score change is positively correlated with the risk level and the proportion is consistent. Through the above normalization and linear mapping operations, a species invasion risk score is finally obtained. This score is within the preset standard range of 0 to 100, which facilitates direct comparison with the preset warning threshold and provides a clear basis for prevention and control personnel to quickly understand the risk level (e.g., 60 points corresponds to moderate risk).

[0102] In this embodiment of the invention, real-time multi-source data is converted into a format consistent with the standardized fusion dataset through a data preprocessing process to obtain a standardized real-time dataset. This ensures that the real-time data and the dataset format on which the species invasion association rules are constructed are consistent. The standardized real-time dataset is input into the species invasion association rules, and the rule engine performs rule-by-rule matching to obtain the matching degree result of each rule. This systematically and standardizedly completes the comparison between real-time data and each association rule, ensuring that the matching process is complete and without omissions or arbitrariness, and making the matching degree results traceable and evidence-based. The predefined weight coefficients of each rule in the species invasion association rules are read, and the matching degree result of each rule is multiplied by its corresponding weight coefficient and then summed to calculate a preliminary risk assessment value, which can reflect the different degrees of impact of different association rules on the risk of species invasion. The preliminary risk assessment value is normalized to obtain a normalized numerical sequence, and then linearly mapped to a preset standard score range to obtain a species invasion risk score. This can eliminate the influence of different calculation dimensions on the assessment value, and at the same time, make the final score within an intuitive and easy-to-understand standard range, which is convenient for subsequent early warning judgment based on the score.

[0103] In a preferred embodiment of the present invention, step 500 above involves performing an early warning judgment based on the comparison result between the species invasion risk score and the preset early warning threshold, thereby obtaining an early warning judgment result; when the early warning judgment result indicates that an early warning needs to be triggered, early warning information containing the early warning location and key data is obtained, including:

[0104] Step 501: Compare the species invasion risk score with the preset warning threshold to obtain the comparison result; perform a warning judgment based on the comparison result to obtain the warning judgment result, specifically including: before carrying out the warning judgment operation, it is necessary to complete the setting of the preset warning threshold and divide the monitoring area into three categories: core ecological area, edge transition area, and densely populated human activity area. Among them, the core ecological area is such as native wetlands and habitats of rare species, the edge transition area is such as the buffer zone within 500 meters outside the core area, and the densely populated human activity area is such as tourist distribution centers and transportation hubs. Determine the spatial boundaries and ecological function positioning of each type of area. The core ecological area needs to be given key protection, the edge transition area undertakes the ecological buffer function, and the densely populated human activity area is more susceptible to human interference.

[0105] Secondly, historical data from the past five years were collected for three types of regions: First, ecological sensitivity data, including the species endemicity index and ecosystem vulnerability score within the region. The species endemicity index within the region is as follows: core ecological areas are usually greater than or equal to 0.8, peripheral transition areas are 0.5 to 0.8, and densely populated human activity areas are less than or equal to 0.5. The ecosystem vulnerability score ranges from 1 to 10, with the core area mostly scoring 8 to 10. Second, the frequency of historical invasions, which includes the number of invasion events occurring annually in each region (3 to 5 times per year in core ecological areas, 2 to 3 times per year in peripheral transition areas, and 1 to 2 times per year in densely populated human activity areas) and the difficulty of recovery after invasion. The recovery difficulty score in the core area is greater than or equal to 8, significantly higher than in other regions. Third, the applicability of species invasion association rules in each region, i.e., the accuracy rate of the rules in predicting invasion events in each region. Due to the stable environment, the accuracy rate of the rules reaches 90% in the core area, 85% in the peripheral area, and 75% in the densely populated human activity area.

[0106] Core ecological areas, due to their high ecological sensitivity, difficulty in restoring historical invasions, and high rule accuracy, require a lower threshold to achieve early warning. Based on the risk score distribution corresponding to their historical invasion events, which are mostly concentrated between 40 and 60 points, the initial threshold is set at 50 points. Edge transition areas have moderate ecological sensitivity, and based on their historical risk score distribution, which ranges from 50 to 70 points, the initial threshold is set at 60 points. Densely populated human activity areas, due to numerous interference factors and lower rule accuracy, require a higher threshold to reduce false alarms. Based on their historical risk score distribution, which ranges from 60 to 80 points, the initial threshold is set at 70 points.

[0107] Subsequently, an assessment was conducted on whether the threshold matched the regional ecological protection priority. For example, if the threshold in the core area was too low, leading to excessive early warning, feedback was then provided on whether the threshold could balance the cost and effectiveness of prevention and control. For example, if the threshold in densely populated areas was too high, leading to delayed early warning, the threshold was fine-tuned based on the feedback. For instance, the threshold in the core area was adjusted to 45 points to strengthen early warning, and the threshold in densely populated areas was adjusted to 75 points to reduce misjudgments.

[0108] Finally, the historical risk scores of each region are compared with the adjusted thresholds to calculate the early warning accuracy and false alarm rate. The early warning accuracy is the ratio of correct early warnings to actual intrusions, and the false alarm rate is the ratio of incorrect early warnings to total early warnings. The requirements are: core area accuracy greater than or equal to 90% and false alarm rate less than or equal to 15%; peripheral area accuracy greater than or equal to 85% and false alarm rate less than or equal to 20%; and densely populated area accuracy greater than or equal to 80% and false alarm rate less than or equal to 25%. If the standards are not met, the threshold values ​​are readjusted until the verification requirements are met. Finally, the preset early warning thresholds for the three types of areas are determined and stored to provide a benchmark for subsequent early warning judgments.

[0109] The monitoring area category of the spatial unit to be assessed is determined: core ecological zone, peripheral transition zone, or densely populated human activity zone. The corresponding preset warning threshold is retrieved from the threshold storage module, such as 45 points for the core ecological zone, 60 points for the peripheral transition zone, and 75 points for the densely populated human activity zone. Further, the species invasion risk score of the spatial unit is directly compared with the retrieved preset warning threshold to form a comparison result: if the risk score is greater than or equal to the preset warning threshold (e.g., a unit in the core ecological zone has a risk score of 50 points, which is greater than 45 points), the warning judgment result is determined to be that a warning needs to be triggered; if the risk score is less than the preset warning threshold (e.g., a unit in the densely populated human activity zone has a risk score of 70 points, which is less than 75 points), the warning judgment result is determined to be that a warning does not need to be triggered.

[0110] Step 502: When the warning judgment result indicates that a warning needs to be triggered, location information, environmental parameters, and species observation data are extracted from the standardized real-time dataset as key data. Specifically, this includes: when the warning judgment result indicates that a warning needs to be triggered, firstly, the time and spatial range of the standardized real-time dataset corresponding to the warning are locked. The time range must be limited to real-time data within the monitoring cycle prior to triggering the warning, and the spatial range must accurately correspond to the spatial unit where the warning is located, ensuring that the extracted data has spatiotemporal relevance and avoiding interference from irrelevant data; furthermore, three types of key data are extracted from the standardized real-time dataset within the locked range: the first is location information, specifically... The first part includes the boundary coordinates and core reference point coordinates of the early warning spatial unit, which must be consistent with the spatial coordinate system of the previously deployed monitoring network to ensure that prevention and control personnel can accurately locate the early warning area. The second part is environmental parameters, which focus on extracting key environmental factor data that match the species invasion association rules, such as real-time monitoring values ​​and statistical averages of temperature, humidity, and soil moisture content over the past 12 hours. These parameters must correspond one-to-one with the threshold parameter types of the environment-species association sub-rules in step 312. The third part is species observation data, which extracts information such as species type, observation quantity, and discovery time recorded in real time within the early warning area. If there are suspected invasive species, their morphological characteristics must be specially marked.

[0111] Step 503: Integrate the key data with the species invasion risk score to obtain structured early warning information containing early warning locations and key data. Specifically, this includes: First, designing a unified format for the structured early warning information. This format must balance information completeness and readability, and can be organized according to a logical hierarchy of early warning basic information, key data, and risk description. The early warning basic information section needs to determine the early warning location, early warning trigger time, and the corresponding monitoring period. The early warning location is the spatial unit coordinates and corresponding region name extracted in step 502, such as spatial unit A3 of the XX core ecological zone. The key data section needs to include the location data extracted in step 502... Information, environmental parameters, and species observation data are categorized and listed. Environmental parameters need to be labeled with the comparison results with the threshold of the association rule, such as a real-time temperature of 26℃, which is higher than the threshold of 24℃ for the association sub-rule. Species observation data need to be labeled to indicate whether they meet the characteristics of a suspected invasive species. The risk description section needs to determine the species invasion risk score and the corresponding risk level, such as a risk score of 85 points, which corresponds to a high-risk level. Furthermore, the above-categorized information is integrated into structured text according to a preset format. This integration process transforms scattered early warning-related data into standardized, hierarchical structured information, enabling prevention and control personnel to quickly grasp the core content.

[0112] In this embodiment of the invention, the species invasion risk score is compared with a preset warning threshold, and a warning judgment is performed. This provides a clear criterion for triggering the warning, ensuring that the warning judgment process is standardized and based on evidence, avoiding warning confusion caused by subjective judgment, and providing clear guidance for whether to initiate subsequent prevention and control actions. When a warning is triggered, location information, environmental parameters, and species observation data are extracted from a standardized real-time dataset as key data. This allows for precise screening of the core information required for prevention and control, avoiding redundant data interference, and enabling subsequent prevention and control personnel to quickly obtain key content related to the warning, thus improving information utilization efficiency. Integrating key data with the species invasion risk score forms structured warning information, organizing scattered warning-related data into a standardized and unified form. This facilitates prevention and control personnel's intuitive understanding of the warning location, risk level, and key supporting data, providing clear data support for subsequent special verification and other prevention and control work, and improving the orderliness of the prevention and control response.

[0113] like Figure 2 As shown, embodiments of the present invention also provide a species invasion detection system based on association relationships, comprising:

[0114] The data acquisition module is used to collect species observation data, environmental monitoring data, and human activity data to obtain an initial multi-source dataset. The initial multi-source dataset is then spatially processed and spatiotemporally aligned to obtain a spatiotemporally aligned multi-source dataset.

[0115] The fusion module is used to perform data fusion processing on spatiotemporally aligned multi-source datasets to obtain a standardized fusion dataset containing ecological feature elements;

[0116] The module is used to construct a monitoring network by deploying multiple spatial reference points within the monitoring area based on the standardized fusion dataset; to divide the monitoring network into spatial units to obtain each spatial unit; to obtain rule correction parameters based on the ecological characteristics of each spatial unit; and to construct species invasion association rules using the rule correction parameters.

[0117] The matching module is used to perform rule matching analysis on real-time acquired multi-source data based on the species invasion association rules to obtain matching degree results; and to calculate the species invasion risk score based on the matching degree results.

[0118] The early warning module is used to perform an early warning judgment based on the comparison result between the species invasion risk score and the preset early warning threshold, and obtain an early warning judgment result; when the early warning judgment result is that an early warning needs to be triggered, early warning information containing the early warning location and key data is obtained.

[0119] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0120] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A species invasion detection method based on association relationships, characterized in that, The method includes: Step 100: Collect species observation data, environmental monitoring data, and human activity data to obtain an initial multi-source dataset. Perform spatial processing and spatiotemporal alignment on the initial multi-source dataset to obtain a spatiotemporally aligned multi-source dataset. Step 200: Perform data fusion processing on the spatiotemporally aligned multi-source dataset to obtain a standardized fusion dataset containing ecological feature elements; Step 300: Based on the standardized fusion dataset, multiple spatial reference points are deployed within the monitoring area to construct a monitoring network; the monitoring network is divided into spatial units to obtain the divided spatial units. Based on the ecological characteristics of each spatial unit after division, rule correction parameters are obtained; using the rule correction parameters, species invasion association rules are constructed, including: Ecological characteristic parameters of each spatial unit after division are extracted, and the ecological characteristic parameters are standardized to eliminate dimensional differences, resulting in a standardized set of characteristic parameters. Based on the standardized feature parameter set, calculate its feature covariance matrix, and decompose to obtain the principal component eigenvectors of the feature covariance matrix. The principal component eigenvectors corresponding to the top k largest eigenvalues ​​are selected as the principal component directions. The standardized feature parameter set is projected onto the principal component directions to obtain the environmental feature vector of each spatial unit. Based on the correlation analysis between the environmental feature vector and historical intrusion record data, the rule correction weight parameter of each spatial unit is calculated. The rule correction weight parameter reflects the ecological sensitivity of the spatial unit and ranges from 0 to 1. The higher the value, the stronger the sensitivity, indicating that the ecological characteristics of the spatial unit have a more significant impact on the risk of intrusion. Based on the rule-corrected weight parameters, the environmental thresholds in the predefined basic association rules are adjusted by weighting to generate environment-species association sub-rules. The basic association rules contain unverified environment-species association thresholds. The adjustment formula is: sub-rule threshold = basic threshold × (1 - rule correction weight parameters × 0.1). Based on the adjusted environment-species association sub-rules, and combined with the predefined basic association rules between species and human activities, a species invasion association rule is constructed. Step 400: Based on the species invasion association rules, perform rule matching analysis on the real-time acquired multi-source data to obtain the matching degree results; calculate the species invasion risk score based on the matching degree results. Step 500: Based on the comparison result between the species invasion risk score and the preset warning threshold, a warning judgment is performed to obtain a warning judgment result; when the warning judgment result indicates that a warning needs to be triggered, warning information containing the warning location and key data is obtained.

2. The species invasion detection method based on association relationships according to claim 1, characterized in that, Step 100 includes: Species observation data were collected through on-site inspections and recordings, resulting in structured observation records containing information on species type, quantity, and geographical location. Environmental monitoring data is collected by deploying an environmental sensor network to obtain a sensor dataset containing coordinate locations and corresponding environmental parameters; Human activity data is collected through entrance gates and visitor registration data to obtain activity logs including the number of visitors, their place of origin, types of items they carry, and the areas they stay in the park. By integrating structured observation records, sensor datasets, and activity logs, an initial multi-source dataset is obtained, which includes species observation data, environmental monitoring data, and human activity data. For the species observation data in the initial multi-source dataset, by calling predefined coordinate transformation parameters, the geographical location information based on different geodetic coordinate systems is uniformly transformed to the target spatial coordinate system to obtain species observation data with unified spatial reference. Based on species observation data with unified spatial benchmarks and environmental monitoring data from initial multi-source datasets, a spatial field of environmental parameters covering the entire monitoring area is obtained. The spatial field of environmental parameters is spatiotemporally correlated and matched with human activity data in the initial multi-source dataset to establish a correspondence between environmental parameters, species observations and human activities. Based on the correspondence, a spatiotemporally aligned multi-source dataset is obtained.

3. The species invasion detection method based on association relationships according to claim 2, characterized in that, Step 200 includes: Data cleaning is performed on spatiotemporally aligned multi-source datasets. By filling in missing values ​​and smoothing outliers, the cleaned multi-source datasets are obtained. The cleaned multi-source dataset is standardized to obtain a numerically standardized multi-source dataset. The numerically standardized multi-source data set is subjected to feature fusion processing to obtain a fused feature set; By integrating the fused feature set with the structured observation records from the numerically standardized multi-source dataset, a standardized fused dataset containing ecological feature elements is obtained.

4. The species invasion detection method based on association relationships according to claim 3, characterized in that, Step 300 includes: Spatial heterogeneity analysis was performed on the environmental parameters and species distribution data in the standardized fusion dataset to identify the characteristics of ecological gradient changes within the monitoring area; Based on the characteristics of ecological gradient change, a differentiated spatial sampling strategy is derived; Based on the differentiated spatial sampling strategy, the distribution of sampling points is adjusted according to the preset spatial reference point quantity constraint to obtain a set of spatial reference point coordinates that meets the requirements of ecological gradient coverage and spatial distribution balance. Based on the spatial reference point coordinate set, a monitoring network covering the entire monitoring area is constructed; The monitoring network is divided into adjacent areas, and the monitoring area is divided into multiple spatial units with spatial reference points as the core. Based on the standardized fusion dataset, ecological characteristic parameters within each spatial unit are calculated, including environmental parameter statistics and species distribution density. By integrating all spatial units and their corresponding ecological characteristic parameters, we obtain the divided spatial units with ecological characteristic annotations.

5. The species invasion detection method based on association relationships according to claim 4, characterized in that, Step 400 includes: Acquire real-time multi-source data, and convert the real-time multi-source data into a format consistent with the standardized fusion dataset through a data preprocessing process to obtain a standardized real-time dataset; The standardized real-time dataset is input into the species invasion association rules, and the rule engine performs rule matching one by one to obtain the matching degree result of each rule; The matching degree result is obtained, the weight coefficients of each predefined rule in the species invasion association rule are read, the matching degree result of each rule is multiplied by its corresponding weight coefficient and then summed to calculate the preliminary risk assessment value. The preliminary risk assessment values ​​are normalized to obtain a normalized numerical sequence; the normalized numerical sequence is then linearly mapped to a preset standard score range to obtain a species invasion risk score.

6. The species invasion detection method based on association relationships according to claim 5, characterized in that, Step 500 includes: The species invasion risk score is compared with a preset warning threshold to obtain a comparison result; a warning judgment is performed based on the comparison result to obtain a warning judgment result. When the early warning judgment result indicates that an early warning needs to be triggered, location information, environmental parameters, and species observation data are extracted from the standardized real-time dataset as key data. By integrating the key data with the species invasion risk score, a structured early warning information containing early warning locations and key data is obtained.

7. A species invasion detection system based on association relationships, wherein the system implements the method as described in any one of claims 1 to 6, characterized in that, include: The data acquisition module is used to collect species observation data, environmental monitoring data, and human activity data to obtain an initial multi-source dataset. The initial multi-source dataset is then spatially processed and spatiotemporally aligned to obtain a spatiotemporally aligned multi-source dataset. The fusion module is used to perform data fusion processing on spatiotemporally aligned multi-source datasets to obtain a standardized fusion dataset containing ecological feature elements; The construction module is used to deploy multiple spatial reference points within the monitoring area to construct a monitoring network based on the standardized fusion dataset; The monitoring network is divided into spatial units to obtain the divided spatial units; rule correction parameters are obtained based on the ecological characteristics of each divided spatial unit; and species invasion association rules are constructed using the rule correction parameters. The matching module is used to perform rule matching analysis on the multi-source data acquired in real time based on the species invasion association rules, and obtain the matching degree result; Calculate the species invasion risk score based on the matching results; The early warning module is used to perform an early warning judgment based on the comparison result between the species invasion risk score and the preset early warning threshold, and obtain an early warning judgment result; when the early warning judgment result is that an early warning needs to be triggered, early warning information containing the early warning location and key data is obtained.

8. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 6.

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

Images