KR102627837B1 - Method and system for risk assessment of alien species using global-scale environmental data and economic loss database - Google Patents

Abstract

A technology related to a method and system for assessing the risk of alien organisms using global scale environmental data and an economic loss database is disclosed. The device for evaluating the risk of alien organisms includes a prediction unit for the probability of occurrence of alien organisms using an artificial intelligence-based prediction model and an alien organism risk assessment unit that calculates the risk of alien organisms in the evaluation area by considering economic loss data. The alien organism appearance probability prediction unit receives global-scale environmental spatial information, extracts first-stage independent variables that are environmental variables, creates second-stage independent variables derived from the extracted first-stage independent variables, and generates the generated second-stage independent variables. You can predict the probability of appearance of alien organisms by inputting it into an artificial intelligence-based prediction model. Using artificial intelligence-based prediction models and databases, evaluation time and costs can be reduced, and objective evaluation of risks from alien organisms is possible. Additionally, it is possible to compare assessment results between countries using global-scale environmental spatial information.

Description

{Method and system for risk assessment of alien species using global-scale environmental data and economic loss database}

The present invention relates to data processing technology using artificial intelligence. In particular, it predicts the probability of appearance of alien organisms from global scale environmental data using an artificial intelligence-based prediction model, and compiles the prediction results of the probability of appearance of alien organisms and an economic loss database. We disclose technology on methods and systems that can quantitatively estimate the risk of each alien organism in conjunction.

Alien organism risk assessment is an assessment of the possibility of establishment or disturbance of the ecosystem if alien organisms are introduced into the domestic natural environment. As many cases of social and ecological damage are reported internationally due to ecosystem disturbance caused by the introduction of alien organisms, the introduction of alien organisms is also managed domestically and their risks are assessed. The Ministry of Environment and the National Institute of Ecology under the Ministry of Environment regularly designate and update new alien organisms of risk by reviewing expert advice and overseas data for each taxon.

The risk assessment of organisms to be cautious of introduction into the country, conducted by the National Institute of Ecology, is based on data on the invasiveness, spread, damage cases, and effects on the surrounding ecosystem of the species being assessed. Therefore, it takes a lot of time and money to collect information suitable for the evaluation criteria and obtain evaluation results, and the subjective judgments of multiple judges are involved in the evaluation process, so the evaluation results may be inaccurate. Another drawback of this method is that although it determines the risk of a specific species, it does not evaluate the degree of risk in individual areas with spatially different environmental conditions.

Meanwhile, in foreign countries, software tools have been developed to evaluate risks to alien organisms using survey methods and statistical-based evaluation methods. Examples of survey methods include the Fish Invasiveness Screening Kit and the European Non-native Species in Aquaculture Risk Analysis Scheme. Examples of statistical-based evaluation methods include Classification & Regression Trees, Bayesian network, and Regression analysis.

In the case of the survey method, similar to the domestic evaluation system, various questions related to risk assessment are provided and additional uncertainty in the response is reflected in the evaluation, but subjective judgment due to the evaluator's prior experience and prior knowledge may be involved. There is a problem with inconsistent results between characters. In addition, in the case of statistical-based evaluation methods, good results can be obtained when evaluating one's own country by using environmental variables limited to a specific country, but it is difficult to use in countries with different environmental conditions.

Publication Patent No. 10-2023-0115128, published on August 2, 2023, relates to “invasive alien species risk management evaluation system and evaluation method using the same.” Here, an invasive alien species risk management evaluation system is disclosed to design survey items and apply an ecological model to evaluate the comprehensive risk that invasive alien species affect the ecosystem and society.

KR 10-2023-0115128 A (2023.08.02.)

An object of the present invention is to provide a method and system for evaluating the risk of alien organisms based on artificial intelligence that can reduce evaluation time and cost compared to the conventional survey method using experts, and exclude the subjective judgment of the evaluator. will be.

Another purpose of the present invention is to provide an artificial intelligence-based prediction model that can quantitatively predict the probability of the appearance of alien organisms from global-scale environmental information.

Another purpose of the present invention is to provide a method and system for evaluating the risk of alien organisms that allows comparison of evaluation results between countries using general-purpose environmental spatial information.

The problems to be solved by the present invention are not limited to the problems mentioned above, and other problems not mentioned can be clearly understood by those skilled in the art from the description below.

According to one aspect of the proposed invention, a device for predicting the probability of appearance of alien organisms includes an input unit, a metadata generation unit for each region, an artificial intelligence-based prediction model, and an output unit.

The input unit receives area information of the area in which the probability of appearance of alien organisms is to be predicted and environmental spatial information related to the area. The region-specific metadata generator extracts first-stage independent variable data, which is an environmental variable for each region, from the environmental spatial information and generates second-stage independent variables derived from the first-stage independent variable data. The artificial intelligence-based prediction model is trained to predict the probability of occurrence of alien organisms using global-scale environmental spatial information consisting of environmental information and alien organism appearance information, including two-stage independent variable data related to the risk of alien organisms. The output unit organizes the output probabilities of alien organisms predicted by the artificial intelligence-based prediction model by alien species by basin and outputs them to the outside.

According to an additional aspect, in the device for predicting the probability of appearance of alien organisms, the first stage independent variable data includes any one or more of geology, soil, climate, topography, and land use, and the second stage independent variable data is the Includes one or more of the average, total sum, distance-weighted average, minimum value, and maximum value of the first stage independent variable data.

According to another aspect of the proposed invention, the device for evaluating the risk of alien organisms includes a foreign organism appearance probability prediction unit and an alien organism risk assessment unit.

The alien organism appearance probability prediction unit receives regional information and environmental spatial information related to the region in which the probability of exotic organisms is to be predicted, extracts first-stage independent variable data, which are environmental variables for each region, from the input environmental spatial information, and , generate second-stage independent variable data derived from the extracted first-stage independent variable data, and input it into an artificial intelligence-based prediction model to predict the probability of appearance of alien organisms.

The alien organism risk assessment department calculates the risk of alien organisms in the area subject to evaluation by considering the economic loss and the probability of the alien organisms appearing.

According to an additional aspect, the device for evaluating the risk of alien organisms further includes an economic loss database that stores economic loss data caused by alien organisms, and a database linkage unit that searches data for each alien organism from the economic loss database.

According to another aspect of the proposed invention, a method for determining a model for predicting the probability of occurrence of alien organisms involves receiving global-scale environmental spatial information consisting of environmental information related to the risk of alien organisms and information on the appearance of alien organisms. A step of extracting first-stage independent variable data, which is an environmental variable, from the environmental spatial information, and determining second-stage independent variable data derived from the first-stage independent variable data, and determining the second-stage independent variable data and the appearance of alien organisms. Creating learning metadata containing information, dividing the learning metadata into a training dataset and a test dataset, selecting an artificial intelligence-based prediction model to learn from among the artificial intelligence-based prediction models that implement the artificial intelligence algorithm, Inputting the training data set into the selected artificial intelligence-based prediction model to train a target artificial intelligence-based prediction model to predict the probability of appearance of alien organisms as a dependent variable, and setting a test data set to the learned artificial intelligence-based prediction model It includes the step of predicting the probability of occurrence of alien organisms by inputting and verifying the learned artificial intelligence-based prediction model by comparing the predicted value with the actual value of the test dataset.

According to an additional aspect, in the method of determining a model for predicting the probability of appearance of alien organisms, the first stage independent variable data is classified into any one of geology, soil, climate, topography, and land use, and the second stage independent variable data is generated using any one of the average, total sum, distance-weighted average, minimum value, and maximum value of the first stage independent variable data.

According to another aspect of the proposed invention, the method for evaluating the risk of alien organisms includes the steps of receiving area information of the area in which the probability of appearance of the alien organism is to be predicted and environmental spatial information related to the area, and the input environment Extracting first-stage independent variable data, which is an environmental variable, from spatial information, generating second-stage independent variable data derived from the extracted first-stage independent variable data, and calculating the probability of appearance of alien organisms using global-scale environmental spatial information. Alien organism risk assessment, which involves predicting the probability of occurrence of alien organisms using an artificial intelligence-based prediction model trained to predict, and calculating the risk of alien organisms in the evaluation area by considering economic losses in the predicted probability of occurrence of alien organisms. Includes steps.

According to an additional aspect, the method for assessing the risk of alien organisms further includes the step of querying economic loss data caused by each alien organism from an economic loss database.

The method and system for evaluating the risk of alien organisms according to the proposed invention can be quantitatively evaluated using global-scale environmental data and economic loss database, which not only reduces evaluation time and cost, but also reduces the evaluator's subjective Objective evaluation is possible by excluding judgment.

Furthermore, the proposed invention enables comparison of evaluation results between countries using a data-based model using universal environmental spatial information with a spatial range of countries around the world.

Figure 1 is a diagram showing the configuration of a device for evaluating the risk of alien organisms using an artificial intelligence-based prediction model according to an embodiment.
Figure 2 is a configuration diagram showing the configuration of a device for determining an artificial intelligence-based prediction model used in a prediction model for the probability of occurrence of alien organisms according to an embodiment.
Figure 3 is a conceptual diagram showing a multi-layer perceptron neural network model among artificial intelligence-based prediction models used in a prediction model for the probability of occurrence of alien organisms according to an embodiment.
Figure 4 is a conceptual diagram showing learning metadata combining environmental variables and species appearance information for each watershed around the world to learn a model for predicting the probability of occurrence of alien organisms according to an embodiment.
Figure 5 is a conceptual diagram showing the probability of species appearance predicted by regional metadata and artificial intelligence-based prediction model that determines species appearance information for domestic watersheds using an alien species appearance probability prediction model according to an embodiment.
Figure 6 is a flowchart showing a method of determining a prediction model for the probability of occurrence of alien organisms using an artificial intelligence-based prediction model according to an embodiment.
Figure 7 is a flowchart showing a method for assessing the risk of alien organisms using an artificial intelligence-based prediction model according to an embodiment.

The foregoing and additional aspects are embodied through embodiments described with reference to the accompanying drawings. It is understood that the components of each embodiment can be combined in various ways within the embodiment or with components of other embodiments as long as there is no other mention or contradiction between them. Based on the principle that the inventor can appropriately define the concept of terms in order to explain his or her invention in the best way, the terms used in this specification and claims have meanings that correspond to the description or proposed technical idea. It must be interpreted as a concept. In this specification, a module or part is a set of program instructions stored in a memory that can be executed on a computer or processor, or can be implemented using a set of electronic components or circuits such as ASIC or FPGA to execute these instructions. Additionally, the operation of each module or part may be performed by one or more processors or devices. Components marked with the same/similar symbols perform the same/similar functions, so descriptions can be omitted. For components with reference symbols whose explanations have been omitted, the content previously described for components with the same or similar symbols can be referred to.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

Figure 1 is a diagram showing the configuration of a device for evaluating the risk of alien organisms using an artificial intelligence-based prediction model according to an embodiment.

According to one aspect of the proposed invention, a device or system for evaluating the risk of alien organisms includes an alien organism appearance probability prediction unit 110 and an alien organism risk assessment unit 180.

The alien organism appearance probability prediction unit 110 receives area information of the area in which the probability of appearance of alien organisms is to be predicted and environmental spatial information related to the area, and generates area-specific metadata from the input environmental spatial information to provide artificial intelligence. Predict the probability of occurrence of alien organisms by inputting them into the base prediction model. Predicted results can be output by region and exotic species.

Exotic organisms refer to organisms (fish, plants, mammals, amphibians, reptiles, invertebrates, etc.) introduced artificially or naturally from abroad. Global-scale environmental spatial information refers to environmental data that is commonly available in each country at home and abroad, rather than data that exists only in certain countries. Environmental spatial information includes environmental information related to the risk of alien organisms, which is an independent variable, and information on the appearance of alien organisms, which is a dependent variable.

The probability of appearance of alien organisms is determined by an artificial intelligence-based prediction model (140). The artificial intelligence-based prediction model 140 determines the probability of appearance of alien organisms, which are dependent variables, from independent variables extracted from environmental spatial information. Artificial intelligence-based prediction models can be implemented using various classification algorithms, including multi-layer perceptron (MLP) and convolution neural network (CNN).

The artificial intelligence-based prediction model 140 is trained to predict the probability of alien organisms appearing using global-scale environmental spatial information. Artificial intelligence-based prediction models are learned using metadata for learning. Metadata for learning can be generated including second-stage independent variable data derived from environmental spatial information and information on the appearance of alien organisms.

To generate second-stage independent variable data, environmental variables, which are first-stage independent variables, are first extracted from environmental spatial information. In addition, a second-stage independent variable can be created by applying various statistical indicators to the first-stage independent variable data. Information on the appearance of alien species can be generated as data on whether a specific species of alien species has appeared in the area where the second-stage independent variable data was created.

Step 1 independent variables can be constructed from data classified by type from environmental spatial information, for example, geology, soil, climate, topography, land use, etc. Stage 2 independent variables are independent variables derived from stage 1 independent variables for use in artificial intelligence-based prediction models.

The second-stage independent variable is a variable to be input into the artificial intelligence-based prediction model 140, and can be set by further classifying the first-stage independent variable, or derived by applying various statistical indicators to the first-stage independent variable. For example, for climate, which is a first-stage independent variable, more detailed precipitation, temperature, etc. can be set as second-stage independent variables. In addition, a second-stage independent variable can be additionally set by applying various statistical indicators such as average, total, distance-weighted average, minimum value, and maximum value to the independent variable.

The generated learning metadata can be divided into a training dataset and a test dataset. For example, you can split the training dataset and test dataset in a ratio of 80:20. The dividing ratio between the training dataset and the test dataset can be changed by the user's choice, such as 70:30. The training dataset is used to learn the artificial intelligence-based prediction model, and the learning dataset is used to verify the artificial intelligence-based prediction model.

The probability of appearance of alien organisms in each region can be predicted by inputting region-specific metadata into the artificial intelligence-based prediction model 140 learned using learning metadata. At this time, regional metadata must use the same two-stage independent variable as the learning metadata.

The alien organism risk assessment unit 180 calculates the risk of alien organisms in the evaluation area by considering the economic loss in the probability of alien organism appearance for each alien organism output from the alien organism appearance probability prediction unit 110. For example, the risk of alien organisms can be determined by multiplying the probability of occurrence of alien organisms by the economic loss. Specifically, the risk of alien species B in a specific watershed A can be calculated using Equation 1 below.

[Equation 1]

Species B risk = Species B occurrence probability × loss cost per area (USD/km ² ) × basin A area

Here, the probability of occurrence of species B is the probability of appearance of alien organism B determined by the alien organism appearance probability prediction unit 110, the loss cost per area is data stored in the database as the cost of loss upon appearance of alien species relative to the surface area of the watershed, and the watershed area is This is the surface area of basin A in the region.

According to an additional aspect, the device for evaluating the risk of alien organisms includes an economic loss database 170 that stores economic loss data caused by alien organisms, and a database linkage unit that searches data for each alien organism from the economic loss database. (not shown) is further included. For example, the economic loss database 170 may store data on economic loss costs per area for each exotic species.

According to an additional aspect, the device for evaluating the risk of exotic organisms may further include a regional risk comparison unit 190 that compares regional risks to exotic organisms. Since the device for assessing the risks of alien organisms uses global environmental spatial information data rather than environmental variables limited to specific countries, the risk comparison unit 190 by region can compare the evaluation results on the risks of alien organisms by region and country. there is.

According to an additional aspect, the alien organism appearance probability prediction unit 110 includes an input unit 120, a region-specific metadata generation unit 130, an artificial intelligence-based prediction model 140, and an output unit 150. do.

The input unit 120 receives area information of the area in which the probability of appearance of alien organisms is to be predicted and environmental spatial information related to the area.

The region-specific metadata generator 130 extracts first-stage independent variable data, which is an environmental variable for each region, from the environmental spatial information and generates second-stage independent variable data derived from the first-stage independent variable data. The region-specific metadata generator 130 extracts environmental variables for each region by matching environmental spatial information to the spatial information of each region. For example, the environmental spatial information with spatial information belonging to each region can be averaged to extract first-stage independent variable data for that region.

The region-specific metadata generator 130 generates region-specific metadata by combining second-stage independent variable data corresponding to environmental variables and region information including the surface area of the region. The generated regional metadata is transmitted to the artificial intelligence-based prediction model (140). The second-stage independent variable data consists of the same fields as the second-stage independent variable data used for learning the artificial intelligence-based prediction model 140.

The artificial intelligence-based prediction model 140 learns to predict the probability of the appearance of alien organisms using global-scale environmental spatial information consisting of environmental information and alien organism appearance information including two-stage independent variable data related to the risk of alien organisms. do. The metadata for each region is input into the artificial intelligence-based prediction model.

The output unit 150 externally outputs the output probability of alien organisms predicted by the artificial intelligence-based prediction model. The output unit 150 can organize the predicted output probabilities of alien organisms by alien species by region and output them to the outside.

According to a variation of the embodiment, an apparatus or system for assessing the risk of alien organisms includes one or more processors and a memory storing instructions for driving the processors. The alien organism appearance probability prediction unit 110 and the alien organism risk assessment unit 180 may be implemented by a single processor, or may be driven by separate processors. In other words, the alien organism appearance probability prediction unit can be independently used as a device for predicting the probability of alien organism appearance.

According to an additional aspect, in the device 110 for predicting the probability of appearance of alien organisms, the first stage independent variable data includes any one or more of geology, soil, climate, topography, and land use, and the second stage independent variable The data includes one or more of the average, total sum, distance-weighted average, minimum value, and maximum value of the first stage independent variable data.

Figure 2 is a configuration diagram showing the configuration of a device for determining an artificial intelligence-based prediction model used in a prediction model for the probability of occurrence of alien organisms according to an embodiment.

According to another aspect of the proposed invention, the alien organism appearance probability prediction model determination device 210, which determines the artificial intelligence-based prediction model used in the alien organism appearance probability prediction unit 110, includes an input unit 220, and a learning It includes a metadata generation unit 230, an artificial intelligence-based prediction model learning unit 240, an output unit 250, and a control unit 260.

The configuration of the input unit 220, the learning metadata generation unit 230, the artificial intelligence-based prediction model learning unit 240, and the output unit 250 is the input unit ( 120), a configuration corresponding to the regional metadata generation unit 130, the artificial intelligence-based prediction model 140, and the output unit 150. For omitted information, refer to the description of the corresponding configuration in FIG. there is.

The input unit 220 receives global-scale environmental spatial information consisting of environmental information related to the risk of alien organisms and information on the appearance of alien organisms.

The control unit 260 selects an artificial intelligence-based prediction model and selects second-stage independent variables to input into the selected artificial intelligence-based prediction model. The control unit 260 controls the learning metadata generator 230 to generate learning metadata.

The learning metadata generator 230 extracts first-stage independent variable data, which is an environmental variable, from environmental spatial information, and generates and determines second-stage independent variable data derived from the first-stage independent variable data. The learning metadata generator 230 generates learning metadata by combining watershed information including the surface area of the watershed, second-stage independent variable data of the watershed, and information on the appearance of alien organisms in the watershed. It is desirable to separate the learning metadata into a training dataset for learning the artificial intelligence-based prediction model and a test dataset for verifying the performance of the learned artificial intelligence-based prediction model. For example, you can split the training dataset and test dataset in a ratio of 80:20.

The control unit 260 transmits the learning metadata to the artificial intelligence-based prediction model learning unit 240 to train the selected artificial intelligence-based prediction model, and verifies whether the learned artificial intelligence-based prediction model accurately predicts the probability of alien organisms appearing. .

The artificial intelligence-based prediction model learning unit 240 inputs a training dataset into the artificial intelligence-based prediction model and trains it to predict the probability of occurrence of alien organisms. Input data from the test dataset into the learned artificial intelligence-based prediction model to obtain the probability of occurrence of alien organisms predicted by the artificial intelligence-based prediction model. The predicted probability of occurrence of alien organisms is compared with the actual value of information on occurrence of alien organisms, and the comparison result is output through the output unit 250.

If the output comparison result for the appearance of alien organisms satisfies a predetermined verification goal, the control unit 260 uses the learned artificial intelligence-based prediction model to use the learned artificial intelligence-based prediction model in the alien organism appearance probability prediction unit 110 (140). ) is decided. If the comparison result does not satisfy the predetermined verification goal, the artificial intelligence-based prediction model is changed by changing the parameters of the artificial intelligence-based prediction model, and then the above processes are repeated to satisfy the verification goal. can be decided. For example, an artificial intelligence-based prediction model can be composed of a deep neural network model using a multi-layer perceptron (MLP).

Figure 3 is a conceptual diagram showing a multi-layer perceptron neural network model among artificial intelligence-based prediction models used in a prediction model for the probability of occurrence of alien organisms according to an embodiment.

The control unit 260 can determine an artificial intelligence-based prediction model suitable for the verification goal by changing the number of hidden layers, input nodes, and the number of neurons in each hidden layer. In this case, an artificial intelligence-based prediction model that meets the verification goal can be determined by learning by changing the hidden layer and the number of neurons in each layer.

Referring to FIG. 3, the multilayer perceptron (MLP) neural network learned to be used in the prediction model for the probability of occurrence of alien organisms may be composed of 322 input nodes and 4 hidden layers. The number of nodes (neurons) in each layer is displayed at the bottom of each layer.

In addition to the multi-layer perceptron (MLP) neural network, artificial intelligence-based prediction models include classification algorithms such as convolutional neural network (CNN), deep belief network (DBN), self-organizing map (SOM), and ensemble models. can be used.

In order to determine an artificial intelligence-based prediction model that meets the verification goal, the control unit 260, in addition to changing the artificial intelligence-based prediction model parameters, changes the type and number of second-stage independent variables of the learning metadata input to the artificial intelligence-based prediction model. You can change it.

The second stage independent variables are the average of monthly minimum temperature in the basin, average of monthly maximum temperature, Sum of Monthly upstream precipitation, and distance-weighted average of monthly minimum temperature ( Distance weighted average of Monthly minimum temperature, Distance weighted average of Monthly maximum temperature, Distance weighted sum of Monthly upstream precipitation, etc. can be used. The control unit can repeat the process of changing and inputting these input variables to determine an artificial intelligence-based prediction model suitable for the verification goal.

Figure 4 is a conceptual diagram showing learning metadata combining environmental variables and species appearance information for each watershed around the world to learn a model for predicting the probability of occurrence of alien organisms according to an embodiment.

The learning metadata input to the artificial intelligence-based prediction model learning unit 240 of the alien organism appearance probability prediction model determination unit 210 includes watershed information, environmental variables, and species appearance information. Watershed information may include the name (N) of the watershed and the area (S) of the watershed. Information on the area (S) of the watershed can be used when the alien species occurrence probability prediction unit 110 calculates the risk of alien species using Equation 1.

Environmental variables can be used as the basin's monthly minimum temperature average, monthly maximum temperature average, monthly precipitation sum, monthly minimum temperature distance-weighted average, monthly maximum temperature distance-weighted average, and monthly precipitation distance-weighted sum. Referring to Figure 4, the monthly minimum temperature average for January is displayed in column C1_1, and the monthly maximum temperature average for January is displayed in column C2_1. Temperature is displayed in degrees Fahrenheit (℉). The monthly minimum temperature average and monthly maximum temperature average from February to December are omitted. In addition, data on environmental variables such as monthly precipitation total, monthly minimum temperature distance-weighted average, and monthly maximum temperature distance-weighted average can be determined as data for the second-level independent variables and combined with metadata as shown in FIG. 4.

Species occurrence information indicates whether the species appears in the watershed. Referring to Figure 4, E1, E2, etc. represent respective species. For example, E1 may represent the gray herring (Alsoa pseudoharengus) species, and E2 may represent the American herring (Alsoa sapidissima) species. Referring to Figure 4, if the value is 1, it indicates that the species appeared in the corresponding watershed. That is, species E1 appeared in the Alabama and Delaware basins, and species E2 appeared in the Alsek, Cedar, and Chehalis basins.

When learning an artificial intelligence-based prediction model using the learning metadata of FIG. 4, the artificial intelligence-based prediction model can be supervised to output the probability of occurrence of a specific species in a specific watershed using species appearance information as a label.

Figure 5 is a conceptual diagram showing the probability of species appearance predicted by regional metadata and artificial intelligence-based prediction model that determines species appearance information for domestic watersheds using an alien species appearance probability prediction model according to an embodiment.

Figure 5(a) shows regional metadata input to the artificial intelligence-based prediction model 140 of the alien species appearance probability prediction unit 110, and Figure 5(b) shows the alien species appearance probability prediction unit 110. ) This is data showing the probability of species appearance by region output from ).

Referring to (a) of FIG. 5, the second-stage independent variable generator 130 includes environmental variables including the second-stage independent variables, watershed information including the surface area of the watershed, and information on whether species appear. Metadata containing the data is generated and delivered to an artificial intelligence-based prediction model. At this time, watershed information and environmental variables have the same structure as the metadata used for learning the artificial intelligence-based prediction model. Because species occurrence information is determined by an artificial intelligence-based prediction model, it is not transmitted to the artificial intelligence-based prediction model.

Referring to (b) of FIG. 5, the results predicted by the artificial intelligence-based prediction model 140 can be expressed as the probability of occurrence of each species for each watershed. For example, the probability of occurrence of species E1 in the Han River-Seoul area is 1.14E-34, and the probability of occurrence of species E2 is 6.91E-14. And, the probability of appearance of species E1 in the Chungju Dam area is 3.72E-22, and the probability of appearance of species E2 is 7.47E-08. In Figure 5 (b), it can be seen that the probability of E2 species (American herring) appearing in Chungju Dam is the highest at 7.47E-08.

Below, the methods performed by the devices described above will be described. For parts omitted to avoid duplication, refer to the description of the previous device.

Figure 6 is a flowchart showing a method of determining a prediction model for the probability of occurrence of alien organisms using an artificial intelligence-based prediction model according to an embodiment.

According to another aspect of the proposed invention, the method of determining an artificial intelligence-based prediction model for predicting the probability of occurrence of alien organisms performed by the alien organism appearance probability prediction model determination unit 210 is performed as follows.

First, global-scale environmental spatial information consisting of environmental information related to the risk of alien organisms and information on the appearance of alien organisms is input through the input unit 220 (S620).

Next, the learning metadata generator 230 extracts a first-level independent variable, which is an environmental variable, from the input environmental spatial information, and determines a second-level independent variable derived from the first-level independent variable. Then, learning metadata including second-stage independent variable data and information on the appearance of alien organisms is generated (S630). Metadata for learning can be divided into training dataset and test dataset.

Then, the control unit 260 selects an artificial intelligence-based prediction model to be learned from among the artificial intelligence-based prediction models that implement the artificial intelligence algorithm, and the artificial intelligence-based prediction model learning unit 240 selects the selected artificial intelligence-based prediction model using learning metadata. Train an artificial intelligence-based prediction model (S640).

The artificial intelligence-based prediction model learning unit 240 inputs the training dataset into the selected artificial intelligence-based prediction model and trains the target artificial intelligence-based prediction model to predict the probability of appearance of an alien organism, which is a dependent variable. Then, the test dataset is entered into the learned artificial intelligence-based prediction model to obtain a predicted probability of the appearance of alien organisms (S650).

The control unit 260 verifies the performance of the learned artificial intelligence-based prediction model by comparing the predicted value of the probability of occurrence of alien organisms obtained from the learned artificial intelligence-based prediction model with the actual value of the test dataset (S655). The performance of classification-based artificial intelligence-based prediction models can be verified using indicators such as accuracy, precision, recall, and fall-out. As a result of verification, if the performance of the learned artificial intelligence-based prediction model does not achieve the verification goal, return to the artificial intelligence-based prediction model selection step (S640), select a new artificial intelligence-based prediction model, and select the selected artificial intelligence-based prediction model. Repeat the learning process.

As a result of the verification, if the performance of the learned artificial intelligence-based prediction model satisfies the verification goal, the learned artificial intelligence-based prediction model is determined as the artificial intelligence-based prediction model to be used in the probability prediction model for the appearance of alien organisms (S660). The determined artificial intelligence-based prediction model can be used to predict the probability of appearance of alien organisms in a device that evaluates the risk of alien organisms.

Figure 7 is a flowchart showing a method for assessing the risk of alien organisms using an artificial intelligence-based prediction model according to an embodiment.

According to another aspect of the proposed invention, in a device for assessing the risk of alien organisms, a method of assessing the risk of alien organisms using an artificial intelligence-based prediction model that predicts the probability of appearance of alien organisms is as follows: It can be done.

First, the input unit 120 receives area information of the area in which the probability of appearance of alien organisms is to be predicted and environmental spatial information related to the area (S720).

Then, the regional metadata generator 130 extracts first-stage independent variable data, which is an environmental variable, from the input environmental spatial information, and generates second-stage independent variable data derived from the extracted first-stage independent variable data. At this time, in order to extract environmental variables for each region, environmental spatial information is matched to the spatial information for each region to extract environmental variables for each region. For example, the first-stage independent variable data for each region can be extracted by averaging the environmental spatial information with the spatial information belonging to each region. Then, the generated second-stage independent variable data is combined with region information to determine metadata for each region (S730). Regional metadata is transmitted to an artificial intelligence-based prediction model (140).

The artificial intelligence-based prediction model 140, which has been trained to predict the probability of appearance of alien organisms using global-scale environmental spatial information, receives the metadata for each region and predicts the probability of appearance of alien organisms for each region (S750).

The alien organism risk assessment unit 180 calculates and evaluates the risk of alien organisms in the evaluation area by considering economic losses and the predicted probability of alien organisms appearing (S780).

According to an additional aspect, the method of assessing the risk of alien organisms may further include a step (S770) of linking the economic loss database 170. Economic loss data caused by each alien organism can be searched from the economic loss database and reflected in the risk of alien organisms in the evaluation area.

According to an additional aspect, the method of assessing the risk of alien organisms may further include a step (S790) of comparing the risks of alien organisms by region. The artificial intelligence-based prediction model of the device that predicts the probability of appearance of alien organisms calculates the probability of appearance of alien organisms using global environmental spatial information data, so the regional risk comparison unit (190) The risks can be compared.

In the above, the present invention has been described through embodiments with reference to the attached drawings, but it is not limited thereto, and should be interpreted to encompass various modifications that can be easily derived by those skilled in the art. The claims are intended to cover these variations.

110: Foreign organism appearance probability prediction unit
120: input unit
130: Regional metadata generation unit
140: Artificial intelligence-based prediction model
150: output unit
170: Economic loss database
180: Alien organism risk assessment department
190: Regional risk comparison department

Claims (8)

An input unit that receives area information of the area in which the probability of appearance of alien organisms is to be predicted and environmental spatial information related to the area;
a region-specific metadata generator that extracts first-stage independent variable data, which is an environmental variable for each region, from the environmental spatial information, and generates second-stage independent variable data derived from the first-stage independent variable data;
An artificial intelligence-based prediction model that is learned to predict the probability of the appearance of alien organisms using global-scale environmental spatial information consisting of environmental information and alien organism appearance information, including two-stage independent variable data related to the risk of alien organisms;
An output unit that organizes the output probabilities of alien organisms predicted in the artificial intelligence-based prediction model by alien species by region and outputs them to the outside; and
An alien species risk assessment unit that calculates the risk of alien species in each region of the evaluation area based on the probability of occurrence of alien species in each region and economic loss predicted in the artificial intelligence-based prediction model;
Including,
The artificial intelligence-based prediction model is,
Using global-scale environmental spatial information consisting of environmental information for each region including second-stage independent variable data related to the risk of alien organisms and information on the appearance of alien species in each region, the environmental information of the input region is It includes a deep neural network that is learned to predict the probability that biological species will appear in each region,
The foreign organism risk assessment department,
Using the probability of species B appearing in each area predicted by the artificial intelligence-based prediction model, the risk of alien species B in watershed A belonging to each area is calculated using the following equation 1,
A device that predicts the probability of appearance of alien organisms and assesses their risk.
[Equation 1]
Species B risk = Species B occurrence probability × loss cost per area (USD/km ² ) × basin A area According to claim 1,
The first stage independent variable data includes one or more of geology, soil, climate, topography, and land use,
The second stage independent variable data includes one or more of the average, total sum, distance weighted average, minimum value, and maximum value of the first stage independent variable data,
A device that predicts the probability of appearance of alien organisms and assesses their risk. delete According to claim 1,
An economic loss database that stores economic loss data caused by invasive species; and
Further comprising a database linkage unit that searches data for each exotic organism from the economic loss database,
A device that predicts the probability of appearance of alien organisms and assesses their risk. A step of receiving regional information of the area in which the probability of appearance of alien organisms is to be predicted and environmental spatial information related to the area;
extracting first-stage independent variable data, which are regional environmental variables, from the input environmental spatial information, and generating second-stage independent variable data derived from the extracted first-stage independent variable data;
Using an artificial intelligence-based prediction model learned to predict the probability of the appearance of alien organisms using global-scale environmental spatial information consisting of environmental information and alien organism appearance information, including two-stage independent variable data related to the risk of alien organisms, Predicting the probability of appearance of a creature;
An alien species risk assessment step in which the risk of alien species in each area of the evaluation area is calculated based on the probability of occurrence of alien species in each area and the economic loss predicted in the artificial intelligence-based prediction model; and
Using global-scale environmental spatial information consisting of environmental information for each region including second-stage independent variable data related to the risk of alien organisms and information on the appearance of alien species in each region, the environmental information of the input region is Including a step of determining the deep neural network trained to predict the probability of biological species appearing in each region as an artificial intelligence-based prediction model,
The alien organism risk assessment step is,
Comprising: calculating the risk of alien species B in watershed A belonging to each region using the probability of species B appearing in each region predicted by an artificial intelligence-based prediction model using Equation 1 below,
A method of predicting the probability of appearance of alien organisms and assessing risk.
[Equation 1]
Species B risk = Species B occurrence probability × loss cost per area (USD/km ² ) × basin A area According to claim 5,
The first stage independent variable data is classified into one of geology, soil, climate, topography, and land use,
The second stage independent variable data is generated using any one of the average, total sum, distance weighted average, minimum value, and maximum value of the first stage independent variable data,
How to determine a model that predicts the probability of occurrence of exotic organisms. delete According to claim 5,
Further comprising: retrieving economic loss data caused by each alien organism from the economic loss database,
A method of predicting the probability of appearance of alien organisms and assessing risk.

Images