Forest pest prediction method and system and readable storage medium

Abstract

The invention provides a forest insect pest prediction method and system and a readable storage medium, and relates to the technical field of forest insect pest prediction.The forest is divided into a plurality of grids, main forest tree types, forest tree characteristics, soil characteristics and meteorological data of each grid are collected, and scientific assessment and management of insect pest risks are achieved. The method mainly comprises the following steps: firstly, carrying out characteristic parameter acquisition on grids, and carrying out similar grid group division based on soil characteristics; secondly, constructing an insect pest prediction model through a random forest algorithm, training by utilizing insect pest data of a sample grid, and predicting insect pest severity levels of other grids; and finally, analyzing the pest severity level change of each grid at the current and historical moments, generating an alarm index, and sending out a corresponding alarm signal according to a preset threshold value. According to the method, the environmental suitability index and the forest vulnerability index are combined, the insect attack grade is determined by adopting the expert scoring method, and the insect attack risk can be effectively monitored and early warned, so that a scientific basis is provided for forest management.

Description

Technical Field

[0001] The present invention relates to the technical field of forest pest prediction, and specifically provides a forest pest prediction method, system and readable storage medium. Background Art

[0002] Forest pest prediction is an important task in forest resource management and protection. With the intensification of global climate change and the impact of human activities, forest ecosystems are facing increasingly serious pest threats. Pests can not only lead to the loss of forest biodiversity, but also affect the ecological balance and the sustainable development of the forest economy. Therefore, timely and accurately predicting the risk of pest occurrence is crucial for formulating effective control measures.

[0003] Traditional pest prediction methods mostly rely on a single data source, such as meteorological data or historical pest records, and often have difficulty in comprehensively capturing the complexity and diversity of pest occurrence. This has led to insufficient accuracy and timeliness of pest early warnings, posing challenges to forest management and resource protection. Prediction models based on historical data often lack sufficient adaptability when facing newly emerging pest species and cannot provide effective early warnings in a timely manner. Therefore, there is an urgent need for a new method to integrate multiple data sources and effectively model pests and their ecological dynamics to improve the accuracy and reliability of pest prediction.

[0004] The deficiencies of the prior art are as follows: There are some significant deficiencies in the prior art in forest pest prediction, mainly reflected in the limitations of data collection and the accuracy of prediction models. Many traditional methods rely on limited monitoring points or empirical data, resulting in insufficient data coverage and inability to comprehensively reflect the actual situation of forest ecosystems. This limitation makes the pest prediction results often inaccurate, prone to misjudgment, and affects forest management decisions. In addition, in the assessment of pest severity, the prior art usually uses a single factor or a simplified model for analysis, and fails to fully consider the combined effects of multiple meteorological conditions and tree characteristics on pest occurrence. The lack of a comprehensive assessment of different environmental suitability makes the flexibility and adaptability of the prediction model insufficient and difficult to cope with the uncertainties brought about by climate change. In this case, traditional methods are difficult to timely warn of pest risks, resulting in managers being unable to take effective prevention and control measures and increasing the risk of losses caused by pests.

[0005] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and therefore it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present invention is to provide a forest pest prediction method, system and readable storage medium to solve the problems raised in the above background art.

[0007] To achieve the above object, the present invention provides the following technical solutions: A method for predicting forest pests, the specific steps including: Step 1: Divide the forest into N grids of equal area, and collect the characteristic parameters of each grid. The characteristic parameters include the main tree species, tree characteristic data, soil property data, and meteorological data; Step 2: Divide all grids into multiple similar grid groups according to the soil property data of each grid, randomly select several sample grids from the similar grid groups, and obtain the forest pest data of the sample grids. The forest pest data includes the proportion of pest-infected trees, the spreading speed of pest-infected area, and the growth speed of pest density; Step 3: Determine the environmental suitability index of each grid by combining the main tree species and meteorological data of the sample grids, analyze the tree characteristic data to obtain the tree vulnerability index of the sample grids, and analyze the tree vulnerability index and forest pest data of the sample grids based on the expert scoring method to determine the pest severity level of each grid; Step 4: Construct a pest prediction model corresponding to each similar grid group based on the random forest algorithm. Use the main tree species, tree characteristic data, and environmental suitability index of the sample grids in the same similar grid group as inputs, and the corresponding pest severity level as the output to train the pest prediction model, and obtain the pest severity level of other grids except the sample grids in the same similar grid group through the trained pest prediction model; Step 5: Analyze the change in the distribution of pest severity levels at the current and historical moments of the N grids, obtain the change amount of each pest severity level, so as to determine the pest situation of the forest and send out corresponding alarm signals.

[0008] Further, it specifically includes: Divide the land area of the forest into N grids of equal area, collect the types of all trees in each grid, take the tree species with the largest number in each grid as the main tree species of the corresponding grid, and collect the tree characteristic data of the trees corresponding to the main tree species in each grid. Take the average value of the tree characteristic data of the trees as the tree characteristic data of the corresponding grid. The tree characteristic data includes tree species, tree density, and tree moisture content; Set the center of each grid as the sampling point, and collect the soil property data of the sampling point of each grid. The soil property data includes the pH value of the soil, soil humidity, and organic matter content; Obtain the meteorological data of the location where the forest is located. The meteorological data of each grid is the meteorological data of the location where the forest is located. The meteorological data includes temperature, humidity, and rainfall.

[0009] Further, obtaining the forest pest data of the sample grids specifically includes: Take each grid as a node, and take the soil property data corresponding to each node as the attributes of the node to construct an undirected graph. If two nodes are adjacent, add an edge in the undirected graph to form an undirected graph network; Place each node in an independent similar grid group. For each node in the undirected graph network, move it successively to the similar grid groups where all its neighbor nodes are located, and calculate the difference in property changes that occur each time. Specifically, it includes: ; where is the difference in property changes between the i-th node and the j-th node, is the average pH value of the i-th node and the j-th node, is the pH value of the i-th node and the j-th node, is the average soil moisture of the i-th node and the j-th node, is the soil moisture of the i-th node, is the average organic matter content of the i-th node and the j-th node, is the organic matter content of the i-th node and the j-th node. Both i and j are nodes in the undirected graph, and i and j are neighbor nodes; Move each node to the neighbor similar grid group with the smallest difference in property changes, and repeat moving the nodes until the belonging of the similar grid groups of all nodes no longer changes. At this time, randomly determine several nodes in each similar grid group as the sample grids of this similar grid group; The calculation formula for the proportion of pest-infected trees is: ; where Q is the proportion of pest-infected trees in the sample grid, is the number of pest-infected trees in the sample grid, is the total number of trees in the sample grid; Randomly select several trees in the sample grid as sample trees, and obtain the proportion of pest-infected area of the sample trees. The calculation formula is: ; where is the proportion of pest-infected area of the x-th sample tree in the sample grid, is the pest-infected area of the x-th sample tree in the sample grid, is the total area of the x-th sample tree in the sample grid. x is the index of the sample tree, and take the average of the proportions of pest-infected areas of all sample trees as the proportion of pest-infected area of this sample grid; The calculation formula for the spreading speed of the pest-infected area is: ; Among them, \(P_v(t)\) is the spreading speed of the pest area in the sample grid at the current moment, is the proportion of the pest area in the sample grid at the current moment, is the proportion of the pest area in the sample grid at the previous moment, represents the time interval between the current moment and the previous moment; Randomly select multiple sampling areas within the pest area of the trees in the sample grid, collect the image data of the sampling areas, obtain the number of pests in each sampling area based on the image processing software, and the formula for calculating the pest density is: ; Among them, is the pest density of the sampling area, \(P_n\) is the number of pests in the sampling area image, and \(a\) is the area of the sampling area; Calculate the pest density of all sampling areas, calculate the average value as the pest density of the corresponding sample grid, and the formula for calculating the growth rate of the pest density is: ; Among them, is the growth rate of the pest density in the sample grid at the current moment, is the pest density in the sample grid at the current moment, is the pest density in the sample grid at the previous moment.

[0010] Furthermore, calculating the environmental suitability index of each grid specifically includes: The formula for calculating the environmental suitability index is: ; Among them, is the environmental suitability index, are the suitability functions of temperature, humidity, and rainfall respectively, are the weight coefficients of their respective corresponding items, , and ; The formula for calculating the temperature suitability function is: ; Among them, represents the temperature, with the unit of °C, is a positive value used to control the opening size of the function, represents the optimal suitability value of the temperature, is the optimal pest control temperature of the trees belonging to the main tree species corresponding to this grid; The formula for calculating the humidity adaptability function is: ; Among them, represents the humidity, is a positive value used to control the opening size of the function, represents the optimal suitable value of humidity, is the optimal humidity for pest control of the trees belonging to the main tree species corresponding to this grid; The calculation formula of the rainfall adaptability function is: ; where R represents rainfall, with the unit of mm, is a positive value used to control the opening size of the function, represents the optimal suitable value of rainfall, is the optimal rainfall for pest control of the trees belonging to the main tree species corresponding to this grid.

[0011] Furthermore, calculating the vulnerability index of the trees in each grid specifically includes: The calculation formula of the vulnerability index of the trees in the sample grid is: ; where, is the vulnerability index of the trees in the sample grid, is the tree density of the trees belonging to the main tree species corresponding to the sample grid, is the water content of the trees belonging to the main tree species corresponding to the sample grid; Based on the expert scoring method, analyze the environmental suitability index and the vulnerability index of the trees in each grid to determine the pest severity level of each grid. The pest severity level includes high pest level, medium pest level, and low pest level.

[0012] Furthermore, constructing the pest prediction model specifically includes: Integrate the main tree species, tree feature data, environmental suitability index, and pest severity level of several sample grids in the similar grid group into a sample set, and divide the sample set into a training set and a test set with a division ratio of 7:3. Use the main tree species, tree feature data, and environmental suitability index in the training set as inputs, and the corresponding pest severity index as the label to train the random forest model to obtain the pest prediction model. Set the mean squared error as the loss function, monitor the value of the loss function, and when the value of the loss function drops to 0.001 within 10 iteration cycles, it is considered that the model has been trained. Then use the test set to evaluate the model, and the evaluation indicators are accuracy, precision, and recall. If the accuracy, precision, and recall are above 70%, it is considered that the model performance is good. If the values of the evaluation indicators are below 70%, then adjust the model parameters; Then, input the main tree species, tree feature data, and environmental suitability index of other grids except the sample grid in the same similar grid group into the pest prediction model after training to obtain the pest severity level output by the model.

[0013] Further, determining the pest situation of the forest specifically includes: Obtain the proportion of grids with a high pest severity level at the current moment. The calculation formula is: ; Among them, is the proportion of grids with a high pest severity level at the current moment, is the indicator function, which takes the value of 1 when is high, otherwise takes the value of 0, is the classification value of the pest severity level of high, medium, and low for the nth grid at the current moment. n is the index of the grid, N is the total number of grids, and ; Similarly, obtain the proportion of grids with a medium pest severity level and the proportion of grids with a low pest severity level at the current moment, expressed as ; Obtain the proportion of grids with a high pest severity level at the previous moment. The calculation formula is: ; Among them, is the proportion of grids with a high pest severity level at the previous moment. When is high, it takes the value of 1, otherwise takes the value of 0, is the classification value of the pest severity level of high, medium, and low for the nth grid at the previous moment; Similarly, obtain the proportion of grids with a medium pest severity level and the proportion of grids with a low pest severity level at the current moment, expressed as ; Calculate the change amount of different pest severity levels in the forest at the current moment. The calculation formula for the change amount of the high pest severity level is: ; Among them, is the change amount of the high pest severity level at the current moment; Similarly, obtain the change amount of the medium pest severity level and the change amount of the low pest severity level at the current moment, expressed as ; Generate an alarm index according to the change amount of different pest severity levels in the forest. The calculation formula is: ; Among them, AI(t) is the alarm index of the forest at the current moment, are the weight coefficients of the corresponding items respectively, ; Compare the alarm index of the forest at the current moment with a preset threshold. If , it is determined that there are many insect pests in the forest at the current moment, and a high insect pest alarm is issued. If , it is determined that the insect pests in the forest at the current moment are average, and an insect pest alarm is issued. If , it is determined that there are few insect pests in the forest at the current moment, and no alarm is issued. Among them, are respectively the high-risk threshold and the low-risk threshold.

[0014] The present invention further provides a forest insect pest prediction system. The forest insect pest prediction system is used to implement the above-mentioned forest insect pest prediction method, and includes: A feature parameter acquisition module, which is used to divide the forest into N grids of equal area and collect the feature parameters of each grid. The feature parameters include the main tree species, tree feature data, soil property data, and meteorological data; A sample selection module, which is used to divide all grids into multiple similar grid groups according to the soil property data of each grid, randomly select several sample grids from the similar grid groups, and obtain the forest insect pest data of the sample grids. The forest insect pest data includes the proportion of insect-infested trees, the spread speed of the insect-infested area, and the growth speed of the insect pest density; An insect pest severity level determination module, which is used to determine the environmental suitability index of each grid by combining the main tree species and meteorological data of the sample grids, analyze the tree feature data to obtain the tree vulnerability index of the sample grids, and analyze the tree vulnerability index and forest insect pest data of the sample grids based on the expert scoring method to determine the insect pest severity level of each grid; An insect pest prediction model construction module, which is used to construct an insect pest prediction model corresponding to each similar grid group based on the random forest algorithm, use the main tree species, tree feature data, and environmental suitability index of the sample grids in the same similar grid group as inputs, and the corresponding insect pest severity level as the output to train the insect pest prediction model, and obtain the insect pest severity level of other grids in the same similar grid group except the sample grids through the trained insect pest prediction model; An insect pest monitoring and alarm module, which is used to analyze the distribution changes of the insect pest severity levels of the N grids at the current moment and historical moments, obtain the change amounts of each insect pest severity level, so as to determine the insect pest situation of the forest and issue corresponding alarm signals.

[0015] The present invention further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and the computer program implements the above-mentioned forest insect pest prediction method when executed by a processor.

[0016] In the above technical solution, the technical effects and advantages provided by the present invention: The forest pest prediction method of the present invention effectively solves the problem that traditional pest monitoring means cannot achieve accurate and timely early warning. By dividing the forest into grids and combining the comprehensive analysis of various environmental characteristic parameters, it can more accurately identify and evaluate pest risks. This data-driven method uses the random forest algorithm to construct a pest prediction model, significantly improving the prediction accuracy of pest levels, thus providing a scientific basis for the formulation of control measures, helping managers take timely countermeasures, and reducing the economic losses caused by pests. In addition, by combining the environmental suitability index and the vulnerability index of forest trees, this solution makes the pest risk assessment more comprehensive, and can comprehensively consider various factors such as meteorological and soil characteristics. This multi-dimensional analysis method not only improves the flexibility of monitoring, but also enhances the response ability of the early warning system, ensuring that an alarm can be issued at the initial stage of pest occurrence, thus effectively protecting the health and sustainable development of the forest ecosystem. This innovative method provides a new idea for forest pest management and significantly improves the efficiency of ecological protection and resource management. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic diagram of the overall method flow of the present invention; Figure 2 is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the following further details the present invention with reference to specific embodiments.

[0019] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar terms used in the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "linked" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative position relationships, and when the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0020] Embodiment: Please refer to Figure 1 , the present invention provides a technical solution: A forest pest prediction method, the specific steps include: Step 1: Divide the forest area equally into N grids, and collect the characteristic parameters of each grid. The characteristic parameters include the main tree species, tree characteristic data, soil property data, and meteorological data. In this embodiment, it specifically includes: Divide the land area of the forest equally into N grids, collect the species of all trees in each grid, take the tree species with the largest number in each grid as the main tree species of the corresponding grid, and collect the tree characteristic data of the trees corresponding to the main tree species in each grid. Take the average value of the tree characteristic data of the trees as the tree characteristic data of the corresponding grid. The tree characteristic data includes tree species, tree density, and tree moisture content. Set the center of each grid as the sampling point, and collect the soil property data of the sampling point of each grid. The soil property data includes the pH value of the soil, soil humidity, and organic matter content. Obtain the meteorological data of the location where the forest is located. The meteorological data of each grid is the meteorological data of the location where the forest is located. The meteorological data includes temperature, humidity, and rainfall.

[0021] The moisture content data of the trees can be monitored by a tree moisture meter. The meteorological data is obtained through the meteorological station at the location of the forest. Determine the tree species in each grid area based on the local forest survey data, and then query relevant literature and data according to the tree species to determine the tree density. Arrange pH sensors and humidity sensors at the sampling points of each grid to collect the pH value and soil humidity of the soil, or sample the soil at the location of the grid and analyze it through the laboratory to obtain the pH value, soil humidity, and organic matter content data of the soil.

[0022] Select the tree species with the largest number in each grid as the main tree species, and calculate the average value of its tree characteristic data, which can better reflect the ecological characteristics of the grid and improve the representativeness of the data. The collected soil property data (such as pH value, humidity, and organic matter content) and meteorological data (such as temperature, humidity, and rainfall) provide the necessary conditions for subsequent evaluation of environmental suitability. Here, the daily rainfall data is collected. These data can help analyze the impact of different environmental factors on forest pests and provide a scientific basis for the potential risk of pest occurrence.

[0023] Step 2: Divide all grids into multiple similar grid groups according to the soil property data of each grid, randomly select several sample grids from the similar grid groups, and obtain the forest pest data of the sample grids. The forest pest data includes the proportion of pest-infected trees, the spreading speed of pest-infected area, and the growth speed of pest density. In this embodiment, obtaining the forest pest data of the sample grids specifically includes: Take each grid as a node and the soil property data corresponding to each node as the attributes of the node to construct an undirected graph. If two nodes are adjacent, add an edge in the undirected graph to form an undirected graph network; Place each node in an independent similar grid group. For each node in the undirected graph network, move it successively to the similar grid groups where all its neighbor nodes are located, and calculate the difference in property changes that occur each time. Specifically, it includes: ; Among them, is the difference in property changes between the i-th node and the j-th node, is the average pH value of the i-th node and the j-th node, is the pH value of the i-th node and the j-th node, is the average soil moisture of the i-th node and the j-th node, is the soil moisture of the i-th node, is the average organic matter content of the i-th node and the j-th node, is the organic matter content of the i-th node and the j-th node. Both i and j are nodes in the undirected graph, and i and j are neighbor nodes; Move each node to the neighbor similar grid group with the smallest difference in property changes. Repeat moving the nodes until the belonging of the similar grid groups of all nodes no longer changes. At this time, randomly determine several nodes in each similar grid group as the sample grids of this similar grid group.

[0024] By dividing the grids into similar groups according to soil properties, areas with similar environmental conditions can be effectively identified, which provides a more accurate and detailed basis for pest prediction. Calculating the difference in property changes helps to understand the environmental changes between adjacent grids, enabling the algorithm to select the most suitable similar grid group for pest data collection and ensuring that the selected samples are representative. The process of gradually moving the nodes enables the grids to adapt, and finally the division of the similar grid groups is more in line with the actual situation, improving the accuracy of the prediction.

[0025] It reflects the degree of change in soil properties (pH value, soil moisture, organic matter content) between node i and its adjacent node j. Specifically, the smaller this value, the more similar the properties of the two nodes. The calculation of this value can reveal the differences between environmental characteristics, helping to identify which adjacent areas are closer in soil properties, thus providing a basis for pest monitoring decisions. The independent variables include pH value, soil moisture, and organic matter content. These soil properties directly affect the growth status of plants, and the health status of plants is closely related to the occurrence of pests, affecting the availability of nutrients, thus affecting the growth status of trees, and further affecting the occurrence of pests; directly affect the growth and health of plants, and too high or too low humidity may lead to a decline in plant stress resistance and increase the probability of pest occurrence; affect soil fertility, and thus affect plant growth. Healthy plants have stronger resistance to pests. If the soil properties of two nodes are similar, that is a small value indicates that the environmental conditions of these nodes have similar effects on pests, and may lead to similar pest occurrence risks. If the soil properties of node i and node j differ greatly, it indicates that the environmental conditions may lead to significant differences in pest occurrence risks. Grouping more similar ones into a similar grid group and randomly selecting samples within it can better describe the characteristics of all grid-related data in this similar grid group.

[0026] The calculation formula for the proportion of pest-infested trees is: ; where Q is the proportion of pest-infested trees in the sample grid, is the number of pest-infested trees in the sample grid, is the total number of trees in the sample grid; Here, a limiting condition can be given. When the pest-infested area proportion of a tree exceeds a certain threshold, for example: 30%, this tree is regarded as a pest-infested tree, or when the pest density in the sampled area of the sample grid reaches a certain threshold, for example: 10 pests per square meter, the trees in the sampled area are regarded as pest-infested trees.

[0027] Q is the proportion of pest-infested trees in the sample grid, and this proportion reflects the proportion of trees affected by pests in the sample grid relative to the total number of trees, and can intuitively reflect the severity of pests. The more pest-infested trees in the sample grid, the higher the corresponding proportion of pest-infested trees.

[0028] Randomly select several trees as sample trees in the sample grid, and obtain the pest-infested area proportion of the sample trees. The calculation formula is: ; where, is the pest-infested area proportion of the xth sample tree in the sample grid, is the pest-infested area of the xth sample tree in the sample grid, is the total area of the x-th sample tree in the sample grid, where x is the index of the sample tree, and the mean value of the pest damage area ratios of all sample trees is taken as the pest damage area ratio of this sample grid; The pest damage area ratio is the pest damage area ratio of a certain tree in the sample grid. This indicator reflects the ratio of the area affected by pests to the total area of a specific tree, indicating the degree of pest damage to the specific tree.

[0029] The calculation formula for the spreading speed of the pest damage area is: ; where Pv(t) is the spreading speed of the pest damage area of the sample grid at the current moment, is the pest damage area ratio of the sample grid at the current moment, is the pest damage area ratio of the sample grid at the previous moment, represents the time interval between the current moment and the previous moment; The spreading speed of the pest damage area represents the change rate of the pest damage area ratio per unit time, reflecting the speed of pest spread. In the same time interval, the higher the pest damage area ratio at the current moment, the greater the change in the pest damage area ratio, indicating a faster spreading speed of the pest damage area.

[0030] Randomly select multiple sampling areas within the pest damage area of the trees in the sample grid, collect the image data of the sampling areas, obtain the number of pests in each sampling area based on image processing software, and the calculation formula for the pest density is: ; where, is the pest density of the sampling area. This indicator represents the number of pests per unit area, indicating the intensity of the pest damage. Pn is the number of pests in the image of the sampling area, and a is the area of the sampling area; Calculate the pest density of all sampling areas, calculate the average value as the pest density of the corresponding sample grid, and the calculation formula for the growth speed of the pest density is: ; where, is the growth speed of the pest density of the sample grid at the current moment, is the pest density of the sample grid at the current moment, is the pest density of the sample grid at the previous moment.

[0031] The growth speed of the pest density reflects the change rate of the pest density per unit time, reflecting the growth trend of the pests. If the pest density at the current moment increases, the greater the difference from the pest density at the previous moment, the faster the growth speed of the pest density, indicating an increase in pest damage.

[0032] Step 3: Determine the environmental suitability index of each grid by combining the main tree species in the sample grid and meteorological data, analyze the tree characteristic data to obtain the tree vulnerability index of the sample grid, and analyze the tree vulnerability index and tree pest data of the sample grid based on the expert scoring method to determine the pest severity level of each grid; In this embodiment, calculating the environmental suitability index of each grid specifically includes: The calculation formula for the environmental suitability index is: ; Wherein, is the environmental suitability index, are the suitability functions of temperature, humidity, and rainfall respectively, are the weight coefficients of their respective corresponding items, , and ; Perform maximum-minimum normalization processing on temperature, humidity, and rainfall, and also perform maximum-minimum normalization processing on the optimal suitability values of temperature, humidity, and rainfall. The calculation formula for the temperature suitability function is: ; Wherein, represents temperature, with the unit of °C, ∈[0,1], used to control the opening size of the function, set to 0.2, represents the optimal suitability value of temperature, set the value of c to 1, 1 represents the best, and 0 represents the worst, is the optimal pest control temperature of the tree species to which the main tree species in this grid belong; Temperature is an important factor affecting the growth, reproduction, and activity of insects. Generally, many forest pests show the best activity and reproduction ability within the range of 15°C to 30°C. When the temperature is below 15°C, the metabolic rate of many insects slows down and their activity decreases, while when it exceeds 30°C, many species may be subjected to heat stress, resulting in death or inhibited reproduction. Therefore, this range reasonably reflects the biological requirements for pest growth. The optimal pest control temperature here can be set to 25 degrees Celsius, but the specific value can be obtained by querying relevant literature and reports according to the type of the main tree species to obtain the optimal pest control temperature.

[0033] The calculation formula for the humidity adaptability function is: ; Wherein, represents humidity, is a positive value, used to control the opening size of the function, set to 0.1, represents the optimal suitability value of humidity, set the value of d to 1, The optimal moisture content for preventing insect pests for the trees of the main tree species corresponding to the grid; Humidity has a significant impact on the growth environment of plants and insects. Low humidity (below 40%) may cause drought in plants, affecting their growth and health, thereby reducing their resistance to pests; while high humidity (above 70%) may cause the growth of mold and other pathogens, and also affect insect activity. Therefore, the optimal humidity for pest control can be set at 55%, but the specific value can be obtained by consulting relevant literature and reports based on the main types of forest trees to obtain the optimal humidity for pest control.

[0034] The calculation formula of rainfall adaptability function is: ; Where R represents rainfall in mm. is a positive value used to control the opening size of the function, set to 0.15. represents the best suitable value of rainfall, and sets g to 1. It is the optimal pest prevention rainfall for the trees of the main forest species corresponding to the grid.

[0035] Precipitation directly affects soil moisture and plant health, which in turn affects the occurrence of pests. Within this range, rainfall can ensure soil moisture supply and promote healthy plant growth, while too little rainfall (less than 50 mm) may cause plant drought and reduce its ability to resist pests; too much rainfall (more than 100 mm) may cause waterlogging, leading to an increase in plant diseases and indirectly affecting pests. Therefore, the optimal rainfall for pest prevention can be set to 75, but the specific value can be obtained by querying relevant literature and reports based on the main types of trees to obtain the optimal rainfall for pest prevention.

[0036] E is a comprehensive indicator that reflects the suitability of environmental conditions (temperature, humidity and rainfall) within a specific grid for the occurrence of forest pests. Temperature is an important factor affecting the growth, reproduction and activity of insects. Many forest pests show optimal growth and reproduction within a specific temperature range. Humidity has a significant impact on plant health and the occurrence of pests. Appropriate humidity can promote plant growth and improve plant resistance; while too low or too high humidity may lead to restricted plant growth, thus affecting its resistance to pests. In addition, some insects are more likely to reproduce and survive in a high humidity environment. Rainfall affects soil moisture and plant growth. Appropriate rainfall helps plants absorb water, thereby enhancing plant growth and health and improving their resistance to pests. However, excessive rainfall may cause root hypoxia or disease, which in turn affects plant health and thus affects the occurrence of pests.

[0037] Within the appropriate range, as the three independent variables of temperature, humidity, and rainfall get closer to the optimal temperature for pest control, the optimal humidity for pest control, and the optimal rainfall for pest control, their respective adaptive function values will increase, and the environmental suitability index will also increase, indicating that the current environment is closer to the optimal conditions for pest control. As the three independent variables deviate further from the optimal pest control values, the corresponding adaptive function values will decrease, and the corresponding environmental suitability index will decrease, indicating that the current environmental conditions are not conducive to pest control.

[0038] Step 4: Build a pest prediction model corresponding to each similar grid group based on the random forest algorithm. Use the main tree species, tree characteristics data, and environmental suitability index of the sample grids in the same similar grid group as inputs, and the corresponding pest severity level as the output to train the pest prediction model, and obtain the pest severity levels of other grids except the sample grids in the same similar grid group through the trained pest prediction model. In this embodiment, calculating the tree vulnerability index of each grid specifically includes: The calculation formula for the tree vulnerability index of the sample grid is: ; Where, is the tree vulnerability index of the sample grid, is the tree density of the tree to which the main tree species corresponding to the sample grid belongs, is the tree moisture content of the tree to which the main tree species corresponding to the sample grid belongs; By calculating the tree vulnerability index, the vulnerability of trees to pests under specific environmental conditions can be quantitatively evaluated. Tree density and moisture content are key factors affecting the biophysical characteristics of trees. Trees with lower density or higher moisture content may be more vulnerable to pest attacks.

[0039] Analyze the tree vulnerability index and tree pest data of each grid based on the expert scoring method to determine the pest severity level of each grid. The pest severity level includes high pest level, medium pest level, and low pest level.

[0040] Different weights can be assigned to the forest pest data and the forest vulnerability index respectively. The specific weights can be set by experts. The experts score the forest pest data and the forest vulnerability index of each grid respectively, and then calculate the comprehensive score through the weights. The larger the forest pest data, the relatively higher score can be given by the experts. On the contrary, the smaller the forest pest data, the lower score should be given by the experts. The larger the forest vulnerability index, the higher score can be given by the experts. On the contrary, the smaller the forest vulnerability index, the lower score should be given by the experts. However, note that the scores of the forest pest data and the forest vulnerability index here are positively correlated with the pest severity level, that is, the larger the forest pest data and the forest vulnerability index, the higher the pest severity level.

[0041] In this embodiment, constructing the pest prediction model specifically includes: Integrate the main tree species, tree characteristic data, environmental suitability index and pest severity level of several sample grids in the similar grid group into a sample set, and divide the sample set into a training set and a test set with a division ratio of 7:3. Use the main tree species, tree characteristic data and environmental suitability index in the training set as inputs, and the corresponding pest severity index as the label to train the random forest model to obtain the pest prediction model. Set the mean square error as the loss function and monitor the value of the loss function. When the value of the loss function drops to 0.001 within 10 iteration cycles, it is considered that the model has been trained. Then use the test set to evaluate the model. The evaluation indicators are accuracy, precision and recall. If the accuracy, precision and recall are above 70%, it is considered that the model performance is good. If the values of the evaluation indicators are lower than 70%, then adjust the model parameters. The parameters to be adjusted include the number of trees in the random forest. One can consider increasing or decreasing the number of trees, adjusting the maximum depth of the trees, or adjusting the minimum sample split number and the minimum sample leaf number; Then input the main tree species, tree characteristic data and environmental suitability index of the other grids except the sample grids in the same similar grid group into the trained pest prediction model to obtain the pest severity level output by the model.

[0042] Different tree species have different physiological characteristics, chemical compositions, and structural features, which directly affect the tree's resistance to pests. For example, some tree species may have natural resistance to specific pests, while others may be more vulnerable. Therefore, using the main tree species as input data can help the model capture the characteristics of different trees, thus more accurately predicting the corresponding pest severity levels. Forest tree characteristic data (such as tree density and moisture content) are important factors affecting the vulnerability of forest trees. Generally speaking, forest trees with higher density usually have better pest resistance, while the moisture content may affect the decay and pest occurrence of forest trees. By incorporating these characteristic data into the model, more detailed inputs can be provided, which helps the model understand the correlation with pest occurrence, thereby improving the accuracy of prediction. The environmental suitability index is calculated comprehensively through environmental factors such as temperature, humidity, and rainfall. These environmental factors affect the occurrence and development of pests. For example, some pests may reproduce faster under specific temperature and humidity conditions, and rainfall may affect soil moisture and plant growth, thus indirectly affecting the occurrence of pests. Therefore, using the environmental suitability index as input data can help the model identify the potential risks of pest occurrence under specific environmental conditions. The pest severity index quantifies the degree of pest impact and reflects the threat of pests to the forest ecosystem. Since the input data already consider tree species, characteristics, and environmental factors, the output pest severity index must be closely related to these input variables. When the input conditions change (such as different tree species or environmental conditions), the pest severity index will also change accordingly. The model captures the complex relationships between these inputs and outputs through training, thereby achieving effective prediction of pest severity levels.

[0043] The expert scoring method is used to evaluate the severity level of pest damage. If the environmental suitability index is higher and the vulnerability of forest trees is lower, the severity level of pest damage is lower. Conversely, if the environmental suitability index is lower and the vulnerability of forest trees is higher, the severity level of pest damage is higher. The random forest model makes predictions by integrating multiple decision trees, which can effectively improve the accuracy of prediction. The presence of specific forest tree species may be closely related to the incidence of certain pests. For example, some insects may prefer specific forest trees or are more likely to thrive in specific forest tree species. Therefore, determining the main forest tree species in the sample grid as input can help the model understand the potential driving factors of the severity level of pest damage. The forest tree density is usually proportional to its strength and pest resistance ability. Forest trees with higher density are usually not easily invaded by pests and are more resistant to the impact of pests compared to low-density forest trees. Therefore, the forest tree density as an input feature has a direct impact on the pest severity index. The environmental suitability index is comprehensively calculated from factors such as temperature, humidity, and rainfall, and these factors all affect the occurrence and development of pests. For example, too high or too low temperature may lead to a decrease in pest activities, while suitable humidity may promote the reproduction of pests, which is then incorporated as one of the considerations for judging the severity level of pest damage. Since the random forest utilizes the principles of diversity and ensemble learning, compared with a single decision tree, the generalization ability of the model is stronger and it can better adapt to complex data. In pest prediction, the input variables may have high-dimensional features. The random forest can effectively process high-dimensional data and identify the important features affecting pests through feature selection techniques, thereby improving the efficiency and interpretability of the model. During the process of evaluating the model, multiple indicators such as accuracy, precision, and recall are selected, which can comprehensively reflect the performance of the model. These indicators each focus on different aspects and can better evaluate the actual effect of the model in pest prediction. Especially when dealing with imbalanced datasets, the combined use of precision and recall is more effective.

[0044] Step 5: Analyze the change in the distribution of the severity level of pest damage at the current and historical moments for N grids, obtain the change amount of each severity level of pest damage, to determine the pest situation in the forest and send out corresponding alarm signals.

[0045] In this embodiment, determining the pest situation in the forest specifically includes: Obtain the proportion of grids with a high pest damage level at the current moment, and the calculation formula is: ; where, is the proportion of grids with a high pest damage level at the current moment, is the indicator function, which takes the value of 1 when is high, otherwise takes the value of 0, is the classification value of high, medium, and low pest damage levels corresponding to the nth grid at the current moment, n is the index of the grid, N is the total number of grids, and ; Similarly, obtain the proportion of grids with a medium pest severity level and the proportion of grids with a low pest severity level at the current moment, expressed as ; Obtain the proportion of grids with a high pest severity level at the previous moment. The calculation formula is: ; Among them, is the proportion of grids with a high pest severity level at the previous moment. When is high, it takes the value of 1, otherwise it takes the value of 0. is the classification value of the pest severity level of the nth grid at the previous moment as high, medium, or low; Similarly, obtain the proportion of grids with a medium pest severity level and the proportion of grids with a low pest severity level at the current moment, expressed as ; By comparing the proportions of the pest severity levels at the current moment and the previous moment, the change of the pest situation can be tracked in real time. This dynamic monitoring helps to timely detect the spread trend of pests and ensure that the manager can quickly take corresponding measures. In the formula , the use of the indicator function transforms the classification problem of the pest level into a numerical problem that can be directly calculated. By counting the number of grids with a high pest severity level, it reasonably reflects the current severity of pests. By comparing the pest severity level at the current moment with that at the previous moment, the change trend of pests can be effectively identified. This method is based on the principle of time series analysis and has strong logic and practicality.

[0046] Calculate the change amount of different pest severity levels in the forest at the current moment. The calculation formula for the change amount of the high pest severity level is: ; Among them, is the change amount of the high pest severity level at the current moment; Similarly, obtain the change amount of the medium pest severity level and the change amount of the low pest severity level at the current moment, expressed as ; According to the changes of different pest severity levels in the forest, generate an alarm index. The calculation formula is: ; Among them, AI(t) is the alarm index of the forest at the current moment, are the weight coefficients of the corresponding items respectively, ; Compare the alarm index of the forest at the current moment with the preset threshold. If , it is judged that there are many pests in the forest at the current moment, and a high pest alarm is issued. If , if the pest infestation in the forest at the current moment is moderate, a pest infestation alarm is issued. If , if the pest infestation in the forest at the current moment is low, no alarm is issued, where are the high-risk threshold and the low-risk threshold respectively.

[0047] AI(t) represents the comprehensive evaluation value of the pest infestation risk in the forest at the current moment. It comprehensively considers the change amounts of high, medium, and low pest infestation levels. By weighting the changes of different levels, it reflects the severity of the overall pest infestation situation. This index can help forest managers quickly judge the severity of the current pest infestation. According to the comparison between the alarm index and the preset threshold, high, medium, and low risk alarms can be issued in a timely manner to ensure a quick response to pest problems. The independent variable directly affects the dependent variable. Specifically, the change amount of the pest infestation level reflects the spread degree of the pest infestation. For example, if the increase amount of the high pest infestation level is large, then AI(t) will increase significantly, indicating a high pest infestation risk. When the value of the independent variable (i.e., the change amounts of different pest infestation levels) increases, the dependent variable AI(t) usually also increases. For example, if increases, it means that the proportion of the high pest infestation level grid rises, directly leading to the increase of AI(t), reflecting the increase of the pest infestation risk. Similarly, if the change amount of the medium pest infestation level or the change amount of the low pest infestation level increases quietly, although the impact is small, it will also increase AI(t) to a certain extent. On the contrary, if the change amount of the pest infestation level is negative, that is, decreases, it will cause AI(t) to decrease, reflecting the reduction of the pest infestation risk.

[0048] In the formula for generating the alarm index AI(t), the weight coefficients and settings reflect the influence degree of different pest infestation severity levels on the overall risk assessment. Specifically, the magnitude relationship of the weights reflects the importance differences of the high pest infestation level, medium pest infestation level, and low pest infestation level in pest infestation early warning. First of all, the high pest infestation level refers to the change in the proportion of the area with the highest pest infestation severity at the current time point. This change directly affects the health status of the entire forest because a high pest infestation level usually means that the pests cause more serious damage to the trees, which may lead to large-scale tree deaths and the imbalance of the ecosystem. Therefore, instead of assigning a weight to the high pest infestation level, it is included in the calculation of the alarm index with a 100% proportion parameter, which can also be regarded as a weight of 1, so as to ensure a more sensitive reflection of potential major risks during early warning. The medium-level pest infestation Although the impact on the forest is lower than that of the high-level pest infestation, it may still cause potential ecological problems. Therefore, a certain weight Its weight is set to be higher than that of low - level pests reflecting its moderate impact on forest health. This setting ensures that in pest warnings, a moderate response can be made to changes in medium - level pests. Low - level pests have a relatively small impact on the forest, so the lowest weight is given . This not only reflects their limited contribution to the overall ecological risk but also avoids affecting the accuracy of the alarm index due to fluctuations in low - level pests in high - risk situations.

[0049] The alarm index is calculated by weighted calculation of changes in different pest levels, comprehensively considering the change amounts of high, medium, and low pest levels. This comprehensive evaluation provides a more comprehensive pest risk assessment than a single indicator. By setting high - risk and low - risk thresholds, the pest risk can be divided into different levels (high, medium, low). This hierarchical management enables forest managers to take corresponding countermeasures according to different risk levels, improving management efficiency. The high - risk and low - risk thresholds can be obtained by analyzing pest data over the past few years to calculate the average value of the alarm index when pest risks occurred in the past few years. The average value is set as the high - risk threshold, and 0.7 times the average value is set as the low - risk threshold.

[0050] Please refer to Figure 2 , the present invention further provides a forest pest prediction system. The forest pest prediction system is used to implement the above - mentioned forest pest prediction method, including: A feature parameter acquisition module, which is used to divide the forest into N grids of equal area and collect the feature parameters of each grid. The feature parameters include the main tree species, tree characteristic data, soil property data, and meteorological data; A sample selection module, which is used to divide all grids into multiple similar grid groups according to the soil property data of each grid, randomly select several sample grids from the similar grid groups, and obtain the forest pest data of the sample grids. The forest pest data includes the proportion of pest - infested trees, the spread speed of pest - infested area, and the growth speed of pest density; A pest severity level determination module, which is used to determine the environmental suitability index of each grid by combining the main tree species and meteorological data of the sample grids, analyze the tree characteristic data to obtain the tree vulnerability index of the sample grids, and analyze the tree vulnerability index and forest pest data of the sample grids based on the expert scoring method to determine the pest severity level of each grid; The pest prediction model construction module is used to construct a pest prediction model corresponding to each similar grid group based on the random forest algorithm. The main tree species, tree feature data, and environmental suitability index of the sample grids in the same similar grid group are used as inputs, and the corresponding pest severity levels are used as outputs to train the pest prediction model. The pest severity levels of other grids except the sample grids in the same similar grid group are obtained through the trained pest prediction model. The pest monitoring and alarm module is used to analyze the change in the distribution of pest severity levels at the current and historical moments of N grids, obtain the change amount of each pest severity level, determine the pest situation of the forest, and send out corresponding alarm signals.

[0051] The present invention further provides a computer-readable storage medium storing a computer program, which when executed by a processor implements a forest pest prediction method as described above.

[0052] The above formulas are all dimensionless and take their numerical values for calculation. The formula is obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0053] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0054] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They may be located in one place or distributed to multiple grid units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0055] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all should be covered by the protection scope of the present application.

Claims

1. A method for predicting forest insect pests, characterized in that, The specific steps include: Step 1: Divide the forest area equally into N grids, and collect the characteristic parameters of each grid. The characteristic parameters include the main tree species, tree characteristic data, soil property data, and meteorological data. Step 2: Divide all grids into multiple similar grid groups according to the soil property data of each grid. Randomly select several sample grids from the similar grid groups, and obtain the forest pest data of the sample grids. The forest pest data includes the proportion of pest-infected trees, the spreading speed of pest-infected area, and the growth speed of pest density. Step 3: Determine the environmental suitability index of each grid by combining the main tree species and meteorological data of the sample grids. Analyze the tree characteristic data to obtain the tree vulnerability index of the sample grids. Analyze the tree vulnerability index and forest pest data of the sample grids based on the expert scoring method to determine the pest severity level of each grid. Step 4: Construct a pest prediction model corresponding to each similar grid group based on the random forest algorithm. Use the main tree species, tree characteristic data, and environmental suitability index of the sample grids in the same similar grid group as the input, and the corresponding pest severity level as the output to train the pest prediction model. Obtain the pest severity level of other grids except the sample grids in the same similar grid group through the trained pest prediction model. Step 5: Analyze the change in the distribution of pest severity levels at the current and historical moments of the N grids to obtain the change amount of each pest severity level, so as to determine the pest situation in the forest and send out corresponding alarm signals.

2. The forest pest prediction method according to claim 1, wherein Specifically, it includes: Divide the land area of the forest equally, divide it into N grids, collect the types of all trees in each grid, take the tree species with the largest number in each grid as the main tree species of the corresponding grid, and collect the tree characteristic data of the trees corresponding to the main tree species in each grid. Take the average value of the tree characteristic data of the trees as the tree characteristic data of the corresponding grid. The tree characteristic data includes tree density and tree moisture content. Set the center of each grid as the sampling point, and collect the soil property data of the sampling point of each grid. The soil property data includes the pH value of the soil, soil humidity, and organic matter content. Obtain the meteorological data of the location where the forest is located. The meteorological data of each grid is the meteorological data of the location where the forest is located. The meteorological data includes temperature, humidity, and rainfall.

3. The forest pest prediction method according to claim 2, wherein Obtaining the forest pest data of the sample grids specifically includes: Take each grid as a node, and take the soil property data corresponding to each node as the attribute of the node to construct an undirected graph. If two nodes are adjacent, add an edge in the undirected graph to form an undirected graph network. Place each node in an independent similar grid group. For each node in the undirected graph network, move it successively to the similar grid groups where all its neighbor nodes are located, and calculate the difference in characteristic changes that occur each time. Specifically, it includes: ； Among them, is the difference in characteristic changes between the $i$-th node and the $j$-th node, is the average pH value of the $i$-th node and the $j$-th node, is the pH value of the $i$-th node and the $j$-th node, is the average soil moisture of the $i$-th node and the $j$-th node, is the soil moisture of the $i$-th node, is the average organic matter content of the $i$-th node and the $j$-th node, is the organic matter content of the $i$-th node and the $j$-th node. Both $i$ and $j$ are nodes in the undirected graph, and $i$ and $j$ are neighbor nodes; Move each node to the neighbor similar grid group with the smallest difference in characteristic changes. Repeat moving the nodes until the belonging of the similar grid groups of all nodes no longer changes. At this time, randomly determine several nodes in each similar grid group as the sample grids of the similar grid group. The calculation formula for the proportion of pest-infected trees is as follows: ； Among them, Q is the proportion of pest-infected trees in the sample grid, is the number of pest-infected trees in the sample grid, is the total number of trees in the sample grid; Randomly select several trees as sample trees in the sample grid, and obtain the proportion of the pest-infected area of the sample trees. The calculation formula is as follows: ； Among them, is the pest damage area ratio of the x-th sample tree in the sample grid, is the pest damage area of the x-th sample tree in the sample grid, is the total area of the x-th sample tree in the sample grid, x is the index of the sample tree, and the mean value of the pest damage area ratios of all sample trees is taken as the pest damage area ratio of this sample grid; The calculation formula for the spreading speed of the pest-infected area is as follows: ； Among them, Pv(t) is the spreading speed of the pest area in the sample grid at the current moment, is the proportion of the pest area in the sample grid at the current moment, is the proportion of the pest area in the sample grid at the previous moment, represents the time interval between the current moment and the previous moment; Randomly select multiple sampling areas within the pest-infected areas of the trees in the sample grid, collect the image data of the sampling areas, obtain the number of pests in each sampling area based on the image processing software, and the calculation formula for the pest density is as follows: ； Among them, is the pest density in the sampling area, Pn is the number of pests in the sampling area image, and a is the area of the sampling area; Calculate the pest density of all sampling areas, calculate the average value as the pest density of the corresponding sample grid, and the calculation formula for the growth rate of the pest density is as follows: ； Among them, is the growth rate of the pest density of the sample grid at the current moment, is the pest density of the sample grid at the current moment, is the pest density of the sample grid at the previous moment.

4. A method for predicting forest pests according to claim 2, characterized in that, Calculating the environmental suitability index of each grid specifically includes: The calculation formula for the environmental suitability index is as follows: ； Among them, is the environmental suitability index, are the suitability functions of temperature, humidity, and rainfall respectively, are the weight coefficients of their respective corresponding items, , and ; The calculation formula for the temperature suitability function is as follows: ； Among them, represents the temperature in °C, is a positive value used to control the opening size of the function, represents the optimal suitable value of the temperature, is the optimal temperature for pest control of the trees belonging to the main tree species corresponding to this grid; The calculation formula for the humidity adaptability function is as follows: ； Among them, represents humidity, is a positive value used to control the opening size of the function, represents the optimal suitable value of humidity, is the optimal humidity for pest control of the trees belonging to the main tree species corresponding to this grid; The calculation formula for the rainfall adaptability function is as follows: ； where R represents rainfall, with the unit of mm, is a positive value used to control the opening size of the function, represents the optimal suitable value of rainfall, which is the optimal rainfall for pest control of the trees belonging to the main tree species corresponding to this grid.

5. The forest pest prediction method according to claim 2, wherein, Calculating the vulnerability index of forest trees in each grid specifically includes: The calculation formula for the vulnerability index of forest trees in the sample grid is as follows: ； Among them, is the vulnerability index of forest trees in the sample grid, is the forest tree density of the trees belonging to the main forest tree species corresponding to the sample grid, is the forest tree moisture content of the trees belonging to the main forest tree species corresponding to the sample grid; Based on the expert scoring method, analyze the environmental suitability index and the vulnerability index of forest trees in each grid to determine the pest severity level of each grid. The pest severity level includes a high pest level, a medium pest level, and a low pest level.

6. The forest pest prediction method according to claim 1, characterized in that Constructing the pest prediction model specifically includes: Integrate the main tree species, forest tree characteristic data, environmental suitability index, and pest severity level of several sample grids in the similar grid group into a sample set, and divide the sample set into a training set and a test set at a ratio of 7:

3. Use the main tree species, forest tree characteristic data, and environmental suitability index in the training set as inputs, and the corresponding pest severity index as labels to train the random forest model to obtain the pest prediction model. Set the mean squared error as the loss function, monitor the value of the loss function, and when the value of the loss function drops to 0.001 within 10 iteration cycles, it is considered that the model has been trained. Then use the test set to evaluate the model, and the evaluation indicators are accuracy, precision, and recall rate. If the accuracy, precision, and recall rate are above 70%, it is considered that the model performance is good. If the value of the evaluation indicator is below 70%, adjust the model parameters; Then input the main tree species, forest tree characteristic data, and environmental suitability index of other grids except the sample grids in the same similar grid group into the pest prediction model after training to obtain the pest severity level output by the model.

7. A method for predicting forest insect pests according to claim 1, characterized in that, Determining the pest situation of the forest specifically includes: Obtain the proportion of grids with a high pest level at the current moment. The calculation formula is as follows: ； wherein, is the proportion of grids with a high pest severity level at the current moment among grids with a high pest level, is an indicator function that takes a value of 1 when is high and 0 otherwise, is the classification value of the pest severity level of the nth grid at the current moment, which is high, medium, or low. n is the index of the grid, N is the total number of grids, and ; Similarly, obtain the proportion of grids with a medium pest severity level and the proportion of grids with a low pest severity level at the current moment, expressed as ; Obtain the proportion of grids with a high pest level at the previous moment. The calculation formula is as follows: ； Among them, is the proportion of grids with a high pest severity level at the previous moment, and when is high, it takes the value of 1, otherwise it takes the value of 0. is the classification value of the pest severity level of high, medium, and low corresponding to the nth grid at the previous moment; Similarly, obtain the proportion of grids with a medium pest severity level and the proportion of grids with a low pest severity level at the current moment, expressed as ; Calculate the change amount of different pest severity levels in the forest at the current moment. The calculation formula for the change amount of the high pest level is as follows: ； Among them, is the change amount of the high pest level at the current moment; Similarly, obtain the change amount of the pest level and the change amount of the low pest level at the current moment, expressed as ; Generate an alarm index according to the change amount of different pest severity levels in the forest. The calculation formula is as follows: ； where AI(t) is the alarm index of the forest at the current moment, are the weight coefficients of the corresponding items respectively, ; Compare the current forest alarm index with a preset threshold. If , it is determined that there are many pests in the forest at the current moment, and a high pest alarm is issued. If , it is determined that the pests in the forest at the current moment are average, and a pest alarm is issued. If , it is determined that there are few pests in the forest at the current moment, and no alarm is issued. Among them, are the high-risk threshold and the low-risk threshold respectively.

8. A forest pest prediction system, characterized in that, The described forest pest prediction system is used to implement the forest pest prediction method described in any one of claims 1-7, and includes: A feature parameter collection module for equally dividing the forest area into N grids and collecting the feature parameters of each grid. The feature parameters include the main tree species, forest tree characteristic data, soil property data, and meteorological data; A sample selection module, which is used to divide all grids into multiple similar grid groups according to the soil characteristic data of each grid, randomly select several sample grids from the similar grid groups, and obtain the forest pest data of the sample grids, where the forest pest data includes the proportion of pest-infected trees, the spreading speed of pest-infected area, and the growth speed of pest density; A pest severity level determination module, which is used to determine the environmental suitability index of each grid by combining the main tree species and meteorological data of the sample grids, analyze the forest tree characteristic data to obtain the forest tree vulnerability index of the sample grids, and analyze the forest tree vulnerability index and forest pest data of the sample grids based on the expert scoring method to determine the pest severity level of each grid; A pest prediction model construction module, which is used to construct a pest prediction model corresponding to each similar grid group based on the random forest algorithm, use the main tree species, forest tree characteristic data, and environmental suitability index of the sample grids in the same similar grid group as the input, and the corresponding pest severity level as the output to train the pest prediction model, and obtain the pest severity level of other grids except the sample grids in the same similar grid group through the trained pest prediction model; A pest monitoring and alarm module, which is used to analyze the change in the distribution of pest severity levels at the current and historical moments of N grids, obtain the change amount of each pest severity level, so as to determine the pest situation of the forest and send out corresponding alarm signals.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the forest pest prediction method according to any one of claims 1-7.

Images