CN-121997043-A - Method, device, equipment and medium for constructing prediction model of locust adAN_SNtive area

CN121997043ACN 121997043 ACN121997043 ACN 121997043ACN-121997043-A

Abstract

The invention provides a method, a device, equipment and a medium for constructing a prediction model of a locust adAN_SNtive area, and relates to the technical field of intelligent agriculture and animal husbandry. The method comprises the steps of constructing a maximum entropy model based on a monitoring index set, constructing a training sample based on a sample value set of the monitoring index set in a sample area and a label of whether locust disasters exist in the sample area, and training the maximum entropy model based on the training sample to obtain a prediction model of a locust-suitable area. The method constructs the maximum entropy model based on the monitoring index set, realizes the accurate simulation of the relationship between the space-time distribution of the locusts and the monitoring index, and is beneficial to improving the prediction accuracy of the prediction model of the subsequent locusts suitable zone. The prediction model of the locust-suitable area can accurately predict the occurrence of locust disasters in a large space range, and improves the efficiency and the accuracy of monitoring the locust disasters.

Inventors

GUO JUNMING
DONG YINGYING
GUO JING
HUANG WENJIANG

Assignees

中国科学院空天信息创新研究院

Dates

Publication Date: 20260508
Application Date: 20251226

Claims (10)

1. The method for constructing the prediction model of the locust adaptive zone is characterized by comprising the following steps of: the method comprises the steps of obtaining a monitoring index set, wherein the monitoring index set comprises at least one monitoring index which has an influence on the formation of locust disasters, the monitoring index is determined based on an environment influence factor or q statistical values of combinations of environment influence factors, and the q statistical values are determined based on the environment influence factor or the combinations of environment influence factors, and the influence category contained in the q statistical values and the influence condition of the environment influence factors on the occurrence of the locust disasters; constructing a maximum entropy model based on the monitoring index set to simulate the relationship between the monitoring index set and the formation of the locust disasters; Constructing a training sample based on a sample value set of the monitoring index set in the sample area and a label of whether the locust disaster exists in the sample area; And training the maximum entropy model based on the training sample to obtain a prediction model of the locust adAN_SNtive area.
2. The method for constructing a predictive model of a locust-like zone according to claim 1, wherein the obtaining a monitoring index set comprises: Acquiring at least one environmental impact factor or a combination of environmental impact factors, wherein the q statistic value of the formation of the locust disasters is obtained, and the environmental impact factors comprise environmental factors which have an influence on a specific life cycle of the locust; And sorting all the environmental influence factors or the combination of the environmental influence factors based on the q statistical values, and selecting a set number of environmental influence factors with higher q statistical values as the monitoring index based on the sorting result.
3. The method for constructing a predictive model of a locust adaptation area according to claim 2, wherein the environmental impact factor is determined based on the following manner: acquiring at least one life cycle of the locust; in each life activity period, if the influence degree of environmental factors on the growth of the locust is obviously different, determining the environmental influence factors based on the environmental factors and the life activity period; and in each life activity cycle, if the environmental factors have the same influence degree on the growth of the locust, determining the environmental influence factors based on the environmental factors.
4. The method for constructing a prediction model of a locust-suitable area according to claim 1, wherein the training the maximum entropy model based on the training sample to obtain the prediction model of the locust-suitable area comprises: Randomly sampling the training samples with the return to obtain a training set and a testing set with preset proportions; Training and testing the maximum entropy model based on the training set and the testing set to obtain a training omission curve and a testing omission curve of the maximum entropy model; And when the training omission curve and the test omission curve form positive correlation, and the omission ratio of the maximum entropy model is reduced to a stable value, determining that the maximum entropy model training is completed, and obtaining a prediction model of the locust adaptive region.
5. The method for constructing a predictive model of a locust-like zone according to claim 1, wherein the method further comprises, after the step of obtaining the predictive model of a locust-like zone: Removing target monitoring indexes from the prediction model of the locust adaptive zone to obtain an updated prediction model of the locust adaptive zone, wherein the target monitoring indexes are any monitoring indexes; and verifying the importance degree of the target monitoring index on the formation of the locust disaster based on the prediction accuracy of the updated prediction model of the locust-suitable area.
6. A method for predicting a locust-like growth region, comprising: Acquiring distribution data of locust disasters and change data of at least one monitoring index of a set area; inputting the distribution data and the change data into a prediction model of the locust adaptive zone, and obtaining a predicted locust adaptive zone in the set area output by the prediction model of the locust adaptive zone; Wherein the prediction model of the locust-suitable region is obtained according to the construction method of the prediction model of the locust-suitable region according to any one of claims 1 to 5.
7. The method for predicting a locust adaptation zone according to claim 6, wherein the prediction model of the locust adaptation zone is used for: Based on the distribution data and the change data, obtaining the probability of locust disasters occurring in at least one grid of the set area, the change amplitude of the probability and Z statistics, wherein the Z statistics are used for representing the significance of the change amplitude of the probability; based on the probability, the variation amplitude of the probability and the Z statistic, identifying the remarkable situation of probability variation in each grid so as to identify the predicted locust adaptive area.
8. The device for constructing the prediction model of the locust adAN_SNtive area is characterized by comprising the following components: The monitoring index acquisition module is used for acquiring a monitoring index set, wherein the monitoring index set comprises at least one monitoring index which has an influence on the formation of locust disasters, the monitoring index is determined based on the environment influence factors or q statistical values of the combination of the environment influence factors, and the q statistical values are determined based on the environment influence factors or the combination of the environment influence factors, the influence types contained in the environment influence factors and the influence conditions of the environment influence factors on the occurrence of the locust disasters; the initial model construction module is used for constructing a maximum entropy model based on the monitoring index set so as to simulate the relation between the monitoring index set and the formation of the locust disasters; the training sample acquisition module is used for constructing a training sample based on a sample value set of the monitoring index set in the sample area and a label of whether locust disasters exist in the sample area; and the training module is used for training the maximum entropy model based on the training sample to obtain a prediction model of the locust adAN_SNtive area.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and running on the processor, characterized in that the processor implements a method of constructing a predictive model of a locust-like area according to any one of claims 1 to 5 when executing the computer program; or the processor, when executing the computer program, implements a method for predicting a locust-suitable area as claimed in any one of claims 6 to 7.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements a method of constructing a predictive model of a locust-adaptive area according to any of claims 1 to 5; or the computer program when executed by a processor implements a method for predicting a locust-suitable area as claimed in any one of claims 6 to 7.

Description

Method, device, equipment and medium for constructing prediction model of locust adAN_SNtive area Technical Field The invention relates to the technical field of intelligent agriculture and animal husbandry, in particular to a method, a device, equipment and a medium for constructing a prediction model of a locust adAN_SNtive area. Background Locust plague is one of the main biological disasters affecting the development of northern agriculture and animal husbandry in China, and the outbreak of the population of the locust plague threatens the grassland ecosystem and the animal husbandry production seriously. Because the population dynamics and spatial distribution of grasslands are complexly affected by a plurality of environmental factors, and the habitat environment thereof has high heterogeneity in time and space, the distribution of the grasslands presents remarkable geographic aggregation characteristics. Therefore, the key habitat factors affecting the locust distribution are accurately identified, an effective adaptive zone prediction model is constructed, and the method is important for realizing effective monitoring and early warning and accurate prevention and control of locust plague, and is an urgent technical requirement in current agriculture and animal husbandry production. In order to meet the technical requirements, a plurality of grassland locust adAN_SNtive area prediction methods are provided in the prior art, and can be mainly divided into three types, namely a statistical model, a machine learning model and a deep learning model. The statistical model generally utilizes remote sensing and geographic information system (Geographic Information System, GIS) technology to process environmental factor data of long-time sequence through methods such as linear regression, harmonic analysis and the like, and extracts trends and seasonal components capable of reflecting long-term changes of habitats. Machine learning models, such as Random Forest (RF), support vector machines (Support Vector Machine, SVM), etc., can effectively handle nonlinear relationships and high-dimensional interactions between variables, and are excellent in locust density prediction and habitat classification. The deep learning model, such as a convolutional Long-Short-Term Memory network (ConvLSTM), can combine the spatial feature extraction capability of the Convolutional Neural Network (CNN) with the time-series processing capability of the Long Short-Term Memory network (LSTM), and directly utilize the space-time raster data of the environmental factors to predict the suitability of the future habitat. However, the prior art approaches still suffer from a number of drawbacks. Firstly, the traditional monitoring method is seriously restricted by geographical accessibility in real-time investigation, is difficult to fully cover a wide grassland area, has low efficiency and poor real-time performance, is easy to report, and cannot meet the requirements of the modern agriculture and animal husbandry on quick monitoring and accurate prevention and control of diseases and insect pests. Disclosure of Invention The invention provides a method, a device, equipment and a medium for constructing a prediction model of a locust adaptive area, which are used for solving the defects that prediction of the locust adaptive area in the prior art cannot cover a large-range area and prediction efficiency is low and realizing accurate prediction of the locust adaptive area in the large-range area. The invention provides a method for constructing a prediction model of a locust adAN_SNtive region, which comprises the following steps: The method comprises the steps of obtaining a monitoring index set, wherein the monitoring index set comprises at least one monitoring index which has an influence on the formation of locust disasters, the monitoring index is determined based on an environment influence factor or q statistical values of combinations of environment influence factors, and the q statistical values are determined based on the environment influence factor or the combinations of environment influence factors, and the influence category contained in the environment influence factor or the combinations of environment influence factors and the influence condition of the environment influence factors on the occurrence of the locust disasters; constructing a maximum entropy model based on the monitoring index set to simulate the relationship between the monitoring index set and the formation of locust disasters; Constructing a training sample based on a sample value set of the monitoring index set in the sample area and a label of whether the locust disaster exists in the sample area; And training the maximum entropy model based on the training sample to obtain a prediction model of the locust adAN_SNtive area. According to the method for constructing the prediction model of the locust adAN_SNtive area provided by the invention, a monit