CN-120950863-B - Distribution early warning method based on termite distribution reasoning model

CN120950863BCN 120950863 BCN120950863 BCN 120950863BCN-120950863-B

Abstract

The invention discloses a distribution early warning method based on a termite distribution reasoning model, which relates to the technical field of termite control and ecological monitoring, and comprises the following steps of acquiring multi-source data associated with a target geographic area, wherein the multi-source data comprises historical termite activity records, environmental factor data and engineering structure parameters; assigning termite samples in the historical termite activity record to one or more specific dietary populations according to a preset termite dietary population classification system to generate a population activity dataset. According to the termite prediction method, termites are classified according to diet groups, and a special prediction model is built, so that the problem of insufficient prediction precision caused by neglecting environmental response differences among groups in the prior art is solved. The method can accurately capture the specific association between different groups and environmental factors, and the generated group distribution probability map can clearly show the spatial distribution state of each group, so that the overlapping of the distribution characteristics is avoided.

Inventors

DENG YIZHONG
XIONG WEI
LI WENTAO
CHEN JUAN
SUN ZHENGHUI
WU HAO
LI GUIHAI
ZHU YANBO

Assignees

湖南江山美生态科技有限公司

Dates

Publication Date: 20260512
Application Date: 20250728

Claims (8)

1. A termite distribution reasoning model-based distribution early warning method applied to a target geographic area, which is characterized by comprising the following steps: Acquiring multi-source data associated with the target geographic area, the multi-source data including historical termite activity records, environmental factor data, and engineering structure parameters; Assigning termite samples in the historical termite activity record to one or more specific diet groups according to a preset termite diet group classification system to generate a group body activity data set; Creating a population-specific predictive model for each of the one or more particular dietary populations; The step of creating a group-specific predictive model includes: extracting, for each particular dietary population, corresponding activity point data from the population-separated activity data set; integrating the environmental factor data and the engineering structure parameters corresponding to the activity point data in the space-time dimension to construct a group exclusive original characteristic data set; performing standardization processing on the group-specific original characteristic data set to generate a group-specific training set for model training; Training based on the group exclusive training set by utilizing a preset machine learning algorithm to obtain the group exclusive prediction model, wherein the model internalizes a specific association rule between the specific diet group and the environmental factors; the step of constructing a group-specific raw feature dataset further comprises: selecting an initial subset of environmental factors from the environmental factor data according to a predetermined community preference knowledge base for each particular dietary community; evaluating the completeness and redundancy of the initial environmental factor subset for representing the survival environment of the specific dietary population to obtain an evaluation result; if the evaluation result indicates that the completeness is insufficient or the redundancy is too high, the initial environmental factor subset is adjusted to form an optimized environmental factor subset; Adopting the optimized environment factor subset or the initial environment factor subset as an environment factor input for constructing the group-specific original characteristic data set; generating a group distribution probability map representing the spatial distribution state of each specific diet group in the target geographic area based on the group exclusive prediction model; and generating and outputting distribution early warning information aiming at a specific diet group according to the group distribution probability map and a preset early warning trigger logic.
2. The termite distribution inference model-based distribution pre-warning method according to claim 1, wherein the predetermined termite diet group classification system is classified according to main feeding objects of termites, and the specific diet group comprises: a population of wood-dwelling termites that primarily ingests wood structures and cellulose; soil termite colony, which mainly feeds humus and organic matters in soil; a civil amphibian termite population having both woodland and earthland feeding characteristics; and, a fungus nursery termite population that cultures fungi as a food source within the nest.
3. The termite distribution inference model based distribution pre-warning method of claim 1, wherein the environmental factor data comprises: climate data including annual average temperature, monthly extreme temperature, annual average rainfall, and relative humidity data; soil data comprising soil type, soil pH value, soil organic matter content and soil water content data; Vegetation data comprising vegetation coverage index, vegetation type, and earth surface withered branch and fallen leaf layer thickness data; and the topography data comprises elevation, gradient and slope data.
4. The termite distribution inference model based distribution pre-warning method of claim 1 wherein the step of evaluating the initial environmental factor subset for completeness and redundancy characterizing the survival environment of the particular dietary population comprises: Applying a feature selection algorithm to calculate a relevance score of each environmental factor in the initial subset of environmental factors to the distribution of activity points of the particular dietary population; calculating mutual information scores or correlation coefficients among all the environmental factors in the initial environmental factor subset; And identifying a low correlation factor and a high redundancy factor according to a preset score threshold and a mutual information threshold, and generating the evaluation result.
5. The termite distribution inference model based distribution pre-warning method of claim 1 wherein the step of generating a cluster distribution probability map characterizing the spatial distribution state of each specific dietary population within the target geographic region comprises: Gridding the target geographic area to generate a group of geographic cells; extracting corresponding environmental factor data and engineering structure parameters of each geographic cell in the group of geographic cells to form a feature vector to be predicted; Aiming at each specific diet group, applying the group exclusive prediction model corresponding to the specific diet group to the feature vector to be predicted, and calculating to obtain the occurrence probability value of the diet group in the geographic cell; Combining the occurrence probability values of all the geographic cells to form a spatial distribution probability layer of the specific diet group; And superposing the spatial distribution probability map layers of all specific diet groups to form the grouping body distribution probability map.
6. The termite distribution inference model based distribution pre-warning method of claim 1 wherein the pre-warning trigger logic comprises: presetting a probability threshold and a space aggregation threshold for each specific diet group; Triggering early warning when the distribution probability of a specific diet group in the group distribution probability map is continuously higher than the probability threshold corresponding to the specific diet group in a certain continuous area and the area or the number of cells in the area reaches the space aggregation threshold; the early warning information comprises early warning diet group types, geographic coordinate ranges of early warning areas and average occurrence probability in the areas.
7. The termite distribution inference model based distribution pre-warning method according to claim 1, wherein the method further comprises the step of dynamically updating: receiving newly-added termite activity monitoring data, wherein the newly-added termite activity monitoring data comprises new activity point positions and corresponding diet group types; adding the newly added termite activity monitoring data and the corresponding environmental factor data to the corresponding group body activity data set and the group exclusive original characteristic data set; Triggering a retraining process of one or more affected group-specific predictive models according to a preset updating strategy to generate updated group-specific predictive models; the update policy includes a trigger policy based on the data increment reaching a predetermined number or a timed trigger policy based on a preset time period.
8. The termite distribution inference model based distribution pre-warning method of claim 7 wherein the step of triggering a retraining process of one or more affected ones of the population-specific predictive models further comprises: before retraining is performed, evaluating performance indexes of the current group exclusive prediction model by using an independent verification data set; executing the retraining process to generate a candidate updated predictive model; evaluating performance metrics of the candidate updated predictive models using the independent validation data sets; comparing the performance index of the current group exclusive prediction model with the performance index of the candidate updated prediction model, and if the performance index of the candidate updated prediction model is better than the performance index of the candidate updated prediction model, replacing the current group exclusive prediction model by adopting the candidate updated prediction model; And recording the operation of replacing the current group-specific prediction model by the candidate updated prediction model, the corresponding performance index change and the time stamp information for triggering the operation.

Description

Distribution early warning method based on termite distribution reasoning model Technical Field The invention relates to the technical field of termite control and ecological monitoring, in particular to a distribution early warning method based on a termite distribution reasoning model. Background In the termite control field, the current technology can collect termite surface activity characteristic data, biological characteristics, environmental preference and other information through multiple channels, integrate engineering structure parameters, soil types and other key parameters, perform standardized processing through statistical analysis or a machine learning algorithm, and select proper mathematical model fitting data to perform training. The termite active region can be displayed by combining GIS technology and remote sensing images to carry out spatial distribution modeling, and a dynamic monitoring system can be established to adjust the prediction result and early warning according to new information. However, the prior art has significant shortcomings in the refinement of the response of different termite diet populations to environmental factors. Termites from different dietary populations, including wood-dwelling, soil-dwelling, civil-amphibious, and fungus-nursery termite populations, have significant differences in environmental requirements for survival and reproduction. In the current model, termites are mostly taken as research objects, and subtle differences of various diet groups in environmental preference cannot be deeply distinguished, and in particular, a technical scheme for establishing a dedicated prediction model for different diet groups is lacking. The model is limited in accuracy when predicting distribution of different termite diet groups in a specific area, can not provide more differentiated guidance for targeted control, and can not meet the requirements of accurate positioning and control of different termites in actual control work. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a termite diet population classification-based distribution early warning method and system, which can improve the distribution prediction precision of different termite diet populations and provide differentiated guidance for targeted control. In order to achieve the purpose, the invention is realized by the following technical scheme that the termite distribution reasoning model-based distribution early warning method is applied to a target geographic area and comprises the following steps: Acquiring multi-source data associated with the target geographic area, the multi-source data including historical termite activity records, environmental factor data, and engineering structure parameters; Assigning termite samples in the historical termite activity record to one or more specific diet groups according to a preset termite diet group classification system to generate a group body activity data set; Creating a population-specific predictive model for each of the one or more particular dietary populations; generating a group distribution probability map representing the spatial distribution state of each specific diet group in the target geographic area based on the group exclusive prediction model; and generating and outputting distribution early warning information aiming at a specific diet group according to the group distribution probability map and a preset early warning trigger logic. Further, the step of creating a group-specific predictive model includes: extracting, for each particular dietary population, corresponding activity point data from the population-separated activity data set; integrating the environmental factor data and the engineering structure parameters corresponding to the activity point data in the space-time dimension to construct a group exclusive original characteristic data set; performing standardization processing on the group-specific original characteristic data set to generate a group-specific training set for model training; training based on the group-specific training set by using a preset machine learning algorithm to obtain the group-specific prediction model, wherein the model internalizes the specific association rule between the specific dietary group and the environmental factor. Further, the preset termite diet group classification system is divided according to main feeding objects of termites, and the specific diet group comprises: a population of wood-dwelling termites that primarily ingests wood structures and cellulose; soil termite colony, which mainly feeds humus and organic matters in soil; a civil amphibian termite population having both woodland and earthland feeding characteristics; and, a fungus nursery termite population that cultures fungi as a food source within the nest. Further, the environmental factor data includes: climate data including annual average temperature, monthly extreme temper