CN-122022021-A - Broadleaf hybrid forest management prediction method based on time sequence every wood data

CN122022021ACN 122022021 ACN122022021 ACN 122022021ACN-122022021-A

Abstract

The invention discloses a broadleaf mixed forest operation prediction method based on time sequence every wood data, which comprises the steps of collecting and preprocessing time sequence data, calculating basic growth potential through basic growth potential quantification logic of an integrated attention mechanism, carrying out competition-microenvironment interaction correction by combining every wood detection ruler data and microenvironment data, introducing biological interaction data and operation measure parameters to construct feedback optimization logic, outputting final growth prediction values, generating an individual operation scheme, and carrying out closed loop iteration dynamic adjustment. The method solves the core problem of the prior art of prediction and operation disconnection, realizes the accurate prediction and individual operation of the growth of the broadleaf hybrid forest, and adapts to the growth dynamics and environmental change of the forest.

Inventors

Su Murong
ZHANG LINGYU
SHEN CHANGQING
TAN SHA
LU CANHUI

Assignees

佛山市云勇林场

Dates

Publication Date: 20260512
Application Date: 20260114

Claims (10)

1. The broadleaf hybrid forest operation prediction method based on time sequence every wood data is characterized by comprising the following steps of: S1, collecting time sequence data, wherein the time sequence data comprises each piece of wood detection rule data, micro-environment data and biological interaction data, and preprocessing the time sequence data to obtain preprocessed time sequence data; S2, calculating basic growth potential through a basic growth potential quantization formula based on each wood detection rule data in the preprocessed time sequence data, wherein the basic growth potential quantization formula integrates time sequence growth rate calculation logic weighted by an attention mechanism; S3, based on each wood detection rule data and micro-environment data in the preprocessed time sequence data, carrying out interactive correction on the basic growth potential through a competition-micro-environment collaborative correction formula to obtain corrected growth potential; S4, optimizing the corrected growth potential through a biological interaction-operation feedback optimization formula based on biological interaction data in the preprocessed time sequence data and combining operation measure parameters, and outputting a final growth prediction value, wherein the operation measure parameters are obtained through inverse solution of the biological interaction-operation feedback optimization formula; S5, generating an individual operation scheme based on the final growth predicted value; and S6, repeating the steps S1 to S5 to carry out dynamic iterative adjustment.
2. The broadleaf mixed forest operation prediction method based on time sequence every wood data, which is disclosed in claim 1, is characterized in that every wood detection rule data comprise a target tree chest diameter, a target tree height, an adjacent tree chest diameter and a horizontal distance between a target tree and an adjacent tree, the micro-environment data comprise a seasonal average temperature, a seasonal precipitation amount and a seasonal total illumination time length, and the biological interaction data comprise a target tree root system mycorrhiza infection rate, a target tree nitrogen fixation tree species number within a 5m range and a target tree total tree species number within a 5m range.
3. The broadleaf mixed forest operation prediction method based on time sequence every wood data according to claim 1, wherein the preprocessing comprises filling missing data by a linear interpolation method, eliminating abnormal data by a 3 sigma criterion, wherein the abnormal data are numerical values exceeding a data mean value by +/-3 times of standard deviation, the acquisition period number of the time sequence data is 3 periods or more, and the adjacent acquisition interval is 1 growing season.
4. The method of claim 1, wherein the dynamic iterative adjustment comprises recalculating the base growth potential, modified growth potential, and final growth prediction based on the newly acquired time series data, and synchronously adjusting the operational measure parameters in the personalized operational plan.
5. The method of claim 1, wherein the basic growth potential quantification formula integrates an attention mechanism, the basic growth potential is calculated by a coupled logic of a relative biomass and a time sequence weighted average growth rate, the relative biomass is determined based on a ratio of a target tree chest diameter, a tree height, a crown size coefficient and a maximum biomass of mature ages of the target tree species, and the time sequence weighted average growth rate highlights the influence of recent data on the growth potential by an attention weight.
6. The broadleaf mixed forest operation prediction method based on time sequence every wood data according to claim 1, wherein the competition-microenvironment cooperative correction formula corrects basic growth potential through interaction logic of adjacent wood competition strength and seasonal microenvironment suitability, the adjacent wood competition strength is calculated based on the ratio of the sum of adjacent wood chest diameters to the target tree chest diameter and the ratio of standard adjacent distance coefficient to the sum of adjacent wood horizontal distances, and the seasonal microenvironment suitability is determined based on weighted summation of the ratio of temperature, precipitation and illumination to corresponding optimal values.
7. The broadleaf mixed forest operation prediction method based on time sequence every wood data according to claim 1, wherein the biological interaction-operation feedback optimization formula optimizes the corrected growth potential through cooperative logic of biological interaction effect coefficients and operation measure parameters, the biological interaction effect coefficients are determined based on weighted summation of ratios of mycorrhiza infection rate to maximum mycorrhiza infection rate and ratios of nitrogen fixation tree species number to total tree species number, and the operation measure parameters comprise m-valve strength and fertilization amount.
8. The method of claim 1, wherein the personalized management scheme comprises a thinning scheme, wherein the thinning scheme determines a thinning time and a thinning object based on the final growth prediction value and the adjacent wood competition strength, and wherein the thinning object is an adjacent wood with an adjacent wood competition strength greater than 0.4.
9. The method of claim 8, wherein the personalized management scheme further comprises a fertilization scheme that determines a fertilization amount and a fertilization time based on the final growth prediction value and the seasonal microenvironment suitability, the fertilization time selecting a growth season having a seasonal microenvironment suitability of greater than 0.7.
10. The method for predicting broadleaf mixed forest operations based on time series per-tree data according to claim 1, wherein the personalized operation scheme further comprises a harvest and sales scheme, wherein the harvest and sales scheme estimates a harvest cycle of a target forest to reach a preset chest diameter based on the final growth prediction value, and sales volume and sales time are determined in combination with the harvest cycle.

Description

Broadleaf hybrid forest management prediction method based on time sequence every wood data Technical Field The invention relates to the technical field of forestry management and prediction, in particular to a broadleaf hybrid forest management prediction method based on time sequence every wood data. Background The method is characterized in that the method is mainly characterized in that the method comprises the steps of carrying out linear prediction on the basis of single forest growth data, carrying out linear prediction on the basis of the single forest growth data, not considering the interactive inhibition or promotion effect between adjacent wood competition and seasonal microenvironment, ignoring the influence of a special biological interaction mechanism of the hybrid forest on the forest growth, leading to obvious deviation of a growth prediction result and an actual growth state, and further comprises the steps of lacking dynamic feedback logic between the growth prediction and the operation measures, wherein the operation scheme is mainly uniformly set, and key parameters such as interval cutting, fertilization and the like cannot be reversely adjusted according to the real-time growth prediction result of the forest, so that the operation measures and the forest growth requirement are disjointed, the resource utilization efficiency is low, and the optimal growth state of a forest individual is difficult to realize. Based on the above problems, a technical solution for realizing accurate growth prediction and dynamic adaptation of personalized operation is needed. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a broadleaf mixed forest operation prediction method based on time sequence per wood data, which comprises the following steps: S1, collecting time sequence data, wherein the time sequence data comprises each piece of wood detection rule data, micro-environment data and biological interaction data, and preprocessing the time sequence data to obtain preprocessed time sequence data; S2, calculating basic growth potential through a basic growth potential quantization formula based on each wood detection rule data in the preprocessed time sequence data, wherein the basic growth potential quantization formula integrates time sequence growth rate calculation logic weighted by an attention mechanism; S3, based on each wood detection rule data and micro-environment data in the preprocessed time sequence data, carrying out interactive correction on the basic growth potential through a competition-micro-environment collaborative correction formula to obtain corrected growth potential; S4, optimizing the corrected growth potential through a biological interaction-operation feedback optimization formula based on biological interaction data in the preprocessed time sequence data and combining operation measure parameters, and outputting a final growth predicted value, wherein the operation measure parameters are obtained through reversely solving the biological interaction-operation feedback optimization formula; and S6, repeating the steps S1 to S5 to carry out dynamic iterative adjustment. Preferably, each piece of wood detection rule data comprises a target tree chest diameter, a target tree height, an adjacent tree chest diameter and a horizontal distance between the target tree and the adjacent tree, the microenvironment data comprises a seasonal average temperature, a seasonal precipitation amount and a seasonal total illumination time length, and the biological interaction data comprises a target tree root system mycorrhiza infection rate, a target tree nitrogen fixation tree species number within a 5m range and a target tree total tree species number within a 5m range. Further preferably, the preprocessing comprises filling missing data by a linear interpolation method, removing abnormal data by a 3 sigma criterion, wherein the abnormal data are numerical values exceeding a data mean value by +/-3 times of standard deviation, the acquisition period number of the time sequence data is 3 or more, and the adjacent acquisition interval is 1 growth season. Further preferably, the dynamic iterative adjustment includes recalculating the base growth potential, the modified growth potential, and the final growth prediction value based on the newly acquired time series data, and synchronously adjusting the operational measure parameters in the personalized operational scheme. Further preferably, the basal growth potential quantification formula integrates an attention mechanism, and basal growth potential is calculated by coupling logic of relative biomass and a time sequence weighted average growth rate, wherein the relative biomass is determined based on the ratio of a target tree chest diameter, tree height, crown amplitude coefficient and a maximum biomass of mature ages of target tree species, and the time sequence weighted average growth rate highl