CN-121977637-A - Tree root growth state detection device and detection method

CN121977637ACN 121977637 ACN121977637 ACN 121977637ACN-121977637-A

Abstract

The invention relates to the technical field of plant monitoring, in particular to a device and a method for detecting the growth state of a tree root system. According to the invention, abnormal point smoothing and standardization processing are carried out by collecting methane, oxygen, carbon dioxide and pH value data of multi-depth tree roots, multi-parameter wave crest and wave trough time points are extracted and matched to generate a soil environment matched set, the phase difference of multi-parameter wave crest and wave trough in the matched set is calculated, stability is analyzed to form a root system metabolic phase difference, further multi-parameter ratio gradients are calculated between adjacent depths, the change directions are judged to generate environment ratio gradients, distribution and dynamic trends of the environment parameters on a vertical section are revealed, the root system growth state is judged by fusing the analysis consistency of the root system metabolic phase difference and the environment ratio gradients, and a neural network is constructed to predict root system growth trend based on the result.

Inventors

HOU YAHUI
CAO FANGYI
CHEN BINGHAN
YE SHAOPING
Wen Yueying
LI ZHIXIONG
WANG YONGYUE
LI SIYING

Assignees

广州市林业和园林科学研究院

Dates

Publication Date: 20260505
Application Date: 20251229

Claims (10)

1. Tree root growth state detection device, its characterized in that, the device includes: The soil environment module is used for collecting data of methane, oxygen, carbon dioxide and pH values of multiple depths of tree root systems, smoothing and standardizing abnormal points, extracting and pairing multi-parameter wave crest and valley time points, generating a soil environment pairing group and transmitting the soil environment pairing group to the phase extraction module; The phase extraction module is used for calculating phase differences between wave crests and wave troughs in the multiple paired groups and analyzing the stability degree of the phase differences based on the soil environment paired groups, generating root metabolism phase differences and transmitting the root metabolism phase differences to the gradient calculation module; The gradient calculation module is used for calculating the ratio gradient of methane, oxygen, carbon dioxide and pH value between adjacent depths through a linear regression model based on the root system metabolic phase difference, judging the gradient change direction, generating an environment ratio gradient and transmitting the environment ratio gradient to the state judgment module; the state judging module is used for analyzing the consistency of the multi-depth environmental ratio gradient and the phase difference stability degree to judge the growth state of the root system based on the root system metabolic phase difference and the environmental ratio gradient, generating a growth state judging result and transmitting the growth state judging result to the trend predicting module; And the trend prediction module is used for constructing a tree root system health model and predicting the root system growth trend through a convolutional neural network based on the growth state judgment result to generate a root system growth trend prediction result.
2. The tree root system growth state detection device according to claim 1, wherein the soil environment pairing group comprises methane peak valley pairing, oxygen peak valley pairing and pH peak valley pairing, the root system metabolic phase difference comprises methane phase difference, oxygen phase difference and phase difference stability coefficient, the environment ratio gradient comprises methane ratio gradient, oxygen ratio gradient, pH ratio gradient and gradient direction, the growth state judgment result comprises stability consistency degree and root system growth state classification, and the root system growth trend prediction result comprises root system growth trend prediction and root system characteristics.
3. The tree root growth state detection device according to claim 1, wherein the soil environment module is specifically: The data analysis submodule collects data of multi-depth methane content, oxygen concentration, carbon dioxide concentration and pH value of the root system of the tree, and the data are arranged according to time sequences, and are uniformly numbered according to the data with the same time dimension to generate an original sequence; The abnormal smoothing sub-module is used for calling the original sequence, detecting abnormal points of the sequences of methane, oxygen, carbon dioxide and pH values, smoothly replacing the values exceeding the reference interval of the parameter time sequence, and carrying out standardized conversion on the processed time sequence in a multi-depth range to obtain a standardized interval; and the peak Gu Diqu submodule detects local peak time points and trough time points in the multi-parameter time sequence of methane, oxygen, carbon dioxide and pH values according to the standardized interval, screens pairing intervals of adjacent peaks and troughs and combines the pairing intervals to generate a soil environment pairing group.
4. The tree root growth state detecting apparatus according to claim 3, wherein the parameter time series reference interval is set according to the average value and the fluctuation amplitude of the monitoring period by detecting the minimum value and the maximum value of the numerical values as boundaries in the complete time series of the same parameter.
5. The tree root growth state detection device according to claim 1, wherein the phase extraction module specifically comprises: The phase difference calculation sub-module is used for calculating the time difference between adjacent peaks and troughs in the same pairing group based on the detected peak time points and trough time points in the soil environment pairing group, and arranging the phase differences under a plurality of pairing groups according to the time sequence order to obtain a phase difference value sequence; The stability analysis submodule calls the phase difference value sequence, calculates the discrete degree of the phase difference value in the same time period according to the difference value distribution in the continuous time period, compares the discrete degree with the interval mean value, and analyzes the phase stability interval; and the metabolism phase generation submodule combines the stability degree of the phase difference value under the time dimension with the corresponding pairing group according to the phase stability interval, screens phase differences of which the stability accords with the interval, and performs summarizing and sequencing to obtain root system metabolism phase differences.
6. The tree root growth state detection device according to claim 1, wherein the gradient calculation module specifically comprises: The ratio extraction submodule detects monitoring data of methane, oxygen, carbon dioxide and pH values in adjacent depth intervals based on the root system metabolic phase difference, and performs pairing ratio calculation on four items of data of each depth layer to obtain a multi-factor ratio sequence; The gradient calculation sub-module is used for calling the multi-factor ratio sequence, applying a linear regression model to adjacent depth intervals, taking the multi-factor ratio sequence as an independent variable to correspond to the depth difference value, calculating a regression coefficient and taking the regression coefficient as a gradient result to obtain a ratio gradient coefficient; And the direction judging sub-module is used for judging the gradient directions of multiple ratios and integrating the gradient direction information of all factors by combining the variation trend among the factors according to the ratio gradient coefficient and corresponding the positive and negative values of the gradient to the variation sequence of the adjacent depth interval, so as to generate the environmental ratio gradient.
7. The tree root growth state detection device according to claim 1, wherein the state determination module specifically comprises: The phase comparison sub-module is used for carrying out point-by-point difference calculation on the two types of data in the same depth interval based on the root system metabolic phase difference and the environmental ratio gradient, sequencing the difference results of multiple depths and counting the deviation degree to obtain a depth difference sequence; The consistency evaluation sub-module is used for calling the depth difference value sequence, calculating the fluctuation amplitude of the difference value between the adjacent intervals according to the variation trend of the difference value between the multiple depths, comparing the fluctuation amplitude with a set consistency threshold value, judging the stability degree between the multiple depths and quantifying the stability degree as the stability consistency degree; And the state generation sub-module is used for judging and integrating a specific growth state according to the stability consistency degree and the directivity information of the environmental ratio gradient at a plurality of depths, and generating a growth state judging result when the stability consistency degree is larger than a threshold A and the number of positive gradient directions exceeds a threshold B.
8. The tree root growth state detecting apparatus according to claim 7, wherein the consistency threshold is set by counting the fluctuation value of the difference between adjacent depth intervals, taking the median of the distribution as a reference value, and adding twice the fluctuation interval limit to the reference value.
9. The tree root growth state detection device according to claim 1, wherein the trend prediction module specifically comprises: the characteristic extraction submodule extracts fluctuation data related to the root system state based on the growth state judging result and combines the fluctuation data into an input sequence, and then carries out convolution and pooling operation of a convolution neural network on the input sequence, screens key change factors and analyzes root system characteristic values as input tensors; The model training submodule calls the root characteristic value to input a multi-layer convolution kernel structure, carries out weight updating and iterative training according to the error between network output and input, and fixes the converged parameter set as a structural index of the health model to obtain the parameter number of the health model; And the trend generation sub-module is used for inputting continuous sequence data corresponding to the growth state judgment result according to the health model parameter, calculating the numerical distribution of the output layer in the time dimension, extracting the change direction, predicting the root system growth trend and generating a root system growth trend prediction result.
10. The method for detecting the growth state of the root system of the tree, which is characterized in that the device for detecting the growth state of the root system of the tree according to any one of claims 1 to 9 is executed and comprises the following steps: S1, collecting data of methane, oxygen, carbon dioxide and pH values of multiple depths of tree root systems, smoothing and standardizing abnormal points, extracting and pairing multi-parameter wave crest and valley time points, and generating a soil environment pairing group; S2, calculating phase differences between wave crests and wave troughs in the multiple paired groups based on the soil environment paired groups, analyzing stability degree of the phase differences, and generating root system metabolism phase differences; S3, calculating the ratio gradient of methane, oxygen, carbon dioxide and pH value between adjacent depths through a linear regression model based on the root system metabolic phase difference, judging the gradient change direction, and generating an environment ratio gradient; S4, analyzing the consistency of the multi-depth environmental ratio gradient and the phase difference stability degree based on the root system metabolic phase difference and the environmental ratio gradient to judge the growth state of the root system, and generating a growth state judging result; And S5, constructing a tree root system health model and predicting the root system growth trend through a convolutional neural network based on the growth state judging result, and generating a root system growth trend predicting result.

Description

Tree root growth state detection device and detection method Technical Field The invention relates to the technical field of plant monitoring, in particular to a device and a method for detecting the growth state of a tree root system. Background The field of plant monitoring technology includes a variety of devices and methods for detecting physiological states, ecological environmental parameters, and soil and root characteristics associated with the growth of plants over multiple growth cycles. The core content comprises biological information acquisition based on a sensing detection principle, plant tissue characteristic analysis based on optical or electrical signals and plant root system and soil interaction condition detection based on geological and ecological parameters. The tree root system growth state detection device and the tree root system growth state detection method are a method and a device for monitoring the distribution state of the tree root system, the root system growth speed and the combination condition of the root system and a soil medium by utilizing a root system penetrating type or non-contact type sensing structure combined with an optical signal reflection intensity measurement mode. The patent theme covers the technical scheme of detecting the root system structural profile through an underground sensing probe, acquiring root system tissue density information through spectral characteristic analysis and recording the dynamic change of the root system by utilizing a time sequence measurement means, and aims to provide a technical path capable of realizing multi-point synchronous acquisition and information analysis in an underground environment. In the prior art, the tree root system monitoring mainly depends on a sensing detection principle, optical or electrical signal analysis and geological parameter detection, and the operation mode of the tree root system monitoring is focused on root system structure contour detection and spectral characteristic analysis, so that limitations exist in practical application. The prior art often ignores integrated monitoring of environmental chemical parameters such as gas concentration and pH value, so that dynamic metabolic processes of root system interaction with soil are difficult to comprehensively capture, for example, the measurement of root system distribution by optical reflection intensity only can be interfered by soil impurities or humidity to reduce accuracy. Although the time sequence measurement means is used for recording root system changes, the stability and time sequence characteristics of metabolic activities cannot be effectively evaluated due to the lack of multi-parameter pairing and phase difference analysis, so that state judgment is biased to a static structure rather than a dynamic process. When the multipoint synchronous acquisition technology is implemented in an underground environment, the multipoint synchronous acquisition technology is limited by physical layout and signal transmission of a sensing probe, and high-resolution multi-depth gradient analysis is difficult to realize, so that accuracy of monitoring the growth speed of a root system is affected. In addition, the prior art relies on the traditional data processing method, and does not introduce a machine learning model for trend prediction, so that the prior art can only provide historical data review but not prospective insight, and limits the application value of the prior art in early warning and decision support. These deficiencies make the prior art possible to generate monitoring blind areas, response delays and resource waste in actual operation, such as increased root health evaluation errors under complex soil conditions, and influence the effectiveness of plant growth management. Disclosure of Invention In order to solve the technical problems in the prior art, the embodiment of the invention provides a tree root growth state detection device and a tree root growth state detection method. The technical scheme is as follows: in one aspect, a tree root growth state detection device is provided, the device comprising: The soil environment module is used for collecting data of methane, oxygen, carbon dioxide and pH values of multiple depths of tree root systems, smoothing and standardizing abnormal points, extracting and pairing multi-parameter wave crest and valley time points, generating a soil environment pairing group and transmitting the soil environment pairing group to the phase extraction module; The phase extraction module is used for calculating phase differences between wave crests and wave troughs in the multiple paired groups and analyzing the stability degree of the phase differences based on the soil environment paired groups, generating root metabolism phase differences and transmitting the root metabolism phase differences to the gradient calculation module; The gradient calculation module is used for calculating the ratio