CN-121978037-A - Quick detection and analysis method for soil nutrients

CN121978037ACN 121978037 ACN121978037 ACN 121978037ACN-121978037-A

Abstract

The invention belongs to the technical field of soil nutrient detection and analysis, and particularly discloses a rapid detection and analysis method for soil nutrients, which comprises the steps of inserting a spectrum detection probe into soil, and synchronously collecting soil multispectral reflection signals, moisture and actual measurement values of compactness of each detection point in situ; the method comprises the steps of respectively carrying out compensation calculation on a moisture absorption band and a scattering effect based on an actual measurement value to obtain a compensation coefficient and correcting an original spectrum signal, dynamically matching a target prediction model from a nitrogen content prediction model library divided by a pre-established moisture-compactness state interval according to a real-time moisture and compactness state, inputting corrected spectrum data into a corresponding model to obtain nitrogen content, and finally integrating position coordinates and nitrogen content of all detection points to generate a soil nitrogen content distribution map of a target detection area through a spatial interpolation algorithm. The invention realizes the rapid and accurate detection and visual space analysis of the soil nitrogen content in the field complex environment.

Inventors

CHEN WENTANG
CHEN XUE
SONG ZHIHUA
CHEN YINING
WANG BINGHAO

Assignees

山东深蓝智谱数字科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260403

Claims (10)

1. A rapid detection and analysis method for soil nutrients is characterized by comprising the following steps: Inserting a spectrum probe into each detection point at a preset depth and angle, and collecting a soil multispectral reflection signal, a soil moisture actual measurement value and a soil compactness actual measurement value; Based on the actual measurement value of soil moisture and the actual measurement value of soil compactness, respectively carrying out compensation calculation on a moisture absorption wave band and a scattering effect to obtain a moisture compensation coefficient and a compactness compensation coefficient, correcting a soil multispectral reflected signal, and generating corrected spectrum data of each detection point; According to the measured value of the soil moisture and the measured value of the soil compactness, matching a target prediction model corresponding to each detection point from nitrogen content prediction models corresponding to the pre-established moisture-compactness state interval; the corrected spectrum data of each detection point is input into a corresponding target prediction model to obtain corrected soil nitrogen content values of each detection point; and integrating the position coordinates of all detection points and the corresponding corrected soil nitrogen content values, and generating a soil nitrogen content distribution map of the target detection area through a spatial interpolation algorithm.
2. The method for rapidly detecting and analyzing soil nutrients according to claim 1, wherein the compensation calculation of the water absorption wave band comprises the following steps: extracting original reflectance values at each characteristic wavelength in a preset strong absorption band range of moisture from the soil multispectral reflected signals; Constructing a local reflectivity curve based on the original reflectivity values at each characteristic wavelength, and performing continuous system removal processing on the local reflectivity curve to obtain an actual measurement value of the soil moisture absorption depth; Substituting the actual measured value of the soil moisture absorption depth into a pre-calibrated relation model of the absorption depth and the moisture, and mapping to obtain a theoretical value of the soil moisture; and taking the ratio of the actual measured value of the soil moisture to the theoretical value of the soil moisture as a moisture compensation coefficient.
3. The method for rapidly detecting and analyzing soil nutrients according to claim 2, wherein the calculation of the actual measurement value of the soil moisture absorption depth comprises the following steps: the method comprises the steps of constructing a local spectrum curve of a moisture absorption characteristic by taking a moisture absorption center wavelength in original reflectivity values at each characteristic wavelength and reference shoulder wavelengths positioned at two sides of a spectrum of the moisture absorption center wavelength as an abscissa and taking corresponding original reflectivity values of the moisture absorption center wavelength as an ordinate; on the local reflectivity curve, connecting the wavelengths of each reference shoulder by straight lines to obtain a continuous system line of the moisture absorption characteristic; Obtaining a continuous system line reflectivity value at the wavelength of the moisture absorption center by carrying out interpolation calculation on the continuous system line; and calculating the ratio of the original reflectance value at the wavelength of the moisture absorption center to the reflectance value of the continuous system line, and taking the difference value between the value 1 and the ratio as an actual measurement value of the moisture absorption depth of the soil.
4. The method for rapidly detecting and analyzing soil nutrients according to claim 1, wherein the compensation calculation of the scattering effect comprises the following steps: Carrying out average value calculation on the actual measurement values of the soil compactness of each detection point to obtain an average value of the soil compactness; taking the ratio of the measured soil compactness value of each detection point to the average soil compactness value as an offset ratio; The negative exponential function of the offset ratio is taken as the compactness compensation coefficient.
5. The method for rapid detection and analysis of soil nutrients according to claim 1, wherein the generating of the corrected spectral data for each detection point comprises: based on the moisture compensation coefficient and the compactness compensation coefficient, calculating a comprehensive spectrum correction coefficient of each spectrum band in the soil multispectral reflected signal; and correcting the reflectivity value of each spectrum band covered by the soil multispectral reflected signal band by utilizing the comprehensive spectrum correction coefficient of each spectrum band, and generating corrected spectrum data.
6. The rapid detection and analysis method for soil nutrients of claim 1, wherein the establishment of the nitrogen content prediction model corresponding to the moisture-compactness state interval comprises the following steps: Preparing a soil sample set covering gradients of different moisture contents and soil compactness, and synchronously collecting soil multispectral reflection signals, actual measured values of soil moisture, actual measured values of soil compactness and standard values of soil nitrogen content of all samples in the soil sample set; dividing a soil sample into a plurality of moisture-compactness state intervals according to the distribution range of the measured value of the soil moisture and the measured value of the soil compactness, and determining a soil moisture range value and a soil compactness range value corresponding to each state interval; For each moisture-compactness state interval, extracting spectral data corrected by all samples falling into the state interval as an input variable, taking a soil nitrogen content standard value corresponding to the state interval as an output variable, and training by a partial least squares regression algorithm to obtain a nitrogen content predictor model of each moisture-compactness state interval; And integrating the nitrogen content predictor models corresponding to all the moisture-compactness state intervals and the soil moisture range values and the soil compactness range values corresponding to the moisture-compactness state intervals to form a nitrogen content prediction model library corresponding to the moisture-compactness state intervals.
7. The method for rapid detection and analysis of soil nutrients according to claim 6, wherein the calculation step of the nitrogen content predictor model comprises the following steps: Calculating pearson correlation coefficients between corrected reflectivity values of all samples in the state interval on each spectral band and soil nitrogen content standard values, sequencing all spectral bands from high to low according to absolute values of the correlation coefficients, and selecting a preset number of spectral bands before ranking as each prediction characteristic band; Extracting reflectivity values of each predicted characteristic wave band from corrected spectrum data of all samples, and constructing a characteristic spectrum data set according to the reflectivity values; randomly dividing samples in the characteristic spectrum data set into a training subset and a verification subset according to a preset proportion, further respectively calculating the mean value and standard deviation of each predicted characteristic wave band in the training subset, and carrying out standardized processing on the spectrum data of the training subset and the verification subset by utilizing the mean value and standard deviation; And training a partial least squares regression model by taking the standardized training subset spectrum data as input and the soil nitrogen content standard value as output to obtain the nitrogen content predictor model of each moisture-compactness state interval.
8. The rapid detection and analysis method for soil nutrients according to claim 1, wherein the matching of the target prediction model corresponding to each detection point comprises: acquiring intermediate values of a soil moisture range and a soil compactness range from each moisture-compactness state interval to jointly form a state interval representative point of each state interval; constructing a soil moisture actual measurement value and a soil compactness actual measurement value as position coordinates of each detection point; Calculating Euclidean distances from the position coordinates of each detection point to representative points of all state intervals, and judging the state interval with the minimum Euclidean distance with the detection point as a matching result; And taking the nitrogen content predictor model corresponding to the matching result as a target prediction model of each detection point.
9. The method for rapid detection and analysis of soil nutrients according to claim 1, wherein the generating of the soil nitrogen content profile of the target detection zone comprises: Dividing the grid of the target detection area, and acquiring the position coordinates of each grid point; Performing association and integration on the position coordinates of all detection points and the corresponding corrected soil nitrogen content values to form a space-attribute data set; searching all detection points located in a preset neighborhood range of each grid point in the space-attribute data set to serve as each effective neighborhood point; based on the position coordinates of all the effective neighborhood points and corrected soil nitrogen content values thereof, calculating to obtain soil nitrogen content estimated values of all grid points through a spatial interpolation algorithm; and generating a soil nitrogen content distribution map based on the soil nitrogen content estimated values of all grid points.
10. The method for rapidly detecting and analyzing soil nutrients according to claim 9, wherein the calculating the estimated value of the nitrogen content of the soil at each grid point comprises: Based on the position coordinates of each grid point and each effective neighborhood point, calculating the Euclidean distance between each grid point and each effective neighborhood point; Determining the distance contribution weight corresponding to each effective neighborhood point according to the Euclidean distance; multiplying the corrected soil nitrogen content value of each effective neighborhood point by the distance contribution weight of the corrected soil nitrogen content value, and summing all the product results to obtain the soil nitrogen content estimated value of each grid point.

Description

Quick detection and analysis method for soil nutrients Technical Field The invention belongs to the technical field of soil nutrient detection and analysis, and relates to a rapid detection and analysis method for soil nutrients. Background Soil nitrogen is a core nutrient element for maintaining crop growth and determining yield, and the rapid detection of the content of the soil nitrogen is of great significance for accurate fertilization. Although the traditional laboratory chemical analysis method has higher precision, the method has the limitation that large-area real-time monitoring is difficult to realize. The nondestructive detection technology based on visible-near infrared spectrum has become an important research direction by virtue of the advantages of rapidness, in situ and repeatability. However, the technology still faces a serious challenge in practical field application, namely, the real-time moisture of the soil can strongly absorb spectral signals in specific wave bands to mask nutrient characteristics. Meanwhile, the change of the soil compactness can change the scattering path and the reflection intensity of light. Currently, the industry has explored facilitated detection schemes for integrated optical sensors. For example, chinese patent application publication No. CN120594427a discloses a soil nutrient detection device for seedling planting, which is provided with a photodetection cylinder equipped with a detection sensor on a detection table and integrated with an automatic liquid preparation unit to be detected, and aims to improve convenience and functionality of equipment operation. This approach represents an attempt to advance to integrated, automated detection. However, the prior art scheme has the following two key defects, so that the actual requirements of in-situ and rapid detection in the field are difficult to meet, firstly, the detection core of the method relies on preparing a soil sample into a uniform liquid to be detected for optical measurement, the original moisture and compaction state of soil are separated, and further, the spectral interference caused by the change of the moisture and compaction degree of the soil in the field at any time cannot be dealt with, so that the problem of unstable precision of a detection model is caused when the detection model is applied to in-situ rapid detection. Secondly, the method mainly focuses on the detection automation and convenience of single-point samples, does not dynamically select an optimal prediction model according to the specific states of detection points, does not have the functions of integrating position coordinate information and nutrient content data of a plurality of detection points and generating a soil nitrogen content distribution map through a spatial interpolation algorithm, and further cannot realize crossing from single-point detection to planar distribution evaluation, so that rapid diagnosis of the nutrient space variation of a target detection area is difficult to meet. Therefore, a detection and analysis method capable of directly working in situ in the field, synchronously correcting the interference of moisture and compactness, adapting to different soil states and finally generating a visual nutrient space distribution map is needed to solve the technical problems. Disclosure of Invention In view of this, in order to solve the problems set forth in the background art, a method for rapidly detecting and analyzing soil nutrients is proposed. The invention provides a rapid detection and analysis method for soil nutrients, which comprises the steps of inserting a spectrum detection probe into soil at each detection point at a preset depth and angle, and correspondingly collecting a soil multispectral reflection signal, a soil moisture actual measurement value and a soil compactness actual measurement value. Based on the actual measurement value of soil moisture and the actual measurement value of soil compactness, the moisture absorption wave band and the scattering effect are respectively compensated and calculated to obtain a moisture compensation coefficient and a compactness compensation coefficient, and then the soil multispectral reflected signal is corrected to generate corrected spectrum data of each detection point. And matching a target prediction model corresponding to each detection point from nitrogen content prediction models corresponding to the pre-established moisture-compactness state interval according to the measured value of the soil moisture and the measured value of the soil compactness. And (3) inputting the corrected spectrum data of each detection point into a corresponding target prediction model to obtain corrected soil nitrogen content values of each detection point. And integrating the position coordinates of all detection points and the corresponding corrected soil nitrogen content values, and generating a soil nitrogen content distribution map of the target detect