Rapid detection method for organic matters in soil based on infrared spectroscopic intelligent identification model

Abstract

The invention provides a rapid detection method for organic matters in soil based on an infrared spectroscopic intelligent identification model. The method comprises the following steps: acquiring a soil sample; acquiring mid-infrared photoacoustic spectral information and carrying out continuous multitime scanning to obtain average spectrum; analyzing the content of organic matters in a soil sample by using a chemical process; preprocessing spectral information; acquiring mid-infrared photoacoustic spectral information of a to-be-detected soil sample; taking average spectrum and carrying out preprocessing; determining the chemical reference value of the to-be-detected soil sample; calculating mahalanobis distance between the to-be-detected soil sample and training samples and establishing a new sequence matrix according to a sequence from smaller distance to bigger distance; establishing a prediction model with different modeling numbers from the new sequence matrix according to the characteristics of the to-be-detected soil sample; and evaluating the parameters of the model, selecting an optimal modeling number and carrying out optimization so as to obtain the intelligent soil identification model. According to the invention, the intelligent identification model according with the spectral characteristics of the spectrum of each sample is established and one model corresponds to one sample, so background interference and variability effect of soil can be effectively reduced; and the intelligent identification model has good universality, is stable and effectively improves prediction accuracy.

Description

technical field [0001] The invention belongs to the field of agricultural informatization, and specifically relates to a fast, accurate and nondestructive determination method for detecting soil organic matter content by establishing an intelligent identification model based on mid-infrared photoacoustic spectroscopy. Background technique [0002] Soil organic matter is an important part of soil and an important indicator of farmland soil fertility and soil quality. Rapid detection of soil organic matter content is of great significance for precise fertilization in farmland. [0003] At present, the traditional determination of soil organic matter content generally adopts potassium dichromate volumetric method, hydration thermal potassium dichromate oxidation-colorimetric method, etc. These methods require pretreatment of samples, complicated operation, high cost, long test period, and It requires a large amount of chemical reagents, is easy to pollute the environment, and i...

AI Technical Summary

Problems solved by technology

[0005] Combining infrared spectroscopy with chemometric methods to construct a prediction model can realize the prediction of soil organic matter content. This type of model is based on a given sample for model construction. The model is single and stable, and the calibration set and verification set are relatively fixed.

However, the soil structure is complex, the background interference is large, and the soil variation coefficient is large. The soil at different sampling points has different characteristics and is easily affected by background interference. Too few or too many modeling samples will affect the accuracy and universality of the model.

Images