NPK Prediction Of Sugarcane Leaves With Spectral Reflectance And Development Of Related Sensor

In the sugarcane planting in Guangxi,excessive and improper use of chemical fertilizers has caused problems such as low international competitiveness and environmental pollution.In order to understand the growth status of sugarcane and improve the utilization rate of chemical fertilizer,this paper established a prediction model of nitrogen(N),phosphorus(P)and potassium(K)content in sugarcane leaves based on the spectroscopy method,and further considered the inter-element The interaction to improve the prediction accuracy of N,P,K content.In this paper,sugarcane was used as the research object.Through the measurement of spectral reflectance(220-1400 nm)of leaves in different growth stages,the feasibility of establishing a prediction model of leaf N,P and K content by various spectral analysis methods was explored.Firstly,the second-order differential(2nd Der)transform is performed on the full-spectrum data,and the feature variables are filtered by CARS-PCA algorithm.The number of characteristic variables of N,P and K is 34,32 and 39 respectively.The 6-fold cross-predicted PLS model for the nitrogen,phosphorus and potassium contents of sugarcane leaves was established by using the characteristic variables.The prediction results of PLS model were 0.8591,0.6769 and 0.9321,respectively.Considering the possible interaction effects of N,P and K in the growth of sugarcane,this paper also introduces a three-regression model with 19 factors to reflect the interaction between the three factors.The PMA and MLR correction prediction models of N,P and K were established by using PCA to extract the 19IF-PCA principal component factors.The prediction coefficients of the MLR correction model of N,P and K content were 0.8908,0.8019 and 0.9439,the prediction accuracy is significantly improved.In addition,based on the principle of spectroscopy,a portable crop growth sensor is designed and developed to realize the nondestructive and rapid detection of crops.