Research On Nutrient Element Detection In Compound Fertilizer Based On LIBS Technology

Nitrogen,phosphorus and potassium in compound fertilizers are important nutrients required by crops.Insufficient or excessive nutrients in products will affect the growth and development of crops and pollute the soil.Therefore,it is of great significance to ensure the quality of compound fertilizer products.In the production process of compound fertilizers,the traditional detection method can not perform real-time online quality monitoring.However,LIBS technology has the advantages of multi-element online real-time detection and sensitive detection.Therefore,this paper uses LIBS technology to the main the nutrient elements nitrogen,phosphorus and potassium were analyzed qualitatively and quantitatively.This paper analyzes the factors that affect the measurement of nutrients in inorganic compound fertilizer by LIBS,including laser energy,ICCD delay,laser repetition frequency and compactness of inorganic compound fertilizer samples.The best experimental conditions for quantitative analysis of nitrogen,phosphorus and potassium are as follows: laser energy is 50 m J,70 m J and 50 m J,respectively;ICCD delay is 0.8 ?s,1.0 ?s and 1.8 ?s,respectively;laser repetition frequency is 5 Hz,4 Hz,4 Hz,respectively;the pressure acting on the sample is 10 MPa,15 MPa,10 MPa,respectively.It is also calculated that the electron number density of the inorganic compound fertilizer plasma is 6.386 ? 1017 cm-3 and the plasma electron temperature is 15353.51 K.The experiment shows that the local thermodynamic equilibrium condition is satisfied.Quantitative analysis of nitrogen,phosphorus and potassium elements using traditional external standard method,Obtained calibration curves for N I 742.4 nm,N I 744.2 nm and N I 746.8 nm,with R2 of 0.86642,0.88832,0.85015,and the calibration curves for K I 766.5 nm and K I 769.9 nm have R2 of 0.82251 and 0.83307 respectively.The calibration curves for P I 255.3 nm have R2 of 0.92213.According to the calibration model established by the traditional external standard method,N I 744.2 nm,P I 255.3 nm,and K I 769.9 nm with good fit are used as analytical lines.The relative error between the predicted concentration and the true concentration is-1.55%,16.2 %,1.39%,respectively.Using the Lorentz function to fit the line shape of the plasma line,the relative errors of nitrogen,phosphorus,and potassium were-1.20%,14% and 4.74%,respectively;R2 was enhanced by 0.00338,0.01937 and 0.04083,which were all improved.Finally,a support vector regression method is used to establish a model.For different nutritional elements in inorganic compound fertilizers,select the entire peak data of the analysis lines N I 744.2 nm,P I 255.3 nm,K I 769.9 nm,and the data is normalized to construct SVR training sets.The grid search method is used to obtain regression models of the optimal penalty factors C in the nitrogen,phosphorus and potassium.The optimal penalty factors C is 1,1.4142 and 8 respectively,and the kernel parameter g is 0.0625.The MSE of the training model is 1.960%,0.043% and 2.335% respectively.R2 is 0.97307,0.99963 and 0.96872 respectively.And the relative errors of the test set samples in the regression model is-1.18%,1.42% and 1.31% respectively.The experiment proves that LIBS has certain feasibility for real-time analysis of nutrients of inorganic compound fertilizer.