Estimation of Elution Induce Time and 80% Elution Time of Polymer-Coated Manure by a Neural Network and C^k Interpolation Scheme
- Development of a Neural Network Simulator for Structure-Activity Correlation of Molecules: Neco (5) -

Tomoko FUKUDA^a,b, Sumie TAJIMA^c, Hisato SAITOH^d, Umpei NAGASHIMA^e*, Haruo HOSOYA^c and Tomoo AOYAMA^f

(Received: September 19, 2000; Accepted for publication: December 6, 2000; Published on Web: January 31, 2001)

We present a method to estimate elution induced time and 80% elution time of polymer-coated manure using 1/f fluctuation analysis of near-infrared reflection spectrum and a perceptron type neural network and/or C^k interpolation scheme in order to reduce the evaluation time of elution induce time and 80% elution time. Usually, the evaluation time of elution induce time and elution speed is more than 20 days.
In the case of the neural network, relative error is almost less than 10%. Furthermore, the slopes of regression lines for the correlation between observed and calculated values are almost 0.9. For the C^k interpolation scheme, the error is a little bit larger than those of the neural network but less than 20%. This method is applicable for actual use because more than 10% error is included in the observed values. This method requires almost 10 minutes so that the evaluation cost of elution induce time and elution speed of polymer-coated manure will be significantly improved.

Keywords: Polymer-Coated Manure, Elution Induce Time, 80% Elution Time, Property Estimation, Neural Network, Multi-dimensional C^k Interpolation Scheme, 1/f Fluctuation

Text in Japanese

PDF file(365KB)