Analysis of Main Influencing Factors and Modeling of Photosynthetic Rate for Cucumber at Initial Flowering Stage

zhang haihui,zhang zhen,zhang siwei,hu jin,xin pingping,wang zhiyong
DOI: https://doi.org/10.6041/j.issn.1000-1298.2017.06.031
2017-01-01
Abstract:Crop photosynthetic rate is under the influence of physiological and ecological interactions,which could impact plants ’ whole growth cycle.Aiming to demonstrate the main affecting factors of photosynthetic rate for cucumber at initial flowering stage and build a high-efficiency photosynthetic rate predicting model by combining the main factors with intelligence algorithm.Firstly,eight typical affecting factors were selected and a multi-factor coupling test was designed.Among the eight factors,photon flux density,temperature and CO2 concentration were set at 16,5,6 gradients,respectively.Under each gradients combination,the values of stomatal conductance,relative humidity and difference of vapour pressure were measured by gas analyzer Li-6400 XT.Besides,chlorophyll was measured by analyzer SPAD-502 Plus and nitrogen was measured by analyzer TYS-4N.Meanwhile,photosynthetic rate was measured by Li-6400 XT.Secondly,correlation analysis method was employed to find out the main affecting factors.Results showed that the five factors of photon flux density, CO; concentration,temperature,stomatal conductance and chlorophyll were correlated with photosynthetic rate of cucumber at initial flowering stage significantly.Then a combination algorithm of genetic algorithm and radial basis function neural network(GA-RBF) was adopted to build photosynthetic rate prediction model under these five main factors,while genetic algorithm(GA) was employed to optimize the propagation speed of radial basis function(RBF) neural network.Finally,XOR checkup method was used to analyze the prediction model performances with the five main affecting factors and the total eight factors.It showed that the model with five main factors had an obviously higher prediction accuracy than the one with eight factors,while the determination coefficient of photosynthetic rate between actually measured and calculated values reached 0.997 6,the maximum absolute error was 1.008 6 μmol/(m;·s),and the mean absolute error was 0.350 9 μmol/(m;·s).As a conclusion,the approach proposed for predicting photosynthetic rate of cucumber at initial flowering stage not only predigested model complexity but also improved the predictionaccuracy, which may hold potential applications for cucumber growth environment regulation in greenhouse.
What problem does this paper attempt to address?