论文部分内容阅读
Greenhouses are productive systems characterized by an intensive but efficient use of primary resources.As one of the major greenhouse crops, the yield and quality of tomato is significantly affected by photosynthesis.It is well known that CO2 enrichment increases the net photosynthetic rate in plant.How to establish optimal model of photosynthetic rate, analyze the association between CO2 and it, accurately manage CO2 concentration have become urgent problems in the field of crop cultivation.To solve the above requirements, the optimal regulation model of CO2 concentration in the seedling stage of tomato by applying Support Vector Machine (SVM) associated with Partial Least Squares (PLS) was proposed in the paper.Firstly, the environmental parameters affecting the growth of tomato, including air temperature, air humidity,photosynthetically active radiation (PAR), CO2 concentration, soil temperature and moisture, were real-time monitored by the wireless sensor nodes while the plants were growing.Theindices of tomatos growth dynamics were obtained by artificial.The photosynthetic rate of tomato was measured by the LI-6400XT.Secondly, the correlationdegree between the environmental influences parameters and photosynthetic rate, and indices of tomatos growth dynamics were calculated by grey relational analysis.The selected main influences parameters were taken as the input of SVM while the output was photosynthetic rate of tomato in the seedling stage.Then, in order to improve the prediction accuracy of the photosynthesis prediction model based on SVM, an improved particle swarm optimization (PSO) was designed to optimize the parameters of the kernel function (g) and penalty factor (C) of the SVM model.The association between CO2 concentration and photosynthetic rate was predicted by the established model, which the CO2 saturation point under different environment conditions was obtained.Lastly, regulation model of CO2 concentration was built by processing multivariate nonlinear regression of the corresponding environment conditions, which achieved by PLS combined with fixed expressions of the environment parameters.The result shows that the correlation coefficient of photosynthesis prediction model was 0.948 and root-mean-square error was 1.196, which proved that the model had a high accuracy and could provided reliable basis for findingCO2 saturation point.Through comparison between simulated values and observed CO2 saturation points, the correlation coefficient of regulation model of CO2 concentration was 0.741, the slop of the fitted line was 1.19,which indicated that the values had good correlation and similarity, meanwhile, CO2 concentration needed for plants growth could be calculatedin a dynamic greenhouse environment.The conclusion provides a theoretical basis for the optimal regulation of an important raw material of tomato photosynthesis.