Research On Decision-Making Method For Greenhouse Cucumber Fertilization Based On DEA And GA-BP Method

Rational application of fertilizer is vital for the greenhouse cucumber to have a good harvest. In addition, it also helps improve product quality, maintain and increase soil fertility and protect the ecological environment. Based on the current circumstances such as excessive fertilization of greenhouse cucumber, lower fertilizer efficiency, lack of quantitative decision of scientific fertilization method and so forth, this research studied fertilization mode of greenhouse cucumber, and explored the fertilization decision method based on GA-BP model through studying fertilization efficiency. This project is to enrich theory and method of greenhouse cucumber fertilization decision, and provide a scientific reference for fertilization. The contents and conclusions are as follows:(1) Use the data envelopment analysis method to study the greenhouse cucumber fertilization efficiency. Analyze and compare fertilizer efficiency among traditional fertilization method, target yield fertilization method and mathematical model fertilization method through the following aspects: the relative efficiency, technical efficiency, scale efficiency and redundancy rate. Based on the existing data, this research had two vital conclusions: different fertilizer application methods affected fertilization efficiency a lot; the mathematical model fertilization method had a higher efficiency than the target yield fertilizing method which had a litter higher efficiency than the traditional fertilization method.(2) The data extraction method of fertilization-output based on DEA model was studied. By applying the'projection principle'of DEA model, eliminating the redundancy of original fertilizer quantity, and getting the corresponding fertilizing amount of N, P₂O₅ and K₂O, this article obtained the effective data which was necessary in establishing the mathematical model of greenhouse cucumber fertilization, from the perspective of optimizing input-output efficiency. 15 samples data from five years of continuous cropping were extracted in the article, and 27 samples data of traditional fertilization method, target yield fertilizing method and mathematical model fertilization method were extracted here. (3) The greenhouse cucumber fertilization mode was established. On the basis of acquiring the valid data of fertilization, use the regression analysis method to construct the greenhouse cucumber fertilization mathematical model. Consider the continuous cropping time and fertilizing method: a. the mathematical model between the output of cucumber of continuous cropping time of 5 years and the amount of the fertilizer of N, P₂O₅ and K₂O was constructed. b. the mathematical model between the amount of the fertilizer of N, P₂O₅ and K₂O and the output of cucumber of traditional fertilization method, target yield fertilizing method and mathematical model fertilization method was constructed. In addition, this article discussed the effect from N, P₂O₅ and K₂O. This provided decision-making basis for production practice.(4) This article studied the fertilization decision-making method of greenhouse cucumber based on GA-BP. To solve the disadvantages of traditional model, such as uncertain model parameters, long test cycle, subjectivity, heavy workload and so forth, applying GA to optimize the weights and threshold of BP, the fertilizer recommendation model was constructed based on GA-BP in which by inputting greenhouse plot nutrient content and target yield, the amount of fertilization of N, P₂O₅ and K₂O would be outputted. The result showed: N, P₂O₅ and K₂O forecasting mean error of the recommendation model were 3.64%, 8.72% and 6.48% respectively, and the forecasting error of GA-BP model was much smaller than the BP network.