Design Of An Integrated Water And Fertilizer Intelligent Model That Is Adaptive To Environmental Changes

As a big agricultural country,China owns plenty of agricultural monitoring data that is rarely utilized.With the development of information technology and the society,data analysis and data mining are paid more attention by people,making use of the agricultural data to develop modern,smart,and technological agriculture.The accuracy and monitoring network distribution of the agricultural data has a significant impact on the performance of the integrated smart irrigation and fertilizing system.Currently,monitoring network comprised of cheap and low accuracy sensors are used in the agricultural field,which may cost a lot of effort to correct once sensor failure happens and cause significant consequence on the accurate control of the system.In this thesis,a solution was made to the self-adaption of the integrated smart system by analyzing the data collected from accurate sensors.Based on the analysis,I found that the yearly distribution of difference between daily highest temperatures almost follows the normal distribution.Based on this feature,feature extraction,smoothing was performed,and then the same pattern was found.Therefore an algorithm was designed for the error analysis and accuracy improvement of agricultural sensor network,as a supplement of the self-adaptive integrated model and check of the data quality.The feasibility and efficiency analysis where 89.13% accuracy and at least 0.3? accuracy improvement Ire observed were conducted.The model is based on conventional specialist system and Long-Short Term Memory(LSTM).The former performs as basic control while our algorithm offers additional information as a secondary specialist system to adjust the basic conventional specialist system,to reach the self-adaption of the system according to the change of environment condition.