The Research And Application Of Horticultural Crop Growth Monitoring Technology For Sustainable Agriculture

With the rapid development of science and technology,agricultural information technology has gradually become an important symbol of the new agricultural science and technology revolution.The agricultural intelligent system has become an important means to enrich agricultural expertise and disseminate agricultural information,promote the rapid development of agriculture and the realization of agricultural informatization and modernization.This thesis is mainly devoted to the research of horticultural crop growth monitoring technology,applying the concept of agricultural sustainability to the system,and taking watermelon as an example to provide a sustainable concept-driven planting decision support program.The main research work is as follows:1)In this paper,probability information gain algorithm is used to determine the importance of disease indicators and rank them according to the influence of specific indicators based on the entropy value.By using the information gain method,the gain brought by each index to the certainty of classification is quantified separately.Once a certain feature is determined,the system will be in a new state,and it is necessary to use the deep iteration method and introduce probability calculation to sort the next important feature.Finally,using a real data set,the information gains of features "abnormality","color","organ",and "shape" are calculated by the proposed method.The results show that the information gains decrease in sequence.2)This paper uses a directed acyclic graph data structure to implement an efficient disease classifier.Using the directed edges and landmarks of the graph,combined with the probability information gain algorithm,the hierarchical structure of the graph and the category of each layer are determined.Each feature and disease need only be stored once as a node of the graph,and the relationship between the feature and the disease is expressed by directed edges.A piece of data in the data set is constructed as a path in the graph,and the final disease classification is completed by means of graph search.Aiming at the problem of incomplete features,the thesis also designs a search graph.The search graph combines the structure of a linked list and a graph to provide possible disease classifications in the absence of features.3)This paper proposes a watermelon growth model which includes the growth and development characteristics of the unique period,as well as the growth and development curves of different organs.Once anomaly occurs during the growth and development of watermelon,an intelligent irrigation and fertilization machine can be turned on or off based on current growth environment.In the first three periods of watermelon growth,the growth state was judged mainly on the characteristics of watermelon growth and development.From the fruit setting period,the fruits and main vines develop rapidly and have continuous growth data.Based on the collected data,a polynomial regression fitting model is used to construct a watermelon growth curve including environmental factors.Finally,this paper optimizes the parameters of the proposed model using gradient descent method to train and obtain a set of optimum parameters.