Research On Supply And Demand Matching Model Of Agricultural Products Based On Deep Learning

In recent years,with the rapid development of artificial intelligence deep learning technology,its prediction research has attracted extensive attention.Deep learning has good generalization ability,and its semi-supervised learning mode can maximize the utilization of multi-feature data,it has superior performance in processing large-scale data.At present,the problems of "increasing production without increasing income" and "difficult to sell but expensive to buy" of agricultural products in China emerge one after another,the imbalance between supply and demand becomes more and more prominent,which directly affects the sustained growth of farmers’ income and the healthy development of agricultural products industry.Under the background of the current agricultural supply-side reform,as an important index reflecting the relationship between supply and demand of agricultural products,the changing trend of price is a key link to study the matching between supply and demand of agricultural products.Therefore,Spatio-temporal evolutionary association rule and deep learning algorithm are used to analyze and forecast the price trend,and a supply-demand matching model based on improved information is established to study the optimal supply-demand matching of agricultural products.The main work is as follows:(1)Considering that static association rules cannot reflect the periodicity and fluctuation of rules,Spatio-temporal evolutionary association rules are proposed,and Arc GIS is used to make Spatio-temporal evolution diagram to analyze the evolution of recommendation degree of strong association rules.The results show that spatio-temporal evolutionary association rules can improve the maximum recommendation degree of strong association rules to more than 80%.It can better reflect the dynamic evolution of agricultural prices in different time and space.(2)To realize automatic feature extraction of the model and consider the influence of long-term and short-term information on the prediction effect of the model,this paper combines CNN’s adaptive feature extraction function with LSTM’s sensitivity to time series data,establishes CNN-LSTM agricultural product price prediction model,and evaluates the model with three evaluation indexes.The example shows that this model has high prediction accuracy and stability,and its results are consistent with the results of Spatio-temporal evolutionary association rule.The changing trend of the relationship between supply and demand of agricultural products with time is analyzed.(3)To fully consider the interests of both sides of supply and demand,combined with fuzzy mathematics theory and information axiom,a matching model of supply and demand of agricultural products based on improved information quantity is proposed,and an enhanced information quantity calculation method is proposed from both quantitative and qualitative attributes,and an optimization model is established to maximize the total transaction matching degree between supply and demand sides.The example shows that the model can provide the optimal supply-demand matching scheme with the maximum matching logarithm.To sum up,this paper mainly explores the efficient and intelligent prediction algorithms and supply-demand matching recommendation model,which provides theoretical support for the production,sales,and allocation of some agricultural products,which not only further deepens the application of deep learning algorithm,but also provides a new idea for solving the supply-demand matching problem of agricultural products.