Anomaly Analysis And Prediction Of Grain Situation Based On Isolated Forest And LSTM

Since ancient times,grain has been a strategic material to maintain social stability and promote people’s living and working in peace and contentment.China is a population superpower,and the food consumption that comes with it is huge.At present,with the economic development of our country and the improvement of People’s living standards,People’s demand for grain is gradually changing from grain quantity to grain quality.Therefore,the quantity and quality safety of grain is directly related to the vital interests of consumers.A complete grain production system and gradually improving food security can also promote the country’s rapid economic development and social harmony and stability.Because of this,the introduction of big data analysis in the field of food is of great significance.However,the data of the grain industry is growing rapidly and its structure is complex,so it is difficult to get and arrange the data.Based on this,this paper plans to construct the data visualization scheme,which aims at the algorithm support needed in the construction process,carries on the thorough exploration,the concrete work is as follows:First,food data collection is difficult.Because of the importance of food for the country,data on the food sector can not be fully disclosed.In this paper,researchers cooperate with the relevant enterprises,relevant government departments to communicate,in the case of permission,access to part of the research data.In collaboration with other researchers,to get their data.At the same time,to get the public data on the network,the data is crawled on the network based on Scrapy,and the data is merged,sorted,and stored.Secondly,the anomaly detection of the big data of grain.One of the research points of this paper is the anomaly detection of grain data.Because the grain data is huge,and the influence factor is complex.In the process of anomaly detection,we need to choose a fast and efficient algorithm.After comparing different algorithms,this paper chooses the iForest algorithm as the basic algorithm and proposes the improvement measures of multi-algorithm fusion according to the shortcomings of the iForest algorithm.To a certain extent,the accuracy of classification is improved.After the comparison of Precision,Recall,and F1,the effectiveness of the proposed anomaly detection algorithm is verified.Thirdly,the forecast of grain temperature is based on grain situation data.The other research point of this paper is the grain temperature forecast.The grain temperature in the granary can reflect the grain condition to a great extent.Therefore,the forecast of grain temperature is very important in the analysis of grain data.In this paper,for the current prediction algorithm,after comparing different algorithms,choose to LSTM model as the basis,after analyzing its advantages and disadvantages,given its shortcomings,using multi-model fusion algorithm to improve,after the experiment in the multi-data set,the validity of the proposed multi-modal fusion algorithm is verified by the evaluation of RMSE and Mae.Fourth,develop a data visualization system.In this paper,after the completion of the above algorithm research,the results of the above algorithms are applied to the data visualization system,the completion of the system development.