畢業論文

打賞
當前位置: 畢業論文 > 計算機論文 >

java遞歸神經網絡大棚溫室短序列預測+源代碼

時間:2019-04-16 20:29來源:畢業論文
自動化溫室的傳統的環境因素調控方式是通過復雜的線性函數模型,而采用神經網絡對于各個環境因素實現預估從而實現調控相較于傳統方式更加方便快捷。神經網絡中的遞歸神經網絡

摘要:隨著社會的進步發展,傳統的農業生產模式愈發不能滿足人們的需求。新型農業設施—溫室大棚,有著不限于時間空間的優點,廣泛的應用于各種環境下的農業生產。而相較于傳統溫室,自動化溫室有著無可比擬的優勢。自動化溫室的傳統的環境因素調控方式是通過復雜的線性函數模型,而采用神經網絡對于各個環境因素實現預估從而實現調控相較于傳統方式更加方便快捷。神經網絡中的遞歸神經網絡在時間序列的預測上有著優異的效果。普通的遞歸神經網絡對于長期依賴的信息沒有良好的學習效果,LSTM網絡是一種改進的遞歸神經網絡模型,能有效的解決這一問題。在自動化溫室的復雜環境下,采用LSTM能夠有較好的預測效果。本文工作主要包括一下三個方面:(1) 本文開始是數據的特征處理,以及數據的不同處理方式。(2)其次對與網絡模型的搭建,通過對溫室環境因素的學習預測,測試模型中不同的參數,使之對比調整使之達到最佳的學習效果。(3)最后是圖形界面的編寫,實現一個可視化界面,更加便于展示操作。34610
畢業論文關鍵詞:自動化溫室;遞歸神經網絡;長短記憶網絡;時間序列
Based on the recursive neural network prediction of short series for greenhouse
Abstract: With the progress of the society development, the traditional mode of agricultural production can't meet the needs of people. New type of facilities agricultural—greenhouses, has a merit of that is not limited to time or space. It is widely used in agricultural production in various kinds of environments. Compared with traditional greenhouse, automated greenhouse has incomparable advantages. Tranditional automated greenhouse controls the environmental factors by complex linear function model. Neural network estimates all of the environmental factors, then realizes the control. So neural network is easier and faster than the tranditional way. Recursive neural network has excellent effect in time series prediction in the numerous neural network. While the ordinary recursive neural network is not good in learning effect for long-term dependence on information, LSTM - an improved recursion network model, can effectively solve the problem. So in case of the automated greenhouse under the complex environment factors, adopt LSTM can have better prediction effect.This paper mainly includes three aspects:(1) Firstly, it’s the characteristics of the data processing and different approaches to process data.(2) Secondly, constructs the network model and test different parameters in the model based on the learning and calculating to the environmental factors of the greenhouse. And makes a contrast adjustment to achieve the best learning effect. (3) Finally, writes a graphical interface, makes it easy for operation by a visual interface.
源Z自-六+維L論W文W網^www.aftnzs.live

Key words: Automated Greenhouse ;Recurrent Neural Network ;Long Short-Term Memory ;Time Series
目 錄
摘要:    1
關鍵詞:    1
Abstract:    1
Key words:    1
1 緒論    2
1.1自動化溫室研究背景及現狀    2
1.2遞歸神經網絡研究背景及現狀    2
1.3 研究內容和技術路線    3
1.3.1研究內容    3
1.3.2 技術路線    4
2 遞歸神經網絡結構及相關函數介紹    4
2.1 RNN    4
2.2 LSTM    7
2.3 激活函數    9
2.3.1 Sigmoid 函數    9
2.3.1 Tanh 函數    10
2.3.2 ReLu 函數    11
2.4 優化函數    11
2.4.1 SGD    12
2.4.2 RMSprop    12
2.4.3 Adam    12 java遞歸神經網絡大棚溫室短序列預測+源代碼:http://www.aftnzs.live/jisuanjilunwen/20190416/32168.html
------分隔線----------------------------
推薦內容
双色球走势图带连线