本文選題：深度學(xué)習(xí) + 精細化預(yù)測��； 參考：《哈爾濱工業(yè)大學(xué)》2017年碩士論文

【摘要】：伴隨著計算機技術(shù)的迅猛發(fā)展,深度學(xué)習(xí)開啟了人工智能新時代。以深度學(xué)習(xí)為代表,伴隨其在計算機視覺、語音識別、自然語言處理等領(lǐng)域取得的突破性進展,新技術(shù)創(chuàng)新帶來的不僅是挑戰(zhàn),同時也給氣象預(yù)測技術(shù)的發(fā)展帶來了機遇。課題針對氣象溫度進行時間序列建模,通過分析國內(nèi)外研究現(xiàn)狀及對時間序列預(yù)測模型的研究與對比,提出了改進深度學(xué)習(xí)框架來進行溫度時間序列預(yù)測的思路。考慮到普通神經(jīng)網(wǎng)絡(luò)中出現(xiàn)的天氣參數(shù)被認為是彼此獨立,時序關(guān)系一般不被考慮的缺點,在對氣象預(yù)測模型的構(gòu)建中,提出了通過滑動時間窗手段改造,讓普通神經(jīng)網(wǎng)絡(luò)也能學(xué)習(xí)到歷史時序特征。實驗表明,在深度前饋網(wǎng)絡(luò)中加入時序特征的天氣預(yù)報模型,效果要明顯優(yōu)于不考慮時序的模型。更進一步,針對實驗中暴露出的前饋神經(jīng)網(wǎng)絡(luò)預(yù)報準(zhǔn)確率隨著預(yù)報時間增長快速下降的問題,提出了通過改造循環(huán)神經(jīng)網(wǎng)絡(luò)(RNN)進行氣溫預(yù)測的方法,并采用專門解決普通循環(huán)神經(jīng)網(wǎng)絡(luò)長時依賴問題的長短時記憶網(wǎng)絡(luò)(LONG SHORT-TERM MEMEORY,LSTM)來構(gòu)建氣溫預(yù)測模型。本文在分析了循環(huán)神經(jīng)網(wǎng)絡(luò)、RNN-LSTM網(wǎng)絡(luò)、RNN-GRU網(wǎng)絡(luò)的基礎(chǔ)上,結(jié)合氣溫預(yù)測實驗?zāi)Ｐ椭谐霈F(xiàn)的過擬合、梯度消失與梯度爆炸等一系列問題,提出使用Re LU激活函數(shù)以及加入正則化手段改進等策略,通過優(yōu)化后的氣溫預(yù)測模型都較以往有更好的收斂效果。在實驗中,還包含了對氣象數(shù)據(jù)集的轉(zhuǎn)換、清洗、屬性選擇、特征提取等工作。在平臺應(yīng)用方面,將實驗搬到谷歌最新的深度學(xué)習(xí)框架TensorFlow-GPU中進行,使用GPU直接參與并行運算,為嘗試復(fù)雜深度模型實驗提供了可能。同時為驗證模型的效果,實驗不僅有對深度學(xué)習(xí)框架之間的比較,還加入了與傳統(tǒng)ARIMA模型的比較。本文提出深度學(xué)習(xí)技術(shù)在精細化氣溫預(yù)測的應(yīng)用研究,解決了一系列深度學(xué)習(xí)技術(shù)在氣象預(yù)測上的具體實現(xiàn)與運用問題,創(chuàng)新了氣溫預(yù)測時序分析方法,拓展了區(qū)域化天氣預(yù)報手段。
[Abstract]:With the rapid development of computer technology, deep learning has opened a new era of artificial intelligence. With the breakthrough in computer vision, speech recognition, natural language processing and so on, the new technology innovation brings not only challenges but also opportunities for the development of meteorological prediction technology. In this paper, time series modeling for meteorological temperature is carried out. Based on the analysis of the current research situation at home and abroad and the research and comparison of time series prediction model, an improved depth learning framework is put forward to predict the temperature time series. Considering the fact that the weather parameters in general neural networks are considered to be independent of each other and the time series relationships are not generally considered, in the construction of meteorological prediction models, a new method is proposed to modify the weather parameters by sliding time windows. So that ordinary neural networks can also learn the characteristics of historical time series. The experimental results show that the effect of the weather prediction model with time series features in the depth feedforward network is obviously better than that without considering the time series. Furthermore, aiming at the problem that the prediction accuracy of feedforward neural network is decreasing rapidly with the increase of forecast time, a method of temperature prediction by modifying the cyclic neural network (RNNN) is put forward. The long term memory network (LONG SHORT-TERM MEORYY LSTM), which is specially used to solve the problem of long time dependence of general circulatory neural networks, is used to construct the temperature prediction model. In this paper, based on the analysis of RNN-LSTM network and RNN-GRU network, a series of problems such as overfitting, gradient vanishing and gradient explosion in the experimental model of temperature prediction are discussed. By using re LU activation function and adding regularization method, the optimized temperature prediction model has better convergence effect than before. In the experiment, the transformation, cleaning, attribute selection and feature extraction of meteorological data sets are also included. In the aspect of platform application, the experiment is carried out in Google's latest depth learning framework (TensorFlow-GPU), and GPU is directly involved in parallel operation, which makes it possible to try out complex depth model experiment. In order to verify the effectiveness of the model, the experiment not only compares the depth learning framework, but also adds a comparison with the traditional ARIMA model. In this paper, the application research of depth learning technology in fine temperature prediction is proposed, which solves a series of problems in the realization and application of depth learning technology in meteorological prediction, and innovates the analysis method of temperature forecasting time series. Regional weather forecast methods have been expanded.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：P45;TP18

【參考文獻】

相關(guān)期刊論文 前10條

1 黃建風(fēng);陸文聰;;基于小波-NAR神經(jīng)網(wǎng)絡(luò)的氣象要素時間序列預(yù)測與天氣指數(shù)彩虹期權(quán)估值[J];系統(tǒng)工程理論與實踐;2016年05期

2 王定成;曹智麗;陳北京;倪郁佳;;日氣溫多元時間序列局部支持向量回歸預(yù)測[J];系統(tǒng)仿真學(xué)報;2016年03期

3 李軍;;氣候變化影響人類健康[J];中華環(huán)境;2016年Z1期

4 白致威;張雷;王杰;黃英;;基于ARIMA模型的云南氣象干旱預(yù)測研究[J];人民長江;2015年15期

5 智協(xié)飛;李剛;彭婷;;基于貝葉斯理論的單站地面氣溫的概率預(yù)報研究[J];大氣科學(xué)學(xué)報;2014年06期

6 李澤椿;畢寶貴;金榮花;徐枝芳;薛峰;;近10年中國現(xiàn)代天氣預(yù)報的發(fā)展與應(yīng)用[J];氣象學(xué)報;2014年06期

7 智協(xié)飛;彭婷;李剛;王佳;王晶;;多模式集成的概率天氣預(yù)報和氣候預(yù)測研究進展[J];大氣科學(xué)學(xué)報;2014年02期

8 段文廣;周曉軍;石永煒;;數(shù)據(jù)挖掘技術(shù)在精細化溫度預(yù)報中的應(yīng)用[J];干旱氣象;2012年01期

9 白玉潔;;改進時間序列模型在降雨量預(yù)測中的應(yīng)用研究[J];計算機仿真;2011年10期

10 胡邦輝;袁野;王學(xué)忠;叢愛麗;;基于貝葉斯分類方法的雷暴預(yù)報[J];解放軍理工大學(xué)學(xué)報(自然科學(xué)版);2010年05期

，

本文編號：1932438