工业应用中的智能方法与大数据分析

需积分: 9 127 浏览量更新于2024-07-18 收藏 13.42MB PDF 举报

"这本电子书《Intelligent Methods and Big Data in Industrial Applications》深入探讨了智能方法和大数据在工业应用中的前沿理论与实践。由多位知名专家编辑，旨在揭示大数据在工程、计算机科学、物理学、经济学和生命科学等多个领域的理论、研究、开发和应用。书中涵盖了从传感器、数字源、社交媒体到互联网交易等各种来源产生的大规模、复杂和分布式数据集的分析与理解。内容涉及计算智能，如神经网络和进化计算等技术。" 本书是"Studies in Big Data"系列的第40卷，由Janusz Kacprzyk教授担任系列编辑，他来自波兰科学院。该系列致力于快速、高质量地发布大数据领域的新进展。内容不仅限于理论探讨，还包含了实际应用的案例，特别是如何将智能方法（如机器学习、人工智能算法）与大数据相结合，解决工业环境中的复杂问题。书中讨论的主题广泛，可能包括但不限于以下几个方面： 1. **大数据处理框架**：介绍如Hadoop、Spark等大数据处理平台，以及它们在处理海量数据时的优势和挑战。 2. **智能算法**：阐述深度学习、神经网络、支持向量机、决策树等机器学习算法在大数据分析中的应用，以及如何优化这些算法以适应大数据的特性。 3. **实时与流式数据分析**：探讨如何在数据流不断生成的情况下进行实时分析，以支持动态决策和预测。 4. **大数据可视化**：讲解如何有效地可视化大数据，帮助决策者理解复杂的数据模式和趋势。 5. **数据挖掘与模式识别**：探讨从大数据中发现有价值信息的技术，包括聚类、分类和关联规则挖掘等。 6. **工业物联网（IoT）与大数据**：讨论物联网设备产生的大量数据如何与智能分析结合，提升工业生产效率和智能化水平。 7. **大数据安全与隐私保护**：介绍在大数据环境下确保数据安全和用户隐私的方法，包括加密、匿名化和访问控制策略。 8. **案例研究**：提供工业领域的具体应用案例，展示智能方法和大数据技术如何解决实际问题，提升业务绩效。 9. **未来趋势与挑战**：展望大数据和智能方法的发展方向，分析面临的挑战，如数据质量问题、算法解释性和能源效率等。通过阅读这本书，读者将能够了解到当前智能方法和大数据的最新研究成果，以及如何将这些技术应用于实际工业环境中，以实现更高效、更智能的运营。无论是研究人员、工程师还是企业决策者，都能从中获取宝贵的见解和灵感。

Part I

Artificial Intelligence Applications

Nonlinear Forecasting of Energy Futures

Germán G. Creamer

Abstract This paper proposes the use of the Brownian distance correlation for

feature selection and for conducting a lead-lag analysis of energy time series. Brow-

nian distance correlation determines relationships similar to those identiﬁed by the

linear Granger causality test, and it also uncovers additional non-linear relationships

among the log return of oil, coal, and natural gas. When these linear and non-linear

relationships are used to forecast the direction of energy futures log return with

a non-linear classiﬁcation method such as support vector machine, the forecast of

energy futures log return improve when compared to a forecast based only on Granger

causality.

Keywords Financial forecasting

·Lead-lag relationship ·Non-linear correlation

Energy ﬁnance

·Support vector machine ·Artiﬁcial agents

1 Introduction

The major contaminant effects of coal and the reduction of natural gas prices since

2005 have led to a contraction in the proportion of coal and an increase in the share

of natural gas used in the production of electricity in the US since the year 2000

(see Fig. 1). According to the US Energy Information Administration [25], natural

gas and coal will account for 43 and 27% of total electricity generation in 2040,

respectively. The share of oil on electricity generation has also decreased since 2000

in the US; however, it is still important at the world level, where it accounts for about

5% [24]. These change of inputs by electric power plants, due to environmental,

political or market considerations, may indicate that several fossil fuel prices are

mutually determined or that one price depends on another one.

Mohammadi [20] ﬁnds that in the case of the US, oil and natural gas prices are

globally and regionally determined, respectively, and coal prices are deﬁned by long-

term contracts. Mohammadi [19], using cointegration analysis, exposes a strong rela-

tionship between electricity and coal prices and an insigniﬁcant relationship between

G. G. Creamer (

)

School of Business, Stevens Institute of Technology, Hoboken, NJ 07030, USA

e-mail: gcreamer@stevens.edu

R. Bembenik et al. (eds.), Intelligent Methods and Big Data in Industrial

Applications, Studies in Big Data 40, https://doi.org/10.1007/978-3-319-77604-0_1

4 G. G. Creamer

Fig. 1 Evolution of net electricity generation by energy source. Source: US Energy Information

Administration

electricity and oil and/or natural gas prices. Asche et al. [3] and Bachmeier and Grifﬁn

[6] ﬁnd very weak linkages among oil, coal and natural gas prices using cointegra-

tion analysis, while crude oil and several reﬁned product prices are integrated [3].

Hartley et al. [12] notice an indirect relationship between natural gas and oil prices.

Furthermore, Aruga and Managi [2] detect a weak market integration among a large

group of energy products: WTI oil, Brent oil, gasoline, heating oil, coal, natural gas,

and ethanol futures prices.

Mjelde and Bessler [18] observe that oil, coal, natural gas, and uranium markets

are not fully cointegrated. Asche et al. [5] indicate that the U. K. energy market

between 1995 and 1998 was highly integrated when the demand was for energy

rather than for a particular source of energy. Brown and Yucel [8] show that oil and

natural gas prices have been independent since 2000; however, when weather and

inventories are taken into consideration in an error correction model, crude oil prices

have an effect on natural gas prices. Similar results are obtained by Ramberg [21]

using cointegration analysis. Amavilah [1] observes that oil prices inﬂuence uranium

prices.

Causality analysis is also used to evaluate the relationship between spot and future

commodity prices. Asche et al. [4]–using a non-linear Granger causality test–shows

that neither the futures nor the spot crude oil market leads the relationship.

Most of the s tudies mentioned are based on cointegration analysis and Granger

causality; however, none of these studies have used a non-linear correlation measure

to evaluate the lead-lag relationship among the fossil fuels.

In this paper, I propose to use the Brownian distance correlation to conduct a non-

linear lead-lag dependence analysis of coal, oil and gas futures log return. I also test if

there is any improvement in the forecast of these energy futures using these non-linear

dependencies compared to a forecast based only on linear relationships such as those

identiﬁed by the Granger causality test. Section 2 introduces the different methods

explored in this study; Sect. 3 presents the data used; Sect. 4 explains in detail the

6 G. G. Creamer

Székely and Rizzo [22] also proposed the Brownian distance covariance, which

captures the covariance on a stochastic process. Distance covariance between the

random vectors X and Y measures the distance between f

and f

X,Y

and is

obtained as:

ν(X, Y ) =



 f

X,Y

(t, s) − f

(t) f

(s)

(3)

where . is the norm, t and s are vectors, f

and f

are the characteristic functions

of X and Y respectively, and f

X,Y

is the joint characteristic function of X and Y.

Empirically, ν(X, Y ) evaluates the null hypothesis of independence H

: f

X,Y

versus the alternative hypothesis H

: f

= f

X,Y

. In this paper, we refer to

this test as the distance covariance test of independence.

Likewise, distance variance is:

ν(X ) =



 f

X,X

(t, s) − f

(t) f

(s)

(4)

Once distance covariance is deﬁned, I obtain the distance correlation R(X, Y )

from the following expression:



(X,Y )

√

(X)ν

(Y ))

,ν

(X)ν

(Y )>0

0,ν

(X)ν

(Y ) = 0

(5)

Distance correlation takes a value of zero in case of independence and one when

there is complete dependence.

In general, this research proposes the evaluation of the non-linear dependence

of any ﬁnancial time series such as the current value of Y (Y

)onthel lagged

value of X (X

t−l

) with the Brownian distance correlation R(X

t−l

, Y

). In par-

ticular, I wish to explore the lead-lag relationship among the time series under

study. If R(X

t−l

, Y

) = 0 and l > 0, then X

t−l

leads the series Y

. Additionally, if

R(X

t−l

, Y

) = 0, R(X

, Y

t−l

) = 0 and l > 0, then there is an unidirectional relation-

ship from X

t−l

to Y

. However, if R(X

t−l

, Y

) = 0, R(X

, Y

t−l

) = 0 and l > 0, then

there is a feedback relationship between X and Y . On the contrary, if R(X

t−l

, Y

) = 0

and R(X

, Y

t−l

) = 0 then there is no lead lag relationship between X and Y [26].

2.3 Support Vector Machine

Support vector machine was proposed by Vapnik [27] as a classiﬁcation method

based on the use of kernels to pre-process data in a higher dimension than the original

space. This transformation allows an optimal hyperplane to separate the data X in

two categories or values.

A hyperplane is deﬁned as:



X : F (X )

= X

β + β



剩余369页未读，继续阅读

J-10

粉丝: 18
资源: 483

工业应用中的智能方法与大数据分析

Security with Intelligent Computing and Big-data Services-Springer(2018).pdf

英文原版-The Intelligent Enterprise in the Era of Big Data 1st Edition

Big Data and Computational Intelligence in Networking-CRC(2018).pdf

英语2200词关于人工智能和大数据的发展前景，看法

驾校管理系统的外文文献

NVIDIA Jetson NX unity

ssm框架英文参考文献

springboot实验室设备管理系统的英文参考文献

驾校管理系统今年英文文献

最新资源