Elsevier

Energy

Volume 122, 1 March 2017, Pages 350-362
Energy

Energy saving and prediction modeling of petrochemical industries: A novel ELM based on FAHP

https://doi.org/10.1016/j.energy.2017.01.091Get rights and content

Highlights

  • The ELM integrated FAHP approach is proposed.

  • The FAHP-ELM prediction model is effectively verified through UCI datasets.

  • The energy saving and prediction model of petrochemical industries is obtained.

  • The method is efficient in improvement of energy efficiency and emission reduction.

Abstract

Extreme learning machine (ELM), which is a simple single-hidden-layer feed-forward neural network with fast implementation, has been widely applied in many engineering fields. However, it is difficult to enhance the modeling ability of extreme learning in disposing the high-dimensional noisy data. And the predictive modeling method based on the ELM integrated fuzzy C-Means integrating analytic hierarchy process (FAHP) (FAHP-ELM) is proposed. The fuzzy C-Means algorithm is used to cluster the input attributes of the high-dimensional data. The Analytic Hierarchy Process (AHP) based on the entropy weights is proposed to filter the redundant information and extracts characteristic components. Then, the fusion data is used as the input of the ELM. Compared with the back-propagation (BP) neural network and the ELM, the proposed model has better performance in terms of the speed of convergence, generalization and modeling accuracy based on University of California Irvine (UCI) benchmark datasets. Finally, the proposed method was applied to build the energy saving and predictive model of the purified terephthalic acid (PTA) solvent system and the ethylene production system. The experimental results demonstrated the validity of the proposed method. Meanwhile, it could enhance the efficiency of energy utilization and achieve energy conservation and emission reduction.

Introduction

Energy is an essential pillar of modern economy and the cornerstone of modern civilization. With the rapid development of economy, energy issues are increasingly prominent. From a development point of view, the energy conservation and emissions reduction now are deemed as a worldwide common themes, especially in the petrochemical industries. Therefore, the improvement of energy efficiency is an important means to achieve economic goals and solve environmental problems. The amount of the industrial ethylene production, which is the pillar of the petrochemical industry, is commonly used as one of the major indicators of a country's industry level. And its energy consumption also accounts the important all-around index for measuring the technical performance of plants. According to the report in 2014 when China National Petroleum Corporation's ethylene production amount was 10420kt/a [1], the average fuel plus power consumption (standard oil) was 571.39 kg/t for producing a ton of ethylene, which costed far more than that of world's advanced level. A huge development in energy conservation and saving is potential in ethylene fabricating industry. Meanwhile, the purified terephthalic acid (PTA), as an important chemical raw materials, also has a pivotal role in the chemical industry and closely relates to our daily life [2], [3]. In China, 70% of the PTA is used to produce polyester products while 20% of them being used to produce Polyethylene terephthalate (PET), and the rest is utilized to produce the packing material. Therefore, the optimization of the PTA production process has become a hot research topic in the petrochemical industry. The relevant organizations of countries have been committed to improve the production equipment increasing energy efficiency continuously. If the production process of petrochemical industries can be accurately analyzed and predicted, the energy conservation and emission reduction of the petrochemical industry can be achieved. In order to better analyze and predict the energy efficiency of the petrochemical industry production, we propose an improved extreme learning machine (ELM) based on fuzzy C-Means algorithm (FCM) integrating analytic hierarchy process (AHP) (FAHP-ELM).

The organization of the remainder is as follows: Section 2 presents the research status of the energy saving and prediction with the ELM and the AHP. The details of the FAHP-ELM are introduced in Section 3. Section 4 presents the comparisons with the other feed-forward neural network by University of California Irvine (UCI) benchmark datasets. Section 5 shows two practical cases study about the energy efficiency analysis and prediction of petrochemical industry based on the FAHP-ELM. Finally, the concluding remarks are given in Section 6.

Section snippets

Related work

In recent decades, as the result of the development of the computer technology, a variety of artificial intelligence methods (expert system, genetic algorithm, neural network etc.) are widely applied for the simulation of production process as well as prediction models, and have achieved good results.

The artificial neural network (ANN), adjusts the connected relationship between many internal nodes to achieve the goal of information processing without regard to the internal mechanism.

FAHP- ELM

This FAHP-ELM aims to achieve data compression by the weighted fusion of high-dimensional data samples, the extraction of the data characteristics, the removal of redundant information and the filtration of noise. With the FAHP fusion data as the input of the ELM, the established FAHP-ELM is suitable for the complex data modeling.

UCI benchmark data set test

In order to verify the feasibility and validity of the proposed FAHP-ELM, the Wine Quality and the Parkinsons telemonitoring data sets in UCI database (For the details of the dataset, one can refer to the following website: http://archive.ics.uci.edu/ml/datasets.html) are selected to do the test, which is shown in Table 1.

First, cluster the input attributes of two data sets using the FCM into 5 classes and 7 classes respectively, and then do the test according to modeling process described in

Case study: production capacity prediction and energy saving analysis of the petrochemical industry

In order to verify the actual value of the FAHP-ELM for petrochemical industries, the acetic acid consumption of the PTA solvent system and the product output of ethylene production can be predicted by the FAHP-ELM, as well as the production status and energy efficiency status of the ethylene production can be analyzed.

Conclusions

The paper proposes the FAHP-ELM prediction model. This proposed model is able to filter redundant information and extract characteristic components. Meanwhile, the FAHP-ELM is also capable of training the blended data, simplifying the network structure of the ELM as well as optimizing the learning performance of networks. Compared with the BP neural network and the ELM, the faster convergence, higher effectiveness and accuracy, and stronger generalization of modeling of the FAHP-ELM are

Acknowledgments

This research was partly funded by National Natural Science Foundation of China (61533003, 61603025), Natural Science Foundation of Beijing, China (4162045) and National Key Technology Support Program (2015BAK36B04).

References (48)

  • G.B. Huang et al.

    Extreme learning machine: theory and applications

    Neurocomputing

    (2006)
  • Y. Miche et al.

    SOM-ELM—self-organized clustering using ELM

    Neurocomputing

    (2015)
  • Z.L. Sun et al.

    Application of BW-ELM model on traffic sign recognition

    Neurocomputing

    (2014)
  • Y. Zhang et al.

    A novel biologically inspired ELM-based network for image recognition

    Neurocomputing

    (2016)
  • S. Salcedo-Sanz et al.

    Feature selection in wind speed prediction systems based on a hybrid coral reefs optimization–Extreme learning machine approach

    Energy Convers Manag

    (2014)
  • P. Gastaldo et al.

    SIM-ELM: connecting the ELM model with similarity-function learning

    Neural Netw

    (2016)
  • Y.M. Han et al.

    Energy consumption hierarchical analysis based on interpretative structural model for ethylene production

    Chin J Chem Eng

    (2015)
  • Y.M. Han et al.

    Energy efficiency evaluation based on data envelopment analysis integrated analytic hierarchy process in ethylene production

    Chin J Chem Eng

    (2014)
  • L. Yagmur

    Multi-criteria evaluation and priority analysis for localization equipment in a thermal power plant using the AHP

    Energy

    (2016)
  • D.L. Pham

    Spatial models for fuzzy clustering

    Comput Vis Image Underst

    (2001)
  • B.A. Pimentel et al.

    Multivariate Fuzzy C-Means algorithms with weighting

    Neurocomputing

    (2016)
  • Z. Zainuddin et al.

    An effective fuzzy C-means algorithm based on symmetry similarity approach

    Appl Soft Comput

    (2015)
  • L.Y. Zhang et al.

    Fuzzy C-Means clustering of incomplete data based on probabilistic information granules of missing values

    Knowledge Based Syst

    (2016)
  • Y.L. He et al.

    Positive and negative correlation input attributes oriented subnets based double parallel extreme learning machine (PNIAOS-DPELM) and its application to monitoring chemical processes in steady state

    Neurocomputing

    (2015)
  • Cited by (0)

    View full text