Elsevier

Measurement

Volume 114, January 2018, Pages 122-131
Measurement

An application of evolutionary system identification algorithm in modelling of energy production system

https://doi.org/10.1016/j.measurement.2017.09.009Get rights and content

Highlights

  • Survey on System Identification (SI) field is undertaken in this study.

  • Evolutionary SI approach of GP is found to be appropriate method for modeling.

  • An application of GP in modeling of energy system such as fuel cell is illustrated.

  • Experimental data validates the GP model and its performance is found satisfactory.

Abstract

The present work introduces the literature review on System Identification (SI) by classifying it into several fields. The review summarizes the need of evolutionary SI method that automates the model structure selection and its parameter evaluation based on only the system data. In this context, the evolutionary SI approach of genetic programming (GP) is applied in modeling and optimization of cleaner energy system such as direct methanol fuel cell. The functional response of the power density of the fuel cell with respect to input conditions is selected based on the minimum training error. Further, an experimental data is used to validate the robustness of the formulated GP model. The analysis based on 2-D and 3-D parametric procedure is further conducted to reveals insights into functioning of the fuel cell. The pareto front obtained from optimization of model reveals that the operating temperature of 64.5 °C, methanol flow rate of 28.04 mL/min and methanol concentration of 0.29 M are the optimum settings for achieving the maximum power density of 7.36 mW/cm2 for DMFC.

Introduction

The art of building mathematical models from the input–output data obtained from the System is known as System Identification. The main purpose of building the models from the system is to understand its behavior and predicts its performance for fault diagnosis. The term modeling has been extensively used in the field of SI. Basically, several types of models, systems and methods can be studied under various fields of SI as shown in Fig. A1 in Appendix A [1]. The systems modelled can be the manufacturing processes, cleaner energy production systems such as fuel cell or such as those involving the study of mechanical and thermal properties of graphene and carbon nanotubes or the stock market and weather phenomenon, etc. [1]. Among these processes, the energy storage systems such as fuel cells and batteries are the potential ones because they reduce the environmental burden arising due to toxic gases emitted by industries and by transportation vehicles. The system complexity and its functioning depends on the multiple set of input–output variables attributed to it. The cost incurred in optimization of system behavior is high because it is not economical to measure the data. The main concept behind analyzing the obtained data from the system is to unravel the useful information that is responsible for its long term efficient performance before the system is replaced. Given with the multi-functional nature of the systems, the need of modeling and optimization has been strengthened to understand its complex behavior.

The work described in this manuscript is divided into seven sections. Section 2 discusses the models and modeling methods classified under various fields of SI. Section 3 discusses the alternative methodology (evolutionary approach of Genetic programming (GP)) suggested in the latest trends in era of SI. Section 4 illustrates the application of GP in modeling of the fuel cell system. Section 5 discussed the model formulation and statistical fit on the experimental data. Section 6 provides the uncertainty analysis, 2-D and 3-D surface analysis and optimization results of GP model. Finally, Section 7 concludes the study with recommendations and novelty in work.

Section snippets

Literature review on SI

This section describes the various types of models, systems and methods being studied in SI. The methods such as those based on statistics, finite element analysis or artificial intelligence can be used to formulate models. Models are also formed using Analysis of Variance (ANOVA) or the hypothesis tests. These models however serve common purpose to understand the complex systems behavior. Models build on only the given input–output continuous data are known as regression models. The methods

Evolutionary system Identification approach of genetic programming

The literature review on SI reveals that evolutionary SI approach of GP can be suitable alternative for modeling complex systems when its behavior is completely unknown. There are hardly any statistical assumptions in the mechanism of GP. The mechanism (Fig. 1) of GP is similar to GA and being extensively used as a structural optimization methodology. Researchers have developed hybrid approach of GP such as GA-GP [30], Clustering-GP [31], FEM-GP [32], GP-OLS [33], GP-SA [34], etc. for improving

Statement of the research

Direct methanol fuel cell (DMFC) is considered as an alternative source for generating cleaner energy [36], [37]. An appropriate value of operating conditions for DMFC affects its efficiency and performance. DMFC is a complex system and therefore, the notion of mathematical modeling and optimization seems to be a suitable alternative to attain the values of these operating conditions. Among the type of mathematical models and methods discussed in Section 2, it will be interesting to explore the

Results for GP based power density model

The GP simulations are performed in MATLAB R2010b with the population size and the maximum number of genes is set at 300 and 15 respectively. For selecting the parent genes from the pool of available solutions, a Lexicographic tournament selection strategy is adopted with a tournament size of 4. The maximum depth of each tree is set to 15. The crossover, mutation and direct reproduction probabilities are taken as 0.85, 0.1 and 0.05 respectively.

The training data samples from the data set (Table

Uncertainty analysis and optimization

The power density model generated by GP is simulated in three uncertain operating conditions to check if the model formulated is providing the range of values within the permissible working conditions of DMFC. In this simulation design, each of the input is assumed to follow a normal distribution with the minimum and maximum values set same as those of the three input conditions. Fig. 4 shows that an approximate normal trend is observed with the power density values of DMFC ranging from 0.2 to

Conclusions

The present study highlights the importance of system identification (SI) approaches in modeling and optimization of complex systems such as the fuel cell. A comprehensive literature study on types of modeling methods and models in several fields of SI is conducted. Important issues such as automation of model structure and its parameter evaluation is addressed by showing the application of an evolutionary approach of GP in modeling of direct methanol fuel cell (DMFC). The models formulated

Acknowledgement

The authors wish to acknowledge that this research has been supported by Shantou University Scientific Research Foundation (Grant No. NTF 16002).

References (43)

  • K. Charoen et al.

    Application of response surface methodology to optimize direct alcohol fuel cell power density for greener energy production

    J. Clean. Prod.

    (2017)
  • Q. Tian et al.

    Robust and stepwise optimization design for CO2 pipeline transportation

    Int. J. Greenhouse Gas Control

    (2017)
  • D. Zhao et al.

    A new stepwise and piecewise optimization approach for CO2 pipeline

    Int. J. Greenhouse Gas Control

    (2016)
  • D. Zhao et al.

    A framework of neural networks based consensus control for multiple robotic manipulators

    Neurocomputing

    (2014)
  • D. Zhao et al.

    Synchronized control with neuro-agents for leader–follower based multiple robotic manipulators

    Neurocomputing

    (2014)
  • A. Garg, K. Tai, B.N. Panda, System identification: survey on modeling methods and models, in: S. Dash, K. Vijayakumar,...
  • M. Deistler, System identification and time series analysis: past, present, and future, Stochast. Theor. Cont. (2002)...
  • M. Chandrasekaran et al.

    Application of soft computing techniques in machining performance prediction and optimization: a literature review

    Int. J. Adv. Manufact. Technol.

    (2010)
  • E. Vladislavleva, G. Smits, Symbolic regression via genetic programming, Final Thesis for Dow Benelux...
  • U. Çaydaş et al.

    Support vector machines models for surface roughness prediction in CNC turning of AISI 304 austenitic stainless steel

    J. Intell. Manufact.

    (2012)
  • S.N. Patra, R.J.T. Lin, D. Bhattacharyya, Regression analysis of manufacturing electrospun nonwoven nanotextiles, J....
  • Cited by (52)

    • An effective framework for wake predictions of tidal-current turbines

      2021, Ocean Engineering
      Citation Excerpt :

      It is unaffordable in terms of computational cost to apply high-fidelity CFD simulations for multi-turbine wake analysis. In recent years, artificial intelligence (AI) approaches have wide applications in different field of engineering applications, such as water level forecasting (Chang, 2013; Deo and Sahin, 2016), water quality prediction (Tanty and Desmukh, 2015; Ravansalar et al., 2016), shear rate prediction in rheological applications (Chin et al., 2019a,b), power characteristic evaluation for cleaner energy system and application in the energy production system (Garg et al., 2017a,b; Huang et al., 2018), etc. Among the AI approaches, artificial neural network (ANN) is a computational technique inspired by the human nerve cells (neurons), owing to its ability to model complicated relationships or to find partners in data.

    • Estimate of in-situ stress and geomechanical parameters for Duvernay Formation based on borehole deformation data

      2021, Journal of Petroleum Science and Engineering
      Citation Excerpt :

      In the 1980s, Aadnøy developed a method to determine the orientation and magnitudes of horizontal in-situ stresses based on the formation breakdown pressure of a circular borehole of arbitrary trajectory (Aadnoy, 1990; Aadnoy et al., 1994). In recent years, artificial neural networks and genetic algorithms have been applied in many fields, including the oil and gas energy industry (Sabir et al., 2018; Ibrahim et al., 2016; Huang et al., 2018). These methods have been used to map the relationship between in-situ stress and the displacements of a borehole wall and to estimate the in-situ stress (Zhang and Yin, 2014a, 2014b, 2015).

    • In-situ stress inversion in Liard Basin, Canada, from caliper logs

      2020, Petroleum
      Citation Excerpt :

      In 1980s, Aadnoy developed a method to determine the orientation and magnitudes of horizontal in-situ stresses based on the formation breakdown pressure of a circular borehole of arbitrary trajectory [11]. In recent years, artificial intelligence method and genetic algorithm based methodologies have been practiced in many fields including energy industry [12,13,14]. In petroleum industry, artificial intelligence method and genetic algorithm have been used to map the relationship between in-situ stress and the displacements of a borehole wall and to calculate the in-situ stress [15,16–18].

    View all citing articles on Scopus
    View full text