A review on Artificial Neural Network approach in manufacturing systems Kumar, V.,1 Shukla, O.J.,2 Soni, G.,3 Kumar, R.4
1,2,3
Department of Mechanical Engineering, Malaviya National Institute of Technology Jaipur, Rajasthan 1
Email:
[email protected] 2
Email:
[email protected] 3
Email:
[email protected] 4
Department of Electrical Engineering, Malaviya National Institute of Technology Jaipur, Rajasthan 4
Email:
[email protected]
Abstract: The use of artificial intelligence in manufacturing handle unprecedented and unforeseen situations under incomplete & imprecise information and perform optimization continues to be a major field of research. Artificial Intelligence based techniques are designed for capturing, representing, organizing and utilizing knowledge by computers and hence play an important role in manufacturing systems. The purpose of this paper is to present a survey of the use of Artificial Neural Network (ANN), an AI technique, in various manufacturing systems aimed at presenting the key research themes and trends of research. The paper uses database of Elsevier and Taylor & Francis publications as a source and reviews the current research on the artificial neural network approach. The applications adopted are: manufacturing process planning, manufacturing scheduling, manufacturing system & assembly and logistics & maintenance. The paper concluded by plotting application-wise and year-wise trends of the research publication on ANN. This is a one kind of study to classify research on the use of ANN technique in manufacturing. Keywords: Artificial Neural Network, Manufacturing systems, Artificial Intelligence.
1. INTRODUCTION The effect of increasing competition in business is relentless pressure on manufacturing organizations. Due to these pressures the manufacturing organizations concentrate all efforts to reduce process variability and stable manufacturing operation. Artificial intelligence (AI) in manufacturing can help to make smart manufacturing organizations. The use of AI approaches handle uncertainty and perform optimization seems to be the primary means of tackling the challenges with the help of its ability to evolve solutions. The number of publications and studies utilizing AI techniques in manufacturing appears to be growing rapidly.
This paper considered 24 research papers, from Elsevier and Taylor & Francis publications, in most of the applications of manufacturing for reviewing an AI approach i.e. artificial neural network (ANN). Some of the authors used ANN in their research papers with combination of genetic algorithm (GA), fuzzy logic etc. (Cheng et al.(1998), Chan. et al.(2000), Zeidi. et al.(2013), Dobrzanski. et al.(2005), Wang. et al.(2003), Han. and Yang.(2006), Tamayo et al.(2009)).
Table. 1 Number of research paper published on ANN according to types of journals
Journal's Name
1
Computer in industry
2
Computer and Industrial Engineering
1
Expert system with applications
2
Engineering Application of Artificial Intelligence
2
European journal of operational research
1
International journal of production research
2
journal of the Chinese institute of industrial engineers International Journal of computer integrated manufacturing
1
1 3 1
1
Material Science and Engineering
1
Machine tools and manufacture
1
Knowledge Based System
1
Simulation Modeling Practice and Theory Robotics and computer integrated manufacturing
2. METHODOLOGY
No. of Papers
Applied Soft Computing
International journal of production economics International Journal of Machine Tools And Manufacture Journal of Material Processing Technology
provides the quantitative data and concludes the paper.
1 1
Neurocomputing
1
Total
24
The table.1 shows the number of research papers published in different journals regarding ANN application in manufacturing. Industrial engineering, computer, production, robotics, manufacturing tools, neurocomputing, operational research etc. are few journals which promptly publish papers on ANN. The paper is organized as follows. Section 2 describes the methodology adopted and section 3
The methodology used in this paper for reviewing research paper is based on the application areas. We have chosen four application areas as follows: 2.1 Manufacturing Process Planning Guh et al. (1999) worked on hybrid intelligent tool (IntelliSPC) in which a neural network based control chart pattern recognition system, an expert system based control chart alarm interpretation system and a quality cost simulation system were integrated for online SPC. Pacella. and Semeraro. (2005) used ANN in Statistical process control (SPC) to develop an ART- based neural network for automating quality control of manufacturing process with statistical background. It detects the unnatural process behavior by detecting change in the state of the process. Neuroth. et al. (2000) uses two case studies, the first case study showed a fuzzy logic control strategy incorporated into an existing conventional control system in order to improve a particular operating aspect. The second case study showed the use of an ANN for the determination of the pipeline friction factor. Wang. et al. (2003) used ANN in electro discharge machining process for modeling and optimizing the Electro- discharge machining. Scheffer. et al (2003) from South Africa, in the same year used ANN process monitoring to monitor the tool in hard turning to control its rapid wearing. Addition to above manufacturing process Guessasma. et al (2004) used ANN in Atmospheric Plasma spray process in which Artificial Neural Network structure coupled with experimental results of an optic sensor to control the powder particle fusion state before the coating formation. Lorenzo. et al (2006) design ANN to aim at predicting fracture occurrence for AISI 1045 steel in different forming operation and found its suitability for proposed technique in dealing with fracture predictions and its effectiveness in forming processes design. After 4 years, in year 2010 Hsieh. (2010) used ANN in process analysis of TFT- LCD industry and developed an integrated intelligent data analysis system to analyze the possible clustering effect with respect to abnormal position (or defect) on LCDTFT panel. 2.2 Manufacturing Scheduling Dobrzanski. et al (2005) used ANN to develop a model which helps in forecasting of yield point and
ultimate tensile strength for steel. These parameters are calculated basing on the chemical composition and technological factors of steel manufacturing. Based on the model, software was developed for searching the optimal chemical composition of steel on a particular technical process to minimize the risk in manufacturing of product. Orsoni. and Bandinelli. (2007) anticipated the impact of local failures on the productivity of the coordinated plants, in the form of actual failure patterns, high levels of resources utilization and throughput can be maintained throughout the production system over time. For such purposes a modular fault predictor based on ANN was devised capable of providing quick estimates of these impacts and of enabling local rescheduling of activities around failure events. Dao. et al (2007) used the combination of Hopfield Neural Networks (HNN) and Tabu (local search) approach to define optimal groups of operations which will facilitate the generation of the route sheet. For each operation, HNN was used to generate the feasible manufacturing alternatives, and the Tabu search technique was applied to identify the best schedule. Hybrid Hopfield Neural Networks (HHNN) and the simulation techniques save time and minimize unbalanced workloads among the operations. Lolas. And Olatunbosun. (2008) demonstrated an ANN system to predict accurately enough the reliability behavior of an automotive vehicle at 6000 km by only knowing information from the initial vehicle’s inspection at 0 km. The ANN manages to learn association between input value and target value of reliability performance. It also recognized and identifies the degradation mechanism and predicts the reliability behavior in their development phase. Noroozi. et al. (2013) used an adaptive learning approach which is inspired in neural networks is devised to optimize the performance of three different types of intelligence algorithms, namely, HGA, HSA, and PSO which have been used for solving scheduling problems with batch processing machine. The adapted learning process uses a weight factor to perturb the parameters of the scheduling problem, whereas traditional algorithms uses static amount of parameters to trap the weaker local area. 2.3 Manufacturing System and Assembly Kullkarni. and Kiang. (1995) presented a neural network based technique called Self- Organizing Map (SOM) network in context of flexible manufacturing system. SOM helps to depict the relationship between different parts and visualized the picture of parts to be grouped together in families. SOM can accept different types of input data for forming part families. In addition Barschdorff. et al
(1997) described two hybrid system for controlling and monitoring of manufacturing system on hardware and software bases. They described different distributed control system incorporated by ANN. Cheng et al (1998) proposed an AI-Internet based approach for implementing design and manufacturing agility in journal bearing design and application. The approach had great potential of quickly providing design expertise and solutions based on customer’s requirements. Chan. et al (2000) designed a flexible Manufacturing System (FMS) which uses simulation and multi-criteria decision-making techniques. The design process consists of the construction and testing of alternative designs using simulation methods. Design selected on the basis of AHP is employed to analyze the output from the FMS simulation models and ANN and fuzzy used to support the FMS design process. cakar and Cil (2004) used ANN for designing of a manufacturing system. They used four different priority rules namely EDD, SPT, CR and FCFS for efficient decision making. As a result four different design alternatives are obtained from trained ANN. Performance measures of a manufacturing system are given to the ANN which then gives a design alternative. Beyond 2013, researchers do not consider planning, implementation, and capital investment issues. They have focused only on forming manufacturing cells comprehensively or non-incrementally but Zeidi. et al. (2013) presented a new multi-objective mathematical model that try to plan conversion of job shops to cellular manufacturing systems with minimization of the total material handling cost and total exceptional elements. For this approach a multi objective optimization within the artificial neural network (ANN) has been constructed. Altiparmak. et al. (2007) developed ANN metamodel for the simulation model of an asynchronous assembly system (AAS) considering three different types of AASs. ANN metamodels were compared with regression metamodels using polynomial and exponential function to estimate production rate when buffer size configuration and workstation failure rates were given. 2.4 Logistics and Maintenance Han. and Yang. (2006) used advance techniques such as ANN and communication technologies to create a new e- maintenance system. This system co-operate among associated areas, sharable information resources and integration of existent advanced techniques on the communication platform. It consists of two subsystems: maintenance centre and local maintenance. This division can effectively reduce maintenance cost, maintenance system design
period, and solve the problem of lack of experts. Tamayo. et al. (2009) used ANN in deliveries optimization. They try to minimize the cost of product recall. The raw material dispersion problem was analyzed, in order to determine a risk-level criterion or ‘‘production criticality’’. This criterion was used subsequently to optimize deliveries dispatch with the purpose of minimizing the number of batch recalls in case of crisis. Yang. and Lu. (2010) used a combined pre-emptive and competitive neural-network approach to solve a multi-objective dynamic dispatching problem in a two-workstation manufacturing environment with parallel machines incorporating a TFT-LCD process. CONCLUSION This paper presented a review of the use of Artificial Neural Network (ANN) in different categories of manufacturing. We particularly focused on manufacturing process planning, scheduling, system & assembly and logistics and maintenance.
10 8 6 4 2 0
no. of papers
Fig.2 Year-wise trends of research publications on ANN
The trend of the publications according to years is as shown in fig. 2. The paper considered the research papers published in years during 1995-2013. Research publications during above mentioned period are divided in to three years duration for clear understanding of trends in research. According to our survey methodology it is found that the maximum number of research papers, based on ANN technique, published in duration 2003-06. An approximate equal number of research papers are published during 1995-98 and 1999-2002. After 2002, publications on ANN increased and then a negative slope from 200306 to 2010-13 depicts lesser work in later years.
REFERENCES 1.
2. Fig. 1 Application-wise trends of research publications on ANN
From the fig. 1, it is found that manufacturing process planning and manufacturing system & assembly have good number of research papers, because ANN is an efficient tool and can be used in modeling of nonlinear relations. (Cakar and Cil (2004)) Manufacturing scheduling have average number of research papers and Logistics and maintenance too have less number of research papers, which shows need of more attention in these areas regarding ANN application.
3.
4.
5.
Noroozi, A., Mokhtari, H., Nakhai, I., & Abadi, K.., Research on computational intelligence algorithms with adaptive learning approach for scheduling problems with batch processing machines. Neurocomputing, 101, 190–203, 2013. Taylor, P., A hybrid hopfield neural networks based simulation approach for optimisation of manufacturing group scheduling, Journal of the Chinese Institute of Industrial Engineers, 24(4), 37–41, 2007. R. S. Ransing and R. W. Lewis., A semantically constrained neural network for manufacturing diagnosis. International Journal of Production Research, 35(9), 2639–2660, 1997. Lorenzo, R. Di, Ingarao, G., & Micari, F., On the use of artificial intelligence tools for fracture forecast in cold forming operations, Journal of material processing technology, 177, 315–318, 2006. Pacella, M., & Semeraro, Q., Understanding ART-based neural algorithms as statistical tools for
6.
7.
8.
9.
10.
11.
12.
13.
14.
manufacturing process quality control, Engineering applications of Artificial Intelligence, 18, 645–662, 2005. Guessasma, S., Salhi, Z., Montavon, G., Gougeon, P., & Coddet, C., Artificial intelligence implementation in the APS process, diagnostic, Material Science and Engineering B, 110, 285–295, 2004. Taylor, P., Artificial neural networks for design of manufacturing systems and selection of priority rules, International journal of computer integrated manufacturing, 17(3), 37–41, 2007. Neuroth, M., Macconnell, P., Stronach, F., & Vamplew, P., Improved modeling and control of oil and gas transport facility operations using artificial intelligence, Knowledge based system ,13, 81–92, 2000. Wang, K., Gelgele, H. L., Wang, Y., Yuan, Q. & Fang, M., A hybrid intelligent method for modelling the EDM process, international journal of machine tools and manufacture, 43, 995–999, 2003. Ã, K. H., Incorporating ANNs and statistical techniques into achieving process analysis in TFT-LCD manufacturing industry. Robotics and Computer Integrated Manufacturing, 26(1), 92–99, 2010. Rezaeian, J., Javadian, N., Tavakkolimoghaddam, R., & Jolai, F., A hybrid multiobjective approach based on the genetic algorithm and neural network to design an incremental cellular manufacturing system. Computers & Industrial Engineering, 66(4), 1004–1014, 2013. Kowalski, M., & Madejski, J., Methodology of the mechanical properties prediction for the metallurgical products from the engineering steels using the Artificial Intelligence methods, Journal of material processing technology,165, 1500–1509, 2005. Han, T., & Yang, B., Development of an emaintenance system integrating advanced techniques, Computer in industry,57, 569– 580, 2006. Tamayo, S., Ã, T. M., & Sauer, N., Deliveries optimization by exploiting production traceability information. Engineering Applications of Artificial Intelligence, 22(4-5), 557–568, 2009.
15. Yang, T., & Lu, J. C., A hybrid dynamic pre-emptive and competitive neural-network approach in solving the multi-objective dispatching problem for TFT-LCD manufacturing, International Journal of Production research, 46(16) 37–41, 2010. 16. Orsoni, A., & Bandinelli, R., Improving the remote scheduling of distributed production with process statistics and AI techniques, Simulation modeling practices and theory, 15, 175–184, 2007. 17. Lee, J. H., Artificial intelligence-based sampling planning system for dynamic manufacturing process, Expert system with application, 22,117-133, 2002. 18. Kulkarni, U. R., & Kiang, M. Y., Dynamic grouping of parts in flexible manufacturing systems - A self-organizing neural networks approach, European journal of operational research, 192-212, 1995. 19. Altiparmak, F., Dengiz, B., & Bulgak, A. A., Buffer allocation and performance modeling in asynchronous assembly system operations : An artificial neural network metamodeling approach, Applied soft computing, 7, 946–956, 2007. 20. Scheffer, C., Kratz, H., Heyns, P. S., & Klocke, F., Development of a tool wearmonitoring system for hard turning, International journal of machine tool and manufacturing, 43, 973–985, 2003. 21. Chan, F. T. S., Jiang, B., & Tang, N. K. H., The development of intelligent decision support tools to aid the design of flexible manufacturing systems, International journal of production economics, 65, 73–84, 2000. 22. Monostori, L., Wijstenkiihler, G. W., Egresits, C., & Kadr, B., Approaches to coupling connectionist and expert systems in intelligent manufacturing, Computer in industry, 33, 5–15, 1997. 23. Lolas, S., & Olatunbosun, O. A., Prediction of vehicle reliability performance using artificial neural networks, Expert system with application, 34, 2360–2369, 2008. 24. Cheng, K., Harrison, D. K., & Pan, P. Y., Implementation of agile manufacturing- an AI and Internet based approach, Journal of material processing technology, 76, 96–101, 1998.