Ijaiem 2014 01 02 114

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 12, December 2013 ISSN 2319 - 4847
Comparative Analysis amongst Prediction of Field Findings through Empirical Model and Optimized Neural Network Mathematical Model for Human Powered Flywheel Motor
Arati R. Lende1 and J. P. Modak2
1
Ex Assistant. Professor, DMIETR, Sawnagi (M), Wardha, MH, India 2 Emeritus Professor and Dean (R&D), PCE, Nagpur, MH, India
Abstract
Neural Network is one of the most emerging tools in fitting function, future forecasting, pattern recognition, etc. This article signifies the effective methodology of Neural Network modeling for prediction of field findings. J. P. Modak and their associates had already carried out lots of investigations over development of applications utilizing Human Powered Flywheel Motor as an energy source. The applications tried so far are mostly rural based such as brick making, low head water lifting, wood turning, wood strip cutting, electricity generation, etc. The productivity of the above mentioned applications had great affection towards rider thereby affecting quality and quality of production. The paper evaluates optimum artificial neural network (ANN) parameters for prediction of experimental findings accurately through sequential variation of each ANN parameter. The most favorable values of each parameter are selected and a mathematical model is extracted in the course of it. This document also compares the prediction amongst ANN based mathematical model and traditionally generated empirical model.
Keywords: Artificial Neural Network (ANN), mathematical model, empirical model.
1. INTRODUCTION TO HUMAN POWERED FLYWHEEL MOTOR (HPFM)

1.1 Working of Human Powered Flywheel Motor Energized Process Unit This machine system comprises three sub systems namely (i) HPFM [11] (ii) Torsionally Flexible Clutch (TFC) (iii) A Process Unit. The process units tried so far are mostly rural based such as brick making machine [1] [3](both rectangular and keyed cross sectioned), Low head water lifting, Wood turning, Wood strips cutting, electricity generation etc. The Figure 1 shows the schematic arrangement of pedal operated flywheel motor which comprises of following elements
Figure 1 Schematics of Human Powered Flywheel Motor R= Rider M = mechanism (01-OA-B-02-01) BSC = Big Sprocket Chain Drive SSC = Small Sprocket Chain Drive GSR = Gear of Speed Rise PSR = Pinion of Speed Rise FW= Flywheel CH = Chain CS = Counter Shaft FS = Flywheel Shaft
Volume 2, Issue 12, December 2013
Page 423

1.2 Study of already available experimental data The various parameters involved [11] in the experimentation are Table 1: Independent Variables and their symbols
Sr. No. 1 2 3 4 5 6 Independent Variable Moment of Inertia of Flywheel Input by the Rider Time Mechanical Efficiency Gear Ratio Angular Velocity of Flywheel Symbol I R T ME G
Table 2: Range of Variation of Independent parameter i.e. Rider(R)

Range of age Height Weight Blood Pressure Pulse rate 20-25Years 155-170cm 40-55 140-70 68-80/min.
The observations recorded during the experimentation are as below Table 3: Experimental observations
Independent Variables Log (I/RT2) 1 23 50 82 141 200 -7.4270 -7.1792 -7.2694 -6.1549 -5.9717 -7.2694 Log (ME) 0.00 0.0662 0.0600 0 0 0 Log (G) 0.3010 0.0010 0.0010 0.301 0.301 0.0569 Dependent variable Log ( T) 3.6305 3.5570 3.5004 3.0767 2.8587 3.4107
1.3 Empirical Model The experimental Independent variables were reduced by evaluating dimensionless pi terms by Buckingham pi theorem and a mathematical equation was generated by traditional method to predict the experimental findings. The equation [12] is as shown. T = 1.288 ( I/RT2)-0.46 (ME)-0.87 (G)0.40
2. EXECUTION OF ARTIFICIAL NEURAL NETWORK MODELING AND ITS OPTIMIZATION

Modeling a system through ANN simulation [9] involves use of ANN parameters appropriately. A topology is nothing but the complete architecture of network formed through the use of ANN parameters. The ANN parameters should be varied systematically in an attempt to identify best topology for a specified problem. The number of layers was restricted to two as the variables involved were high in number. A table for evaluation of modeling technique is formed [5] as below. The shaded column indicates the variation of that particular parameter and shaded row shows the slandered value of that parameter.
Page 424

Table 4: Sequence of variation of ANN Parameters
Training Number T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15 T16 T17 T18 T19 T20 T21 T22 T23 T24 T25 T26 T27 Hidden layer Size 20 50 100 150 250 300 500 600 700 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 Type of Training Function trainlm trainlm trainlm trainlm trainlm trainlm trainlm trainlm trainlm trainb trainbfg trainlm trainbr traingdm traingb traincgf traincgp trainlm trainlm trainlm trainlm trainlm trainlm trainlm trainlm trainlm trainlm Performance Function mse mse mse mse mse mse mse mse mse mse mse mse mse mse mse mse mse mse mae sse mae mae mae mae mae mae mae Types of transfer function Layer1 tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig tansig logsig tansig tansig tansig tansig tansig Layer2 purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin purelin logsig Purelin Purelin Purelin Purelin Type of Learning Algorithm learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd learngd Learncon Learngd Learnh Learnk
3. VARIATION IN PREDICTION OF ANN MODEL WITH VARIATION OF ANN PARAMETERS

The graphs for each program are generated which illustrate the effect of each variation on prediction of model. Skipping few of the bad predictions all other graphs are shown as below The percentage error in prediction is also plotted to compare and select the best of the topology amongst these topologies.
Figure 2 Predictions with 20 Neurons
Page 425

Figure 3 Percentage errors in predication with 20 Neurons
Page 426

Page 427

Figure 14 Predictions with training Function trainlm
Figure 15 Percentage errors in predication with training Function trainlm
Page 428

Figure 16 Predictions with training Function trainbr
Figure 17 Percentage errors in predication with training Function trainbr
Figure 18 Predictions with training Function traincgb
Figure 19 Percentage errors in predication with training Function traincgb
Figure 20 Predictions with training Function traincgp
Page 429

Figure 21 Percentage errors in predication with training Function traincgp
Figure 22 Predictions with performance Function mse
Figure 23 Percentage errors in predication with performance Function mse
Figure 24 Predictions with performance Function mae
Figure 25 Percentage errors in predication with performance Function mae
Page 430

Figure 26 Predictions with performance Function sse
Figure 27 Percentage errors in predication with performance Function sse
Figure 28 Predictions with transfer Function tansig, purelin
Figure 29 Percentage errors in predication with transfer Function tansig, purelin
Figure 30 Predictions with transfer Function logsig, purelin
Page 431

Figure 31 Percentage errors in predication with transfer Function logsig, purelin
Figure 32 Predictions with learning Function learncon
Figure 33 Percentage errors in predication with learning Function learncon
Figure 34 Predictions with learning Function learngd
Figure 35 Percentage errors in predication with learning Function learngd
Page 432

Figure 36 Predictions with learning Function learnh
Figure 37 Percentage errors in predication with learning Function learnh The training program T28 was selected on the basis of minimum percentage error occurred in prediction of evidences.
4. DEDUCTION OF MATHEMATICAL MODEL THROUGH ANN MODEL

Once a model is trained the values of weights and biases along with the function of the input layer and hidden layer defines the prediction of a model. The mathematical model generated from the algorithm T28 is as follows
Output = 1* (LW * (2/(1 + exp(-2 * (IW * b + Ib))) - 1) + Ob) Mathematical model

Where, LW = Weight Matrix of Output layer of ANN Ob = Bias Matrix of Output Layer of ANN IW = Weight Matrix of Input layer of ANN Ib = Bias Matrix of Input Layer of ANN b = input matrix Purelin = Function of Output layer of ANN Tansig = Function of input layer of ANN The matrix dimensions of variables above variables are too large to show in this paper.
5. COMPARING PREDICTION OF EVIDENCES WITH ANN MATHEMATICAL MODEL AND EMPIRICAL MODEL
The figures shown below give comparative analysis amongst prediction of evidences with ANN mathematical model and previously drawn empirical model.
Figure 38 Comparison between experimental evidences (red) and prediction through empirical model (blue)
Page 433

Figure 39 Comparison between experimental evidences (red) and prediction through ANN mathematical model (blue)
Figure 40 Percentage error in prediction through ANN mathematical model (red) and Empirical equation (blue) The figure 38 shows the prediction of experimental evidences (red) with previously drawn empirical model (blue) which is much deflected compared to the experimental findings. On the contrary artificial neural network based mathematical model show much better results as in figure 39. The figure 40 compares percentage error in prediction at every stage of the experimentation which also implies that neural prediction (red) is much better than the empirical model (blue).
6. CONCLUSION
The paper carries a systematic method of optimization of artificial neural network model and comes out with a audacious solution for prediction of experimental findings. The plots carried for each variation of ANN parameter clearly signifies its effect on prediction of the model. The mathematical model deduced may be utilized for future research in development of a controller for Human Powered Flywheel Motor.
References
[1] Modak J. P. and Askhedkar R. D. Hypothesis for the extrusion of lime flash sand brick using a manually driven Brick making machine, Building Research and Information U.K., V22,NI, Pp 47-54, 1994 [2] Modak J. P. and Bapat A. R. Manually driven flywheel motor operates wood turning machine, Contempory Ergonomics, Proc. Ergonomics Society annual convension13-16April, Edinburg, Scotland, Pp 352-357, 1993. [3] Sohoni V. V., Aware H. V. and Modak J. P. Manual Manufacture of Keyed Bricks, Building Research and Information UK, Vol 25, N6, 1997, 354-364. [4] Modak J. P.Design and Development of Manually Energized Process Machines having Relevance to Village/Agriculture and other productive operations, Application of manually energized flywheel motor for cutting of wood strip, Human Power, send for Publications. [5] H. Schenck Junior Theory of Engineering Experimentation, MC Graw Hill, New York. [6] A. R. Lende, Modelling of pedal driven flywheel motor by use of ANN, M. Tech. Thesis, PCE, Nagpur [7] S. N. Shvanandam, Introduction to Neural Network using Matlab 6.0, McGraw Hill publisher. [8] Stamtios V. Kartaplopoulos , Understanding Neural Networks and Fuzzy Logics, IEEE Press [9] Neural Network Toolbox TM 7 Users Guide R2010a, Mathworks.com [10] Rudra Pratap, Getting Started with Matlab7, Oxford, First Indian Edition 2006. [11] A. R. Bapat, Experimental Optimization of a manually driven flywheel motor, M.E. Thesis, VNIT, Nagpur. [12] A. R. Bapat, Experimentation of Generalized experimental model for a manually driven flywheel motor, PhD Thesis, VNIT, Nagpur.
Page 434

[13] A. R. Lende, J. P. Modak Modellling and Simulation of Human Powered Flywheel Motor for field data in the course of artificial neural network- a step forward in the development of artificial intelligence
AUTHOR
Prof. (Ms) A. R. Lende received her Bachelor degree in Mechanical Engineering from BDCOE, Sewagram in 2004 and Master degree in Mechanical Engineering Design from PCE, Nagpur in 2007. She had registered her Ph.D. in 2009. She had given her teaching services to MIT, Kothrud, Pune and DMIETR, Wardha for three and two years respectively. The author also has an industrial experience of one year in the field of design and implementations. She also had worked in many mechanical engineering design projects and published papers in that region.
Dr. J. P. Modak is an Emeritus Professor and Dean (R&D) PCE, Nagpur. He has guided number of Ph.D. and PG Projects in the field of Design engineering, mathematical modeling, Artificial Intelligence, Vibration, etc. He is also consultant for many industries. He have delivered number of key not lectures, expert lectures at various conferences and workshops. He became a guiding light for researchers for driving research in systematic manner. The author had contributed through large number of papers in reputed international Journals & Conferences for developing research and path of research.
Page 435

Ijaiem 2014 01 02 114

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Ijaiem 2014 01 02 114

Hochgeladen von

Copyright:

Verfügbare Formate

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Keywords: Artificial Neural Network (ANN), mathematical model, empirical model.

1. INTRODUCTION TO HUMAN POWERED FLYWHEEL MOTOR (HPFM)

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Table 2: Range of Variation of Independent parameter i.e. Rider(R)

2. EXECUTION OF ARTIFICIAL NEURAL NETWORK MODELING AND ITS OPTIMIZATION

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

3. VARIATION IN PREDICTION OF ANN MODEL WITH VARIATION OF ANN PARAMETERS

Figure 2 Predictions with 20 Neurons

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 3 Percentage errors in predication with 20 Neurons

Figure 4 Predictions with 50 Neurons

Figure 5 Percentage errors in predication with 50 Neurons

Figure 6 Predictions with 20 Neurons

Figure 7 Percentage errors in predication with 150 Neurons

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 8 Predictions with 300 Neurons

Figure 9 Percentage errors in predication with 300 Neurons

Figure 9 Predictions with 500 Neurons

Figure 10 Percentage errors in predication with 500 Neurons

Figure 18 Predictions with 600 Neurons

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 11 Percentage errors in predication with 600 Neurons

Figure 12 Predictions with 700 Neurons

Figure 13 Percentage errors in predication with 700 Neurons

Figure 14 Predictions with training Function trainlm

Figure 15 Percentage errors in predication with training Function trainlm

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 16 Predictions with training Function trainbr

Figure 17 Percentage errors in predication with training Function trainbr

Figure 18 Predictions with training Function traincgb

Figure 19 Percentage errors in predication with training Function traincgb

Figure 20 Predictions with training Function traincgp

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 21 Percentage errors in predication with training Function traincgp

Figure 22 Predictions with performance Function mse

Figure 23 Percentage errors in predication with performance Function mse

Figure 24 Predictions with performance Function mae

Figure 25 Percentage errors in predication with performance Function mae

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 26 Predictions with performance Function sse

Figure 27 Percentage errors in predication with performance Function sse

Figure 28 Predictions with transfer Function tansig, purelin

Figure 29 Percentage errors in predication with transfer Function tansig, purelin

Figure 30 Predictions with transfer Function logsig, purelin

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 31 Percentage errors in predication with transfer Function logsig, purelin

Figure 32 Predictions with learning Function learncon

Figure 33 Percentage errors in predication with learning Function learncon

Figure 34 Predictions with learning Function learngd

Figure 35 Percentage errors in predication with learning Function learngd

Volume 2, Issue 12, December 2013

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Figure 36 Predictions with learning Function learnh