Printer Friendly

ARTIFICIAL NEURAL NETWORK MODEL FOR PREDICTING DRAFT AND ENERGY REQUIREMENTS OF A DISK PLOW.

Byline: S. A. Al-Hamed, M. F. Wahby, S. M. Al-Saqer, A. M. Aboukarima and Ahmed A. Sayedahmed

ABSTRACT

In this study, artificial neural network (ANN) model was developed for predicting draft and energy requirements of a disk plow. The ANN model utilizes ten input parameters: plowing depth and speed, sand content, silt content, clay content, soil moisture content, disk diameter and angle, tilt angle and soil density. The model provides the draft, unit draft and energy requirements of disk plows as the predicted output. The ANN model was trained based on data from literature and tested on actual data from field experiments. The architecture of the ANN model consisted of one hidden layer with 8 nodes. Standard backpropagation-based algorithm was used to train the network. The results showed that correlation coefficients for testing points were 0.934, 0.933 and 0.915 for draft, unit draft and energy requirements, respectively.

The promising results obtained indicate that the newly developed ANN model can be considered as a practical and reliable tool for predicting disk plow performance criteria under wide range of conditions. Using the network weights obtained from the ANN model, new formulations were presented for the calculation of draft, unit draft and energy requirements. Furthermore, as an added benefit, these formulations can be implemented with any programming language or spreadsheet program making them an attractive choice for routine analyses.

Key words: Artificial neural network, disk plow, draft, unit draft, energy requirement.

INTRODUCTION

Tillage is the mechanical manipulation of the soil by disturbing its original structure in the plow layer in order to promote tilth i.e. desired soil physical condition in relation to plant growth (Karmakar, 2005). The disk plow has always been the basic tillage implement on the farm. It is still useful and widely employed for primary tillage of virgin, stony and wet soils, cut through crop residues and roll over the roots (Boydas and Turgut, 2007). Disk plows have disks inclined to the rear for additional penetration (Vozka, 2007). The angle of attachment of the disk to the direction of travel is called the disk angle (Vozka, 2007). Another angle is the tilt angle of the disk. It is the slant (tilt) backward of the disk from the vertical (Bukhari et al., 1992). The disk angles vary from 42deg to 45deg and the tilt angles vary from 15deg to 25deg. The disk diameters are commonly between 60 and 70 cm.

Performance efficiency of the tillage process is measured in terms of draft or input energy (Gill and Vanden Berg, 1967). The availability of data on the draft requirement of tillage implements is also an important factor while selecting suitable tillage implements for a particular farm situation. The factors affecting forces on disk plow include type of soil, the bearing of disk, scraper type, tilt angle, disk angle, forward speed, depth, width of cut, soil density and others as reported by many studies (Sommer et al., 1983; Godwin et al., 1987; Panigrahi et al., 1990; Bukhari et al., 1992; Morad, 1992; Shirin et al.,1993; Manian et al, 2000; Ismail, 2002; Abu-Hamdeh and Reeder, 2003; Mamkagh, 2009; Osman et al.,2011). At smaller disk angles, the draft tends to increase because of greater contact area between the furrow wall and the convex (rear) side of the disk.

Collecting draft data under wide range of field conditions is a tedious and time-consuming job. Therefore, draft prediction models are required to predict the draft of tillage implements such as the disk plow under different soil and operating conditions (Roul et al., 2009). Further, mathematical solutions of soil-tool interaction based on empirical and semi-empirical models may be help tool for designers and researchers in the field of tillage implements (Karmakar, 2005).

The literature studies have shown that the relationship between the dependent and independent variables affecting performance parameters of a disk plow was modeled by regression analysis as reported by Ismail (2002) or by dimensional analysis as reported by Olatunji (2011). Recently, Karmakar (2005) reported that artificial neural networks (ANNs) have been used as possible approach to solve problems in the area of soil- tool interaction. It is noteworthy that there is a growing interest in modelling draft and energy requirements of tillage implements using ANN due to complexity and unavailable analytical models for all tillage implements. On the other hand, there is a shortage in empirical and analytical models for predicting draft force of disk plows due to the inherent characteristics of input factors, especially disk geometry.

To overcome the problems related to analytical and empirical draft models, attempts have been made to develop ANN models for predicting the draft requirement of tillage implements from soil conditions, tool geometry and working conditions.

Several authors found ANN predictions for draft, pull and energy requirements of tillage implements to be an effective tool, as shown in studies by Hassan and Tohmaz (1995), Tohmaz and Hassan (1995), Kushwaha and Zhang (1997), Zhang and Kushawaha (1999), Al- Janobi et al. (2001), Aboukarima et al. (2003), El Awady et al. (2003), Aboukarima (2004), Aboukarima and Saad (2006), El Awady et al. (2004), Aboukarima (2007), Roul et al. (2009), Al-Janobi et al. (2010), Aboukarima (2013) and Saleh and Aly (2013).

Rahman et al. (2011) developed an ANN model to predict energy requirement of a tillage tool from the laboratory data. The ANN model was trained and tested with soil moisture content, plowing depths, and forward operating speeds as input parameters. The measured energy requirement for a tillage tool in silty clay loam soil was used as output parameter. Their results showed that the variation of measured and predicted energy requirement was small.

The literature studies have shown that empirical models can be useful alternative and practical tool for predicting both draft and energy requirement of tillage implement under different conditions. Therefore, the objective of this study was to develop, evaluate and validate a new ANN model to predict draft, unit draft and energy requirements of a disk plow. Several common and readily available factors were used as inputs to the ANN. Using these inputs, the ANN was trained based on literature data and later validated with data from field experiments.

MATERIALS AND METHODS

a) Field experiment site and procedure: The field test was carried out at Heef Alqahtani farm, Riyadh, Saudi Arabia which is 468.97 m above sea level and lies on longitude 47.14deg East and latitude 24.33deg North. The purpose of the field experiment was to determine draft force for disk plow to acquire data for developing and testing the ANN model. The soil at the site was loam sandy. Average soil density and soil moisture content were 1.84 g/cm3 and 5.8 %db, respectively. All laboratory and field tests were conducted according to the recommendation of the Regional network for Agricultural machinery (RNAM, 1983).

Cone index values were obtained by taking penetrometer readings over the plowing depth. The cone used was of ASAE standard with a 30deg cone angle and a diameter of 12.83 mm. Table 1 shows soil characteristics at the site. Disk plow (Nardi, mounted category II, weight 362 kg, Italy), model MF 38, serial No. TDPE48/D, with 3 disks with 36 cm disk diameter and distance between disks was 60 cm, the disk angle and tilt angle were measured and the angles set at 45deg and 16deg, respectively, were used in the experiment. The disk plow was hitched to a Fendt tractor model 306 LSA. The auxiliary tractor was Fendt tractor model 312LSA. Three plowing speeds were obtained by changing gears of the tractor.] An experimental block 60 m long by 3 m wide was used for each treatment. A small block of approximately 10 m long by 3 m wide in the beginning of each tested block was used to enable the tractor and plow to reach the required plowing speed and plowing depth. The depth of cut was measured with a steel tape from the bottom of the furrow to the surface level of the soil at eleven randomly selected places. The horizontal force (draft) was measured using a load cell (model Omega with capacity 0-10,000 lb) using the method described in (PAES, 2001). The draft was recorded within the distance of 60 m. The plowing speed was calculated by measuring of distance of five turns of the tractor rear wheel with time. On the same field, the plow was lifted out the ground and the rear tractor was pulled to record the idle draft force. The difference gave the draft of the implement. Raw data of the field tests are presented in Table 2.

b) Data Collection: Available draft data of disk plows in the literature, which directly related to the subject, were collected from (Bukhari et al., 1992; Olatunji, 2011; Abu-Hamdeh and Reeder, 2003; Kheiralla et al., 2004; Al-Janobi and Al-Suhaibani, 1998; Makki and Mohamed, 2008; El-Shazly et al., 2008; Vozka, 2007). The above mentioned studies performed field or laboratory experiments using different disk plows in soils having different moistures, bulk densities and textures with different changeable working conditions and disk geometries. Collected data set consisted of 130 data points. Table 3 lists the summary statistics for the inputs used for training and testing the developed ANN model. Table 4 lists the corresponding statistics for the outputs.

Table 1. Mean characteristics of the soil at the site of the experiment.

###Unit###Value

Sand###(%)###83.2

Silt###(%)###9.8

Clay###(%)###7.0

Soil bulk density at depth (0-25 cm)###(g/cm3)###1.52

Soil moisture content at depth (0-25 cm)###(% db)###6.44

Cone index at depth (0-25 cm)###(kPa)###1517

Table 2. Raw data of the field tests during plowing with disk plow.

###Distance of five turns of the tractor###Time of five turns of the###Plowing###Draft###Plowing

###rear wheel###tractor rear wheel###speed###depth

###(m)###(sec)###(km/h)###(kN)###(cm)

###24.35###60.65###1.45###5.09###23

###22.97###36.43###2.27###5.43###23

###28.50###36.17###2.84###5.86###23

c) Calculation of energy and unit draft: The width of cut (w, cm) may be adjusted on standard disk plows by changing the angle of the disks with respect to forward motion (Dumitru, 2009). In this study, actual width of cut (w) of a disk can be expressed as (Alam, 1989):

(Equation) where Beta is disk angle (deg), Alpha is tilt angle (deg), D is disk diameter (cm), d is plowing depth (cm). The total plowing width (W, cm) is calculated as: W = N ' w . (2) where N is the number of disks on the plow. The tillage energy was calculated according to Smith (1993) as follows (Equation) where F is draft force and V is plowing speed. Field efficiency is the most frequently used factor in determining the effective field capacity (Lar et al., 2011). Thus, in this study, the effective field capacity (EFC) was calculated according to the following equation EFC (ha/h) TFC (ha/h) where Eta is the field efficiency and for tillage primary it was ranged from 70 to 85% (Powell, 2000). In this study, Eta was assumed to be 0.8. Further, TFC denotes theoretical field capacity and was calculated according to the following equation:

(Equation) Unit draft (UD) is calculated according to the following equation: (Equation)

Table 3. Statistics for the inputs used in training and testing the ANN model.

Statistic###Plowing###Plowing###Soil moisture###Disk###Tilt###Sand###Silt###Clay###Disk###Soil

###depth###speed###content###angle###angle###diameter###density

###(cm)###(km/h)###(%db)###(deg)###(deg)###(%)###(%)###(%)###(cm)###(g/cm3)

Mean###16.06###4.27###16.57###45###20###55.08###16.79###28.14###63###1.44

Standard

###3.95###1.97###6.66###2.83###2.34###28.41###16.04###19.22###4.77###0.15

deviation

Kurtosis###-0.85###1.64###-0.51###6.79###-0.93###-1.50###-0.30###-1.17###0.15###-1.34

Skewness###-0.44###1.34###0.37###2.58###0.05###-0.13###1.19###0.18###-1.29###0.32

Minimum###6.70###1.20###4.90###40.00###15.00###9.00###4.00###4.55###53.00###1.22

Maximum###23.40###10.00###28.00###55.00###25.00###90.85###48.00###70.00###66.00###1.67

Count###130###130###130###130###130###130###130###130###130###130

Table 4. Statistics for the targets (outputs) used in training and testing the ANN model.

###Statistics###Draft###Unit draft###Energy

###(N/disk)###(N/cm2/disk)###(kW.h/ha/disk)

###Mean###3828.95###2.48###12.88

###Standard deviation###2033.12###1.60###7.02

###Kurtosis###-1.07###0.53###-0.85

###Skewness###0.52###1.08###0.53

###Minimum###920.00###0.46###3.04

###Maximum###8000.00###8.02###29.81

###Count###130###130###130

d) ANN Background: ANNs are considered as artificial intelligence modelling techniques. They have a highly interconnected structure that aims to mimic the connectivity of brain cells. ANN consists of a large number of processing elements called neurons, which are arranged in different layers in the network: an input layer, an output layer and one or more hidden layers (Kumar and Singh, 2008). The basic working mechanism of a neuron is shown in Figure 1, where the neuron in the network receives input signals, processes them and sends an output signal (Haykin, 1999). Each neuron is connected with at least one other neuron and each connection is represented by a real number called a weight.

The weights are adjusted iteratively so that the network attempts to produce the desired output (Safa et al., 2009). Mathematically, this can be represented as (Haykin, 1999):

(Equation) where wk represents the weight vector, xk is the input vector (k = 1, 2..., m), bk is the bias, f is the transferfunction, and yk is the output. The logistic sigmoid transfer (activation) function was chosen and it is defined for any variable S as:

(Equation) One of the well-known advantages of ANN is that the ANN has the ability to learn from the sample set, which is called training set, in an iterative learning process. Once the architecture of network is defined, a learning algorithm is used to search for the weights that produce the desired output with minimal error. Back- propagation neural networks represent a supervised learning method, requiring a large set of complete records, including the target (output) variables. As each observation from the training set is processed through the network, an output value is produced from output nodes. These values are then compared to the actual values of the target variables for this training set observation and the errors are calculated at each step of the algorithm (Kumar and Singh, 2008).

Feed forward ANNs are currently being used in a variety of applications with great success. Their first main advantage is that they do not require a user- specified problem solving algorithm (as is the case with classic programming) but instead they "learn" from examples, much like human beings. Their second main advantage is that they possess inherent generalization ability, when trained properly. This means that they can identify and respond to patterns that are similar but not identical to the ones with which they have been trained (Anantachar et al., 2010).

e) Disk plow performance modeling with ANN: In the present work, the ANN model with standard back- propagation algorithm was developed using a commercially available software Qnet 2000 (Vesta Services, 2000). The ANN developed in the present study was characterized by three layers: an input layer, one corresponding target vectors are used to train the network until it can approximate a function which associates input vectors with specific output vectors. The inputs to the ANN model in this study were plowing depth, plowing speed, sand content, silt content, clay content, soil moisture content, disk diameter, disk angle, tilt angle and soil density. The outputs of the ANN were draft, unit draft and energy requirement of disk plows.

The data set (a total of 130) were randomized and were used in training and 22 points were randomly selected by an in-built algorithm in the software for the testing set. The test points provide an independent measure of how well the network can be expected to perform on data not used to train it. Prior to their use in the ANN model, the input and the output values were normalized between 0.15 and 0.85 according to the following equation: (Equation) where t is the original values of input and output parameters, T is the normalized value; tmax and tmin are the maximum and minimum values of the input and the output parameters, respectively.

Three and four layers ANN structures were investigated and the number of neuron in the hidden layers was also varied. Further, different values of the learning coefficient and the momentum factor, different transfer functions were used in training the network. The best ANN structure and optimal values of the network parameters were obtained on the basis of lowest error on training data by trial and error. The final learning rate was 0.05, a momentum factor of 0.8, logistic function (sigmoid transfer function) of neuron activation, 100,000 training cycles were used. These configurations gave a training error of 0.024. Root mean square error (RMSE) during training is shown in Figure 2. Table 5 presents network statistics from Qnet software.

Using the weights obtained from the trained ANN model, new formulations were presented in Appendix (A) for the calculation of draft, unit draft and energy requirements of a disk plow. These formulations can be employed with any programming language or spreadsheet program (Ahmed et al., 2012), thus facilitating their use in routine analysis.

f) Determination of errors in disk plow performance: The accuracy of ANN predictions was evaluated using different error statistics as follows:

(Equation) where Eiobs and Eipre are observed and predicted values, Nt is the number of data points, MAE is the mean absolute error and RMSE is the root mean square error. In addition, the coefficient of determination (R2) was selected to measure the linear correlation between the observed and the predicted values. The optimal coefficient of determination value is unity.

Table 5.Network statistics from Qnet software.

###Training data

###Output node###Standard deviation###Bias###Maximum error###Correlation coefficient

###Draft (N/disk)###282.136###-2.261###1226.592###0.990

###Unit draft (N/cm2/disk)###0.210###-0.003###1.202###0.991

Energy requirement (kW.h/ha/disk)###0.867###-0.001###4.336###0.992

###Testing data

###Draft (N/disk)###794.456###-0.997###2029.314###0.934

###Unit draft (N/cm2/disk )###0.583###0.016###1.797###0.933

Energy requirement (kW.h/ha/disk)###2.793###-0.175###7.229###0.915

RESULTS AND DISCUSSION

In this study, ANN model has been developed with 10 neurons in the input layer (plowing depth, plowing speed, sand content, silt content, clay content, soil moisture content, disk diameter, disk angle, tilt angle and soil density), 8 neurons in the hidden layer and 3 neurons in the output layer for the prediction of draft, unit draft and energy requirements of disk plows,

Error criteria such as R2, RMSE and MAE of training and testing phases are given in Table 6. For training phase, the coefficient of determination (R2), the root mean square error (RMSE) and the mean absolute error (MAE) were 0.983, 0.21 N/cm2/disk and 0.137 N/cm2/disk, respectively for unit draft. For the testing set, the corresponding values were 0.8713, 0.584 N/cm2/disk and 0.394 N/cm2/disk, respectively for unit draft.

Figures 3 through 5 show the relationships and coefficients of determination between the observed and the predicted values of draft, unit draft and energy requirements, respectively, using ANN model during testing phase. From these figures, it is clear that the

Table 6. Error criteria during training and testing process of ANN model.

###RMSE###MAE###R2

###Performance parameters

###Training###Testing###Training###Testing###Training###Testing

###Draft (N/disk)###282.136###794.456###187.171###567.845###0.9803###0.8733

###Unit draft (N/cm2/disk)###0.210###0.584###0.137###0.394###0.9830###0.8713

Energy requirement (kW.h/ha/disk)###0.867###2.793###0.587###1.910###0.9846###0.8374

points, during the testing process, were not uniformly scattered around the regression lines and coefficients of determination (R2) was 0.8733.,0.8713 and 0.8374, respectively as illustrated in Figures (3) through (5).

The Qnet algorithm computed the contribution percent which indicates how the change in each input changes the output prediction. The contribution percentage of the ten input variables to the outputs was calculated using the developed ANN model and results are illustrated in Figure 6 for draft, unit draft and energy requirements. This figure can be used to ascertain the relative contributions (and importance) of each of the ten input parameters. As evident from Figure 6, the input parameter with the largest contribution is the soil moisture content. This input parameter contributes about 23.44%, 17.36%, and 18.49%, respectively, for the predicted output obtained for draft, unit draft and energy requirements of disk plows by the developed ANN model.

To validate the developed ANN model, the field experiments data were tested with the newly developed ANN model to predict draft, unit draft and energy requirements of disk plows. Figure 7 depicts the relationship between plowing speed vs. field and predicted values of draft of a disk plow. It is clear that the predicted pattern behaves as field pattern (i.e. increasing plowing speed results in increasing draft force of a disk plow). However, the coefficient of determination (R2) as shown in Figure 8 between the actual data from the field experiment and the predicted values from the developed ANN model of draft of a disk plow was 0.983 which indicates accurate prediction of such draft.

Conclusion: ANN model was developed for the prediction of the performance parameters (draft, unit draft and required energy) of the disk plow. The appropriate architecture of the neural network was 10-8-3. Based on the results, the ANN model appears capable of providing accurate predictions of the disk plow's performance.

Acknowledgement: The authors express their gratitude to the National Plan for Science, Technology and Innovation Program, King Saud University, Saudi Arabia for financially supporting this research effort as a part of the project entitled "Modelling of energy consumption during seed bed preparation operation based on soil mechanical properties", No. 09-SPA876-02.

REFERENCES

Aboukarima, A. M. (2013). Evaluation of using a combination of simulated and experimental data to predict draft force of a moldboard plow. Misr J. Ag. Eng., 30 (1): 1-28. Aboukarima, A.M. (2004). Artificial neural networks configuration for predicting performance of tillage implements under Egyptian conditions. PhD Thesis, Agric. Eng. Dept., Faculty of Agric., Ain Shams University, Egypt.

Aboukarima, A.M. (2007). Draft models of chisel plow based on simulation using artificial neural networks. Misr J. Agric. Eng. 24 (1):42-61.

Aboukarima, A.M., and A.F. Saad (2006). Assessment of different indices depicting soil texture for predicting chisel plow draft using neural networks. Alexandria Science Exchange J. 27(2): 170-180.

Aboukarima, A.M., M.N. El Awady, A.G. El Kabany and M.H.A. Kabeel (2003). Plows performance under Egyptian conditions depicted by artificial neural networks. Misr J. Agric. Eng. 20 (4): 919- 936.

Abu-Hamdeh, N.H. and R.C. Reeder (2003). A nonlinear 3D finite element analysis of the soil forces acting on a disk plow. Soil and Tillage Research, 74:115-124.

Ahmed, M. I., M. B. Sharif, M. Yousaf, A. H. Khan and I. Rashid (2012). Optimization of reinforced concrete frames by using spreadsheet. Pakistan J. Sci., 64(3):196-202. Alam, M. M. (1989). Soil reaction forces on agricultural disk implements. PhD Thesis, University of Newcastle, UK. Al-Janobi, A., S. Al-Hamed and A.M. Aboukarima (2010). An excel spreadsheet to estimate performance parameters for chisel plow-tractor combination based on trained an artificial neural network. Bulletin USAMV Agriculture, 67(2):1-

9. Al-Janobi, A.A., A.M. Aboukarima and K.A. Ahmed (2001). Prediction of specific draft of different tillage implements using neural networks. Misr J. Agric. Eng. 18 (3): 669-714.

Al-Janobi, A.A. and S. A. Al-Suhaibani (1998). Draft of primary tillage implements in sandy loam soil. Applied Engineering in Agriculture, 14(4): 343- 348.

Anantachar, M., P.G.V. Kumar and T. Guruswamy (2010). Neural network prediction of performance parameters of an inclined plate seed metering device and its reverse mapping for the determination of optimum design and operational parameters. Computers and Electronics in Agriculture, 72(2): 87-98. Boydas, M.G. and N. Turgut (2007). Effect of tillage implements and operating speeds on soil physical properties and wheat emergence. Turk J Agric For, 31:399-412.

Bukhari, S., G.R. Mari, M. Zafaruhah, J.M. Baloch and M.S. Panhwar (1992). Effect of disk and tilt angle on field capacity and power requirements of mounted plow. Agricultural Mechanization in Asia, Africa, and Latin America, 23(2):9-13.

Dumitru, M. (2009). Studies and researches on tillage machines. Research J. of Agricultural Science, 41 (2):422-426. El Awady, M.N., A.G. El Kabany, M.H.A. Kabeel and A.M. Aboukarima (2003). Relative importance of variables affecting chisel-plow performance using neural networks. The 11th Conf. of Misr Society Agric. Eng., 15-16 October, Rice Mech. Centre, Meet El Deeba, Kafer El Sheikh Governorate: 395-407. El Awady, M.N., A.G. El Kabany, M.H.A. Kabeel and A.M. Aboukarima (2004). Predicting unit draft of tillage implements using statistical models and neural networks. The 12th Conf. of Misr Society Agric. Eng., 15-16 October, Alexandria Univ.: 139-249.

El-Shazly, M.A., M.M. Morad, M.M. Ali and K.I. Wasfy (2008). Optimization of disk plow performance under Egyptian conditions. Misr J. Agric. Eng. 25(1): 15-37. Gill, W.R. and G.E. Vanden Berg (1967). Soil Dynamics in Tillage and Traction. Agriculture Handbook No. 316. United States Department of Agriculture. Godwin, R.J., D.A. Seig and M. Allott (1987). Soil failure and force prediction for soil engaging disks. Soil Use and Management, 3(3):106-114.

Hassan, A.E. and A.S. Tohmaz (1995). Performance of skidder tires in swamps-comparison between statistical and neural network models. Transaction ASAE, 38 (5): 1545-1551.

Haykin, S. (1999). Neural Networks: a comprehensive foundation, Second Edition, Prentice Hall, 842pp.

Ismail, K.M. (2002). Statistical treatment of disk tool data for predicting soil draft. Misr J. Agric. Eng. 19(2):455-466.

Karmakar, S. (2005). Numerical modelling of soil flow and pressure distribution on a simple tillage tool using computational fluid dynamics. PhD Thesis, Department of Agricultural and Bioresource Engineering, University of Saskatchewan, Canada, 220p.

Kheiralla, A.F., A. Yahya, M. Zohadie and W Ishak (2004). Modelling of power and energy requirements for tillage implements operating in Serdang sandy clay loam, Malaysia. Soil and Tillage Research, 78:21-34. Kumar, A.J.P. and D. K. J. Singh (2008). Artificial neural network-based wear loss prediction for a390 aluminium alloy. J. of Theoretical and Applied Information Technology, 4(10):961-964. Kushwaha, R.L. and Z. X. Zhang (1997). Artificial neural networks modelling of soil-tool interaction. ASAE Paper No. 97-3067, ASAE, St. Joseph, Michigan, USA: 1-11.

Lar, M.B., Z. K. Pour and G. R. Bamimohammadi (2011). Field efficiency and its use for energy coefficient determination. J. of American Science, 7(8):599-603. Makki, E.K. and A. E. Mohamed (2008). Tillage Implements Performance and Effect on Some Soil Physical Properties. Agricultural Mechanization in Asia, Africa, and Latin America, 39(2):9-13.

Mamkagh, A.M. (2009). Effect of plowing speed, disk angle and tilt angle on farm tractor wheel slip and on plowing depth using disk plow. Jordan J. of Agricultural Sciences, 5(3):352-360. Manian, R., R.V. Rayan and K. Kathirvel (2000).

Influence of operating and disk parameters on performance of disk tools. Agricultural Mechanization in Asia, Africa, and Latin America, 31(2):19-26, 38. Morad, M.M. (1992). Effect of some operating parameters on disk plow performance. Misr J. Agric. Eng. 9 (1): 108-115.

Olatunji, O.M. (2011). Evaluation of plow disk performance on sandy loan soil at different soil moisture levels. Research J. of Applied Sciences, Engineering and Technology, 3(3): 179-184. Osman, A.N., L. Xia and Z. DongXing (2011). Effects of tilt angle of disk plow on some soil physical properties, work rate and wheel slippage under light clay soil. International J. of Agricultural and Biological Engineering, 4(2):29-35.

PAES (2001). Philippine Agricultural Engineering Standard (PAES 121: 2001 Agricultural Machinery - Disk Plow - Specifications,9 p. Panigrahi, B., J.N. Mishra and S. Swain (1990). Effect of implement and soil parameters on penetration depth of a disk plow. Agricultural Mechanization in Asia, Africa, and Latin America, 21(2):9-12 .

Powell, G. (2000). Selection and matching of tractors and implements., DPI's Agency for Food and Fibre Sciences. File No: FS0305. Date created: September 2000. Revised: N/A. http://www2.dpi.qld.gov.au/fieldcrops/3492.htm l. Rahman, A., R.L. Kushawaha, S.R. Ashrafizadeh and S. Panigrahi (2011). Prediction of energy requirement of a tillage tool in a soil bin using artificial neural network. ASABE Paper No. 111112. American Society of Agricultural and Biological Engineers, St. Joseph, Michigan, USA.

RNAM (1983). Test Codes and Procedures for Farm Machinery. Test Codes and Procedure for Plows. Regional Network for Agricultural Machines, Technical Series No. 12, Philippines. Roul, A.K., H. Raheman, M.S. Pansare and R. Machavaram (2009). Prediction of draft requirement of tillage implements in sandy clay loam soil using an artificial neural network. J. of Biosystems Engineering, 104(4): 476-485. Safa, M., S. Samarasinghe, and M. Mohsen (2009).

Modelling fuel consumption in wheat production using neural networks. 18th World IMACS / MODSIM Congress, Cairns, Australia 13-17 July 2009:775-781. Saleh, B.,and A. A. Aly (2013). Artificial neural network model for evaluation of the plowing process performance. International J. Control Automation and Systems, 2(2):1-11.

Shirin, A.K.M., V.M. Salokhe and M. Hoki (1993). Effects of disk and working parameters on the performance of a disk plow in clay soil. Agricultural Mechanization in Asia, Africa, and Latin America, 24 (4): 9-12.

Smith, L.A. (1993). Energy requirements for selected crop production implements. Soil and Tillage Research, 25:281-299.

Sommer, M.S., S. H. Chen and J. F. Bierl (1983). Disk blade performance. ASAE Paper No. 83-1537, American Society of Agricultural and Biological Engineers, St. Joseph, Michigan, USA, 12p. Tohmaz, A.S. and A. E. Hassan (1995). Application of artificial neural networks to skidder traction performances. J. of Terramechanics, 32 (3): 105- 114.

Vesta Services, Inc. (2000). Qnet2000 Shareware, Vesta Services, Inc., 1001 Green Bay Rd, STE 196, Winnetka, IL 60093. Vozka, P. (2007). Comparison of alternative tillage systems. MSc Thesis, School of Applied Sciences, National Soil Resources Institute, Cranfield University, UK, 110p. Zhang, Z.X., and R. L. Kushawaha (1999). Application of neural networks to simulate soil-tool interaction and soil behaviour. Canadian Agricultural Engineering, 41(2): 119-125.

Appendex (A): The following equations were used to predict draft, unit draft and energy requirements of a disk plow based on training ANN model: Step 1: Normalizing the original input variables as follows: (Equations) Step 2: Computing the sum of input signals as follows: (Equations) Step 3: Applying the sigmoid transfer function on the sum of input signals as follows: (Equations) Step 4: Computing the sum of hidden signals as follows: (Equations) Step 5: Computing the normalized output signals as follows: (Equations) Example: Predict draft (N/disk) of a disk plow, plowing speed was 7.85 km/h, plowing depth was 10 cm in soil having 79% sand, 10% clay and 11% sill. Disk diameter was 66 cm, disk angle was 45deg, tilt angle was 22deg, the soil moisture content was 9.5 % (d.b) and the soil density was 1.58 g/cm3.

Then by applying equation (33-a), the draft value is obtained and equals to 2378.51 N/disk. The observed draft was 2433.33 N/disk. So, the error was (2433.33-2378.51)/2433.33 100 =2.25%.
COPYRIGHT 2013 Asianet-Pakistan
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:Journal of Animal and Plant Sciences
Article Type:Report
Date:Dec 31, 2013
Words:5357
Previous Article:PERCENTAGE UPTAKE OF HEAVY METALS OF DIFFERENT MACROPHYTES IN STAGNANT AND FLOWING TEXTILE EFFLUENT.
Next Article:CLONAL DIFFERENCES OF BLACK POPLAR CUTTINGS FOR MORPHO- PHYSIOLOGICAL AND BIOCHEMICAL RESPONSES TO SOIL WATER DEFICITS.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters