International Journal of Advanced Scientific and ... - RS Publication

International Journal of Advanced Scientific and Technical Research Available online on http://www.rspublication.com/ijst/index.html

Issue 5 volume 6, September-October 2015 ISSN 2249-9954

Using Collocation Neural Network Technique to Estimation the Concentration of Nickel in Soil Saad Ali Ahmed College of Education for Pure Science Ibn Al-Haitham, Baghdad University. Email: [email protected]. edu.iq

Abstract The aim of this paper is to presents a parallel processor technique to estimate the concentration of Nickel in soils. We suggest collocation neural network as an alternative accurate technique. The proposed network is learned by back propagation with Levenberg-Marquardt training algorithms. The next objective of this paper was to compare the performance of aforementioned algorithms with regard to predicting ability. Then applied the suggest technique to estimate the concentration of Nickel in Baghdad soils in Iraq. Keywords: Artificial neural networks (ANN), Training algorithm, Soil, Nickel.

1. Introduction The heavy metals in the soil have the peculiarities of long accumulate time, complexity of existent state and toxicity. Their bioavailability is harmful for the plants, human beings and environment [1]. Nickel (Ni) is one of harmful elements in soil [2, 3]. It can harm human health via food chain. The harm of heavy metals is related to the total concentration of heavy metals and especially to the chemical species of heavy metal [4-6]. So, the researches of heavy metals species and their distribution in soil sample have important guiding significance to the researches on the effects of heavy metals to human health and environment conservation. There are many method which an effective chemical operational procedure for species analysis [7-9]. But it is difficult to choose the highly selective extractants and the interference among various species is unavoidable in the sequential extraction procedure [10]. Another problem is that the sequential extraction procedure will finish in a week approximately. So, it is necessary to develop a fast and sensitive method for determining various species of heavy metals in soil simultaneously. Artificial neural network (ANN) a simplified mathematical model of the human brain; it can be implemented by both electric elements and computer software. It is a parallel distributed processor with large numbers of connections; it is an information processing system that has certain performance characters in common with biological neural networks [11]. It has been applied in many fields [12-15]. In this paper we suggest collocation neural network (CNN) to estimate the concentration of Nickel in soils and to illustrate the efficiency and accuracy of suggest technique we used this technique to estimate Ni in soil of Baghdad city in Iraq. Page 93

RSPUBLICATION, [email protected]



2. Architecture of the Collocation Neural Network In the proposed work, the suggested design of CNN is as three layer network: input layer which consist three input unit, hidden layer consist eight hidden unit with tansig. sigmoid transfer function and output layer consist one output unit. The suggested network is learned by back propagation with Levenberg-Marquardt training algorithms (see Figure 1). In the proposed approach the model is expressed as the sum of two terms: the first term satisfies the initial concentration (C0) and contains no adjustable parameters. The second term can be found by using CNN which is trained so as to estimate the concentration of Ni and such technique called collocation neural network. In this section we will illustrate how our approach can be used to estimate the concentration of Ni. Let 𝑦𝑡(𝑥, 𝑝) denotes a trial concentration with adjustable parameters 𝑝, correspond to the weights and biases of the neural architecture. In our proposed approach, the trial concentration 𝑦𝑡 employs a CNN and we choose a form for the trial function 𝑦𝑡(C) such that it satisfies the initial concentration (C0). This is achieved by writing it as a sum of two terms: 𝑦𝑡 (C𝑖, 𝑝) = C0 + 𝐺(𝑥, t, 𝑁(C, 𝑝)) , (1) where 𝑁(C, 𝑝) is a multi-layer CNN with parameters 𝑝, depth x, time t and 𝑛 input units fed with the input vector C. The weights and biases are to be adjusted in order to deal with the minimization problem. A trial solution can be written as: 𝑦𝑡 (C, 𝑝) = C0 + (𝑥 – x0) (t – t0) 𝑁(𝑥,𝑝) , (2) where 𝑁(𝑥, 𝑝) is the output of a CNN with one three input unit for 𝑥, t, C0 and weights 𝑝.

Figure 1: Architecture of suggested CNN

3. Choosing of Training Set The choosing of training set is the key of accurate discretionarily of the concentration of Ni in soil samples. For the choosing of the training set, from 40 soil samples from different sections were obtained by atomic absorption spectrometry (AAS). We choose training set was consisted of 20 samples of concentrations of Ni, which were selected discretionarily from 40 soil samples. The effect of training set on estimating result was tested, we choose testing set was consisted of 10 samples of concentrations of Ni were selected discretionarily in four samples and choose validation set was consisted of 10 samples of concentrations of Ni, were the soil samples selected from the industrial sites, communities and agricultural land in Baghdad city. The results illustrated in Figure 2.

Page 94




Figure 2: Comparison between the results of CNN by the training, testing and validation sets & AAS

4. Selection of Learning Rate Learning rate, η, is a learning step (step size), where η (0 Commercial, but in Open area soil, we see that Industrial > Commercial > residential. For more detail Ni content varies from 72.63 to 85.32 mg/kg. The observed values are higher than the world average concentration of Ni in soil which is around 20 mg/kg [16]. Results exceed the calculated world mean of unpolluted soil (34 mg/kg) [17]. Further, considering the analyzed values of rural (control) as 35.75, indicated than Ni median concentration in the all samples of the study area was observed to be 2.2 times than Ni content in the rural soil. It is evident that local solid waste and anthropogenic activities such as burning of fuel contribute to the increase in Ni content in the soil of the study area. It may be noted that many domestic cleaning products, e.g. soap, 100 - 700 mg Ni/kg; powdered detergents, 400 - 700 mg Ni/kg and powdered bleach, 800 mg Ni/kg, may prove to be important sources of Ni in the urban soils [16].

6. Conclusions The concentration of nickel can be estimated by collocation neural network. The 40 soil samples from the different sections of Baghdad city in Iraq, were estimated by this technique. The prediction errors of this technique are less than 10% compared with those of AAS. This technique is fast, convenient, sensitive, and can eliminate the interference among various species. The determination of sample by AAS is finished in 7 d; however the determination of sample by the proposed technique is finished in 1 d. So the CNN is an effective technique for the determinations of various species content of heavy metals in soil simultaneously.

References [1] Martin-Dupoint, F., Gloaguen, V., Granet, R., Guilloton, M., Morvan, H., and Krausz, P., (2002), Heavy metal adsorption by crude coniferous barks: a modeling study, J. Environ. Sci. Health A37, pp: 1063–1073.

Page 96




[2] Pichtel, J., Sawyerr, J. H. T., and Czarnowska, K., (1998), Spatial and Temporal Distribution of Metals in Soils in Warsaw, Poland,” Environmental Pollution, Vol. 98, No. 2, pp: 169-174. [3] Kim, K. and Kim, S., (1998), Heavy Metal Pollution of Agricultural Soils in Central Regions of Korea, Water Air and Soil Pollution, Vol. 82, pp: 109-122. [4] Odoh, R., Agbaji, E. B., Dauda, M. S., and Oko, O. J., (2014), Assessment of Soils in the Vicinity of Rice Mill Industry, Otukpo, Benue State for Potential Heavy Metal Contaminations, International Journal of Modern Analytical and Separation Sciences, Vol. 3, No. 1, pp: 1-12. [5] Gray, C.W., McLaren, R. G., and Roberts, A. H. C., (2003), Atmospheric Accessions of Heavy Metals to Some New Zealand Pastoral Soils, The Science of the Total Environment, Vol. 305, pp: 105-115. [6] Banat, K. M., Howari, F. M., and Al-Hamad, A. A., (2005), Heavy Metals in Urban Soils of Central Jordan: Should We Worry about Their Environmental Risks, Environmental Research, Vol. 97, pp: 258-271. [7] Semhi, K., Khirbash, S. A., Abdalla, O., Khan, T., Dupley, J., Chaudhuri, S., and Saidi, S. A., (2010), Dry Atmospheric Contribution to the Plant-Soil System around a Cement Factory: Spatial Variations and Sources-A Case Study from Oman, Water Air and Soil Pollution, Vol. 205, pp: 343-357. [8] Jeon, C., Nah, I.W., Hwang, K.-Y., (2007), Adsorption of heavy metals using magnetically modified alginic acid, Hydrometallurgy, Vol. 86, pp:140–146. [9] Kamel, M. M., Ibrahm, M. A., Ismael, A. M., and El-Motaleeb, M. A., (2004), Adsorption of some heavy metal ions from aqueous solutions by using kaolinite clay, Ass. Univ. Bull. Environ. Res., Vol. 7, pp: 101–110. [10] Ntui, N. T., Hassan, U. F., and, Ushie, O. A., (2014), Determination of Heavy Metals Concentration in Cow Dung of Grazing Cattle in Bauchi Urban Area, Nigeria, International Journal of Modern Analytical and Separation Sciences, Vol. 3, No.1, pp: 13-19. [11] Tawfiq, L. N. M., (2013), Improving Gradient Descent method For Training Feed Forward Neural Networks, International Journal of Modern Computer Science & Engineering, Vol. 2, No. 1, pp: 1-25. [12] Yetilmezsoy, K., and Demirel, S., (2008), Artificial neural network (ANN) approach for modeling of Pb(II) adsorption from aqueous solution by Antep pistachio (Pistacia Vera L.) shells, Journal of Hazardous Materials, Vol. 153, pp: 1288–1300. [13] Khonde, R. D. and Pandharipande, S. L., (2011), Application of Artificial Neural Network for Standardization of Digital Colorimeter, International Journal of Computer Applications, ICCIA-5, pp: 1-4. [14] Tawfiq, L. N. M., and Ali, M. H., (2012), Fast Feed Forward Neural Networks to Solve Boundary Value Problems, Lap lambert Academic Publishing. [15] Pandharipande, S. L., Deshmukh, A. R., and Kalnake, R., (2013), Artificial Neural Network Modelling For Estimation of Concentration of NI (II) and CR (VI) Present in Aqueous Solution, International Journal of Advances in Engineering & Technology, Vol. 5, Issue 2, pp. 122-131. [16] Alloway, B. J., (1995), Heavy Metals in Soils, Blackie Academic and Professional, London. doi:10.1007/978-94-011-1344-1. [17] Kabata-Pendias, A., and Pendias, H., (2001), Trace Element in Soils and Plants, CRC Press, London. Page 97
