Application Of Artificial Neural Network Biology Essay

Published:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Water eminence model is a helpful utensil to assess the prospect state of river water through assessment of real pollution loading or singular management options. Based on the human brain physiology research, artificial neural network (ANN) which replicates the structure and mechanism of the human brain is a type of dynamic information processing system that finally accomplishes positive functions of the human brain. Now a day, there are hundreds of neural network approaches and also Back Propagation neural network is one of the most frequently used techniques in current days, whose application in environmental eminence assessment of environmental science field was presented here. Back Propagation neural network was used to create the landscape water eminence assessment system. Here also presented an optimization method of Back Propagation artificial neural network, which generally used MATLAB programming whose toolbox afforded a function of Back Propagation network to avoid complex mathematical calculations and cumbersome code editor. On the base of sampling and examining of water eminence constantly, artificial neural network was used here to create landscape water assessment model in order to estimate the landscape water eminence neutrally and speedily.

Keywords: Back Propagation artificial neural network, landscape water eminence assessment, MATLAB

INTRODUCTION

With the growth of city as well as people's increasing necessities on environmental worth, the Landscape Rivers and canals play a significant role in civilizing and beautifying living environment. Yet, these landscape rivers and canals, which are commonly static or closed slow-flow water bodies, are simple to be polluted for being small water areas, having little environmental capacity and small self-purification capacity. By research proves that more than 90% of the landscape water bodies in Gujarat are subject to varying degrees of pollution. Hence, protection, subrogation and supplementary of landscape water has become a significant question in many cities particularly the cities which be short of water critically.

Landscape River of 33 km long studied, after the conclusion of the landscape river, the superposition impact of the non-point source pollution in surface runoff and sewage discharge of the individual enterprises result to serious water eminence corrosion, and at the same time, low environmental capacity and ecological carrying capacity of landscapes also greatly reduces landscape function. Hence, it is essential to build an inclusive assessment of the polluted landscape water body, and take the assessment into daily management to afford a basis for landscape river pollution control and water eminence management.

Now a day, between all the water eminence assessment approaches, the classic approaches are single-factor evaluation method, multi-factor key evaluation method, fuzzy mathematical evaluation method, gray system evaluation method, analytic hierarchy process (AHP), matter-element analysis, artificial neural network evaluation method and recently proposed approaches such as water quality identification key method, gray fuzzy clustering evaluation method. Each has their individual compensations and drawbacks. For example, the result of key assessment is frequently incompatible with the data of water eminence operational examined, and parameters of different key formula are frequently dissimilar too. The drawback of gray assessment is small motion. With complex assessment process and poor operability, Fuzzy assessment results are not equivalent. And AHP has troubles of low resolution and illogical assessment results. Hence, Back Propagation artificial neural network becomes one of the mainly important and widely used models due to its particular compensations.

Now a day, between all the water eminence assessment approaches, the classic approaches are single-factor evaluation method, multi-factor key evaluation method, fuzzy mathematical evaluation method, gray system evaluation method, analytic hierarchy process (AHP), matter-element analysis, artificial neural network evaluation method and recently proposed approaches such as water quality identification key method, gray fuzzy clustering evaluation method. Each has their individual compensations and drawbacks. For example, the result of key assessment is frequently incompatible with the data of water eminence operational examined, and parameters of different key formula are frequently dissimilar too. The drawback of gray assessment is small motion. With complex assessment process and poor operability, Fuzzy assessment results are not equivalent. And AHP has troubles of low resolution and illogical assessment results. Hence, Back Propagation artificial neural network becomes one of the mainly important and widely used models due to its particular compensations.

Back Propagation artificial neural network initially can improve computing speed of algorithm, secondly it is conductive to overcoming the prejudice brought about other algorithms of water eminence assessment, and thirdly it can resolve the non-linear complex relationship problems of research object.

Back Propagation artificial neural network initially can improve computing speed of algorithm, secondly it is conductive to overcoming the prejudice brought about other algorithms of water eminence assessment, and thirdly it can resolve the non-linear complex relationship problems of research object.

WATER EMINENCE EXAMINING AND DETERMINATION OF THE ASSESSMENT POINTERS

LANDSCAPE WATER EMINENCE EXAMINING

CHOICE OF SAMPLING SITES

Six sampling sites are chosen on the base of field research survey and the layout principle of sampling combining the need of contents. On the upstream river, there is a rainfall pumping location. As a major pollution source of the landscape river, pumping location mostly discharges flood season rainwater into the rivers as well as the water accumulated in pipeline. Thus sampling site one is chosen on the outlet section of the rainfall pumping location, and two to six sampling points are correspondingly preferred along water flow direction. Among them, sampling site two is preferred near an industrial outfall, point three and four are selected close to each other, in order to learn the role of aquatic plants on the degradability of pollutants, and section three is selected on the reach with lush aquatic plants, while section four is on the reach fundamentally lacking aquatic. Also site two, five and six are selected on the downstream river. Examining on these Sampling site is carried constantly.

INVESTIGATION AND MONITORING WATER EMINENCE KEYS

According to Surface Water Environment eminence Standard of Gujarat, India combined with the actual pollution situation of the landscape river as well as the analysis monitoring and experimental conditions, eight water eminence keys including water temperature, dissolved oxygen (DO), pH, total nitrogen, ammonia nitrogen, total phosphorus, and chemical oxygen demand (COD), and chloride are monitored. All the keys are examined according to the standard method of India. Table 1 shows part of the monitoring data.

Table - 1 Standard Monitoring results of the landscape river water eminence in November 2012

Sampling Site

Dissolve Oxygen (Mg / Ltr)

NH3-N (Mg / Ltr)

Chloride (Mg / Ltr)

Chemical Oxygen Demand (Mg / Ltr)

TN (Mg / Ltr)

TP (Mg / Ltr)

1

4.30

6.16

442

16

3.67

0.60

2

3.58

3.92

464

24

2.98

0.52

3

3.47

6.16

802

70

4.26

0.57

4

2.28

1.12

542

40

3.19

0.51

5

7.38

0.28

162

4

4.55

0.59

6

3.28

5.04

536

28

2.79

0.51

ASSESMENT CRITERIA AND DETERMINATION OF THE ASSESMENT KEYS

India surface water environmental eminence standard as Table 2 shown is used as the landscape water eminence assessment criteria. Complementary to the standard, it can be seen that among all the data examined at the six sampling sites, the concentration of TN, COD, NH3-N do exceed the criteria for class V. Generally speaking, when the concentration of total phosphorus and inorganic nitrogen in water body reaches 0.02mg/L and 0.3mg/L respectively, it marks that water is eutrophicated. Therefore, four water eminence keys including DO, NH3-N, Chloride, COD, TN, COD, and TP are selected as water eminence assessment keys.

According to the Indian surface water environmental eminence standard and pollution situation of the landscape river, assessment criteria with four keys of five grades is projected for calculating the landscape water eminence, as shown in Table 2.

Table - 2 Assessment criteria of landscape water pollution degree table type techniques

Evaluation Index

Index

Dissolve Oxygen (Mg / Ltr)

NH3-N (Mg / Ltr)

TN (Mg / Ltr)

TP (Mg / Ltr)

Classification Index

I

15

0.5

0.5

0.1

II

20

1.0

1.0

0.2

III

20

1.5

1.5

0.3

IV

30

2.0

2.0

0.4

V

40

2.5

2.5

0.5

ESTABLISHING LANDSCAPE WATER EMINENCE ASSESMENT MODEL WITH BACK PROPAGATION NETWOTK

CREATING THE METHOD

Back Propagation artificial neural network is one of the most important and commonly used models. The most well-known design characteristic of Back Propagation artificial neural network is that the network weight is accustomed constantly through making the quadratic total of the error between the network model output weight and the consistency sample output to reach the preferred.

Back Propagation artificial neural network is a feed forward neural network which has an input layer, one or more hidden layers and one output layer. The connection between neurons in the adjacent layers is similar and non-connected. Apart from for input layer, there exists non-linear connection between the input layer and output layer in each neuron. The following S-function, as in equation 1 is taken, which is frequently used in Back Propagation neural network.

(1)

The middle subject of Back Propagation algorithm is adjusting the weight of the network to the total error smallest. Algorithm of Back Propagation model learning method consists of positive propagation and back propagation. Through repeated positive propagation and back propagation, neurons weights and thresholds of layers are customized constantly to decrease the error function till the error function is no longer sinking. Heart of the learning process is to struggle for the least value of the error function, which guides sample set constantly. Before accomplishment the smallest error, each training time, the right of the value of the error function changes along the steepest fall direction, finally converges to the least point, and after that saves information of connection weights achieved from the training samples and bias value, in order to deal with the unidentified samples. Determining the network structure is formative the network layer number and neurons number of different layers. presently, there are still no standardized calculating formulas for number of hidden layers and neurons, while many national or international scholars make a number of empirical formulas, such as "2N+1" proposed by Hecht-Nielsen, in which N represents node in the input neurons.

The computation steps of Back Propagation neural network can be explained as follows:

(1) Initialize the network weights. Weight initialization mode is for the 1 / K, where K is associated to the joint point of all the former layer node number.

(2) Train data set. The signal is broadcasted through the network to calculate the starting levels from the first lay of each node output until the output of output layer of each node are acquired.

(3) Determine error value of each node of the output layer.

(4) Determine error value of each node of the preceding layers (for S-function) which is achieved by error back-propagation through the layer by layer, until error value of each node of each layer is designed.

(5) Adjust weight and threshold value.

(6) Determine the error function value, until it arrives at the programmed least error value.

(7) Save weights and thresholds resulting from training in order to process the input of the predicted samples, thus the output of the forecasted samples can be considered.

Model launching is the key to the achievement of the network training. Because the Back Propagation network itself has a few limitations, optimizations are made on the network during model founding process as follows. Firstly, the hidden layers and the hidden layer neuron nodes that four input layer nodes, namely four keys of water eminence assessment are determined; one hidden layer of nine initial nodes; one output layer of five nodes, that is to say, there are five grades of the water eminence assessment. The network is exposed in Figure 1. Secondly, pretreatment on the sampling data, normalizing the training sample set and the desired target output with PREMNMX of MATLAB are necessary. And last, the structure of the network is completed. Function "Newff" is used to create Back Propagation network; function "logsig" and "purelin" are neurons transfer functions of each layer; and function "trainlm" is the network training function.

Fig-I.TIF

Fig. 1 Neural network structure diagram of landscape water eminence assessment

NETWORK PREPARATION

Taking the five water eminence grading standards in Table 2 as training samples, the five nodes preferred outputs of the consequent output layer are shown in Table 3.

Table 3 Expectation output of output layer node

I

II

III

IV

V

1

0

0

0

0

0

1

0

0

0

0

0

1

0

0

0

0

0

1

0

0

0

0

0

1

Back Propagation network training process curve of trainlm (network training) function is exposed in Figure 2. Take the training coefficient rate η as 0.01, and train constantly on the computer until the error total of all training samples is no more than the required value(ε ), here ε =0.001. After that stop learning and save the weights and thresholds among layers derived from training. Standard sample output is exposed in Table 4.

Fig-2.TIF

Fig. 2 Back Propagation network training process curve of training function

Table 4 Standard Sample Output

I

II

III

IV

V

0.9858

0.0009

0.0035

0.0025

0.0027

0.0008

1.0004

0.0003

0.0087

0.0006

0.0002

0.0007

0.9811

0.0057

0.0008

0.0009

0.0011

0.0040

0.9922

0.0028

0.0005

0.0005

0..0011

0.0022

0.9828

ASSESSMENT RESULTS AND STUDY

Taking the average of the examining COD, ammonia nitrogen, total nitrogen, and total phosphorus at the six sampling locations of the landscape river (as shown in Table I) as model input variables, assessment results are shown in Table 5.

Table 5 Water eminence evaluation network results of landscape

Sampling sites

I

II

III

IV

V

Control Level

1

0.0598

0.0911

0.0639

0.0576

1.5007

V

2

0.0299

0.0379

0.1411

0.0841

1.4599

V

3

0.0257

0.05417

0.1245

0.0745

1.3700

V

4

0.0229

0.0712

0.0999

0.0628

1.3655

V

5

0.1115

0.1109

0.1145

0.0511

1.1511

V

6

0.1176

0.0167

0.0139

0.0433

1.0111

V

According to Table 5, all the top weights of network output between the six sampling sites are in Grade V water eminence, and the weight in Grade I、II、III、IV is extreme less than it in Grade V. Therefore the landscape river is measured polluted dangerously because it doesn't meet the requisite of the landscape water eminence as an entire. In universal case, landscape water eminence at least wants to reach Grade V standard. In addition, with the reduction of the highest weight of the five sampling sites from point 1 to 5, it shows that water eminence regularly improves along the stream path of river. The weights of point 1 and 2 are very similar, which are mainly attributed to the comparable water eminence in the two points. Point 1 is chosen next to the rainfall pumping site which discharged flood season rainwater into the landscape water body, and there may be enterprise pollution in point 2. It can be accomplished that the water plants are helpful to the improvement of river water eminence throughout evaluating water eminence of point 3 and 4. The water eminence of point 5 and 6 which are taken in the river downstream is superior to that of other points but not yet meets the standards. On the whole, the landscape river studied is dangerously polluted because of the high weight of Grade V among all the sampling sites. And the in general of landscape water eminence does not reach the standard.

In addition, gray fuzzy clustering assessment technique is also used to assess this landscape water eminence in my research, and the conclusion is that the results of these two techniques are similar.

Conclusion

Water pollution problems have become a more and more prominent issue which dangerously irritating the survival and expansion of civilization. Hence, the difficulty of water environment is no longer narrowed to a particular area or a period of time, but has become global, cross-century focus. Now a day, urban landscape water management will also become a research hotspot. In order to use and protect water bodies more successfully, creating comprehensive assessment on water eminence is required.

In order to reflect the present condition of water eminence precisely, appreciate and clutch the force of water pollution and water environment trend, and ultimately protect and provide a scientific basis for water resources planning and management, water eminence assessment is made based on environmental monitoring information. Water eminence assessment is a qualitative and quantitative assessment on water eminence and environment which according to certain assessment criteria and techniques. Complete water eminence assessment approaches counting single-factor assessment technique, multi-factor key assessment technique, fuzzy mathematical assessment technique, gray system assessment technique, analytic hierarchy process, matter-element analysis, artificial neural network assessment technique as well as recently proposed such as water eminence classification key technique, gray fuzzy clustering assessment technique are the typical approaches at present.

Artificial neural network is a highly non-linear mapping, which can study a mapping between a big numbers of models. Using artificial neural network for river water eminence assessment, issues caused by other techniques, such as fuzzy comprehensive assessment and gray clustering techniques can be successfully avoided, such as man-made influence in weight assignment and purpose of membership function.

The neural networks, throughout learning and training to control system internal variation instead of using mathematical equations to state the application connection among the input and output, is well suited to compact with non-linear water environmental problems. In addition, the application of this technique is easy and appropriate for the daily management of urban water eminence.

Writing Services

Essay Writing
Service

Find out how the very best essay writing service can help you accomplish more and achieve higher marks today.

Assignment Writing Service

From complicated assignments to tricky tasks, our experts can tackle virtually any question thrown at them.

Dissertation Writing Service

A dissertation (also known as a thesis or research project) is probably the most important piece of work for any student! From full dissertations to individual chapters, we’re on hand to support you.

Coursework Writing Service

Our expert qualified writers can help you get your coursework right first time, every time.

Dissertation Proposal Service

The first step to completing a dissertation is to create a proposal that talks about what you wish to do. Our experts can design suitable methodologies - perfect to help you get started with a dissertation.

Report Writing
Service

Reports for any audience. Perfectly structured, professionally written, and tailored to suit your exact requirements.

Essay Skeleton Answer Service

If you’re just looking for some help to get started on an essay, our outline service provides you with a perfect essay plan.

Marking & Proofreading Service

Not sure if your work is hitting the mark? Struggling to get feedback from your lecturer? Our premium marking service was created just for you - get the feedback you deserve now.

Exam Revision
Service

Exams can be one of the most stressful experiences you’ll ever have! Revision is key, and we’re here to help. With custom created revision notes and exam answers, you’ll never feel underprepared again.