I certify that an Examination Committee met on 06 December 2007 to conduct the final examination of SHAHRAM KARIMI GOOGHARI on his Doctor of Philosophy thesis entitled “ReservoirInflowForecastingUsingArtificialNeural Networks and AdaptiveNeuro-FuzzyInferenceSystemTechniques” in accordance with Universiti Pertanian Malaysia (Higher Degree) Act 1980 and Universiti Pertanian Malaysia (Higher Degree) Regulations 1981. The Committee recommends that the candidate be awarded the relevant degree. Members of the Examination Committee are as follows:
This paper is dealing with the development of a forecastingsystem based on ANFIS, which differs from the traditional ArtificialNeural Networks (ANN) in that it is not fully connected and not all the weights or nodal parameters are modifiable. The model uses a hybrid learning algorithm to identify the parameters for the Sugeno- type fuzzyinference systems. It applies a combination of the least-squares method and the back-propagation gradient descent method for training the FuzzyInferenceSystem (FIS) membership function parameters to match the given training data set. Specifically, a back-propagation algorithm is used to optimize the fuzzy sets of the premises and a least-squares procedure is applied to the linear coefficients in the consequent terms. In addition, it uses a testing data set for checking the model over fitting. ANFIS is a multilayer neuralnetwork-based fuzzy consisted of five layers, in which the training and predicted values are represented by the input and output nodes and the nodes functioning as membership functions (MFs) and rules are presented in the hidden layers. Its topology is shown in Figure 1. During the learning phase of ANFIS, the parameters of the membership functions are changing continuously in order to minimize the error function between the target output and the calculated values. ANFIS has a feed-forward neuralnetwork structure where each layer is a neuro-fuzzysystem component. ANFIS is capable to learn and generalize the training data. The consequents of the Takagi-Sugeno (TS) fuzzy rules are linear combinations of their preconditions in this method. ANFIS can simply be defined as the combination of ANNs and fuzzy logic. This combined system has the abilities of deducing knowledge from given, learning, generalization, adaptation and parallelism. FIS is a framework based on fuzzy set theory and fuzzy if-then rules. The structure of FIS has three main components: a rule base, a database, and a reasoning mechanism. This model is called first-order Sugeno model.
Another model that is used in this work is Neuro-Fuzzy, which is the combination of fuzzy logic and neural networks in order to solve wide variety of real world problems in an effective manner. This combination is for removing the limitation of each model. Since neural networks are good at recognizing patterns and not good at explaining how they achieve their decisions. Fuzzy logic systems that can give inexact reasons, and explain their decisions well but not good at reaching the rules they use to make those decisions . The ability to model a problem domain using a linguistic model instead of complex mathematical is the main advantage of using the Neuro-Fuzzy combination . Therefore, these techniques are complementary to be used together .
The main purpose of the current research is comparing the results of ArtificialNeuralNetwork (ANN) with AdaptiveNeuro-FuzzyInferenceSystem (ANFIS) with regard to determination of the importance of soil properties affecting clay dispersibility. After taking samples from two depths of 0-40 and 40-80 cm, the spontaneous and mechanical dispersions of clay were recorded using both weighing and turbidimetric methods. To determine the degree of importance of soil properties affecting clay dispersibility, first ANNs and ANFIS in MATLAB Software were determined, using all research variables. After determining less effective properties and omitting them, the mentioned networks with the remaining variables including percentage of clay and sand, soil reaction, Electrical Conductivity (EC) and Sodium Adsorption Ratio (SAR) were measured and the degree of importance of each variable in clay dispersibility was determined. Finally, the results of ANNs and ANFIS were compared by calculation of validation parameters. Existence of high correlation between calculated values for weighing and turbidimetric methods showed a linear relationship between the two methods. In general, in both depths and for both weighing and turbidimetric methods, the sensitivity of clay dispersibility to the percentage of the clay, sand and SAR, was higher than any other variable. Although the results obtained from the validation statistics indicate high accuracy of both ANN and ANFIS models, the last model showed relatively better results as compared to ANN model.
IMAGES are often corrupted by impulse noise during image acquisition and transmission over communication channel. Noise elimination and enhancement are essential features in digital image processing. Because the performances of subsequent image processing tasks are strictly dependent on the success of the noise removal operation. However, this is a difficult task because the noise removal operator is imposed with the requirement of preserving useful information in the image while efficiently removing the noise. Digital images are valuable sources of information in many research and application areas including astronomy, biology, medicine, remote sensing, materials science, etc. During image acquisition, digital images are frequently corrupted by noise due to number of imperfections in the imaging process. The image corruption is usually introduced by a nonideal imaging system (sensor noise, limited system accuracy, finite precision, quantization of image data, etc.) or an imperfect medium between the original scene and the imaging system (random scattering, absorption, etc.). The currently available nonlinear filters cannot simultaneously satisfy both of these criteria. The existing filters either suppress the noise at the cost of reduced noise suppression performance. Neural networks and fuzzy systems had been investigated to the problems in digital signal processing. A Neuro-FuzzySystem is a flexible system trained by heuristic learning techniques derived from neural networks can be viewed as a 3-layer neuralnetwork with fuzzy weights and special activation functions is always interpretable as a fuzzysystem uses constraint learning procedures is a function approximation (classifier, controller). In this paper, AdaptiveNeuro-fuzzyInferenceSystem (ANFIS) is presented, which is a fuzzyinferencesystem implemented in the framework of adaptivenetwork. This ANFIS training algorithm is suggested by Jang. By using hybrid learning procedure, the proposed ANFIS can construct an input-output mapping which is based on both human knowledge (in the form of fuzzy if- then rules) and learning. The proposed work is carried in two stages. In first stage, noisy image (i.e. received or captured by camera) is denoised by new tristate switching median filter. The denoised image is further enhanced by ANFIS. . The structure of the proposed hybrid filter, to make the process robust against noise, is a combination of nonlinear switching median filter and neuro-fuzzynetwork. The internal parameters of the neuro-fuzzynetwork are adaptively optimized by training. The proposed filter is evaluated under
In this section, the sensitivity analysis is performed based on the eﬀect of each individual non-dimensional input parameter on the output value. In other words, in order to understand the correlation of each input parame- ter with the output parameter using ANN and ANFIS models, ANN1, and ANFIS1 were constructed based on the raw data as their input parameters. In addition, in ANN2 and ANIS 2, (S/A) was added as an additional parameter to the already existing ones in ANN1 and ANFIS1. Therefore, the eﬀect of (S/A) on the prediction accuracy of the model was investigated. In the same way, all the other non-dimensional elements i.e. (W/C), (W/T), (A/C), (R/R) were added to ANN1 and ANFIS 1 individ- ually and constructed in a separate network. The details of these ANN and ANFIs models are shown in Table 4. 184.108.40.206. Sensitivity analysis using ANN model. Figs. 10–15 show the relationship between the target and output com- pressive strength of ANN1 to ANN6, respectively, for both the training and validation steps. According to these ﬁg- ures, the performance of ANN1 to ANN6 is shown based on correlation coeﬃcient (R).
ANFIS is a fuzzyinferencesystem (FIS) implemented in the framework of an adaptivefuzzyneuralnetwork. It combines the explicit knowledge representation of an FIS with the learning power of artificialneural networks. The objective of ANFIS is to integrate the best features of fuzzy systems and neural networks. Using a given input/output data set, ANFIS constructs a FIS whose membership function parameters are tuned (adjusted) using either a back propagation algorithm alone or in combination with a least squares type of method . This adjustment
The hydrologic behavior of rainfall-runoff process is very complicated phenomenon which is controlled by large number of climatic and physiographic factors that vary with both the time and space. The relationship between rainfall and resulting runoff is quite complex and is influenced by factors relating the topography and climate. In recent years, artificialneuralnetwork (ANN), fuzzy logic, genetic algorithm and chaos theory have been widely applied in the sphere of hydrology and water resource. ANN have been recently accepted as an efficient alternative tool for modeling of complex hydrologic systems and widely used for prediction. Some specific applications of ANN to hydrology include modeling rainfall-runoff process. Fuzzy logic method was first developed to explain the human thinking and decision system by . Several studies have been carried out usingfuzzy logic in hydrology and water resources planning . Adaptiveneuro- fuzzyinferencesystem (ANFIS) which is integration of neural networks and fuzzy logic has the potential to capture the benefits of both these fields in a single framework. ANFIS utilizes linguistic information from the fuzzy logic as well learning capability of an ANN. Adaptiveneurofuzzyinferencesystem (ANFIS) is a fuzzy mapping algorithm that is based on Tagaki-Sugeno-Kang (TSK) fuzzyinferencesystem  and . ANFIS used for many applications such as, database management, system
AdaptiveNeuroFuzzyInferenceSystem is the combination of best features of fuzzyinferencesystem and artificialneuralnetwork. ANFIS serve as a basis for constructing a set of fuzzy if-then rules with appropriate membership functions to generate the stipulated input-output pairs. This is a fuzzyinferencesystem with a back propagation that tries to reduce the error and improves the performance. The block diagram of ANFIS architecture is given in Fig.3.
ArtificialNeuralNetwork (ANN) or neural networks are a kind of information processing paradigm. These are inspired by the working of biological nervous systems such as brain. ANN process information in the same manner as biological nervous systems do. Neural networks are composed of highly interconnected processing elements called neurons. These neurons operate in parallel to solve a specific problem. Patterns to be analyzed are presented to the network via the 'input layer'. Getting a specific target output from a particular input requires neural networks to be adjusted, or trained.
In this study, we have been able to develop an architecture model that combines the knowledge-base and reasoning features of fuzzy logic (FL), with self-learning capacity of artificialneuralnetwork (ANN) to diagnose Hepatitis B disease by computing the intensity levels of its attack. Medical field refers to diagnosis as an approach of recognizing a disease through the analysis of underlying physiological symptoms. Hepatitis is a chronic liver disease that can easily decompensate to cirrhosis disease if its severity is not properly checked with appropriate tools. This study can improve the quality of liver disease diagnosis and provide appropriate information for medical practitioners in the process of administering treatment. This work could be further expanded for improved optimal solutions by integrating Genetic Algorithm (GA) with ANFIS.
Recently developed methods used Artificial Intelligence (AI) methods, like NeuralNetwork (NN), AdaptiveNeuro-FuzzyInferenceSystem (ANFIS), Support Vector Machine (SVM) and Radial Basis Function (RBF), in forecasting the demand within the short-term time frame. On the other hand, only historical load data and temperature are used as model input parameters. Very few used relative humidity in addition to load and temperature. There might be other weather variables that affect the load consumption different from temperature and relative humidity, and the relationship between the load and these variables needs to be investigated.
Heart signals allow for a comprehensive analysis of the heart. Electrocardiography (ECG or EKG) uses electrodes to measure the electrical activity of the heart. Extracting ECG signals is a non- invasive process that opens the door to new possibilities for the application of advanced signal processing and data analysis techniques in the diagnosis of heart diseases. With the help of today’s large database of ECG signals, a computationally intelligent system can learn and take the place of a cardiologist. Detection of various abnormalities in the patient’s heart to identify various heart diseases can be made through an AdaptiveNeuro-FuzzyInferenceSystem (ANFIS) preprocessed by subtractive clustering. Six types of heartbeats are classified: normal sinus rhythm, premature ventricular contraction (PVC), atrial premature contraction (APC), left bundle branch block (LBBB), right bundle branch block (RBBB), and paced beats. The goal is to detect important characteristics of an ECG signal to determine if the patient’s heartbeat is normal or irregular. The results from three trials indicate an average accuracy of 98.10%, average sensitivity of 94.99%, and average specificity of 98.87%. These results are comparable to two artificialneuralnetwork (ANN) algorithms: gradient descent and Levenberg Marquardt, as well as the ANFIS preprocessed by grid partitioning.
(Acedański, 2013).In prediction models, inefficient use of resources increases prediction error. Of the reasons of inefficient use of resources is the lack of proper parameters in models, the lack of proper modeling techniques, and the lack of use of indicators that has capable of predicting sufficiently in the forecasting process (Günay, 2018). A survey of the GDP in recent years (Fig. 1) shows that the share of the industrial sector has fallen (World Bank). In addition, significant fluctuations have been experienced in the economic environment, which have reduced the predictive power of variables and created uncertainty about the future (Zamanzadeh, 2010). This is why researchers must provide an efficient way to predict industrial production through the planning and managing of the effective parameters to manage of the future trends.
The aim of this research is to develop a predictive model to estimate the groundwater head in Safwan-Zubair area by using an adaptiveneuralfuzzyinferencesystem (ANFIS). This area represents the southern sector of the Iraqi Desert, an arid region with scarce and limited resources. The data required for building the ANFIS model are generated using MODFLOW model (V.5.3). MODFLOW model was calibrated based on field measurements during one year. MODFLOW model generated (3797) hydraulic head values during each month. 70% of these values (2658 samples) was used for training, 30% of these values (1139 samples) was used for checking. The accuracy of the ANFIS models are compared with previous work based on artificialneuralnetwork (ANN) technique. Different combination of successive hydraulic heads and recharge rates of groundwater is used as input variables. There is no significant increase in the estimation accuracy when adding another input variable (recharge rate). Because the amount of this variable is very little, so its influence on the results was imperceptible. A comparison of ANFIS and ANN shows that the ANFIS model performs preferable than the ANN model on the checking phase. ANFIS model combines both fuzzy logic basics and neural networks; thus their properties can be utilized in one frame. It can be concluded, the ANFIS model appears to be more convenient than the ANN model for predicting groundwater hydraulic head from related input data.
The first artificialneuralnetwork (ANN) was invented in 1958 by psychologist Frank Rosenblatt called ‘perceptron’. ANN is a computational model, which replicates the function of a biological network composed of neurons. ANN is often used to model complex nonlinear functions in various applications. The basic unit in the ANN is the neuron. Neurons are connected to each other by links known as synapses. Andersen et al. (1990) first applied ANN in the welding area to predict weld bead shape for the gas tungsten arc welding (GTAW) process . Researchers have also applied ANN to develop predictive models for FSW joints [18-23]. A multilayer perceptron ANN system has three layers which are input, hidden, and output layers. The input layer consists of all the input factors. Information from input layer is then processed in the hidden layers, and then followed by the output layer (Figure 1). Details on the neuralnetwork modeling approach are given elsewhere .
For decades, due to the increase in electrical power demand, we are in need of enlarging transmission capacity by installing 500-kV extra high-voltage power transmission lines in both AC and DC. The impact of electric fields surrounding the transmission line depends strongly on conductor surface potentials, while load currents flowing through the transmission line result in magnetic field distribution. For the 500 kV systems, high current density transmission is the main purpose. It can cause electrical hazards to people or their livestock nearby. Especially when the power system was faulted, the short-circuit current is much higher than the normal current loading. This paper focuses on utilization of efficient computing techniques to estimate the magnetic field distribution. Obtained estimate solutions can lead to assessment of electrical hazards for 500-kV power transmission systems. This paper presents an online estimation of magnetic fields for live transmission line right of way worker using Generalized Regression ArtificialNeuralNetwork.
coefficient higher than 0.9 was obtained. In addition, with the trial and error method, training results showed that the ANN with three hidden layers has the best performance. Consequently, the developed ANN model for predicting density biodiesel blends is shown in Figure 5 and the training parameters can be found in Table 2. The developed network architecture has a 2-3-1 configuration with two neurons in the input layer indicating temperature and volume fraction of biodiesel. Three hidden layers with varying neurons and ten neurons in the output layer representing density are used.
As mentioned earlier, different structures of the ANN models were employed for estimating VW in order to select the suitable parameter set combination of each model for each crop. The statistical measurements of the results obtained by each ANN models in training, validation and testing sets for different crops individually which are summarized in Table 2. According to Table 2, the best number of neurons in hidden layer for MLP model and the best amount of spread (σ) for RBF and GRNN models are mentioned in column 3. The best percent of all data that were used for each set of training, validation, and testing ANN models have been shown in column 4. Column 7 shows the best network among the applied ANN models for each crop.
Reservoir water release decision is one of the critical actions in determining the quantity of water to be retained or released from the reservoir. Typically, the decision is influenced by the reservoirinflow that can be estimated based on the rainfall recorded at the reservoir’s upstream areas. Since the rainfall is recorded at several different locations, the use of temporal pattern alone may not be appropriate. Hence, in this study a spatial temporal pattern was used to retain the spatial information of the rainfall’s location. In addition, rainfall recorded at different locations may cause fuzziness in the data representation. Therefore, a hybrid computational intelligence approach, namely the AdaptiveNeuroFuzzyInferenceSystem (ANFIS), was used to develop a reservoir water release decision model. ANFIS integrates both the neuralnetwork and fuzzy logic principles in order to deal with the fuzziness and complexity of the spatial temporal pattern of rainfall. In this study, the Timah Tasoh reservoir and rainfall from five upstream gauging stations were used as a case study. Two ANFIS models were developed and their performances were compared based on the lowest square error achieved from the simulation conducted. Both models utilized the spatial temporal pattern of the rainfall as input. The first model considered the current reservoir water level as an additional input, while the second model retained the existing input. The result indicated that the application of ANFIS could be used successfully for modeling reservoir water release decision. The first model with the additional input showed better performance with the lowest square error compared to the second model.