Adaptive Network Based Fuzzy Inference System Structure Research Articles

Islanding detection method based on S transform and ANFIS

In distributed generation systems, islands of power generation will disturb grid recovery and power quality. Therefore, various islanding detection technologies have been proposed, and most are remote detection, local active, and passive methods. Most islanding detection technologies are currently implemented with high computational complexity, inaccurate detection, affecting power quality (active methods), and other problems. This paper proposes an islanding detection method based on S transform (ST) and Adaptive-Network-based Fuzzy Inference System (ANFIS). It uses ST for electrical signal processing and then extracts features which are used as the inputs of the ANFIS system. ANFIS starts training and learning to obtain classification capabilities and accurately determines the islanding events and outputs the results. The innovation of this method is the combination of ST and ANFIS structure. ST has better time-frequency characteristics and signal processing capability. ANFIS can automatically start training and learning in order to enhance the capability. The combination of these two methods can greatly improve the accuracy of islanding detection. The ST and ANFIS based islanding detection method proposed in this paper are simulated in MATLAB Simulink. According to the different characteristics of islanding and non-islanding conditions, full consideration is given to the comprehensiveness of the condition design, and 20 sets of data are designed. Through the training of 12 sets of data, and the testing of 8 sets of data, with equal quantity for the islanding and non-islanding cases, the output of this method is completely correct. Also, the proposed method is compared to ANFIS based and artificial neural network based islanding detection methods. The results show that the proposed ST-ANFIS based islanding detection method can accurately detect the islanding and non-islanding conditions with higher accuracy.

Adaptive network based fuzzy inference system (ANFIS) training approaches: a comprehensive survey

In the structure of ANFIS, there are two different parameter groups: premise and consequence. Training ANFIS means determination of these parameters using an optimization algorithm. In the first ANFIS model developed by Jang, a hybrid learning approach was proposed for training. In this approach, while premise parameters are determined by using gradient descent (GD), consequence parameters are found out with least squares estimation (LSE) method. Since ANFIS has been developed, it is used in modelling and identification of numerous systems and successful results have been achieved. The selection of optimization method utilized in training is very important to get effective results with ANFIS. It is seen that derivate based (GD, LSE etc.) and non-derivative based (heuristic algorithms such us GA, PSO, ABC etc.) algorithms are used in ANFIS training. Nevertheless, it has been observed that there is a trend toward heuristic based ANFIS training algorithms for better performance recently. At the same time, it seems to be proposed in derivative and heuristic based hybrid algorithms. Within the scope of this study, the heuristic and hybrid approaches utilized in ANFIS training are examined in order to guide researchers in their study. In addition, the final status in ANFIS training is evaluated and it is aimed to shed light on further studies related to ANFIS training.

A unique intelligent algorithm for optimization of human reliability and decision styles: a large petrochemical plant

The importance of decision-making failures in complex systems such as petrochemical plants has been well recognized. Decision making failures are mostly resulted from human error. In this paper, efficiency of human operator is assessed based on combination of decision style (DS) and human reliability. At first, a standard questionnaire is designed to collect required data. The reliability of the collected data is investigated by Cronbach’s alpha (about 0.75). Also, the efficiency is ranked according to the impact of decision styles on human reliability factors. Operators of control rooms in a petrochemical plant are respond to a human reliability-decision styles (HR-DS) hybrid questionnaire. The indicators related to decision styles and human reliability are used as inputs and outputs, respectively in a unique adaptive network-based fuzzy inference system (ANFIS) algorithm. Decisive, hierarchical, flexible and integrated are the standard categorization of human decision styles. The optimum structure of ANFIS is selected based on minimum mean absolute percentage (MAPE). The best MAPE is achieved about 26%. Relative analysis shows that the efficient decision style achieved from ANFIS is consistent with plant’s dominant style. The results of this study help managers to enhance the system performance by using the best operator for critical positions. This study presents the first neuro-fuzzy algorithm for improvement of decision making and human reliability in a large petrochemical plant.

Training ANFIS by using the artificial bee colony algorithm

In this study, a new adaptive network-based fuzzy inference system (ANFIS) training algorithm, the artificial bee colony (ABC) algorithm, is presented. Antecedent and conclusion parameters existing in the structure of ANFIS are optimized with the ABC algorithm and ANFIS training is realized. Identification of a set of nonlinear dynamic systems is performed in order to analyze the suggested training algorithm. The ABC algorithm is operated 30 times for each identification case and the average root mean square error (RMSE) value is obtained. Training RMSE values calculated for the four examples considered are 0.0325, 0.0215, 0.0174, and 0.0294, respectively. In addition, test error values for the same cases are respectively computed as 0.0270, 0.0186, 0.0167, and 0.0435. The results obtained are compared with those of known neuro-fuzzy-based methods frequently used in the literature in identification studies of nonlinear systems. It is shown that ANFIS can be trained successfully by using the ABC algorithm for the identification of nonlinear systems.

PAPR Reduction in OFDM Signals: An Adaptive-Network-Based Fuzzy Inference Approach

In this paper, an adaptive-network-based fuzzy inference system (ANFIS) based scheme is analyzed and proposed for reducing the peak-to-average power ratio (PAPR) in multicarrier signals under additive white Gaussian noise and multipath fading (Raleigh) channel environment. This scheme involves training of ANFIS structure in time domain using Orthogonal Frequency Division Multiplexing signals with low PAPR, such as those obtained by approximate gradient project---null subcarrier switching (AGP---NCS) method. Once the ANFIS module is trained, the proposed scheme approximately offers similar reduction in PAPR as compared to AGP---NCS method, with significantly less convergence time and computational complexity. he results show that proposed approach is not only less complex but also maintains the data rate and bit error rate performance compared with other conventional schemes.

Modeling of TiN Coating Thickness Using ANFIS

In this paper, an approach in predicting thickness of Titanium Aluminum Nitrite (TiN) coatings using Adaptive Network Based Fuzzy Inference System (ANFIS) is implemented. The TiN coatings were coated on tungsten carbide (WC) using Physical Vapor Deposition (PVD) magnetron sputtering process. The N2 pressure, argon pressure and turntable speed were selected as the input parameters and the coating thickness as an output of the process. Response Surface Methodology (RSM) was used to design the matrix in collecting the experimental data. In the ANFIS structure, three bell shapes were used as input membership function (MFs). The collected experimental data was used to train the ANFIS model. Then, the ANFIS model was validated with confirmatory test data and compared with other prediction models in terms of the root mean square error (RMSE), residual error and prediction accuracy. The result indicated that the developed ANFIS model result was the lowest RMSE7 and average residual error, besides the highest in prediction accuracy compared to the other models. The result also indicated that the limited experimental data could be used in training the ANFIS model and resulting accurate predictive result.

Predicting Mathematics 1 Course Success by Using Hierarchical Adaptive Network Based Fuzzy Inference System

Predicting Mathematics 1 course success of students is very important to prepare them before the semester. It is difficult to obtain solution because of the non-linear property of data set. Fuzzy logic is one of the common methods for the problems which involve numeric values. In fuzzy logic, it is important to determine membership functions and their parameter's values correctly. This can be done by an expert or can be learned with a data set. In this study, we aimed to predict the Mathematics 1 course success of 434 students who enrolled to Engineering Faculty of Pamukkale University in 2007-2008 academic year by using their university exam data. For this, the adaptive-network-based fuzzy inference system (ANFIS) which combines the important characteristics of artificial neural network and fuzzy logic was used. In training section, nine parameters which are selected from sixteen parameters in data set with different combinations were given to the ANFIS. When an ANFIS structure with nine input parameters has at least three membership functions for each input, it will have at least 3^9 fuzzy rules. Because of this, the training part is too slow and too much memory is needed. Instead of this inefficient structure, a hierarchical method was proposed. In this method, the ANFIS is partitioned to the sub-systems. Each sub-system performs some part of input parameters and sends their result to the final ANFIS structure to obtain the overall system output. After testing with one-third of data set, two best prediction results with ratio 77.77% and 78.47% are obtained. When these results are analyzed, it is seen that 64 successful students from 85 students and 48 unsuccessful students from 59 students in Mathematics 1 course were predicted truly in the result with ratio 77.77%. Similarly, 69 successful students from 85 students, and 44 unsuccessful students from 59 students were predicted truly in the result with ratio 78.47%.

Diagnosis of working conditions of an aluminum reduction cell based on wavelet packets and fuzzy neural network

In this paper, fuzzy neural network is combined with wavelet packet analysis for diagnosis of working conditions of aluminum reduction cells. The sample data is pre-processed using best wavelet packet basis for the forecast and then an adaptive-network-based fuzzy inference system (ANFIS) is established for diagnosis of working conditions. The wavelet packet analysis was used to extract the characteristic of signal according to the frequency spectrum characteristics of voltage vibration signal of aluminum reduction cells. The signals were decomposed into eight frequency bands and the information pre-conditioned was used as an energy characteristic vector. The structure of ANFIS is given and the membership function is developed according to the actual situation. All simulated working conditions are emulated on 350 KA pre-baked aluminum reduction cells. The feasibility of this novel method is proved by the simulation results.