China
Africa
Explore chapters and articles related to this topic
Power Amplifier Modeling Based on Neural Network
Published in Jingchang Nan, Mingming Gao, Nonlinear Modeling Analysis and Predistortion Algorithm Research of Radio Frequency Power Amplifiers, 2021
Fuzzy neural network has attracted extensive attention in different fields owing to its favorable nonlinear function approximation ability, learning adaptability and parallel information processing ability. As a kind of fuzzy neural network, an Adaptive Neural Fuzzy Inference System (ANFIS) combines the learning mechanism of neural networks and the linguistic reasoning ability of fuzzy systems, and is also capable of self-organization, self-learning and logical reasoning. ANFIS is applied to nonlinear system modeling. Literature [9] proposes a method for device and circuit modeling based on the fuzzy logic, with parameter adjustment by means of the least square method and BP algorithm. Literature [10] proposes to apply a fuzzy neural network to PA predistortion. Literatures [11] and [12] propose a method for RF PA behavioral modeling based on an adaptive fuzzy neural network, but its applications are limited due to disadvantages such as complex structure and parameter learning, slow convergence speed, high risk of falling into local optimum, etc.
An Artificial Intelligence System Based Power Estimation Method for CMOS VLSI Circuits
Published in K. Gayathri Devi, Mamata Rath, Nguyen Thi Dieu Linh, Artificial Intelligence Trends for Data Analytics Using Machine Learning and Deep Learning Approaches, 2020
Govindaraj Vellingiri, Ramesh Jayabalan
Table 1.9 shows the power consumption comparison between ANFIS and BPNN, while Table 1.10 gives the details about BPNN and ANFIS error percentage. Among the 11 different BPNN algorithms, Trainscg function is best suited for ISCAS'89 sequential benchmark circuits. ANFIS, with hybrid learning and subtractive clustering, performs better than BPNN. ANFIS is a combination of fuzzy inference system and artificial neural network, which is its main advantage. Error, which is usually the sum of the square difference between actual output and desired output, is reduced at each iteration. ANFIS has the capability to generate an FIS that gives a linear relationship between the input and output data. Hence, ANFIS' exclusive characteristics appear to be a better choice for estimation of power in CMOS VLSI circuits.
Adaptive Neuro-fuzzy Inference System (ANFIS) Modelling in Energy System and Water Resources
Published in Kaushik Kumar, J. Paulo Davim, Optimization Using Evolutionary Algorithms and Metaheuristics, 2019
P. A. Adedeji, S. O. Masebinu, S. A. Akinlabi, N. Madushele
ANFIS model integrates ANN and fuzzy inference system (FIS) principles. It combines the strengths of both techniques on historical data with input-output fuzzification for the purpose of building a knowledge-based system. ANFIS has found its application in the past two decades in many fields: energy systems, medical diagnosis, econometrics, education, geology and so on. The integration between ANN and the Takagi-Sugeno-based FIS has also gained prominence in the field of water and energy resources. ANFIS has been applied in reservoir forecasting (Chang and Chang 2006; Hipni et al. 2013), groundwater level forecast (Sreekanth et al. 2011; Moosavi et al. 2014; Zare and Koch 2018) and water resource allocation (Abolpour et al. 2007) as well as in energy systems (Gayen and Jana 2017).
Weldability analysis and ANFIS modelling on laser welding of Inconel 718 thin sheets
Published in Materials and Manufacturing Processes, 2022
Pasupuleti Thejasree, P. C. Krishnamachary
ANFIS is a hybrid system that incorporates the abilities of learning of ANN and exceptional representation on knowledge and extrapolation competencies of fuzzy logic that can self-modify its membership function (MF) to accomplish a preferred performance. An adaptive network, which includes more or less most types of ANN models, can be engaged for interpreting the fuzzy inference system (FIS). ANFIS achieves the rule of hybrid learning and accomplishes complexity in decision-making. ANFIS has been established to be an efficient tool for altering the MF’s of FIS. ANFIS is an easy and convenient data learning procedure that utilizes an FIS model to transform a certain input into a desired output. This prediction comprises MF’s, fuzzy logic operators, and “if-then rules.” There are two kinds of fuzzy systems, usually recognized as the “Mamdani” and “Sugeno” type models. There are five key processing phases in the ANFIS operation: input fuzzification, fuzzy operators, application method, output combination, and defuzzification. The fuzzy controller’s design goal is for reducing and accomplishing better performance in the presence of disturbances and uncertainties. The design of ANFIS is a multilayer feed-forward network in which each node performs a particular function (node function) on incoming signals.
Prediction of municipal solid waste generation: an investigation of the effect of clustering techniques and parameters on ANFIS model performance
Published in Environmental Technology, 2022
Oluwatobi Adeleke, Stephen A. Akinlabi, Tien-Chien Jen, Israel Dunmade
ANFIS has found wide applications in different fields owing to its accuracy, adaptive nature, swift learning ability, computational speed and its ability to capture the non-linearity in the complex system. Several fields of applications of the ANFIS model include energy consumption [36], wind energy [37], petroleum engineering [38], agriculture [39], biomass and bioenergy [40,41], stock-market [42] and manufacturing [43]. The flexible computational structure of ANFIS allows its features and parameters such as number of rules, membership-function types and the method of generating the Fuzzy Inference System to be varied in order to improve its performance. The efficiency and accuracy of most soft-computing techniques of which ANFIS belongs are contingent on the optimal selection of model parameters [37]. Therefore, a careful choice of clustering techniques and parameters is an important step in modelling using ANFIS as it influences the prediction accuracy of the model significantly.
Hybrid intelligent strategy for multifactor influenced electrical energy consumption forecasting
Published in Energy Sources, Part B: Economics, Planning, and Policy, 2019
Azim Heydari, Davide Astiaso Garcia, Farshid Keynia, Fabio Bisegna, Livio De Santoli
In ANFIS, the most prominent features of neural networks, as well as fuzzy systems, are integrated according to the literature (Jang 1992). There are if-else rules in the ANFIS structure, and training is characterized by input-output pairs of neural network and fuzzy learning algorithms. By using these methods, ANFIS can produce complex nonlinear mappings (İnan et al. 2007). The ANFIS structure, characterized by both fuzzy-logic model and neural network, can function in noisy and imprecise settings (Liu and Ling 2003). The process of neural network training is used in ANFIS to adjust the membership function and important parameters producing the target datasets (Wu, Hsu, and Chen 2009). By using this method, more accurate results can be obtained compared to the mean square error, as it incorporates expert decisions. The hybrid ANFIS learning algorithm integrates back-propagation learning algorithm, as well as the least squares method. Samples with two inputs and one single output can simplify the operation. The ANFIS structure was developed using five layers as shown in Figure 2.