A modified gradient-based neuro-fuzzy learning algorithm and its convergence

Wu, Wei; Li, Long; Yang, Jie; Liu, Yan

doi:10.1016/j.ins.2009.12.030

Cited by 64 publications

(22 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Changing these parameters will give the various contour of the function as required in accordance with the data set for the problem defined. Not only does this novel form shorten the training time, but also accelerate the running speed of the network significantly [6].…”

Section: 2structure Of Fuzzy-neural Networkmentioning

confidence: 99%

Mobile Robot Local Path Planning Based on Improved T-S Fuzzy Neural Network

Zhang¹,

Shen²,

Sun³

2015

Proceedings of the 2015 International Industrial Informatics and Computer Engineering Conference

View full text Add to dashboard Cite

Abstract. Aiming at the problems that BP algorithm has slow convergence rate and is likely to fall into local minimum point. This paper presents a modified fuzzy neural network which takes the distances between the robot and obstacles as input variables, and the velocity and the steering angle of the robot as output variables. Besides, the accurate environmental information can be obtained by fusing the information of ultrasonic sensors, and the network parameters can be adjusted by training the network. With the application of Matlab simulation software, the result shows that the improved fuzzy neural network has good performances in optimizing the convergence speed, learning speed and stability of the network.

show abstract

Section: 2structure Of Fuzzy-neural Networkmentioning

confidence: 99%

Mobile Robot Local Path Planning Based on Improved T-S Fuzzy Neural Network

Zhang¹,

Shen²,

Sun³

2015

Proceedings of the 2015 International Industrial Informatics and Computer Engineering Conference

View full text Add to dashboard Cite

show abstract

“…While the activation function in a "regular" artificial neural network provides outputs using the weighted inputs, the output of a neuro-fuzzy Sugeno system is a weighted average of all the fuzzy rules outputs [15]. The adaptive neuro-fuzzy system was made for the purpose of trajectory optimal selection and it has 5 layers, two inputs, six rules and membership functions associated to the input and output variables (Fig.…”

Section: Feedback Gains Choosing Rulesmentioning

confidence: 99%

Automatic expert system for fuzzy control of robot trajectory in joint space

Tudor

Moise

2013

2013 IEEE International Conference on Mechatronics and Automation

View full text Add to dashboard Cite

Trajectory control for manipulation robots in joint space is used both in industrial and research applications. This paper shows how an expert system can be developed to control a manipulation robot and to generate a trajectory for one single mobile robot joint, supposing all the other joints are locked in fixed positions. The automatic expert system includes a control system for trajectory tracking to simulate a joint trajectory with the provided data. The expert system learning module uses adaptive neuro-fuzzy learning techniques, based on the gradient descent and the least mean squares methods. The learning module improves the expert system performances to optimally choose the trajectory type. The experimental tests for the prototype expert system prove the efficiency of the neuro-fuzzy rules for trajectory selection. The performances of the expert system are also determined by the heuristic rules used to modify the closed loop gains used in the control system to simulate the robot response and to generate the trajectory. The efficiency of the joint trajectory control is shown by making a comparative analysis between the obtained trajectories and those simulated as a response of the control system. Index Terms -automatic expert system, neuro-fuzzy learning, heuristic methods, robot trajectory, control system.

show abstract

“…A variety of system structures and learning algorithms are available for neuro-fuzzy methods [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26]. Learning of the classical neuro-fuzzy systems is based on the gradient descent method [9].…”

Section: Introductionmentioning

confidence: 99%

“…Learning of the classical neuro-fuzzy systems is based on the gradient descent method [9]. It is modified to avoid non-firing or weak firing [10,11] and to improve learning efficiency [12]. Genetic algorithms are applied to a neuro-fuzzy with radial-basis-function-based membership for the automatic generation of fuzzy rules [13].…”

Section: Introductionmentioning

confidence: 99%

Simplified Real-, Complex-, and Quaternion-Valued Neuro-Fuzzy Learning Algorithms

Hata¹,

Akhand²,

Islam³

et al. 2018

IJISA

View full text Add to dashboard Cite

Abstract-The conventional real-valued neuro-fuzzy method (RNF) is based on classic fuzzy systems with antecedent membership functions and consequent singletons. Rules in RNF are made by all the combinations of membership functions; thus, the number of rules as well as total parameters increase rapidly with the number of inputs. Although network parameters are relatively less in the recently developed complex-valued neuro-fuzzy (CVNF) and quaternion neuro-fuzzy (QNF), parameters increase with number of inputs. This study investigates simplified fuzzy rules that constrain rapid increment of rules with inputs; and proposed simplified RNF (SRNF), simplified CVNF (SCVNF) and simplified QNF (SQNF) employing the proposed simplified fuzzy rules in conventional methods. The proposed simplified neuro-fuzzy learning methods differ from the conventional methods in their fuzzy rule structures. The methods tune fuzzy rules based on the gradient descent method. The number of rules in these methods are equal to the number of divisions of input space; and hence they require significantly less number of parameters to be tuned. The proposed methods are tested on function approximations and classification problems. They exhibit much less execution time than the conventional counterparts with equivalent accuracy. Due to less number of parameters, the proposed methods can be utilized for the problems (e.g., real-time control of large systems) where the conventional methods are difficult to apply due to time constrain.

show abstract

A modified gradient-based neuro-fuzzy learning algorithm and its convergence

Cited by 64 publications

References 34 publications

Mobile Robot Local Path Planning Based on Improved T-S Fuzzy Neural Network

Mobile Robot Local Path Planning Based on Improved T-S Fuzzy Neural Network

Automatic expert system for fuzzy control of robot trajectory in joint space

Simplified Real-, Complex-, and Quaternion-Valued Neuro-Fuzzy Learning Algorithms

Contact Info

Product

Resources

About