15.07.2014 Views

Conference Program of WCICA 2012

Conference Program of WCICA 2012

Conference Program of WCICA 2012

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Conference</strong> <strong>Program</strong> <strong>WCICA</strong> <strong>2012</strong><br />

tion dynamics are used to establish the reference model implicating essential<br />

nonlinearities for GEBF-FNN based ship motion model (GEBF-<br />

FNN-SMM) identification. The GEBF-FNN-SMM starts without fuzzy<br />

rules and online recruits efficient fuzzy rules via rule node generation<br />

criteria and parameter estimation. The resultant GEBF-FNN-SMM reasonably<br />

captures essential dynamics since the checking process validates<br />

the prediction performance with high accuracy. Finally, in order to<br />

demonstrate that the GEBF-FNN-SMM scheme is effective, simulation<br />

studies are conducted on zig-zag maneuvers. Moreover, comprehensive<br />

comparisons are carefully presented. Simulation results indicate<br />

that the GEBF-FNN-SMM achieves promising performance in terms <strong>of</strong><br />

approximation and prediction.<br />

◮ FrB02-2 16:10–16:30<br />

A Probabilistic Fuzzy Controller with Operant Learning for Robot Navigation,<br />

pp.368–373<br />

Gao, Yuanyuan<br />

Ruan, Xiaogang<br />

Song, Hongjun<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Beijing Univ. <strong>of</strong> Techbnology<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Fuzzy logic system (FLS) promises an efficient way for obstacle avoidance.<br />

However, it is difficult to maintain the correctness, consistency,<br />

and completeness <strong>of</strong> a fuzzy rule base tuned by a human expert. In<br />

this paper, a novel approach termed probabilistic fuzzy controller with<br />

operant learning (PFCOL) for robot navigation is presented. Operant<br />

learning (OL) is a form animal learning way. The key feature <strong>of</strong> this<br />

approach is that it combines a probabilistic stage and a stochastic perturbation<br />

generator module into FLS to handle problems. At last, the<br />

ultimate output is determined by these two uncertain stages. This imitates<br />

animal learning method <strong>of</strong> generating stochastic behavior in the<br />

complex and uncertain environment. The simulation results show that<br />

the proposed PFCOL method can automatically generate approximate<br />

actor to adapt complex circumstances. Through studies on obstacle<br />

avoidance and goal seeking tasks by a mobile robot verify the approach<br />

is superior in generating efficient fuzzy inference systems.<br />

◮ FrB02-3 16:30–16:50<br />

The Optimization <strong>of</strong> Fuzzy Rules Based on Hybrid Estimation <strong>of</strong> Distribution<br />

Algorithms, pp.561–565<br />

Luo, Xiong<br />

Bai, Xue<br />

Univ. <strong>of</strong> Sci. & Tech. Beijing<br />

Univ. <strong>of</strong> Sci. & Tech. Beijing<br />

Optimization <strong>of</strong> fuzzy rules based on numerical data is an important issue<br />

in the optimization design <strong>of</strong> fuzzy system. In this paper, based on<br />

an improved estimation <strong>of</strong> distribution algorithm, an optimization learning<br />

method COR MUMDA for fuzzy rules is proposed. This method can<br />

generate fuzzy rules directly from numerical data. The method learn<br />

fuzzy rules mainly based on MUMDA (multi-group univariate marginal<br />

distribution estimation algorithm). Unlike the general estimation <strong>of</strong> distribution<br />

algorithms, MUMDA can increase the diversity <strong>of</strong> the population<br />

and avoid sticking at local optima. In addition, the elite genetic<br />

strategy is used to generate the next population. In this way, it reduces<br />

the possibility <strong>of</strong> losing the optimal solutions. To verify the efficiency <strong>of</strong><br />

this algorithm, the simulation experiments are performed. The comparative<br />

results <strong>of</strong> three classic examples are given.<br />

◮ FrB02-4 16:50–17:10<br />

On the Definition <strong>of</strong> Type-2 Fuzzy Sets, pp.601–605<br />

Mo, Hong<br />

Zhou, Min<br />

Changsha Univ. <strong>of</strong> Sci. & Tech.<br />

Changsha Univ. <strong>of</strong> Sci. & Tech.<br />

This paper introduces four kinds <strong>of</strong> definition <strong>of</strong> type-2 fuzzy sets(T2<br />

FS), and provides their difference and connection,then presents a new<br />

definition <strong>of</strong> type-2 fuzzy set to make it to be understood more easily,<br />

In finial, we modify the definition and formula <strong>of</strong> footprint <strong>of</strong> uncertainty<br />

(FOU) <strong>of</strong> T2 FS and give the relation <strong>of</strong> FOU and T2 FS.<br />

◮ FrB02-5 17:10–17:30<br />

Estimation <strong>of</strong> Hand Force from Surface Electromyography Signals using<br />

Artificial Neural Network, pp.584–589<br />

Srinivasan, Haritha<br />

Sauvik, Das Gupta<br />

Oklahoma State Univ.<br />

Oklahoma State Univ.<br />

Sheng, Weihua<br />

Chen, Heping<br />

Oklahoma State Univ.<br />

Texas State Univ.<br />

Haptic technology has many real world applications such as rehabilitation<br />

robotics, telepresence surgery, gaming, virtual reality and humanrobot<br />

interaction. Force plays an important role in the above mentioned<br />

haptic applications. In this paper, we propose a method to estimate<br />

force from surface Electromyography (SEMG) signals using Artificial<br />

Neural Network (ANN). The haptic device is modeled to act as a virtual<br />

spring. The neural network is trained with EMG data from wrist flexion<br />

action as input and force values from the haptic device as target. The<br />

results shown in this paper illustrate the neural network performance in<br />

estimating the force values in real-time.<br />

◮ FrB02-6 17:30–17:50<br />

Supervisor Design with Petri Nets for Asymmetrical System, pp.628–<br />

632<br />

FENG, Aixiang<br />

LUO, Xiong-lin<br />

China Univ. <strong>of</strong> Petroleum<br />

China Univ. <strong>of</strong> Petroleum<br />

Asymmetrical processes are common nonlinear systems, where the<br />

switching between two different operating modes depends on whether<br />

the system input or output is increasing or decreasing. The existing<br />

intelligent control methods for asymmetical system can’t explain the relationship<br />

between continuous and discrete part <strong>of</strong> the system .With<br />

the idea <strong>of</strong> hybrid systems, a supervisory control method is developed<br />

to distinguish asymmetry phenomena <strong>of</strong> the thermal process <strong>of</strong> a furnace.<br />

In order to ensure the stability <strong>of</strong> arriving at the setting point, a<br />

supervisory controller modelled by the extended Controlled Petri nets<br />

is designed to track errors and the direction <strong>of</strong> the output.Then the<br />

system can be switched to an appropriate mode , where the tracking<br />

determines the switch.Test result verifes the validity <strong>of</strong> this method.<br />

FrB03 15:50–17:50 Room 203C<br />

Control Design<br />

Chair: Wang, Guo-sheng<br />

Co-Chair: XIE, Wei<br />

Acad. <strong>of</strong> Armored Force Engineering<br />

South China Univ. <strong>of</strong> Tech.<br />

◮ FrB03-1 15:50–16:10<br />

Quadratically Stabilizing Observer-based Controller Design for LPV<br />

Plant , pp.1073–1076<br />

XIE, Wei<br />

South China Univ. <strong>of</strong> Tech.<br />

This paper presents a new quadratic stability-preserving state-space<br />

realization and quadratically stabilizing observer-based controller design<br />

for Linear Parameter Varying (LPV) plant, which is combined with<br />

a set <strong>of</strong> local vertex LTI plants. A two-step procedure is taken to design<br />

a quadratically stabilizing observer-based controller for LPV plant. First,<br />

according to the stabilizabilities <strong>of</strong> local LTI plants, state-feedback gain<br />

matrices and Lyapunov matrices could be solved for these local plants<br />

with linear matrix inequality technique, independently. With these Lyapunov<br />

matrices, a proper state space realization <strong>of</strong> LPV plant is provided.<br />

Second, based on the state space realization, a quadratically<br />

stabilizing observer-based controller is obtained.<br />

◮ FrB03-2 16:10–16:30<br />

A Closed-loop Evaluation for Regulatory Control Structure <strong>of</strong> Multivariable<br />

System, pp.1083–1088<br />

LUO, Xiong-lin<br />

Ren, Li-hong<br />

China Univ. <strong>of</strong> Petroleum<br />

China Univ. <strong>of</strong> Petroleum<br />

In multivariable processes, control structure selection represented as<br />

the pairing <strong>of</strong> manipulated variables and controlled variables is a major<br />

concern during the design <strong>of</strong> multi-loop regulatory PID control system.<br />

Each pairing method has its application scope, and it is necessary to<br />

analyze and evaluate the closed-loop application effect <strong>of</strong> the system<br />

pairing. Based on the dynamic transmission ratio between manipulated<br />

variables and controlled variables under the closed-loop state, the<br />

degree <strong>of</strong> coupling for each loop is calculated for the system pairing,<br />

and the closed-loop evaluation which is achieved by the determination<br />

<strong>of</strong> whether the degree <strong>of</strong> coupling is within the threshold is proposed<br />

to determine whether the result is satisfied. This method can not only<br />

be used to compare several different pairing results during the design<br />

102

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!