Aiming at high overshoot and steady-state error in fuzzy controller of underwater robots, a new method based on data mining technique was presented. Apply Boolean association rule data mining to mine the polling list of fuzzy control from the database of manual operation record, and simulation and pool experiments were carried out on ship detection underwater robot to verify the feasibility and superiority. The results show that the controller has lower overshoot and good robustness to external disturbances, and the polling list of fuzzy control can be constructed automatically by Boolean association rule data mining, which improves the accuracy and the precision of motion control for underwater robots.
[1]
Jiawei Han,et al.
Data Mining: Concepts and Techniques
,
2000
.
[2]
Guoqing Chen,et al.
Mining generalized association rules with fuzzy taxonomic structures
,
1999,
18th International Conference of the North American Fuzzy Information Processing Society - NAFIPS (Cat. No.99TH8397).
[3]
Yu Jin,et al.
Data Mining Technique
,
2000
.
[4]
J. A. Adam,et al.
Probing beneath the sea: Sending vessels into environments too harsh for humans poses challenges in communications, artificial intelligence, and power-supply technology
,
1985,
IEEE Spectrum.