CONVERGENCE ANALYSIS OF DISCRETE-TIME NEURAL NETWORK FOR SOLVING QUADRATIC PROGRAMMING PROBLEMS
LU Yang ,LI Dewei ,XI Yugeng ,LU Jianbo
Author information+
Department of Automation, Shanghai Jiao Tong University, Key Laboratory of System Control and Information Processing, Ministry of Education, Shanghai 200240
The convergence property of discrete-time neural network for quadratic programming is analyzed. By choosing a proper Lyapunov function, a sufficient ondition for global convergence is obtained. The convergence rate under the condition is also investigated through a in-depth discussion about full-row-rank inequality constraint left matrix condition nd non-full-row-rank inequality constraint left matrix condition, respectively, and the exponential convergence property for both full-row rank and non-full-row rank inequality constraint left matrix conditions under the mentioned sufficient condition is proved. Simulation result
verifies the validity of the theoretical results obtained in this paper.
LU Yang ,LI Dewei ,XI Yugeng ,LU Jianbo.
CONVERGENCE ANALYSIS OF DISCRETE-TIME NEURAL NETWORK FOR SOLVING QUADRATIC PROGRAMMING PROBLEMS. Journal of Systems Science and Mathematical Sciences, 2012, 32(11): 1343-1353 https://doi.org/10.12341/jssms11991