Theoretics of Inverse-Free (IF) Fuzzy ZNN (Zhang Neural Net) with Adaptive SG (Sampling Gap) and Expected Precision Treating Time-Dependent QP (Quadratic Program) Applied to Robot

For solving time-dependent quadratic program (TDQP), some conventional models have been developed in recent years. However, the matrix inverse is usually needed in those conventional models. Not only that, the corresponding discrete models have a time-independent sampling gap. As we know, a small sampling gap corresponds to small computation errors and more computation consumption. Accordingly, a large sampling gap corresponds to large computation errors and less computation consumption. Hence, selecting an appropriate sampling gap is usually meaningful and significant. Specifically, as for the time-dependent problem, an appropriate time-dependent sampling gap is meaningful. In this paper, an inverse-free continuous Zhang neural net (IFCZNN) model is first developed for solving TDQP. A fuzzy system with two inputs and one output is designed to adaptively adjust the sampling gap under the expected precision. Aided by the fuzzy system, an inverse-free (IF) discrete model is further proposed, which is termed IF fuzzy discrete Zhang neural net (IFFDZNN) model. In addition, simulative and physical experiments about robot control are performed to illustrate the applicability of the IFFDZNN model.