Zhang neuronet solving complex-valued time-varying linear inequalities

In this paper, a special method called Zhang neuronet (ZN) is proposed and investigated for online solution of complex-valued time-varying linear inequalities (CVTVLI). Instead of employing a norm-based energy function in traditional gradient neuronet (GN) and related methods, the given ZN model is designed using a vector-valued error function and takes advantage of the first-order time-derivative information of time-varying coefficients involved in CVTVLI. Through adjusting the value of design parameter γ appropriately, superior convergence performance is achieved for the proposed ZN model for dealing with such a time-varying problem. Then, theoretical and simulative results are given to illustrate and substantiate the convergence property of the proposed ZN model. Besides, a GN model is developed and exploited to for the same CVTVLI solving. The comparison on transient behaviors of these two models further shows the efficacy and superiority of the proposed ZN model.