Data-Driven Stochastic Optimal Iterative Learning Control for Nonlinear Non-Affine Systems With Measurement Data Loss

Abstract

In this paper, the measurement data loss is considered and two data-driven stochastic optimal iterative learning control (DDSOILC) methods are presented directly for nonlinear network systems. Specifically, an iterative dynamic linearization (IDL) is adopted to construct the linear incremental input output relationship of the repetitive nonlinear network system between two consecutive iterations. In the sequel, a lifted IDL is obtained by defining two super vectors of inputs and outputs over the entire finite time interval. Then, a lifted IDL-based DDSOILC scheme is proposed where the random data loss is described by a Bernoulli distribution of random variable. The results are extended by using a non-lifted IDL where the input-output relationship is described pointwisely. The learning gains of the proposed two methods are iteration-time-variant and can be iteratively estimated using real-time data. The proposed two methods do not depend on any explicit model. Moreover, the proposed non-lifted IDL-based DDSOILC can use more control information than the proposed lifted IDL-based one, and thus it can achieve a better control performance. Both theoretical analysis and simulations verify the efficiency and applicability of the two proposed methods.INDEX TERMS Data-driven control, stochastic optimal ILC, nonlinear network systems, measurement data loss.

Data-Driven Model-Free Adaptive Positioning and Anti-Swing Control for Bridge Cranes

“…( 16) provides the basis for whether the subsequent bridge crane system can be converted into a linear model with standard input and output changes, and whether the model-free adaptive control method can be used. So, the following assumptions are made to analyze (16).…”

“…The controller design is only conducted through offline or real-time online input and output data of the controlled system. At present, the data-driven control method has been developed and improved continuously, and has been recognized symbolically at home and abroad [11], [12], among which PID control [13], [14], iterative learning control [15], [16], iterative feedback tuning [17], [18], approximate dynamic programming [19], [20] and other methods have been widely used.…”

Model-free adaptive iterative learning integral terminal sliding mode control of exoskeleton robots

“…A reset mechanism is also combined with (24) to make its estimation capacity stronger (Liang et al, 2019)where truetrueΦ¯^(k,1) denotes the value of the estimated PPDs, that is, truetrueΦ¯^(k,r)=[normalΦ^1(k,r),normalΦ^2(k,r),normalΦ^3(k,r)Δtrued^(k,r)], at the first iteration, that is, r = 1. ϵ > 0 is a small-valued parameter, and ϕ¯^ij(k,r) is the ( i , j )th component of truetrueΦ¯^(k,r).…”

“…A reset mechanism is also combined with (24) to make its estimation capacity stronger (Liang et al, 2019…”

Multi-lagged-input information enhancing quantized iterative learning control

“…Fortunately, a new IDL method (Chi et al, 2015) has been proposed by using multi-lagged-input signals with corresponding multiple iteration-time-varying parameters, that is, the control inputs from zero time instant to the current time instant have been utilized together with the corresponding parameters over the interval from zero to the current time instant. However, the problem of data quantization has not been considered in all the ILC methods (Chi et al, 2017; Lin et al, 2018; Liang et al, 2019b; Meng et al, 2015) proposed by using the multi-lagged-input IDL technique.…”