The performance of data-driven models relies heavily on the amount and quality of training samples, so it might deteriorate significantly in the regions where samples are scarce. The objective of this paper is to deve...The performance of data-driven models relies heavily on the amount and quality of training samples, so it might deteriorate significantly in the regions where samples are scarce. The objective of this paper is to develop an online SVR model updating strategy to track the change in the process characteristics efficiently with affordable computational burden. This is achieved by adding a new sample that violates the Karush–Kuhn–Tucker conditions of the existing SVR model and by deleting the old sample that has the maximum distance with respect to the newly added sample in feature space. The benefits offered by such an updating strategy are exploited to develop an adaptive model-based control scheme, where model updating and control task perform alternately.The effectiveness of the adaptive controller is demonstrated by simulation study on a continuous stirred tank reactor. The results reveal that the adaptive MPC scheme outperforms its non-adaptive counterpart for largemagnitude set point changes and variations in process parameters.展开更多
In this paper, we study a basic class of first order sampled-data control systems with unknown nonlinear structure and with sampling rate not necessarily fast enough, aiming at understanding the capability and limitat...In this paper, we study a basic class of first order sampled-data control systems with unknown nonlinear structure and with sampling rate not necessarily fast enough, aiming at understanding the capability and limitations of the sampled-data feedback. We show that if the unknown nonlinear function has a linear growth rate with its 'slope' (denoted by L) being a measure of the 'size' of uncertainty, then the sampling rate should not exceed 1/L multiplied by a constant (≈ 7.53) for the system to be globally stabilizable by the sampled-data feedback. If, however, the unknown nonlinear function has a growth rate faster than linear, and if the system is disturbed by noises modeled as the standard Brownian motion, then an example is given, showing that the corresponding sampled-data system is not stabilizable by the sampled-data feedback in general, no matter how fast the sampling rate is.展开更多
基金Supported by the National Basic Research Program of China(2012CB720500)Postdoctoral Science Foundation of China(2013M541964)Fundamental Research Funds for the Central Universities(13CX05021A)
文摘The performance of data-driven models relies heavily on the amount and quality of training samples, so it might deteriorate significantly in the regions where samples are scarce. The objective of this paper is to develop an online SVR model updating strategy to track the change in the process characteristics efficiently with affordable computational burden. This is achieved by adding a new sample that violates the Karush–Kuhn–Tucker conditions of the existing SVR model and by deleting the old sample that has the maximum distance with respect to the newly added sample in feature space. The benefits offered by such an updating strategy are exploited to develop an adaptive model-based control scheme, where model updating and control task perform alternately.The effectiveness of the adaptive controller is demonstrated by simulation study on a continuous stirred tank reactor. The results reveal that the adaptive MPC scheme outperforms its non-adaptive counterpart for largemagnitude set point changes and variations in process parameters.
基金This work is supported by the National Natural Science Foundation of China and the National Key Project of China.
文摘In this paper, we study a basic class of first order sampled-data control systems with unknown nonlinear structure and with sampling rate not necessarily fast enough, aiming at understanding the capability and limitations of the sampled-data feedback. We show that if the unknown nonlinear function has a linear growth rate with its 'slope' (denoted by L) being a measure of the 'size' of uncertainty, then the sampling rate should not exceed 1/L multiplied by a constant (≈ 7.53) for the system to be globally stabilizable by the sampled-data feedback. If, however, the unknown nonlinear function has a growth rate faster than linear, and if the system is disturbed by noises modeled as the standard Brownian motion, then an example is given, showing that the corresponding sampled-data system is not stabilizable by the sampled-data feedback in general, no matter how fast the sampling rate is.
