Closed-loop identification is important and necessary to various model-based advanced process control strategies, whose performance depends greatly on the informative property of the data set. Switching control is an ...Closed-loop identification is important and necessary to various model-based advanced process control strategies, whose performance depends greatly on the informative property of the data set. Switching control is an important method in process control. Therefore, this paper studies the informative property of a data set in a single-input single-output (SISO) closed-loop system with a switching controller. It is proved that this data set is informative if the controller switches among at least two modes (i.e., feedback laws). Our result does not require any assumption on the way of switch and removes the constraints on the switching manner required in some classical literature. Finally, simulation case studies based on a continuous stirred-tank reactor (CSTR) process are given to validate the results.展开更多
In machine learning and statistics, classification is the a new observation belongs, on the basis of a training set of data problem of identifying to which of a set of categories （sub-populations） containing observa...In machine learning and statistics, classification is the a new observation belongs, on the basis of a training set of data problem of identifying to which of a set of categories （sub-populations） containing observations （or instances） whose category membership is known. SVM （support vector machines） are supervised learning models with associated learning algorithms that analyze data and recognize patterns, used for classification and regression analysis. The basic SVM takes a set of input data and predicts, for each given input, which of two possible classes fon^as the output, making it a non-probabilistic binary linear classifier. In pattern recognition problem, the selection of the features used for characterization an object to be classified is importance. Kernel methods are algorithms that, by replacing the inner product with an appropriate positive definite function, impticitly perform a nonlinear mapping 4~ of the input data in Rainto a high-dimensional feature space H. Cover＇s theorem states that if the transformation is nonlinear and the dimensionality of the feature space is high enough, then the input space may be transformed into a new feature space where the patterns are linearly separable with high probability.展开更多
基金Supported by the National Basic Research Program of China (2010CB731800)the National Natural Science Foundation of China (60974059, 60736026, 61021063, 60904044, 61290324)Tsinghua National Laboratory for Information Science and Technology (TNList) Cross-discipline Foundation
文摘Closed-loop identification is important and necessary to various model-based advanced process control strategies, whose performance depends greatly on the informative property of the data set. Switching control is an important method in process control. Therefore, this paper studies the informative property of a data set in a single-input single-output (SISO) closed-loop system with a switching controller. It is proved that this data set is informative if the controller switches among at least two modes (i.e., feedback laws). Our result does not require any assumption on the way of switch and removes the constraints on the switching manner required in some classical literature. Finally, simulation case studies based on a continuous stirred-tank reactor (CSTR) process are given to validate the results.
文摘In machine learning and statistics, classification is the a new observation belongs, on the basis of a training set of data problem of identifying to which of a set of categories （sub-populations） containing observations （or instances） whose category membership is known. SVM （support vector machines） are supervised learning models with associated learning algorithms that analyze data and recognize patterns, used for classification and regression analysis. The basic SVM takes a set of input data and predicts, for each given input, which of two possible classes fon^as the output, making it a non-probabilistic binary linear classifier. In pattern recognition problem, the selection of the features used for characterization an object to be classified is importance. Kernel methods are algorithms that, by replacing the inner product with an appropriate positive definite function, impticitly perform a nonlinear mapping 4~ of the input data in Rainto a high-dimensional feature space H. Cover＇s theorem states that if the transformation is nonlinear and the dimensionality of the feature space is high enough, then the input space may be transformed into a new feature space where the patterns are linearly separable with high probability.
