Luo Xionglin

Bio: Luo Xionglin is an academic researcher from China University of Petroleum. The author has contributed to research in topics: Model predictive control & Optimization problem. The author has an hindex of 5, co-authored 28 publications receiving 90 citations.

Modeling of soft sensor for chemical process

Abstract: In the commercial chemical process,many primary product variables cannot be measured online,and soft sensor is an important means to solve this problem.Soft sensing modeling is the core issue of soft sensor.The relationship between soft sensing modeling and identification and nonlinear modeling is presented.The dynamic relationship between quality variables and variables that are easy to measure exists between the increments,and identification depends on incremental data,while soft sensing modeling depends on the measured data to get the relationship.Nonlinear modeling establishes the static relationship between these variables,ignoring the dynamic characteristics,which soft sensing modeling should take into account.With deeper understanding of the chemical process properties,the types and structures of soft sensing model have undergone a great change in the last decades,and soft sensing modeling method evolves from mechanism modeling to data-driven modeling,from linear modeling to nonlinear modeling,and from static modeling to dynamic modeling.The development of the soft sensing modeling method is reviewed.The advantages and disadvantages of the proposed methods are analyzed,and the applications of these methods are shown.In the end,the hot issues and the directions of development of soft sensing modeling method are presented.

Reserch on Perceptron Learning Algorithm

Abstract: This paper introduces some perceptron algorithms and their variations,gives various pseudo-codes,pionts out advantage among algorithms.It gives mistake bound's theorems of perceptrons algorithm in linearly separable and unlinearly separable situation.It studies their mistake bounds and works out their bounds.It shows their various applications in the online optimization,reinforcement learning and bandit algorithm,and discusses the open problems.

Predicting protein structural classes with autoencoder neural networks

Abstract: Autoencoder neural networks was firstly introduced by G.E.Hinton to reduce the dimensionality of data.In this paper, we propose a new way to classify protein structural classes using autoencoder neural networks. The optimum configurations with respect to the size of hidden layers are identified. The problem of training a deep autoencoder for classifying protein structural classes is addressed. Stacked autoencoder is used for reducing the convergence time of training. We design a series of experiments to testify the effectiveness of using autoencoder networks to tackle the problem of predicting protein structure. The experimental results show that our proposed method is competitive with state-of-the-art SVM methods in predicting protein structure class.

Multi-model predictive control of nonlinear system based on self-tuning model

Abstract: To handle the unsymmetrical dynamic characteristics of some nonlinear systems in chemical process,a multi-model predictive control method was proposed based on self-tuning model.Through second order Taylor expansion,the current working point was used to modify the local linear models at equilibrium points to form the multiple self-tuning linear models of nonlinear system.And by the integration with state feedback predictive control and the employment of a switching function based on output errors,the multi-model predictive controller was composed.The simulated results of pH control show the efficiency of the method.

Research on denoising sparse autoencoder

Abstract: Autoencoder can learn the structure of data adaptively and represent data efficiently. These properties make autoencoder not only suit huge volume and variety of data well but also overcome expensive designing cost and poor generalization. Moreover, using autoencoder in deep learning to implement feature extraction could draw better classification accuracy. However, there exist poor robustness and overfitting problems when utilizing autoencoder. In order to extract useful features, meanwhile improve robustness and overcome overfitting, we studied denoising sparse autoencoder through adding corrupting operation and sparsity constraint to traditional autoencoder. The results suggest that different autoencoders mentioned in this paper have some close relation and the model we researched can extract interesting features which can reconstruct original data well. In addition, all results show a promising approach to utilizing the proposed autoencoder to build deep models.

Deep model with Siamese network for viable and necrotic tumor regions assessment in osteosarcoma.

Abstract: PURPOSE To achieve automatic classification of viable and necrotic tumor regions in osteosarcoma, most of the existing deep learning methods can only design a simple model to prevent overfitting on small datasets, which leads to the weak ability of extracting image features and low accuracy of the models. In order to solve the above problem, a deep model with Siamese network (DS-Net) was designed in this paper. METHODS The DS-Net constructed on the basis of full convolutional networks is composed of an auxiliary supervision network (ASN) and a classification network. The construction of the ASN based on the Siamese network aims to solve the problem of a small training set (the main bottleneck of deep learning in medical images). It uses paired data as the input and updates the network through combined labels. The classification network uses the features extracted by the ASN to perform accurate classification. RESULTS Pathological diagnosis is the most accurate method to identify osteosarcoma. However, due to intraclass variation and interclass similarity, it is challenging for pathologists to accurately identify osteosarcoma. Through the experiments on hematoxylin and eosin (H&E)-stained osteosarcoma histology slides, the DS-Net we constructed can achieve an average accuracy of 95.1%. Compared with existing methods, the DS-Net performs best in the test dataset. CONCLUSIONS The DS-Net we constructed can not only effectively realize the histological classification of osteosarcoma, but also be applicable to many other medical image classification tasks affected by small datasets.

Short-term output power forecasting of photovoltaic systems based on the deep belief net:

Abstract: Photovoltaic power is now a major green energy resource, and its generated power can be directly connected to the power grid. However, the stability of power grid may be affected by the random and ...