Fault diagnosis is of great importance to enhance the reliability and security of the complex chemical processes. However, many fault diagnosis methods cannot extract correlation information among attributes in the feature extraction procedure, resulting in the degradation of diagnosis performance. In this paper, a novel framework is proposed for fault diagnosis of the chemical processes. First, in the framework, a enhanced kernel principal component analysis (eKPCA) is proposed to obtain the key features of the fault based on the Hotelling’s T2 and squared prediction error (SPE) statistics. Second, to learn the correlation information of the features, an enhanced Naive Bayesian model (eNBM) is constructed with multi-variate Gaussian kernel function. Third, dragonfly algorithm (DA) is employed to find the optimal smoothing parameter to improve the performance of eNBM. The framework is applied to make fault diagnosis of Tennessee Eastman process (TE). Experimental results show that the proposed framework is more effective compared to the already existed traditional approaches such as deep learning.

Online learning in financial management is becoming more common since it has the convenience and flexibility for students. How to predict students’ online learning performance has become a topic of general interest in today's education research. Therefore, a new predictive model is proposed to predict students’ online learning performance. Furthermore, based on students' behavior and their academic information, the proposed model can improve accuracy. The proposed model is compared with other models, and gets best experiment results. The experimental results show that the proposed model outperforms other predictive models.