Entity

Time filter

Source Type


Tao X.-M.,Harbin Engineering University | Fu D.-D.,Harbin Engineering University | Liu F.-R.,Harbin Power Vocational Technology College | Liu Y.,Harbin Engineering University
Kongzhi yu Juece/Control and Decision | Year: 2012

A novel multi-scale parallel artificial immune clone algorithm for unsupervised clustering(MSPAICC) is presented, in which, evolutions of subgroups are performed in parallel with the different mutation strategies. The mutation capability of an individual is determined by the competition among subgroups and subgroup fitness value. The larger mutation operator is used to quickly localize the global optimal space at the early evolution, while the smaller mutation operator whose scale gradually reduces are adopted to improve the local search ability at the later evolution. The experimental results show the proposed method can improve clustering performance and the robustness compared with other clustering algorithms. Source


Tao X.-M.,Harbin Engineering University | Liu F.-R.,Harbin Power Vocational Technology College | Liu Y.,Harbin Engineering University | Fu D.-D.,Harbin Engineering University
Kongzhi yu Juece/Control and Decision | Year: 2011

To deal with the problem of single-scale mutation, premature convergence and slow search speed, a clone selection algorithm(CSA) with directional multi-scale Gaussian mutation is proposed. To implement the share of information between antibodies, the directional evolution mechanism is utilized to induce the antibodies to evolve to the best solution region. The special multi-scale Gaussian mutation operators are introduced to make antibodies explore the search space more efficiently. The large-scale mutation operators can be utilized to quickly localize the global optimized space at the early evolution, while the small-scale mutation operators can implement local accurate minima solution search at the late evolution, which can make the algorithm explore the global and local minima thoroughly at the same time. The comparison of the performance of the proposed approach with other CSAs with different mutations is experimented. The experimental results show that the proposed method can not only effectively solve the premature convergence problem, but also significantly speed up the convergence and improve the stability. Source


Xu Y.,Harbin Power Vocational Technology College | Zhu Q.-Y.,Harbin Institute of Technology
Dongli Gongcheng Xuebao/Journal of Chinese Society of Power Engineering | Year: 2011

CuO/AC desulphurizers were prepared using cocoanut shell activated carbon as the support and a fixed-bed flue gas desulfurization (FGD) reactor was set up, so as to study the influence of following factors on FGD performance of the desulphurizer, such as the calcination temperature, Cu loading capacity, desulphurizing temperature and gas component, etc. Results show that for higher FGD performance of CuO/AC desulphurizer, the calcination temperature would better be kept at about 250°C, Cu loading capacity in 5%-7.5%, desulphurizing temperature in 200-250°C and O 2 concentration in flue gas at a moderate degree. Regenerated desulphurizer would have better performance if the regeneration temperature is kept at 300°C, however, its desulphurization reactivity is worse than the original CuO/AC desulphurizer. Source


Tao X.-M.,Harbin Engineering University | Liu F.-R.,Harbin Power Vocational Technology College | Tong Z.-J.,Harbin Engineering University | Yang L.-B.,Harbin Engineering University
Zhendong yu Chongji/Journal of Vibration and Shock | Year: 2010

The performance of traditional support vector machine (SVM) drops significantly when it is applied to the problem of learning from unbalanced datasets where the normal instances heavily outnumbers the fault instances. To address this problem, a novel fault detection approach was proposed based on a variant of synthetic minority over-sample technique (SMOTE) combined with different error cost-sensitive SVM. As the SVM decision boundary is determined only by a small quantity of support vectors, consequently, based on SMOTE, a new minority over-sample method was presented, in which only the minority examples near the borderline are over-sampled. In order to solve the noise effect, the different error cost-sensitive SVM based on K-nearest neighbors (KNN) was adopted to remedy the problem of noise positive instances. The proposed algorithm was applied in bearings fault detection and its results were compared with those by the algorithms of traditional SVM, different classes of cost-sensitive SVMs (SVM-C) and SVM+SMOTE. The experimental results show the approach can achieve better detection performance than other methods. Source


Tao X.,Harbin Engineering University | Liu F.,Harbin Power Vocational Technology College
Proceedings of the 29th Chinese Control Conference, CCC'10 | Year: 2010

To solve the problems of difficultly obtaining abnormal samples in bearings fault detection application and overfitting of conventional classifications due to the abnormal data imbalanced, a novel one-class detection model based on AR model self-correlation's entropy characteristic is presented in this paper. The AR model is employed to extract the normal samples' parameter characteristics and consequently normal samples' AR sub-space is established. The self-correlation of errors generated by other samples' being projected onto the AR model space is calculated. The entropy of the previously calculated self-correlation is used as the metrics of similarity with the normal sub-space. The experiments show that the proposed approach can efficiently overcome the drawback of computational complexity and detection rate's sensitivity to the length of samples of conventional detection methods based on AR' parameters. The single fault detection and diagnosis schemes based on AR self-correlation's entropy also are proposed in this paper. The model's threshold value settings are also analyzed and the determination approach based on the Particle Swarm Optimization is investigated in this paper. The proposed detection scheme is compared against MLP and detection techniques based on AR parameters as features in the experiments. The results illustrate effectiveness of the investigated techniques with some concluding remarks. Source

Discover hidden collaborations