Nini Maharana; Ameeya Kumar Nayak

This paper presents a least-squares finite element approach for the solution of reaction-diffusion pellet problem for slab geometry by considering some special types of reactions which are applicable for chemical transformation of natural gas components. The nonlinear terms are treated with Newton's linearisation technique. A detailed comparison with other weighted residual methods such as collocation, sub-domain and Galerkin methods are also examined. The issues related to chemical reactor problems and their error analysis is discussed both theoretically and numerically. The least-squares method suffers convergency effect by using the Picard's method instead of Newton's iteration method as obtained by Galerkin and orthogonal collocation methods.]]>

Wei Pan; Yongquan Zhou; Zhiming Li

This paper proposed an improved multi-verse optimisation (IMVO) algorithm based on exponential function inflation size. The main inspirations of IMVO are based on cosmic expansion, and inflation never ends. In the entire universe, we can think it is exponential growth. Exponential function inflation size is introduced to enhance accuracy and increase convergence rate of the multi-verse optimisation (MVO) algorithm. The numerical simulation experiments and comparisons are carried out based on a set of ten benchmark functions. The IMVO algorithm is compared with multi-verse optimisation (MVO), moth-flame optimisation (MFO), artificial bee colony (ABC) algorithm, bat algorithm (BA), differential evolution (DE) algorithm and dragonfly algorithm (DA). The experiment results show that the IMVO has not only the higher accuracy but also the faster convergence speed.]]>

Yanmin Zhao; Shuai Chen; Tinggui Chen

In order to solve the data analysis problem in scientific research management universities, this paper puts forward a K-means clustering method based on artificial immune system. Firstly, the hidden rules and knowledge in the scientific data are found out in order to provide decision opinion for the development of education and scientific research. And then, the key research domain and creative characteristics of the subject can be obtained. Finally, the simulation results show that the method is feasible and effective.]]>

Weiliang Ding; Gongfa Li; Ying Sun; Guozhang Jiang; Jianyi Kong; Honghai Liu

In order to promote the accuracy and complexity in the recognition of sEMG signals by classifiers, this paper tells a method based on fused D-S evidential theory. Three features are discussed in the choice of parameters, which includes AR model coefficient, cepstral coefficients and time-domain integral absolute value. D-S evidential theory gets information based on information fusion of multi feature sets and multi classifiers. In recognition phase, many groups of data are used for the training and the rest is for the test. Through the compare of the accuracy in different parameters, the result is shown according to the experiment about the data fusion in D-S evidential theory. Six actions are set to be the samples. According to three characters, the recognition accuracy is compared. The result shows that the fused data method of D-S evidential theory has better accuracy and robustness. The further study is to determine the optimal fusion feature set to make more accurate and higher robustness of the classification.]]>

Xianfu Cheng; Junyi Luo; Dongshan Gao

For leader-follower characteristic between platform parameters and customisation parameters in product family design, a bi-level model and leaderfollower correlation optimisation model are developed. The adaptability of platform parameters is considered, and they are regarded as association parameters to transfer between the upper and lower in the model. After the values of platform variables in the upper level model are assigned, each sub-plan problem in the lower level model can be solved without relying on platform parameters under certain conditions. Simultaneously, incidence relation of platform parameters could be taken as relaxation constraints, and genetic algorithm is applied to solve the leader-follower correlation optimisation model of the product family. Finally, an example on optimal design of product family for drum group is used to demonstrate the feasibility of the established model and proposed method.]]>

Mahmoud Behroozifar

In this paper, a method based on Bernstein polynomial is proposed for numerical solution of an one-dimensional heat equation subject to non-local boundary conditions. Properties of Bernstein polynomial and operational matrices of integration, differentiation and the product are stated and utilised to reduce the solution of the given elliptic partial differential equation to the solution of the system of algebraic equations. Illustrative numerical tests are given to compare the efficiency and accuracy of the new method with other methods.]]>

Yunfang Peng; Yajuan Han; Yuyan Xiao; Saif Ullah

Grouping of jobs from the same family on bottleneck resource is significant to reduce setup time of jobs. But, the grouping may result in delay of jobs. How to balance the contradiction between utilisation of bottleneck resource and delivery time is a great challenge in production scheduling. This paper considers a single machine scheduling problem in which each job to be scheduled belongs to a family and family setup times exist. This paper is aimed to identify a schedule of families of jobs to minimise total tardiness with respect to the given due dates of the jobs. A mathematical model is formulated for this problem and a heuristic solution approach is presented. Computational experiments show that the proposed heuristic algorithm performs better both in terms of solution quality and computational time.]]>

Wei-Hong Yang; Jia-Rui Liu; Yi Zhang

Cuckoo search algorithm is a recent proposed evolutionary algorithm. However, the local search capability is not efficient when dealing with some multi-modal numerical problems. In this paper, a new local-enhanced cuckoo search algorithm is designed aiming to solve it. Different from the standard version, the local search for each cuckoo is attracted by global best position found by entire swarm. Furthermore, to provide a balance between exploitation and exploration, an inertia weight is introduced. To test the performance, four other algorithms are employed to compare, and simulation results show it achieves the best performance when solve numerical optimisation benchmarks.]]>

Amel Belabbaci; Bachir Djebbar

This paper addresses the optimisation of quadratic functions on convex polyhedrons. We propose a method and an algorithm for optimising a strictly concave function. This algorithm uses a good initialisation by searching the nearest vertex of a convex set to an external point in order to surround the area where the optimal solution can be located. The optimal solution may be the nearest vertex found or a boundary point obtained by the projection of the critical point onto a separating hyperplane passing through the nearest vertex. This method and this algorithm can be adapted for the convex quadratic problem. In this case, the optimal solution is the farthest vertex from the critical point.]]>

Hasnaa El Mandili; Benayad Nsiri

The possibility of removing a customer before being served is one that is often absent in the theory of waiting queues. This reminds us of the concept of negative customer, which was set forth by three leading scientists: Gelenbe, Boucherie-van Dijk and Henderson. Negative customers are likely to eliminate positive ones and make them disappear. They cannot build up in the queue, and are simply lost if they arrive to an empty queue (Gelenbe, 1991; Gelenbe et al., 1991). In other contributions (Boucherie and van Dijk, 1994), we find a type of negative customers that can build up in the queue, as opposed to Gelenbe negative customers. We introduce a new type of negative client, having the same properties mentioned above, but characterised by the fact that they may remove one or more regular clients, and leave the system after being server as a regular customer. We study the existence of stationary distribution, the intensity of transition, along with the quasi-reversibility of this new type of negative customer.]]>