A common way of computing all efficient (Pareto optimal) solutions for a biobjective combinatorial optimisation problem is to compute first the extreme efficient solutions and then the remaining, non-extreme solutions. The second phase, the computation of non-extreme solutions, can be based on a “k-best” algorithm for the single-objective version of the problem or on the branch-and-bound method. A k-best algorithm computes the k-best solutions in order of their objective values. We compare the performance of these two approaches applied to the biobjective minimum spanning tree problem. Our extensive computational experiments indicate the overwhelming superiority of the k-best approach. We propose heuristic enhancements to this approach which further improve its performance. 相似文献
The importance of batch reactors in today's process industries cannot be overstated. Thus said, it is important to optimise their operation in order to consistently achieve products of high quality while minimising the production of undesirables. In processes like polymerisation, these reactors are responsible for a greater number of products than other reactor types and the need for optimal operation is therefore greater.
An approach based on an offline dynamic optimisation and online control strategy is used in this work to generate optimal set point profiles for the batch polymerisation of methyl methacrylate. Dynamic optimisation is carried out from which controller set points to attain desired polymer molecular end point characteristics are achieved. Temperature is the main variable to be controlled, and this is done over finite discrete intervals of time.
For on-line control, we evaluate the performance of neural networks in two controllers used to track the derived optimal set points for the system. The controllers are generic model control (GMC), ([P.L. Lee, G.R. Sullivan, Generic model control, Comput. Chem. Eng. 12(6) (1998) 573–580]) and the neural network-based inverse model-based control (IMBC), ([M.A. Hussain, L.S. Kershenbaum, Implementation of an inverse model based control strategy using neural networks on a partially simulated exothermic reactor, Trans. IchemE 78(A) (2000) 299–311]). Although the GMC is a model-based controller, neural networks are used to estimate the heat release within its framework for on-line control. Despite the application of these two controllers to general batch reactors, no published work exists on their application to batch polymerisation in the literature. In this work, the performance of the neural networks within each controller's algorithm for tracking and setpoint regulation of the optimal trajectory and in robustness tests on the system is evaluated. 相似文献
Optimization of the wire bonding process of an integrated circuit (IC) is a multi-objective optimization problem (MOOP). In
this research, an integrated multi-objective immune algorithm (MOIA) that combines an artificial immune algorithm (IA) with
an artificial neural network (ANN) and a generalized Pareto-based scale-independent fitness function (GPSIFF) is developed
to find the optimal process parameters for the first bond of an IC wire bonding. The back-propagation ANN is used to establish
the nonlinear multivariate relationships between the wire boning parameters and the multi-responses, and is applied to generate
the multiple response values for each antibody generated by the IA. The GPSIFF is then used to evaluate the affinity for each
antibody and to find the non-dominated solutions. The “Error Ratio” is then applied to measure the convergence of the integrated
approach. The “Spread Metric” is used to measure the diversity of the proposed approach. Implementation results show that
the integrated MOIA approach does generate the Pareto-optimal solutions for the decision maker, and the Pareto-optimal solutions
have good convergence and diversity performance. 相似文献
This paper focuses on hierarchical classification problems where the classes to be predicted are organized in the form of
a tree. The standard top-down divide and conquer approach for hierarchical classification consists of building a hierarchy
of classifiers where a classifier is built for each internal (non-leaf) node in the class tree. Each classifier discriminates
only between its child classes. After the tree of classifiers is built, the system uses them to classify test examples one
class level at a time, so that when the example is assigned a class at a given level, only the child classes need to be considered
at the next level. This approach has the drawback that, if a test example is misclassified at a certain class level, it will
be misclassified at deeper levels too. In this paper we propose hierarchical classification methods to mitigate this drawback.
More precisely, we propose a method called hierarchical ensemble of hierarchical rule sets (HEHRS), where different ensembles
are built at different levels in the class tree and each ensemble consists of different rule sets built from training examples
at different levels of the class tree. We also use a particle swarm optimisation (PSO) algorithm to optimise the rule weights
used by HEHRS to combine the predictions of different rules into a class to be assigned to a given test example. In addition,
we propose a variant of a method to mitigate the aforementioned drawback of top-down classification. These three types of
methods are compared against the standard top-down hierarchical classification method in six challenging bioinformatics datasets,
involving the prediction of protein function. Overall HEHRS with the rule weights optimised by the PSO algorithm obtains the
best predictive accuracy out of the four types of hierarchical classification method. 相似文献
A microarray machine offers the capacity to measure the expression levels of thousands of genes simultaneously. It is used
to collect information from tissue and cell samples regarding gene expression differences that could be useful for cancer
classification. However, the urgent problems in the use of gene expression data are the availability of a huge number of genes
relative to the small number of available samples, and the fact that many of the genes are not relevant to the classification.
It has been shown that selecting a small subset of genes can lead to improved accuracy in the classification. Hence, this
paper proposes a solution to the problems by using a multiobjective strategy in a genetic algorithm. This approach was tried
on two benchmark gene expression data sets. It obtained encouraging results on those data sets as compared with an approach
that used a single-objective strategy in a genetic algorithm.
This work was presented in part at the 13th International Symposium on Artificial Life and Robotics, Oita, Japan, January
31–February 2, 2008 相似文献