Abstract. The performance of neural nets can be improved through the use of ensembles of redundant nets. In this paper, some of the available methods of ensemble creation are reviewed and the "test and select" methodolology for ensemble creation is considered. This approach involves testing potential ensemble combinations on a validation set, and selecting the best performing ensemble on this basis, which is then tested on a final test set. The application of this methodology, and of ensembles in general, is explored further in two case studies. The first case study is of fault diagnosis in a diesel engine, and relies on ensembles of nets trained from three different data sources. The second case study is of robot localisation, using an evidence-shifting method based on the output of trained SOMs. In both studies, improved results are obtained as a result of combining nets to form ensembles.

No abstract

A multi-net fault diagnosis system designed to provide an early warning of combustion-related faults in a diesel engine is presented. Two faults (a leaking exhaust valve and a leaking fuel injector nozzle) were physically induced (at separate times) in the engine. A pressure transducer was used to sense the in-cylinder pressure changes during engine cycles under both of these conditions, and during normal operation. Data corresponding to these measurements were used to train artificial neural nets to recognise the faults, and to discriminate between them and normal operation. Individually trained nets, some of which were trained on subtasks, were combined to form a multi-net system. The multi-net system is shown to be effective when compared with the performance of the component nets from which it was assembled. The system is also shown to outperform a decision-tree algorithm (C5.0), and a human expert; comparisons which show the complexity of the required discrimination. The results illustrate the improvements in performance that can come about from the effective use of both problem decomposition and redundancy in the construction of multi-net systems.

The development of a neural net system for fault diagnosis in a marine diesel engine is described.

A case study is presented in which data from two different sources are combined to create robust diagnostic systems. A diesel engine is used as an exemplar of the class of mechanical machines. Several commonly occurring faults symptomatic of early stages of fault development are induced in the engine. Data in the form of cylinder pressures and vibration are acquired. Orthogonal wavelet transforms, principal component analysis and time domain information are used to extract features from the data. Several artificial neural net classifiers are developed using these data. Statistical models are used to evaluate the diversity within the methodologies used to create the classifiers. The 'diversity' metrics are used to propose the most effective majority voting system.