Point-wise confidence interval estimation by neural networks: A comparative study based on automotive engine calibration.

In developing neural network techniques for real world applications it is still very rare to see estimates of confidence placed on the neural network predictions. This is a major deficiency, especially in safety-critical systems. In this paper we explore three distinct methods of producing point-wise confidence intervals using neural networks. We compare and contrast Bayesian, Gaussian Process and Predictive error bars evaluated on real data. The problem domain is concerned with the calibration of a real automotive engine management system for both air-fuel ratio determination and on-line ignition timing. This problem requires real-time control and is a good candidate for exploring the use of confidence predictions due to its safety-critical nature.

Finding relevant search engine results:a minimax linear programming approach

For a submitted query to multiple search engines finding relevant results is an important task. This paper formulates the problem of aggregation and ranking of multiple search engines results in the form of a minimax linear programming model. Besides the novel application, this study detects the most relevant information among a return set of ranked lists of documents retrieved by distinct search engines. Furthermore, two numerical examples aree used to illustrate the usefulness of the proposed approach.

Establishing a framework for dynamic risk management in 'intelligent' aero-engine control

The behaviour of control functions in safety critical software systems is typically bounded to prevent the occurrence of known system level hazards. These bounds are typically derived through safety analyses and can be implemented through the use of necessary design features. However, the unpredictability of real world problems can result in changes in the operating context that may invalidate the behavioural bounds themselves, for example, unexpected hazardous operating contexts as a result of failures or degradation. For highly complex problems it may be infeasible to determine the precise desired behavioural bounds of a function that addresses or minimises risk for hazardous operation cases prior to deployment. This paper presents an overview of the safety challenges associated with such a problem and how such problems might be addressed. A self-management framework is proposed that performs on-line risk management. The features of the framework are shown in context of employing intelligent adaptive controllers operating within complex and highly dynamic problem domains such as Gas-Turbine Aero Engine control. Safety assurance arguments enabled by the framework necessary for certification are also outlined.

The potential use of bio-ultracarbofluids in a standard diesel engine

The replacement of diesel fuel by ultra-carbofluids was perceived to offer the potential to decrease the emissions of environmental pollutants such as carbon dioxide, carbon monoxide, hydrocarbons (HC's) and smoke. Such ultracarbofluids consist of a suspension of coal in fuel oil and water generally in the ratio of 5: 3: 2 plus a small amount of stabilising additive. The literature relating to the economies of coal and fuel oil production, and the production and properties of charcoal and vegetable oils has been critically reviewed. The potential use of charcoal and vegetable oils as replacements for coal and fuel oil are discussed. An experimental investigation was undertaken using novel bio-ultracarbofluid formulations. These differed from an ultracarbofluid by having bio-renewable charcoal and vegetable oil in place of coal and fuel oil. Tests were made with a Lister-Petter 600cc 2-cylinder, 4-stroke diesel engine fitted with a Heenan-Froude DPX 1 water brake dynamometer to measure brake power output, and Mexa-321E and Mexa-211E analysers to measure exhaust pollutants. Measurements were made of engine brake power output, carbon dioxide, carbon monoxide, hydrocarbons and smoke emissions over the speed range 1000 to 3000 rpm at 200 rpm intervals. The results were compared with those obtained with a standard diesel reference fuel. All the bio-ultracarbofluid formulations produced lower brake power outputs (i.e. 5.6% to 20.7% less brake power) but substantially improved exhaust emissions of CO2, CO, HC's and smoke. The major factor in the formulation was found to be the type and amount of charcoal; charcoal with a high volatile content (27.2%) and present at 30% by mass yielded the best results, i.e. only slightly lower brake power output and significantly lower exhaust pollutants.

Investigation of gas engine driven heat pump systems

Heat pumps are becoming increasingly popular, but poor electricity generating efficiency limits the potential energy savings of electrically powered units. Thus the work reported in this thesis concerns the development of a range of gas engine driven heat pumps for industrial and commercial heating applications, which recover heat from the prime mover, normally rejected to waste. Despite the convenience of using proprietary engine heat recovery packages, investigations have highlighted the necessity to ensure the engine and the heat recovery equipment are compatible. A problem common •to all air source heat pumps is the formation of frost on the evaporator, which must be removed periodically, with the expenditure of energy, to ensure the continued operation of the plant. An original fluidised bed defrosting mechanism is proposed, which prevents the build-up of this frost, and also improves system performance. Criticisms have been levelled against the rotary sliding vane compressor, in particular the effects of lubrication, which is essential. This thesis compares the rotary sliding vane compressor with other machines, and concludes that many of these criticisms are unfounded. A confidential market survey indicates an increasing demand for heat pumps up to and including 1990, and the technical support needed to penetrate this market is presented. Such support includes the development of a range of modular gas engine driven heat pumps, and a computer aided design for the selection of the optimum units. A case study of a gas engine driven heat pump for a swimming pool application which provided valuable experience is included.

Formation of droplets from the liquid wall film passing through the inlet port of a spark ignition engine

The available literature has been surveyed to determine the parameters affecting fuelling requirements of spark ignition engines and their relation to engine performance and emissions. Theories and experiment relating to two phase and multi-component flows have also been examined and the techniques employed in the measurement of droplet sizes and liquid wall films have been reviewed. Following preliminary steady flow visualisation experiments to examine the trajectories of droplets discharging from the valve port an extensive practical investigation of the spectrum of droplet sizes formed by the break up of the wall film has produced results which have been correlated in terms of the important fuel and airflow parameters. It is concluded that the Sauter mean diameter of droplets formed by the break up of the wall film will vary between 70 and 150 m, depending on the operating conditions of the engine. The spectra of droplet sizes measured show that a significant proportion of the total mass of the wall film breaks into drops which will be too large to burn completely and, by comparison with measurements of unburned hydrocarbon emissions from engines supplied with a homogeneous mixture of air and gaseous hydrocarbons, it is concluded that the droplets from the wall film are likely to increase emissions by 50%.

Modelling the behaviour of engine assembly workers

Presents a prototype modelling methodology that provides a generic approach to the creation of quantitative models of the relationships between a working environment, the direct workers and their subsequent performance. Once created for an organisation, such models can provide a prediction of how the behaviour of their workers will alter in response to changes in their working environment. The goal of this work is to improve the decision processes used in the design of the working environment. Through improving such processes, companies will gain better performance from their direct workers, and so improve business competitiveness. This paper first presents the need to model the behaviour of direct workers in manufacturing environments. To begin to address this need, a simplistic modelling framework is developed, and then this is expanded to provide a detailed modelling methodology. There then follows a description of an industrial evaluation of this methodology at Ford Motor Company. This modelling methodology has been assessed in this case study and has been found to be valid in this case. There are many challenges that this theme of research needs to address. The work described in this paper has made an important first step in this area, having gone some way to establishing a generic methodology and illustrating its potential value. Our future work will build on this foundation.