Engineering Asset Management (EAM) has been becoming evermore popular in the domain of physical asset management globally. In ensuring that companies are able to achieve the desired performance, managers need to adopt a systemic approach towards monitoring and assessing the asset management performance continually. Publicly Available Specification (PAS) 55, the new standard for Asset Management (AM) was developed by the British Standards Institute (BSI) for this purpose. Seeking to this standard, the International Standards Organization (ISO) released the ISO 55000-2 in 2014. As a result, industries have become more focused on converting AM practice to physical or engineering assets. Although BSI PAS 55 and ISO 55000-2 facilitate the benchmarking of AM performance, unfortunately, these standards do not detail activities, methods or procedures for assessing or evaluating individual performance. Neither do they clarify documentation in terms of formatting, explicit content and performance measurement. The standards do not provide clear directions on how one could select or modify the processes in order to ensure that the practitioners in question have complied with the standards. To overcome these deficiencies, a number of utility service providers are forced to obtain recognition via certification of physical or engineering based asset management by hiring well-known consultancy companies.

Most utility service providers are large corporations and have sufficient resources to hire well-known consultancy companies. However, this is not the case for small and medium-sized enterprises (SMEs)—they are usually unable to afford hiring renowned consultancy companies to guide them in obtaining required certification or receiving more professional suggestions from them to improve their EAM performance. Therefore, one of the purposes of our research effort is to build an intelligent system capable of automatically classifying and ranking performance levels of a given type of SME-based companies and then identify the EAM practice that is most suitable for that company via benchmarking against other similar companies in AM-based performance.

The first step in this process was to design a tailor-made questionnaire and used it while conducting a survey of the performance of AM-based information management in compliance with AM standards. The questionnaire was used to collect data on the level of adoption in AM-based information management as well as to assess Operation and Maintenance (O&M) practices with or without the adoption of AM related standards. The second step in evaluating EAM performance was to design an intelligent system that could automatically evaluate the level of achievement of a particular building owner or a building services provider in terms of AM-based information management. By using the designed Key Performance Indicators (KPIs), the system can automatically evaluate the performance of the surveyed company in accordance with AM standards. Once the level of performance of a particular company was known, we moved on to the third step which involved assigning the company to a group of companies performing AM-based information management at a similar level. A method called Support Vector Machines (SVMs) was used to automatically associate the company with a group, based on its current performance. Since the computational requirement of SVMs was high owing to the need to find a range of required parameters, an automatic process was developed. The process utilized an intelligent way of determining the optimal SVM parameters. Survey data were then collected from a range of companies and used to validate the effectiveness of this novel optimization approach through the application of a Bayesian inference method. A comparison study was then conducted by comparing SVMs employing randomly selected parameters with those employing optimized parameters. The application of this methodology shows that the results generated by using the new SVMs with Bayesian inference increased the accuracy in classification as well as minimizing the computational resource. Further, it is found that the larger the database of SMEs utilizing the intelligent surveying system, the more robust is the accuracy of AM performance self-evaluation and benchmarking.

To encourage SMEs companies to join this survey without incurring the high cost of hiring an expensive consultancy company, another intelligent method, called improved density- and distance-based clustering (IDDC), was implemented. By using this method, we could automatically assign the company to a specific group. In this context, three performance indexes were developed while a systematic design procedure rendered the clustering of different SMEs become more practical and efficient. The IDDC can simplify the measurement procedures for AM and provide a basis for benchmarking, for measuring and ranking the AM performance.

The major research contribution of this thesis is the design of an intelligent system for self-performance evaluation and then automatically classifying each surveyed company according to the degree of compliance with AM's standards. In summary, with the help of the questionnaire and intelligent system, this thesis reports the design and implementation of a simplified, low-cost and effective way for self-evaluation of SMEs for the purpose of benchmarking themselves.