استفاده از شبکه های بیزی برای تجزیه و تحلیل خطر عملیات سوئیچ عایق هواییMV
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|29071||2010||9 صفحه PDF||سفارش دهید||5653 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Reliability Engineering & System Safety, Volume 95, Issue 12, December 2010, Pages 1358–1366
Electricity distribution companies regard risk-based approaches as a good philosophy to address their asset management challenges, and there is an increasing trend on developing methods to support decisions where different aspects of risks are taken into consideration. This paper describes a methodology for application of Bayesian networks for risk analysis in electricity distribution system maintenance management. The methodology is used on a case analysing safety risk related to operation of MV air insulated switches. The paper summarises some challenges and benefits of using Bayesian networks as a part of distribution system maintenance management.
During the last 10–15 years, electricity distribution companies throughout the world have been ever more focused on asset management as the guiding principle for their activities (see e.g. ,  and ). The concept of asset management in general has developed during this same period of time, using input from a number of industrial sectors—such as water supply, transportation and energy supply. All of these sectors share a reliance on an infrastructure of physical assets that provides the foundation for their businesses  and . Asset management in electricity distribution companies is about the complex balancing of cost, performance and risk—taking into account different aspects such as economic performance, quality of supply, safety and environmental impact  and . Together with handling of cost and performance, the management of risk is therefore a key issue for electricity distribution companies, and there is now an increasing awareness towards taking risk analysis into account in the decision making context . For some of the risks there are methods and tools already used within the electricity distribution sector—such as economical risk analyses and quality of supply risk analyses (reliability analyses). For others – and more intangible risks, for example safety issues – there is less culture and practice for performing structured risk analyses to support decisions. Maintenance activities are important parts of the asset management scheme, to control the distribution companies’ risk . Historically, maintenance activities have been decided based largely on existing practice, producers’ recommendations and to some extent direct regulation from authorities, with little application of formal analyses to support or reject the existing paradigms . However, there is now an increasing trend among electricity distribution companies on developing maintenance strategies where different aspects of risk are sought included in a holistic way  and . Electricity distribution companies recognise that there are significant potential for improvement in their analytical approaches within maintenance assessment, and there is a need to test methods to support risk analysis in this context. Such methods can contribute to more optimised spendings on maintenance activities. Bayesian networks is a risk modelling and analysis approach that has been applied for various types of analyses for different purposes in different industrial sectors (see e.g. ,  and ). Due to its versatility and ability to represent complex relations, it is also of interest to use this approach in electricity distribution company decision support. This paper presents a methodology for using Bayesian networks for risk analysis in electricity distribution company maintenance management. The methodology emphasises on analysing intangible risks, like safety. Such risks are especially important in medium voltage (MV) distribution systems, where the impact of failures on reliability is not as widespread as on higher voltage levels. Hence intangible risks constitute more prominent decision criteria in MV systems . The methodology is illustrated through a case analysing safety risk related to the operation of MV air insulated switches. Section 2 presents background concerning MV electricity distribution systems, risk analysis and Bayesian networks. Section 3 introduces the proposed methodology, while in Section 4 the methodology is used on the case. Section 5 discusses some of the main results, while Section 6 summarises the paper with some concluding remarks.
نتیجه گیری انگلیسی
This paper describes a methodology where Bayesian networks are used for risk analysis where the results contribute to establishing a risk informed basis for making maintenance decisions. Bayesian networks are an appealing method for quantitative risk analysis in electricity distribution system maintenance management due to their versatility for different risk problems. The electricity distribution companies today are motivated by the customers and the regulators to cut costs so that tariffs could be lowered. Reducing maintenance and reinvestments are cost-cutting options that have to be weighed against increased risk that cost reductions might impose on safety, quality of supply, etc. Electricity distribution companies might therefore do well to incorporate more analytical approaches to prescribe maintenance strategies (for example through Bayesian networks as shown in this paper) to increase the understanding of where to focus the companies’ efforts, and to optimise the spending of maintenance resources. The purpose of the risk modelling and analysis has not been to create an “objective and true” model of the problem at hand, but rather to increase understanding of the risk problem, and to rate alternative strategies according to a relative scale. This will contribute to a structured framework for risk communication and decision making. The qualitative and quantitative input used in such models will generally be based on expert judgments, because no relevant statistics are available. Generally, to provide input data, one should look into what sources are available—both from statistical analyses and from expert judgements or, preferably, the combination of both. This is one aspect which obviously should be emphasised in future asset management practices when working with such quantitative models. The application of QRA in general and Bayesian networks in particular requires a new way of thinking and new competences among electricity distribution company engineers—which will take some time to establish and gain confidence in.