智能电表集中抄表系统入户子系统 第3页
quickly the cost of ‘diversionary works’, i.e. the cost of moving and rearranging existing external network plant and cables (for example, to accommodate new building developments or roads). It is menu-driven, very easy to use, and based on shared models and data which detail the external plant costs and installation times. A key feature of BECS is that while the application software is widely distributed and runs on users’ desktop PCs, the ‘core’ data on costs and installation times is updated, controlled and distributed from a central point, ensuring enterprise-wide consistency.
5.2 IDACS
IDACS is an investment decision and control system for the access network. It aims to implement the principles of customer segmentation and nominates priorities for investment to cater for growth in customer demand or exhausted network capacity. Specifically, it is being developed to abstract data from a large number of enterprise-wide databases, present this in coherent form to the planner, and suggest capacity growth priorities based on various demand criteria. Again, it is a VB application with network links to BT mainframe and client/server systems containing the core data.
5.3 TASC
TASC is a graphical, menu-driven, semi-automatic planning tool for the production of authorised ‘standard’ cases, i.e. the approval of financial business cases for access network planning schemes. It is initially aimed at
simplifying the planning and cost-benefit analysis of the provision of passive optical networks (PONs) and conventional copper networks. The cost and net present
value of the PON and copper schemes is calculated and apportioned over time and customers, based on input circuit demand forecasts and outline costs, and bills of material are produced for both schemes. The most labour-intensive steps in the production of the standard financial approval case (FAC) calculations and document production are completely automated, and the result is major time-savings in FAC submission preparation (e.g. 2 days work can be reduced to 15 minutes). Just as importantly, the improved productivity and standardisation of processes leads to more cost-effective targeting of PON technology on the best-suited customer areas. Development of TASC was completed in 18 man-months, and the tool has been produced in Delphi to obtain the code-reuse benefits of fully object-oriented Visual Pascal.
5.4 GenOSys
Lastly, the GenOSys genetic optimisation system is a sophisticated tool, again coded in Delphi, which employs genetic algorithms to minimise the cost of greenfield copper access network developments against a well-defined set of possible geographical locations and network planning rules. Genetic algorithms, employing the principles of selecting the fittest from many solutions ‘bred’ over numerous generations, are computationally efficient and very resistant to entrapment in the ‘false optima’ which abound in complex solution spaces. Again the system has proved its worth in massive improvements in planning productivity, mainly as a result of the easy-to-use GIS. The genetic algorithm optimisation engine effectively guarantees that the lowest-cost network solution is planned, and is capable of optimising highly complex ‘objective functions’ which could describe any aspects of network ‘total’ performance. Approximately 15 man-years of R&D has been invested in this system, but the final applications development was completed in around 4 man-years. The next section describes in more detail how GenOSys has embodied innovative methods in software engineering and artificial intelligence — object-oriented technology, genetic algorithms and rule-enforcement inference engines — and applied them to the practical problem of designing copper greenfield networks.
6. An expert system for greenfield planning — GenOSys
This section describes how innovative methods have been applied to the problem of automating the copper greenfield planning process within BT in the UK. The work has involved collaboration between BT Laboratories and Sunderland University and is based on leading-edge techniques in artificial intelligence, specifically the development of customised genetic algorithms well-suited to solving access planning problems quickly on desktop PCs. AI techniques offer a number of advantages; in particular they can address ill-defined problems (e.g. missing or incoherent data and mathematically intractable problems), solve problems to which conventional algorithms cannot be realistically applied (e.g. very large search spaces), and in some cases learn from experience. The set of tools described here has been successfully proven in test-bed trials and the business benefits projected to accrue from their field deployment is about £10 million per annum, or about 10% of the total annual investment made in copper greenfield networks.
6.1 The detailed copper greenfield planning process
In BT the primary method of providing service from a local exchange to the customer is by means of copper pairs. The continuing investment in copper technology —particularly for new housing development — makes it desirable to automate as many parts of the infrastructure design process as possible. Currently, the entire area of copper greenfield activity represents an annual investment of about £100 million per annum to the company. A simplified diagram of the access network is shown in Fig 1.
上一页 [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] ... 下一页 >>
智能电表集中抄表系统入户子系统 第3页下载如图片无法显示或论文不完整，请联系qq752018766
