AN EXPERT SYSTEM FOR SECONDARY DISTRIBUTION-SYSTEM DESIGN

This paper presents the development and use of SECOND, a rule-based expert system (ES) for radial type, electric power secondary distribution system (single-phase/three-phase, overhead/underground) design accommodating a projected load growth over a specified time horizon. SECOND is implemented in an integrated environment involving Prolog, conventional languages (FORTRAN, C, etc.) and a graphic user interface toolkit (GUI). Prolog is chosen for the knowledge base and the inference processes because of built-in support for predicate calculus and first order logic, built-in search mechanism and built-in backtracking. SECOND uses conventional languages (FORTRAN/C) for algorithm-driven numerical claculations, since Prolog is not efficient for this purpose. It also incorporates a window-based user interface (implemented in C) using the GUI toolkit: SunView. The window handling functions are performed by a set of predicates included in a Prolog library developed for this purpose. The data and information on secondary distribution system components is encoded in state frames (SF), which organize numerical and non-numerical data into defined slots. This information is then used by the rule-based procedures in a systematic manner. The interaction between the user and the ES is established through a window user interface. At the beginning of the consultation process, the user inputs names of database files and other design data in response to queries from the ES. The user can choose between static design (no load growth) and dynamic design (projected load growth over a specific time period). For the dynamic design, the ES generates a set of feasible initial designs constrained by possible patterns (poles/transformers per block), thermal limit, voltage spread, available conductor and transformer sizes. For each initial design at year 1, the ES generates all feasible growth paths satisfying specified constraints, by moving along the time axis in yearly increments. If at a particular year, any of the technical constraints are violated, the ES revises the system reinforcement parameters and moves on to subsequent years until the study period is exhausted. The ES then evaluates the cost of the growth paths for each initial design and recommends one with the least cost. Sample designs recommended by the ES for an example system are presented. The advantages of using the ES approach to power distribution system design are discussed.