ANN BASED METHOD FOR OPTIMAL CONTROL OF POWER DISTRIBUTION SYSTEMS

Abstract

Distribution Automation System (DAS) provides means for maximizing utilization of distribution system so that the available capacity is optimally utilized causing minimum inconvenience to the customers. It includes load factor improvement, minimization of losses and maximizing system reliability. To incorporate the above DAS includes automation of various functions such as load management by switching and monitoring individual loads at each of the customer's end and load transfer, i.e. switching load from one feeder to another. Load management is the ability of power system to monitor and control a variety of loads in a network without creating any inconvenience to the customers in the residential, commercial and industrial sectors. This is done by shifting the time and amount of use of electrical power and energy of the distribution system. For maximizing the energy delivery capability of power systems, the individual customer loads are managed. During severe emergencies which result in insufficient resources to meet load, distributed intelligence in load control, i.e. an automatic load shedding program throughout the affected area can prevent load area collapse. By rescheduling the dispersed generators and controlling the storage devices (commonly known as dispersed generators and storage devices DSGDs), the need for new generating capacity can definitely be delayed for few years. In a country like India where generation capacity is already low if the load management is implemented as one of the several distribution automation function, the cost/b~nefit position will be better than for a simple management system only. Direct load control is considered as a possible tool to help such load management. The principle objectives of load management are to reduce the average cost of electricity, generally improve load factor and to reduce the need for generation capacity by shifting electricity use from peak periods. In present dissertation work an ANN based procedure has been developed to obtain optimal operating strategy for distribution systems having various load control options like dispersed generators, dispersed storage devices and load curtailment option at selected buses. Two sample power distribution systems have been considered, one is a 4 bus system while another is 13 bus system. In both cases, ANN architecture has been designed so that the load patterns and maximum allowable load curtailment at selected buses can be varied according to the requirement. The software package OPLCP[17] has been used to generate training and test patterns for the network. At the end results and computational time taken in execution of two methods are compared. A large saving in time has been observed.