A fuzzy SMART based dynamic decision making system: a voltage control case study

Abstract

Most real-life decisions involve Dynamic Decision Making (DDM) that is characterised by the need to make multiple and interdependent decisions in an environment that changes as a function of the decision maker’s actions, environmental events, or both. Some examples can be found in management of transportation networks and in controlling of power systems. To assist humans in these difficult decision making scenarios, computer-based decision making systems have been developed. Most of them rely on dynamic programming algorithms. Unfortunately, heavy computational burden makes them not suitable for application to large systems and precludes finding a solution in a limited time which is an important determinant of the performance. This paper describes the development of a fuzzy Simple Multi-Attribute Rating Technique (SMART) based dynamic decision making system which incorporates the merits of human decision making mechanisms and operational research methods to find an optimal solution taking the least amount of time in a dynamic environment. To illustrate the proposed framework, we apply it to a practical voltage control problem in abnormal scenarios in power systems. The proposed system (see Figure 1) includes three components: a ‘voltage monitor’ to monitor abnormal voltage profiles based on a power flow algorithm, an ‘evaluator’ to evaluate the effectiveness of candidate control actions based on a SMART algorithm, and a ‘decision maker’ to search an optimal voltage control schedule based on a fuzzy linear programming algorithm. We present the test results on a benchmark 9-bus test power system under a dynamic scenario caused by load demand variations. The results show that the proposed approach can quickly find optimal decisions for maintaining an acceptable voltage profile in a dynamically changing power transmission environment. Furthermore, it can reduce the number of unnecessary control actions in comparison to a traditional sensitivity based method. The proposed framework can be easily applied to other similar dynamic decision making problems, such as ordering system in industry.