Research On Optimal Placement Of Phasor Measurement Units

Wide Area Measurement System (WAMS) based on synchronized phasor measurement technology has been gaining increasingly interests due to its great value in power system dynamic monitoring, potential applications in system modeling and validation and system wide protection and control. The wide installation of PMU at every bus will lead to a simplified linear state estimation which provides faster calculation speed and higher numerical stability. But due to the cost limitation and the fact that each PMU can measure not only the bus voltage but also the currents along all the lines incident to the bus, only 1/5 to 1/3 buses in the grid need to be equipped with PMUs to make the entire system observable. Hence, how to make the entire system observable with minimal numbers of PMUs while fulfilling certain constraints becomes a hotspot in the field of WAMS.Firstly the architecture and applications of WAMS and linear state estimation based on PMUs measurement are introduced. And then the current status of optimal PMU placement methods considering the observability of the entire system or some other constraints is reviewed. Analysis and comparison shows that current optimal PMU placement methods are not fully satisfied with the practical needs. This paper presents the notion of PMU placement domain. Using the idea of domain, the PMU placement problem can be simplified and become more flexible. The placement scheme has the ability to resist high risk cascading failure in power grid by varying the domain. And then a flexible optimal PMU placement algorithm based on 0-1 linear integer problem is proposed. Considering the fact that zero injection node is not existed in high voltage power grid, an optimal PMU placement method accounting for cases of any line tripped or loss of a single PMU is provided firstly. Taking the zero injection node into consideration greatly increase the complexity and nonlinearity of PMU placement. This method overcomes this problem by converting the zero injection nodes into branches with known power flow. Finally, the conception of depth of unobservability is defined and an optimal PMU placement algorithm accounting for multi-stage installation is proposed. Using this method, installation scheduling can easily be made.All of the PMU placement methods mentioned above are modeled with 0-1 linear integer program and can account for multi-constraints, such as the restriction of communication conditions, the installed PMUs, the requirement of direct monitoring some critical infrastructures and so on. These methods are validated by IEEE system, New England system and Zhejiang power grid. The effectiveness and flexibility of the proposed algorithms are justified by the simulation results.