Measurement Placement Optimation Of Power System State Estimation

In the case of known the network topology and metering scheme, the system is fullyobservable that is the necessary condition of the state estimate. Precisely estimated resultsand good numerical stability are important condition of the power system state estimationto give full play. Good measurement configuration system is the foundation of the above.This paper focuses on the measurement placement optimization. The main work is asfollows:Firstly, to solve the problem that the observability test only reflects the possibility ofoutput state variables but can not reflect the convergence, accuracy or numerical stabilityof the state estimation. We quantify the system observability and put forward the conceptof the degree of observability in this paper. At the same time, two indicators are quantified:precision indicator and numerical stability indicator. In addition, we found that thenumerical stability indicator can reflect the computational speed of the state estimation toa certain extent. Based on the above findings, the metering scheme can be optimized fromthe accuracy, speed and numerical stability.Secondly, power system often encounters in the complete absence of one or severalnodes, which seriously affect the reliability of the observability, and even cause the stateestimation software can not play a role. For this problem, we combined the N-1reliabilitytest principle with electric betweenness to define the reliability indicators and the degreeof reliability. According to these two indications, we quantify the reliability of theobservability and optimize the metering scheme. That can guarantee full observability of apower system when a station data is completely missing.Finally, we delve into the metering scheme of impact of bad data processingcapability. A unified approach to the optimization of measurements placements employednumerical algorithms to observability checking and bad data processing capabilityidentification. First, we define the node injection radix measurements and measurementcategories. And according to the above definitions the coefficient matrix can be solved.Then the analysis the column vectors of the coefficient matrix that can determine each measurement classification. Furthermore, according to each class of measuring containsmeasurements number, we can check the observability and identify the criticalmeasurements and critical couple that can determine the bad-data processing capability.Then we can improve the metering scheme which is unobservable or does not have thebad-data processing capability...