Kalman Filter based Novel Centralized Dynamic State Estimation in Multi-Machine Power System Incorporating DFIGs

Abstract

The omnipresent, conventional synchronous generators are playing a pivotal role since many years to meet ever increasing demand of electrical energy. A steep growth in energy requirements is now met by power generation mix of fossil fuel based and renewable energy based generators. The stochastic nature of renewable generations, expanding power networks, complex interactions among the system components and loads etc. impose need for superior monitoring and control at power system level for its stable operation. For efficient operation and control of the power system, it is essential for the energy management system (EMS) operator to have an accurate information about every generator's dynamic states and power system behaviour. Diminishing fossil fuels and environment concerns advocated nations to gradually adopt renewable energy sources. Wind energy, better on multiple aspects among other renewable options, is dominating today in the power networks. Among wind energy generators, doubly fed induction generators (DFIGs) are widely accepted due to its operational exibility, small converter size, better power control and low cost. The parallel operation of DFIGs (or wind farm) with pervasive synchronous generators brings in enhanced system dynamics. This condition strongly dictates dynamic state estimation (DSE), not only to infer information about synchronous generators but concurrently know the states of wind generator(s). The performance of DFIGs also dependent on the converter control circuitry and its feedback loop. This thesis is the record of work that appropriately models the generators and the network, suitable for the adoption by the Kalman _lter based algorithms to perform the DSE. With the help of the availability of centralized measurement data, the states of all the generators in the multi-machine system can be established simultaneously. Subsequently, the dynamic states of the DFIG are used for its rotor power control under specific conditions. The synchronous generators, usually considered as a voltage source in the literature are presented by relevant state-space model for stability analysis. On the other hand, widely accepted DFIG based wind generator is presented as current source state-space model. As the models of both the generators are far apart, it is necessitated to bring both on the same platform. The thesis contains the work that shows the possibility of model unification of both kind of generators. Employing traditional DFIG current-source state model, a current-source state model of synchronous generators is proposed and validated using standard software platform. Highlighting feature of the proposed mathematical model is its applicability to power system with no limits on number of synchronous generators and DFIGs. Considering modelling intricacies, the use of these models is recommended to achieve concurrent DSE in a multi-machine power system. Employing current source models of synchronous generators and with substantial penetration of DFIG in multi-machine system, approach for concurrent DSE of synchronous generator and DFIG is presented. The mathematical model is simulated in MATLAB / Simulink platform for the validation. The power system dynamic conditions realized in the MATLAB / Simulink model are then treated as the data available from the phasor measurement units (PMUs) (with and without noise). This is used for the extended Kalman filter (EKF) and unscented Kalman filter (UKF) based DSE algorithms. Centralized dynamic state estimator based on EKF and UKF are employed for the faithful state predictions for all the generators under power system dynamic conditions and results are presented. Application of dynamic states in real time is equally important to achieve better control and operation of DFIG. This apparatus, normally operate in hostile condition whether on-shore or off-shore, can undergo internal sensor erratic operation. Under such conditions, use of dynamic states obtained using EKF, is proposed to have errorfree, continuous and smooth operation of DFIG. The results are embodied in the thesis. As an offshoot of main work, comparative performance of EKF and UKF with different PMU measurement data update rates under discontinuous measurement is analysed. Additionally, use of weighted least square estimation (WLSE) algorithm as an alternate to load ow under bad measurement condition is deliberated with results.