Mixed Measurements Based Distributed State Estimation For Power System

The accurate information of the operating state of the power system is quite important for the power system dispatching, security analysis and control. However, since the power system is normally very large and its structure is also very complex, it is difficult to obtain the status of the power system precisely. Generally, based on the measurement of the SCADA, the state of the power system is estimated using the optimal state estimation algorithm. With the development of the modern power system, the construction and management of the power system become more and more complex, and it is subsequently difficult to run the power system safely. In this condition, the accuracy and the efficiency of the state estimation (SE) have to be improved, so as to supply better information of the power system for the on-line stability analysis and operation control.Compared with the static state estimation, the dynamic state estimation (DSE) can predict the status of the power system in the next time step, which supports the implementation of the on-line system security evaluation, status prediction and preventive control. Hence, it is more suitable for engineering application, and worth to be investigated further. With the development of the wide area measurement system (WAMS), the phasor of the state variables can be measured with higher accuracy and quicker updating speed. If the measurement of the WAMS can be employed in the DSE, the accuracy of the state estimation can be significant improved. Presently, for the reasons of the economics and techniques, the WAMS system has not covered all the power grid and the status of the power system can not be fully measured by the WAMS system. Hence, how to apply the measurement of the WAMS to the DSE effectively needs to be studied further. Additionally, since the computation burden of the DSE is quite heavy, the efficiency of the computation of the DSE needs to be improved so as to satisfy the on-line system analysis. The distributed computation technique is a branch of the parallel process technique, and developed very quickly in the past few decades. Since the structure of the distributed computation is similar to that of the control system of the power grid, it is quite suitable for the power system analysis. The researches carried out in this thesis are as follows.Firstly, the history of the power system state estimation is reviewed, and the SE algorithms are also presented. The characteristics of the modern power grid and the challenge of the security of the power system are analyzed. Based on the results of the analysis, it is pointed out that the DSE and distributed state estimation are very important for operation and control of the modern power system.Researches are carried out to construct a WAMS system, and the phasor measurement unit (PMU) and communication are developed. For the PMU, the errors in the phasor angle caused by the frequency deviation and parameters of the generator are compensated, and the position signal of the rotor of the synchronous generator is employed in the measurement of the rotor angle. Hence, the accuracy of the phasor angle is improved. For the communication system, the improved model of the SPDNet is built to evaluate the time delayed, and a digital communication channel with 2M bandwidth mixed with SPDNet is proposed for communication in the WAMS system.The algorithms of the distribution of PMUs are studied, and the distribution of the PMUs in the practical power grid is also analyzed. Four steps of the distribution of the PMUs are concluded at the first time, and the distribution of the PMUs in the practical power grid is being carried out from step 2 to step 3. Based on the analysis result, the definition of the linear observation and the model of PMU are proposed. Using the model of PMU, the investment function of the distribution of PMU is built. The fitness function is constructed, which try to obtain maximum ratio between the benefit and the investment. Simulations are carried to demonstrate the application of the fitness function in the distribution of PMUs. The optimization of the distribution of the PMUs is valuable for the engineering implementation.The mixed measurements based SE and DSE are reviewed. A mixed measurement based DSE is proposed in this thesis. The mixed measurement system is built firstly, and a algorithm of the DSE is also proposed. The computations of the SE are distributed geographically, so that the transformations of the measurement are carried independently and simultaneously in each part of the power grid. Hence, the efficiency of the DSE can be improved significantly. Since the time delay of the measurement has significant effect on the SE, the random characteristic of the time delay is considered in the DSE. Using the statistic characteristics of the time delay, the statistic matrixαof the time delay of the measurement for each area is introduced to release the problem of different time delay in different area, and the capability of the SE can be improved.Finally, the multi-agent is introduced in the research in this thesis. The definition of the multi-agent is presented, and the structure of the multi-agent based distributed state estimation is proposed. The model of each agent is built, and their functions are also designed. The operation model of the multi-agent system is optimized, and the efficiency of the system can be improved. The core functions and the algorithm of the agent are packed, and the computation burden of the distributed state estimation is distributed by coordinating the agents. The distributed state estimation can be implemented by appending software package, while the structure of the distributed state estimation system keeps constant.