Study On Dynamical State Estimation Problems For Some Class Of Power Systems With Unconventional Measurements

In the past few decades,power systems have been undergoing significant evolution in response to the rising complexity of loads,the increasing penetration of stochastic/intermittent renewable generation units as well as the growing limitations on greenhouse gas emissions.On the other hand,the unconventional measurements subject to time-delay,packet loss,outlier and cyber-attack bring enormous challenges on the safe and stable operation of the power systems.It should be noted that the conventional power system state estimation methods may hardly meet the demanding requirements on the situation awareness of the power grids under new situations.As such,for the purpose of handling new problems,new phenomena and new challenges of the modern power systems,enhancing the real-time,accuracy and reliability of the power grid situation awareness,and promoting the stable and safe operations of the power systems,there is an urgent need for us to develop a series of novel techniques and methods for the dynamic state estimation of power systems under unconventional measurements.In this thesis,we mainly focus on the study of dynamic state estimation problems for power systems such as the power quality system,renewable energy microgrid,quasi-steady power system and the multi-machine power system.The content of this thesis can be classified into four parts.In the first part,the power quality disturbance tracking problem with multiplicative noises and correlated noises is investigated.In the second part,we mainly focus on the dynamic state estimation problems for renewable energy microgrid with unconventional measurements such as fading measurements and sensor saturations,and some novel recursive state estimator design methods are developed.The third part studies the dynamic state estimation problem for a class of quasi-steady power systems,where the sampling rates of the sensors and the update rate of the power system are allowed to be different.Meanwhile,in this part,a novel transformation method is developed to convert the multi-rate system into a single one,and the fusion estimation strategy is then designed.In the fourth part,we mainly focus on the dynamic state estimation problems for multi-machine power systems,where the unconventional measurements such as event-based measurements,measurement outliers and randomly occurring hybrid cyber-attacks on the measurements are considered.In order to better tackle the large-scale nature of the multi-machine power systems and the non-Gaussian noises,the decentralized dynamic state estimation schemes are proposed by resorting to the model-decoupling approach and particle filtering approach.In addition,a novel online outlier detection and localization algorithm is proposed to process the measurement outliers,and a likelihood function modification method is developed to mitigate the influences of the randomly occurring hybrid cyber-attacks on the state estimation performance.Specifically,the main contents of this thesis are outlined as follows.·The estimation problem for the power quality disturbances in the presence of multiplicative noises and correlated noises is studied.Based on the engineering practice,the explicit model of power quality disturbances with multiplicative noises and correlated noises are proposed.With respect to the proposed estimator,the state covariance,the one-step prediction error covariance and the estimation error covariance are first derived.Subsequently,the estimator gain is obtained by minimizing the estimation error covariance at each time instant.Finally,simulation experiments under three scenarios are provided.·The dynamic state estimation problem for the islanded microgrids subject to fading measurements is addressed.An explicit model of the islanded microgrid with fading measurements is proposed.The fading phenomenon of measurements occurs in a random way and the attenuation coefficients are characterized by a set of random variables with known probability distributions.Attention is focused on the design of a recursive state estimator in the presence of fading measurements.In terms of the solutions to two sets of matrix difference equations,an upper bound on the estimation error covariance is firstly obtained,and such an upper bound is then minimized through adequately designing the estimator gain.Finally,simulation is conducted on the islanded microgrid.·The distributed state estimation problem is studied for renewable energy microgrids with sensor saturations.A system model for the microgrids with sensor saturations is proposed.Attention is focused on the design of a distributed recursive estimation scheme such that,in the presence of the sensor saturations,an upper bound of the estimation error covariance is guaranteed.Subsequently,such an upper bound is minimized by appropriately designing the gain matrices of the corresponding state estimator.In particular,the sparsity of the gain matrices resulting from network topology is handled by using a matrix simplification method.Moreover,the performance evaluation of the designed distributed state estimator is conducted by analyzing the exponential boundedness of the estimation error in the mean square sense.Finally,simulation experiments under two cases are carried out on a renewable energy microgrid which contains two distributed generation units.·The fusion estimation problem is investigated for a class of multi-rate power systems with randomly occurring delays in supervisory control and data acquisition(SCADA)measurements.The power system is measured by the SCADA and the phasor measurement unit(PMU),and the state updating period of the power system is allowed to be different from the sampling periods of the SCADA and the PMU.The phenomenon of the randomly occurring SCADA measurement delays is characterized by a set of Bernoulli distributed random variables.To facilitate the state estimator design,a new approach is developed to transform the multi-rate power system into single-rate one.First,two local state estimators are designed,respectively,based on the SCADA and the PMU measurements such that upper bounds of the local estimation error covariances are guaranteed at each sampling instant,and such upper bounds are subsequently minimized by appropriately designing the gains of both local state estimators.Then,the asynchronous estimates from the local state estimators are fused by recurring to the covariance intersection fusion scheme.Finally,a simulation experiment is carried out on the IEEE 14-bus system.·The joint state and unknown input estimation problem for multi-machine power systems under the event-triggered mechanism is concerned.The field voltage and the mechanical torque of the synchronous generator,which are generally difficult to be measured in engineering practice,are treated as unknown inputs to be estimated in this investigation through developing easy-to-implement algorithms.For phasor measurement units(PMUs)that are deployed to monitor the dynamic behavior of the synchronous generator,the massive PMU-based signal transmissions are orchestrated by using an event-based transmission scheme with aim to light the communication burden.Attention is focused on the design of an event-based recursive joint state and unknown input estimator,where the state estimation error is unrelated with the unknown inputs.Upper bounds of the estimation error covariances of both the unknown input and the state are first guaranteed at each sampling instant,and such upper bounds are subsequently minimized by appropriately designing the corresponding estimator gains.Finally,simulation experiments under three cases are carried out on the IEEE 39-bus system.·The decentralized dynamic state estimation(DSE)problem is investigated for a class of multi-machine power systems with non-Gaussian noises and measurement outliers.The model decoupling approach is adopted to facilitate the decentralized DSE for large-scale power systems.The particle filtering technique plays a key role in the developed DSE scheme with aim to tackle the strong nonlinearities and the non-Gaussian noises.To mitigate the negative impact from the measurement outliers on the DSE performance,a novel sliding-window-based online algorithm is proposed to detect and further locate the possible outliers based on the historical measurement data.Finally,simulation experiments are carried out on the IEEE-39 bus system.·The decentralized secure DSE problem for a class of multi-machine power systems with non-Gaussian noises and randomly occurring cyber-attacks is addressed.A model decoupling approach is developed for synchronous generators to facilitate the decentralized DSE for large-scale interconnected power systems.Hybrid cyber-attacks,which include denial-of-service attacks,bias injection attacks and replay attacks,are considered to be occurring in a random way.Attention is focused on the design of a secure DSE algorithm to mitigate the negative influences from the nonlinearity/non-Gaussianity and the random occurrences of the cyber-attacks on the DSE performance.Specifically,ⅰ)the particle filtering technique is adopted in the developed DSE scheme to tackle the nonlinearity/non-Gaussianity of the power systems;and ⅱ)a modified likelihood function is established to compensate for the impacts of the measurements corrupted by the randomly occurring cyber-attacks.Finally,simulation experiments with five scenarios are carried out on the IEEE 39-bus system.