Intelligent Control and Protection Methods for Modern Power Systems Based on WAMS

Continuously growing demand for electricity, driven by deregulated power markets, has forced power systems to operate closer to their security operation limits. Meanwhile, the increasing penetration of large scale renewable energy may impact the operation of power systems by bringing more uncertainties. Under these circumstances, Wide Area Measurement System (WAMS) is widely applied in modern power systems, which is composed of a number of phasor measurement units (PMUs) that can provide high resolution real–time measurements synchronized by global positioning systems (GPS). WAMS can be used to (1) effectively assess the vulnerability of power grids; (2) provide online dynamic security assessment (DSA) with high reliability; (3) intelligently control and prevent power systems from the risk of insecurity and instability; (4) accurately identify the parameters of power system models and provide adaptive corrective control schemes.

Power system is a complex non-linear dynamic system. The key to power system vulnerability assessment is to find the vulnerable elements which could cause largearea power outages under attacks. For the aspect of structural vulnerability assessment, several vulnerability indices i.e. structural vulnerability index (SVI), contingency vulnerability index (CVI) and operational vulnerability index (OVI) are proposed to evaluate the impact of distributed generation (DG) on power system vulnerability. The assessment shows that DG units are able to shorten the electrical distance between power sources and power loads, alleviate long-distance large-capacity transmission, improve the reliability of power system after contingencies, and increase transmission efficiency. For the aspect of dynamic vulnerability assessment, critical clearing time (CCT) is computed by screening of a number of contingencies in various operating conditions. By statistical analysis, vulnerable areas in terms of transient stability are identified. Furthermore, the result of CCT computation in different typical scenarios can evaluate the impact of wind power on power system transient stability. Other influencing factors to power system transient stability are also evaluated, e.g. power output of generators in central power plants (CPP), load consumption level and the power exchange in high voltage direct current (HVDC) links. Both structural and dynamic vulnerability assessment, aiming at providing an early awareness of power system insecurity, are conducted by simulations in the DIgSILENT model of western Danish power system.

DSA is the assessment of the ability of a certain power system to withstand a defined set of contingencies and to survive in the transition to an acceptable steadystate condition. Among pattern recognition techniques, decision trees (DT) using the algorithm of classification and regression trees (CART) is applied in DSA of western Danish power system. It not only provides the results of security assessment but also reveal the principles learned by DTs for security assessment. The systematic approach adopts new methodology that trains contingency-oriented DTs on daily basis using the database generated by importance sampling method. The number of time-domain simulations necessary for importance sampling is significantly reduced, so computation burden is highly reduced, which makes the online DSA possible for even considering large scale integration of uncertain power generation from windfarms and other DG units.

Based on the result of DSA, intelligent contingency control scheme guided by paralleled DTs is able to draw the system from insecure zone to the secure zone. In this approach, two DTs work in tandem, i.e. Observation DT (ODT) and Prevention DT (PDT). Fed with real-time wide area measurements, ODT of measurable variables is employed for online DSA to identify potential security issues and PDT of controllable variables provides online decision support on preventive control strategies against those issues. Finally, optimization of preventive control is conducted to find the most economical trajectory of preventive control. The proposed approach is comprehensively verified by a number of credible contingencies in western Danish power systems with annual data of operating conditions.

In order to prevent voltage collapse, the online identification of load characteristic using PMU measurement-based approach is proposed to evaluate the proximity to voltage instability. Based on different load characteristics, adaptive corrective control schemes were implemented. The method can correctly predict and prevent the voltage collapse, and minimize the amount of load shedding.