Q2. What have the authors stated for future works in "Probabilistic design of power-system special stability controls1" ?
Admittedly, this goal is very ambitious and needs further research to be fully reached. As far as future work is concerned, the first step will consist of taking full advantage of the database.
Q3. What is the first step of this work?
The first step of this work consists of extending the decisiontree induction method, in order to build automatically, instability detection rules that offer a good compromize between selectivity and degree of anticipation (Geurts and Wehenkel, 1998b).
Q4. What is the second direction of the research?
The second research direction is aimed at developing machine-learning methods that are able to exploit the temporal nature of the databases more efficiently.
Q5. What happens when the defense plan trips lines to isolate it from the system?
Loss of synchronism extends to the whole study region, and the defense plan trips lines to isolate it from the system, and activates load shedding in France.
Q6. Why is the need for more regular studies increasing?
due to the faster changes occurring in today’s power systems, the need for more regular studies is increasing, while human expertise may quickly become obsolete and even misleading.
Q7. Why is it becoming more difficult to expand transmission systems?
since it is becoming increasingly difficult to expand transmission systems within reasonable delays (e.g., due to ecological constraints), and since the operation with high security margins conflicts with economic efficiency, the present trend in power systems is to rely more and more on special emergency control schemes.
Q8. What is the main purpose of the design of special emergency control schemes?
The design of such special emergency control schemes implies the identification of the main failure modes of the system, the determination of appropriate mitigating control actions, and the design of automatic triggering devices.
Q9. What is the criterion for preventing voltage collapse in region L?
Predicting voltage collapse in region LSince region L is weak in terms of voltage stability, and tap-changer blocking may act too late to avoid voltage collapse, it would be interesting to find an anticipatory criterion to trigger emergency control.
Q10. What is the main shortcoming of the deterministic approach?
above all, the principal shortcoming of this deterministic approach is its inability to take into account the stochastic nature of the causes of power system failures, and the unavoidable modeling uncertainties.
Q11. What is the main approach to emergency control?
The approach consists of three steps:(i) modeling probabilistically the causes potentially leading to extreme conditions (weakened operating conditions, abnormal operation of protections, multiple disturbances) as well as uncertainties (load behavior, external systems2),(ii) using parallel Monte-Carlo simulations to sample scenarios according to this information (random combinations of operating conditions, dynamic modeling hypotheses (e.g., relay settings, malfunctions, external systems, load2), and disturbances),(iii) building up a database of simulation results, collecting key variables and their temporal behavior, and using data-mining techniques to extract from this synthetic information about the main breakdown modes and possible ways to improve emergency control schemes.
Q12. What is the basic approach to determining the causes of insecurity?
The approach proceeds by iterating through the following elementary steps:Study specification: setting up a detailed probabilistic model of the possible causes of insecurity: multiple disturbances, bad coordination and/or mal-operation of protective devices, over-optimistic preventive security strategies due to uncertainties in modeling parameters.
Q13. How many time steps can be used to access the data?
On a workstation with 256MB of main memory it is thus possible to access simultaneously up to 50 million attribute values: about 1500 scenarios, described by 200 temporal attributes, with an average number of 150 time steps per scenario.
Q14. What is the main purpose of the data mining process?
The data-mining process is itself composed of successive steps, aimed at extracting more and more refined information from the simulation database.
Q15. What are the attributes used to determine the severity of the scenarios?
Some of these attributes are to be used in order to define the severity of the scenarios, i.e. to measure the consequences in terms of loss of load and generation.