Dynamic Expansion Model Research Articles

Lifelong Compression Mixture Model via Knowledge Relationship Graph

Task-Free Continual Learning (TFCL) represents a challenging scenario for lifelong learning because the model, under this paradigm, does not access any task information. The Dynamic Expansion Model (DEM) has shown promising results in this scenario due to its scalability and generalisation power. However, DEM focuses only on addressing forgetting and ignores minimizing the model size, which limits its deployment in practical systems. In this work, we aim to simultaneously address network forgetting and model size optimization by developing the Lifelong Compression Mixture Model (LGMM) equipped with the Maximum Mean Discrepancy (MMD) based expansion criterion for model expansion. A diversity-aware sample selection approach is proposed to selectively store a variety of samples to promote information diversity among the components of the LGMM, which allows more knowledge to be captured with an appropriate model size. In order to avoid having multiple components with similar knowledge in the LGMM, we propose a data-free component discarding mechanism that evaluates a knowledge relation graph matrix describing the relevance between each pair of components. A greedy selection procedure is proposed to identify and remove the redundant components from the LGMM. The proposed discarding mechanism can be performed during or after the training. Experiments on different datasets show that LGMM achieves the best performance for TFCL.

Transmission expansion under uncertainty in the Chilean System via Minmax Regret with GA

Transmission planning has been broadly studied through the development of minimum cost models, adapting to the forecasts of generator and consumer requirements. This research considers that a single planner determines the transmission system expansion, while the generation expansion is decided by different independent market agents, and generation supply becomes an uncertainty, hampering the planner's decision making. Under this context, the planner's and the generation market agents' individual decisions are in a constant struggle because the planner does not decide installing a line unless there is certainty that a power station will be installed, and vice versa. The methodology proposed develops a dynamic expansion model that minimizes the maximum regret, allowing the definition of a robust transmission system expansion. In addition, a heuristic tool is used through genetic algorithms to provide the planner with a ranking of the best configurations. The tool proposed will be useful for the planner, allowing to decide considering the uncertainty of the decisions of the generation market agents.

Closure to discussion of "voltage stability analysis: V-Q power flow simulation versus dynamic simulation"

of each allele represents the number of blocks of fixed capacity that are being added to the existing circuit. The representation of other situations, such as replacing the existing circuit for another one with different technology, or different voltage, would require the use of more than one allele for the respective circuit (let call them as “sub-alleles”) as well as the implementation of a rule that tells the GA to turn off the effect of one sub-allele to let the other take place. For instance, if one wants to analyze the expansion of a given circuit at two different voltage levels, say 230 kV and 500 kV, two sub-alleles should be set up: one representing the number of blocks of expansion at 230 kV and the other at 500 kV. The aforementioned rule should ask the GA to fix to zero either one of the sub-alleles to let the other influence in the chromosome’s fitness. Even though we did not implement such an option, we believe that it can be successful. Concerning the second issue, even though a GA for finding optimum dynamic expansion plans could be developed and implemented, it should be mentioned that considering the uncertainties, the amount of information that is no longer available for the public and the decentralization of decisions in a competitive market of electricity supply, the effort of implementing a GA-based, dynamic expansion model, would not be worthwhile. In our opinion, the dynamic of the expansion (dating of new installations) is defined by complementary studies such as stability and reliability assessment and short-circuit analysis. However, among the set methodologies, methods and programs for the expansion planning, a static integer transmission expansion model is necessary to pick up major interconnections and facilities that will be present in most of the scenarios. A short-run, market-driven expansion model will never foresee those large-scale projects. Regarding the third issue, in a market based on physical bilateral contracts with short-term transactions, the long-term view about the problem should be replaced by a short-term view, whose emphasis is to manage congestion through prices. Nevertheless, if the interest is focused on the long-term analysis of the transmission expansion under such a market of contracts, a GA-based algorithm could be used, if multiple scenarios of load, dispatch and bilateral transactions are considered.