Ka Wing Chan

Equilibrium-Inspired Multiple Group Search Optimization With Synergistic Learning for Multiobjective Electric Power Dispatch

This paper proposes a novel multiple group search optimizer (MGSO) to solve the highly constrained multiobjective power dispatch (MOPD) problem with conflicting and competing objectives. The algorithm employs a stochastic learning automata based synergistic learning to allow information interaction and credit assignment among multi-groups for cooperative search. An alternative constraint handling, which separates constraints and objectives with different searching strategies, has been adopted to produce a more uniformly-distributed Pareto-optimal front (PF). Moreover, two enhancements, namely space reduction and chaotic sequence dispersion, have also been incorporated to facilitate local exploitation and global exploration of Pareto-optimal solutions in the convergence process. Lastly, Nash equilibrium point is first introduced to identify the best compromise solution from the PF. The performance of MGSO has been fully evaluated and benchmarked on the IEEE 30-bus 6-generator system and 118-bus 54-generator system. Comparisons with previous Pareto heuristic techniques demonstrated the superiority of the proposed MGSO and confirm its capability to cope with practical multiobjective optimization problems with multiple high-dimensional objective functions.

R(λ) imitation learning for automatic generation control of interconnected power grids

The goal of average reward reinforcement learning is to maximize the long-term average rewards of a generic system. This coincides with the design objective of the control performance standards (CPS) which were established to improve the long-term performance of an automatic generation controller (AGC) used for real-time control of interconnected power systems. In this paper, a novel R(@l) imitation learning (R(@l)IL) method based on the average reward optimality criterion is presented to develop an optimal AGC under the CPS. This R(@l)IL-based AGC can operate online in real-time with high CPS compliances and fast convergence rate in the imitation pre-learning process. Its capability to learn the control behaviors of the existing AGC by observing system variations enable it to overcome the serious defect in the applicability of conventional RL controllers, in which an accurate power system model is required for the offline pre-learning process, and significantly enhance the learning efficiency and control performance for power generation control in various power system operation scenarios.

Multi-Agent Correlated Equilibrium Q(λ) Learning for Coordinated Smart Generation Control of Interconnected Power Grids

This paper proposes an optimal coordinated control methodology based on the multi-agent reinforcement learning (MARL) for the multi-area smart generation control (SGC) under the control performance standards (CPS). A new MARL algorithm called correlated Q(λ) learning (CEQ(λ)) is presented to form an optimal joint equilibrium strategy for the coordinated load frequency control of interconnected control areas, and a SGC framework is proposed to facilitate information sharing and strategic interaction among multi-areas so as to enhance the overall long-run performance of the control areas. Furthermore, a novel time-varying equilibrium factor is introduced into the equilibrium selection function to identify the optimum equilibrium policies in various power system operation scenarios. The performance of CEQ(λ) based SGC strategy has been fully tested and benchmarked on a two-area power system and the China Southern Power Grid. Comparative studies have not only demonstrated the superior equilibrium optimization and dynamic performance of the proposed SGC strategy but also confirmed its fast convergence and high flexibility in designing the equilibrium factor for the desirable operating state of correlated equilibria.

Hybrid simulation of large electrical networks with assymmetrical fault modelling

Abstract Owing to the increasing attention placed on dynamic security assessment in the light of recent blackouts, hybrid simulation, involving the interfacing of electromagnetic transients (EMT) simulators and transient stability (TS) simulators for the more accurate representation of very large electric networks, has been attracting considerable attention lately. The goal of hybrid simulation is to represent complex systems, such as FACTS devices and HVDC terminals, in detail, while ensuring an acceptable representation of the complete system such that control systems and switching devices are well represented. There has been significant success in this area. Most approaches, however, concentrate on contingencies that preserve the symmetry of the system. Asymmetrical faults, for example, have not been dealt with in the literature on hybrid simulation. In this paper we deal with how two very different simulators can be coordinated to work together and produce accurate solutions, even with asymmetrical disturbances. Based on the interaction protocol proposed in the paper, case studies were performed on 9-bus systems and 39-bus systems to assess the performance of the integration on both EMT and TS aspects. Very good results were achieved.

Optimal Scheduling of Biogas-Solar-Wind Renewable Portfolio for Multi-Carrier Energy Supplies

This paper proposes a multisource multiproduct framework for coupled multicarrier energy supplies with a biogas–solar–wind hybrid renewable system. In this framework, the biogas–solar–wind complementarities are fully exploited based on digesting thermodynamic effects for the synergetic interactions of electricity, gas, and heating energy flows, and a coupling matrix is formulated for the modeling of production, conversion, storage, and consumption of different energy carriers. The multienergy complementarity of biogas–solar–wind renewable portfolio can be utilized to facilitate the mitigation of renewable intermittency and the efficient utilization of batteries, and a multicarrier generation scheduling scheme is further presented to dynamically optimize dispatch factors in the coupling matrix for energy-efficient conversion and storage, while different energy demands of end-users are satisfied. The proposed methodology has been fully tested and benchmarked on a stand-alone Microgrid over a 24-h scheduling horizon. Comparative results demonstrate that the proposed scheme can lower the battery charging/discharging actions as well as the degradation cost, and also confirm its capability to accommodate high penetration of variable renewables.

A Hierarchical Scheme for Utilizing Plug-In Electric Vehicle Power to Hedge Against Wind-Induced Unit Ramp Cycling Operations

Increasing wind power (WP) integration is forcing conventional units to go through more frequent and significant cycling operations, which would accelerate wear and tear to unit components and eventually affect the units lifespan. In this context, this paper proposes a hierarchical scheme to control the power of plug-in electric vehicles (PEVs) to mitigate unit ramp cycling (URC) operations. A general-form representation of the URC operation is proposed for the first time. At the top level of the hierarchical scheme, a system net load variation range (NLVR) is constructed first to capture the uncertainty in WP forecasts, and then the PEV power is scheduled to reshape the NLVR so as to minimize the URC operations that can be caused by the possible net load realizations in the NLVR. Based on updated WP forecasts, the middle-level dispatch model exempts overscheduled anti-URC regulation onus on PEVs to promote PEV charging. At the bottom level, a decentralized PEV charging control strategy is used to implement the PEV power dispatch instruction. Simulation results verify that the proposed scheme can avert the URC operations effectively, while preserve most of the desired PEV charging energy. Simulation results also show that the proposed scheme is more capable of withstanding WP forecast errors compared with its deterministic version and a benchmark scheme.