No abstract

Thermostatically-controlled-loads (TCLs) have been regarded as a good candidate for maintaining the power system reliability by providing operating reserve. The short-term reliability evaluation of power systems, which is essential for power system operators in decision making to secure the system real time balancing, calls for the accurate modelling of operating reserve provided by TCLs. However, the particular characteristics of TCLs make their dynamic response different from the traditional generating units, resulting in difficulties to accurately represent the reliability of operating reserve provided by TCLs with conventional reliability model. This paper proposes a novel multi-state reliability model of operating reserve provided by TCLs considering their dynamic response during the reserve deployment process. An analytical model for characterizing dynamics of operating reserve provided by TCLs is firstly developed based on the migration of TCLs' room temperature within the temperature hysteresis band. Then, considering the stochastic consumers' behaviour and ambient temperature, the probability distribution functions of reserve capacity provided by TCLs are obtained by cumulants. On this basis, the states of reserve capacity and the corresponding probabilities at each time instant are obtained for representing the reliability of operating reserve provided by TCLs in the LZ-transform approach. Case studies are conducted to validate the proposed technique.

The increase in penetration of inverter-based resources provide us with more flexibility in frequency regulation of power systems in addition to conventional linear droop controllers. Because of the fast power electronic interfaces, inverterbased resources can be used to realize complex control functions and potentially offer large gains in performance compared to linear controllers. Reinforcement learning has emerged as popular method to find these nonlinear controllers by parameterizing them as neural networks.The key challenge with learning based approach is that stability constraints are difficult to enforce on the learned controllers. In addition, optimizing the controllers are nontrivial because of the time-coupled dynamics of power systems. In this paper, we propose to explicitly engineer the structure of neural network based controllers such that they guarantee system stability for all topologies and parameters. This is done by using a Lyapunov function to guide their structures. A recurrent neural network based reinforcement learning architecture is used to efficiently train the weights of controllers. The resulting controllers only use local information and outperform linear droop as well as strategies learned purely by using reinforcement learning.

No abstract