Jianzhong Wu’s research while affiliated with Cardiff University and other places

Accurate battery state estimation is important for the operation of energy storage systems, yet existing methods struggle with the complexity and dynamic nature of battery conditions. Conventional techniques often fail to extract relevant spatial and temporal features from basic battery data effectively, leading to insufficient situational awareness in battery management systems. To address this gap, we propose a Hierarchical and Self-Evolving Digital Twin (HSE-DT) method that enhances battery state estimation by coordinating multiple estimation techniques in a hierarchical framework and enabling adaptive updating through transfer learning. The model integrates a Transformer–Convolutional Neural Network (Transformer-CNN) architecture to process historical and real-time data, capturing dynamic state variations with high precision. Simulations indicate that the values of root mean square error (RMSE) for state of charge (SOC) and state of health (SOH) are lower compared to other algorithms, being less than 0.9% and 0.8%, respectively. Its hierarchical structure allows the integration of different estimation models, and the self-evolving method allows the method to adapt to changes in different operating conditions. The experimental results show that the method can estimate the battery state with high accuracy and stability, thus enhancing multi-faceted situational awareness.

The iron and steel industry contributes approximately 25% of global industrial CO emissions, necessitating substantial decarbonisation efforts. Hydrogen-based iron and steel plants (HISPs), which utilise hydrogen-based direct reduction of iron ore followed by electric arc furnace steelmaking, have attracted substantial research interest. However, commercialisation of HISPs faces economic feasibility issues due to the high electricity costs of hydrogen production. To improve economic feasibility, HISPs are jointly powered by local renewable generators and bulk power grid, i.e., by a grid-assisted renewable energy system. Given the variability of renewable energy generation and time-dependent electricity prices, flexible scheduling of HISP production tasks is essential to reduce electricity costs. However, cost-effectively scheduling of HISP production tasks is non-trivial, as it is subject to critical operational constraints, arising from the tight coupling and distinct operational characteristics of HISPs sub-processes. To address the above issues, this paper proposes an integrated resource-task network (RTN) to elaborately model the critical operational constraints, such as resource balance, task execution, and transfer time. More specifically, each sub-process is first modelled as an individual RTN, which is then seamlessly integrated through boundary dependency constraints. By embedding the formulated operational constraints into optimisation, a cost-effective scheduling model is developed for HISPs powered by the grid-assisted renewable energy system. Numerical results demonstrate that, compared to conventional scheduling approaches, the proposed method significantly reduces total operational costs across various production scales.

As the utilization of power electronic-based components in power systems continues to grow, a comprehensive understanding of their dynamics becomes increasingly important for system design, control and protection analysis. To meet practical needs, the high-fidelity but time-consuming electromagnetic transient (EMT) simulations are often required. To improve the performance of these simulations, a highly efficient splitting state-space method with numerical error control is proposed that reduces the computation workload. The method employs a generic decoupling principle to split the state-space equations of the converter-integrated power system and introduces the exponential splitting formulas of multiple orders accuracy to solve and then compose the splitting state-space equations. The decoupling principle is designed based on separation of time-varying portions of the state matrix, which is realized by locating the smallest subcircuit topology that is switch state-dependent, through automatic switch grouping and switch adjacent state variables (SASV) identification. A family of exponential splitting schemes is employed to accelerate the demanding matrix exponential calculation. The splitting state-space method undergoes comprehensive testing across various cases, including a distribution network with DC load, an LLC resonant converter, a large-scale wind farm, and an MMC circuit. The accuracy of the proposed method is thoroughly evaluated, and its efficiency is validated.

Truck mobile charging stations (TMCS) are emerging as an effective solution to bridge the gap between supply and demand for electric vehicle (EV) charging. However, traditional business models face barriers due to high initial costs and low utilization rates, hindering operator participation. To this end, this paper introduces a bi-level optimization model for TMCS leasing to balance the TMCS operator (TMCO) and charging facility operators (CFOs). The upper-level objective maximizes the profit of TMCO, focusing on TMCS fleet size, differentiated pricing for long-term and short-term rentals, and scheduling grid energy arbitrage during idle periods. The lower level aims to maximize the profit of CFOs by determining rental quantities and durations based on leasing offers. A distributionally robust optimization (DRO) approach is employed to address the uncertainties in EV charging demand, using chance constraints with the Wasserstein distance to capture forecast errors. The probabilistic constraints are transformed into tractable linear constraints through conditional value-at-risk (CVaR) approximation. The model is solved by the genetic algorithm (GA) at the upper layer and the nested column-and-constraint generation (NC&CG) algorithm at the lower layer. Case studies show that the model effectively balances the objectives of TMCO and CFOs. With adaptive pricing and TMCS allocation strategies, the model ensures the TMCO’s profitability while improving CFOs’ economics.

Soft open points (SOPs) are power electronic devices placed at normally open points of electricity distribution networks. With millisecond-level control, SOPs are promising in constraint management of distribution networks facing the significant uncertainties from renewable power generation and customer behaviors (such as electric vehicle travelling behaviors). This paper develops a novel feasible operation region (FOR)-based method for optimal SOP control. The FOR, denoted as the allowable range of nodal power injections of distribution networks, can be used to replace the power flow equations and network constraints in a conventional optimal power flow (OPF)-based model. Due to the one-to-one correspondence between FOR boundaries and thermal/voltage constraints, FOR-based constraint management method can adapt to various measurement conditions. Moreover, the FOR constraints can be converted into a format based on line flows and node voltages, allowing for the use of real-time measurements of these operating parameters rather than the measurements of nodal power load/generation that are normally not accessible online. The proposed method is validated on the IEEE 33-node distribution network and IEEE 123-node distribution network. The performance of the method is also compared with that of conventional OPF-based control.