Currently, triple phase-shift (TPS) modulation has attracted more and more attention of researchers as an advanced modulation strategy for dual active bridge converter (DAB). Since it has three degrees of freedom, it can realize better performance both in soft switching ranges and power efficiency. However, how to choose these three degrees of freedom to realize optimal power efficiency of DAB converter becomes a concern for researchers. Generally, there exist two difficulties to apply efficiency-oriented TPS modulation. The first difficulty lies in the analysis process in which the main task is to figure out the relationships between modulation parameters and power loss. The three modulation parameters in TPS bring difficulties in analysis and deduction process, which suffers from high computational burden and low accuracy. Additionally, the second difficulty lies in the real-time realization of TPS modulation. If a look-up table is applied to store the optimized modulation parameters, it is highly likely that its discrete nature will result in unsatisfactory modulation performance. Therefore, this article proposes an efficiency-oriented automatic TPS (ATPS) modulation approach, which utilizes neural network, particle swarm optimization, and fuzzy inference system, respectively, in its three stages. The proposed ATPS is able to mitigate labor in computational burden with a highly automatic fashion. Finally, this proposed ATPS has been validated with 1-kW hardware experiments.

Robust fuzzy model predictive control of discrete nonlinear systems is investigated in this work. A recently developed Takagi-Sugeno (T-S) fuzzy approach which uses nonlinear local models is adopted to approximate nonlinear systems. A critical issue that restricts the practical application of classical model predictive control is the online computational cost. For model predictive control of T-S fuzzy systems, the online computational burden is even worse. Especially for complex systems with severe nonlinearities, parametric uncertainties, and disturbances, existing model predictive control of T-S fuzzy systems usually leads to a very conservative solution or even no solution in some occasions. However, more relaxed results can be achieved by the proposed fuzzy model predictive control approach which adopts T-S systems with nonlinear local models. Another advantage is that, online computational cost of the optimization problem through solving matrix inequalities can be significantly reduced at the same time. Simulations on a numerical example and a two-tank system are presented to verify the effectiveness and advantages of the proposed method. Comparisons among several T-S fuzzy approaches are illustrated and show that the best settling time is achieved via the proposed method.

In this work, model predictive control of discrete T-S fuzzy systems subjected to bounded time-varying delay and persistent disturbances is investigated. The Razumikhin approach is adopted for time-delay systems because it involves a Lyapunov function associated with the original non-augmented state space of system dynamics when compared to the Krasovskii approach. As such, the Razumikhin approach has a good potential to avoid the inherent complexity of the Krasovskii approach especially in the presence of large delays and disturbances. Based on which, both online and off-line model predictive control approaches for systems with time-varying delay are provided. In addition, persistent disturbances are considered that robust positive invariance and input-to-state stability under such circumstances are realized. In the off-line approach, a sequence of explicit control laws that correspond to a sequence of robust constraints sets are computed off-line. And it is proved that system states including all possibly delayed states can be steered to the terminal constraint set in finite time. Moreover, it allows the exact time delay to be unknown in the proposed two approaches. In particular, for systems with time-varying delay the special positively invariant set theory and finite-time control theory based on the Razumikhin approach are directly revealed via the proposed off-line approach.

Dual active bridge (DAB) topology has been recognized as the key circuit for the next generation of high-frequency-link power conversion systems. In order to realize desired operating performance, DAB modulation strategies need to be selected carefully. Different modulation strategies have been considered and combined into a hybrid one, which is able to fully optimize performance. However, to develop a hybrid modulation strategy, the conventional methods including harmonic model and piecewise model have difficulty in balancing modeling accuracy and manpower burden. Although recent data-driven modulation approaches can automate the design process, the accuracy of data-driven models decreases nontrivially with the existence of outliers and without sufficient data. Hence, to alleviate human-dependence and improve modeling accuracy, this paper proposes an AI-based design with data trimming (AI-DT) for hybrid phase shift modulation. Two two-degree-of-freedom modulation strategies are considered for the sake of optimal current stress performance. AI-DT firstly adopts the one-class support vector machine (SVM) to exclude outliers. Second, the state-of-the-art extreme gradient boosting (XGBoost), which is insensitive to training data size, is adopted to build data-driven current stress models for the DAB converter. After that, differential evolution (DE) algorithm helps to choose modulation strategy and optimize modulation parameters for optimal current stress. Generally, the proposed AI-DT is developed in an automated fashion, which largely relieves manual computational complexity while exhibiting satisfactory modulation accuracy. The effectiveness of the proposed AI-DT approach has been experimentally verified with a 1kW hardware prototype, realizing optimal current stress under full operating conditions with more than 96.5% as peak efficiency.

In this work, two efficient robust fuzzy model predictive control algorithms are investigated for discrete nonlinear systems with multiple time delays and bounded disturbances. The famous T-S fuzzy systems are utilized to represent nonlinear systems. Instead of Lyapunov-Krasovskii functional, Lyapunov-Razumikhin function is adopted to deal with time delays because it involves invariant sets in the original state space of the system. A sequence of explicit control laws corresponding to a sequence of constraint sets are computed off-line that the online computational burden associated with classical model predictive control algorithms is significantly reduced. In particular, the set invariance theory behind Razumikhin approach, which is more complicated than the one for non-delayed systems, is directly observed. And it is proved that all (delayed) states can enter the terminal set in finite time. Moreover, robust positive invariance and input-to-state stability for time-delay systems concerning disturbances are realized. In addition, an online optimization algorithm is also provided based on the off-line computed ellipsoidal sets. Thus the conservatism induced by Razumikhin approach is relaxed, while the computational cost is not significantly increased.

The stable operation of air handling units (AHU) is critical to ensure high efficiency and to extend the lifetime of the heating, ventilation, and air conditioning (HVAC) systems of buildings. In this paper, an online data-driven diagnosis method for AHU in an HVAC system is proposed and elaborated. The rule-based method can roughly detect the sensor condition by setting threshold values according to prior experience. Then, an efficient feature selection method using 1D convolutional neural networks (CNNs) is proposed for fault diagnosis of AHU in HVAC systems according to the system’s historical data obtained from the building management system. The new framework combines the rule-based method and CNNs-based method (RACNN) for sensor fault and complicated fault. The fault type of AHU can be accurately identified via the offline test results with an accuracy of 99.15% and fast online detection within 2 min. In the lab, the proposed RACNN method was validated on a real AHU system. The experimental results show that the proposed RACNN improves the performance of fault diagnosis.