Key points are not available for this paper at this time.
Condition monitoring and fault diagnosis play a crucial role in detecting early-stage issues in wind power conversion and generation systems, thereby ensuring efficient and reliable operation. This study proposes a novel condition monitoring framework that integrates a quantum long short-term memory (QLSTM) network with a Transformer model for robust fault detection. Unlike conventional machine learning operating in Euclidean space, quantum machine learning, powered by variational quantum circuits, captures high-order interactions among variables in Hilbert space, enabling explicit modeling of complex couplings without excessive parameterization. The framework also establishes the relationship between the selected input variables and the required qubits to optimize the quantum model architecture. To enhance detection sensitivity, a dynamic thresholding method based on Bayesian inference and a sliding window (BISW) mechanism is developed to adaptively balance short-term variations and long-term trends in residuals. The framework is validated using real-world SCADA data containing representative wind turbine faults, including gearbox bearing, generator winding, and pitch control system failures. Experimental results show that the proposed QLSTM-Transformer-BISW model achieves the highest modeling accuracy against benchmark methods and successfully detects faults 123.83 h, 77.5 h, and 122.67 h, respectively, before turbine shutdowns, without false alarms. These findings demonstrate that the proposed framework offers the potential of quantum machine learning as a new paradigm for intelligent renewable energy systems.
Zhang et al. (Sat,) studied this question.