Fault-tolerant synchronization of quaternion-valued BAM neural networks with time-varying delay subject to multiple-actuator failure.

The synchronization of quaternion-valued bidirectional associative memory neural networks (QVBAMNNs) subject to actuator faults is studied in this work. A fault-tolerant control approach is suggested by considering both actuator effectiveness loss and bias faults, which are rarely addressed simultaneously. A mathematical fault model is formulated to characterize the influence of actuator failures on the control inputs. Under this framework, suitable guidelines are established to guarantee global exponential, fixed-time, finite-time, and predefined-time synchronization. In addition, the corresponding settling times are analyzed. Simulations are shown to confirm the theoretical results.