It has been found experimentally that learning during wakefulness leads to a net enhancement of synaptic strength, accompanied by the neural dynamical transition from tonic to bursting firing, while the net synaptic strength decreases to a baseline level during sleep, accompanied by the transition from bursting to tonic firing. In this paper, we establish a model of synaptic plasticity, which can realize synaptic strength changes and neural dynamical transitions in wakefulness-sleep cycle by using the coupled Hindmarsh-Rose neurons. Through numerical simulation and theoretical analysis, it is further found that the average synaptic weight of the neural network can reach a stable value during either prolonged wakefulness or prolonged sleep, which depends on the ratio of some specific parameters in the model. Particularly, the synaptic weight exhibits a stable log-normal distribution observed in a real neural system, when the average synaptic weight reaches a stable value. Moreover, the fluctuation of this weight distribution is positively correlated with the fluctuation of noise in the synaptic plasticity model. The provided model of the synaptic plasticity and its dynamics results can provide a theoretical reference for studying the physiological mechanism of synaptic plasticity and neuronal firings during the wakefulness-sleep cycle, and they are expected to have potential applications in the development of therapeutic interventions for sleep disorders.