To investigate the effect of water on the mechanical properties and acoustic emission (AE) characteristics of coal in the failure and deformation processes. A uniaxial compressive test with the aid AE monitoring was conducted on coal samples. The characteristics of the AE signals were further analyzed using fractal analysis. The results show that saturated coal samples have substantially reduced mechanical properties such as uniaxial compressive strength (UCS), dissipation energy, peak stress, and elastic modulus. The stress-strain curve demonstrated distinct stage characteristics under loading, and also the response AE signals of natural and saturated coal samples were consistent throughout. A novel Grassberger Procaccia (GP) algorithm was utilized for the first time to find the AE fractal characteristics of coal samples in different stages using phase-space theory. The results indicate that AE energy's first and second phases have no fractal properties. In contrast, the third stage has evident fractal characteristics. In the fourth stage, there is a rapid decline in the fractal dimension, followed by a gradual decrease or increase in the fifth stage. In stable crack propagation stages, the AE shows evident fractal characteristics, indicating that coal is stable. During the accelerated crack propagation stage, the correlation dimension decreases rapidly, continues to decline further, and begins to increase again in the fifth stage. Consequently, the coal begins to collapse, potentially resulting in a coal disaster and dynamic failure. It is, therefore, possible to accurately predict coal and rock dynamic failures by observing the subsequent change in the fractal dimension of the AE signals in response to different stages of loading.