Abstract: Objective Horizontal well staged fracturing and volume stimulation have become one of the effective approaches for the efficient development of deep coal-bed methane (CBM) reservoirs. Nevertheless, the distinctive rock mechanical properties, reservoir filtration characteristics, and micro-fracture development features of deep coal are notably distinct from those of unconventional shale, tight sandstone, and shallow coal. The existing fracturing experience cannot be simply replicated, and it is necessary to conduct targeted research on the optimization of fracturing parameters. Methods In this paper, the deep CBM reservoir in the Baijiahai area of the Junggar Basin is taken as the target. A horizontal well multi-cluster perforation staged fracturing model for composite geological structures such as top and bottom plates, partings, and coal was established. The influence of beddings and cleats on fracture propagation and reservoir filtration was considered. The key parameters of the reservoir were corrected by utilizing the data from pressure diagnosis, which enhanced the accuracy of the model, comprehensively characterizing the physical and mechanical properties of the deep CBM reservoir. The fracturing fluid system suitable for deep CBM reservoirs was optimized through numerical simulation, and a design method for the optimal combination of pumping parameters under multiple objectives was established. The optimization results were verified by the micro-seismic data of the test well. Results and Conclusions The results indicate that the variable viscosity slickwater system is more suitable for deep CBM reservoirs fracturing compared to the low-viscosity slickwater system and gel system, with advantages in terms of fracture length, fracture width, and fracture length balance. The optimal parameter combination for fracturing in the target block is obtained: 3 clusters per stage + cluster spacing of 19 m + proppant volume per meter of 2.8 m³ + pumping rate of 16 m³/min. The micro-seismic monitoring results of the test well demonstrate that after optimization, the fracture half-length and stimulated reservoir volume (SRV) has increased by 57.2%, 12.3%. A comparison of fracture geometries under different hydraulic fracturing parameter combinations reveals that, in deep coal reservoirs with severe interlayer barriers, fractures typically cannot propagate in the height direction. Therefore, increasing fracture length and width is necessary. Increasing proppant concentration and injection rate significantly enhances fracture length and width, which is crucial for improving reservoir stimulation.