Urban water distribution networks must use pressure management to reduce water leakage by modifying storage tank pressure levels in response to variations in water demand. Since each demand node usually restricts the maximum pressure that may be applied, addressing pressure issues at individual nodes is also crucial. To overcome these difficulties, a brand-new Convolutional Neural Network (CNN) Pressure Optimization Model is proposed. This model collects real-time data on water levels and pressure by utilizing level and pressure sensors, and a Backtracking Search Optimization (BSO) model is used to process the data. Subsequently, the optimized data is employed to execute accurate flow control protocols and detect possible leakage points. Major advantages are achieved with the BSO-CNN approach, including lower operating costs, more efficiency, and less water pressure needed. Furthermore, the model demonstrates a noteworthy decline in leakage rates, attaining a noteworthy reduction of roughly 31.5 % in the water distribution network. Improving the sustainability and performance of urban water distribution systems can be achieved through the proper integration of predictive modeling and advanced optimization techniques.