The application of smoothed particle hydrodynamics (SPH) encounters challenges related to consistency, stability, and accuracy. Inconsistencies in SPH arise from non-uniform particle distribution and a lack of neighboring particles at the boundary, leading to numerical instability and inaccurate particle approximations. This instability often manifests as a zig-zag pattern in field distribution, referred to as the hourglass instability mode. Various correction frameworks have been proposed to address these issues. One such framework is the conservative reproducing kernel SPH, designed to ensure zeroth- and first-order consistency in the method. This method can accurately reproduce any linear function and its gradient up to machine precision and achieves a second-order convergence rate using the L2 norm error in fluid dynamics applications. Another technique based on modified SPH and the modified finite particle method aims for higher-order consistency . This method has been tested on 2-D and 3-D problems, demonstrating accurate results and the ability to model highly complex flow patterns. In this research, an alternative method to restoring SPH consistency is introduced. The kernel function is corrected to achieve zeroth-, first-, and second-order consistency, and the kernel's gradient is derived directly through multiple applications of the chain rule. The method's performance is verified through gradient calculations and solving partial differential equations, revealing a significant improvement in the convergence rate of SPH.