Slope Safety Prediction and Application Based on Harris Hawks Optimization-Backpropagation Neural Network (HHO-BP)

Landslide instability is a severe geological hazard that poses significant threats to infrastructure and the safety of human lives and property. To enhance prediction accuracy under complex geological conditions, this study proposes a slope safety prediction model integrating the Harris Hawk Optimization (HHO) algorithm with BP neural networks. Based on landslide mechanisms, an index system was established incorporating six key influencing factors: effective cohesion of soil strips, effective internal friction angle, self-gravity, pore water pressure at strip bases, strip width, and slope angle. Using 200 sets of slope engineering data from Ya’an region in Sichuan Province as samples, the HHO-optimized BP neural network achieved effective prediction of slope safety factors. Comparative analyses with GA-BP and PSO-BP models demonstrated that the HHO-BP neural network model achieved the highest prediction accuracy of 96.46% for 30 slope safety factors in test sets, along with faster convergence speed. These findings can be applied to safety assessment and risk prevention for slopes along the Ya’an-Kangding Expressway.