Gait parameters play an important role in diagnosing children suffering from cerebral palsy (CP). Traditional gait analysis systems, including gait laboratories equipped with optical motion capture, force plates, and pressure sensors, are widely used to measure these parameters. However, these systems require expensive setups and are unsuitable for day to day use, limiting their practicality for continuous monitoring. In contrast, an inertial measurement unit (IMU) has the ability to integrate a tri-axial accelerometer and tri-axial gyroscope to provide a low-cost solution for estimating gait parameters. The existing algorithm for gait analysis suffers from lower accuracy of gait parameters for children suffering from CP due to incorrect detection of gait events. This paper proposes a novel particle filtering algorithm based on the Amari alpha divergence (AAD) to estimate gait parameters from IMU data. By fusing measurement data from wearable accelerometers and gyroscopes attached to the L5 vertebra, the Amari alpha divergence based particle filter (AADPF) aims to improve the accuracy of gait parameters. Sensor fusion refines gait events, leading to more accurate estimation of gait parameters such as stride length, cadence, stance time, etc. To show the general prediction capability of the AADPF, a simulation study is carried out on a nonlinear system. The simulation results show lower mean square error as compared to the Kalman filter. Further, an experiment is conducted to collect the data from ten children suffering from CP and ten control children using a single IMU. The results show that the proposed method leads to significant improvement in accuracy of gait parameters when compared with existing algorithms.