Data Fusion in Redundant Inertial Measurement Unit Using a Fruit-Fly-Optimized Weighted Least Squares Algorithm

The weighted least squares (WLS) method is one of the most popular methods for the data fusion of redundant inertial measurement unit (RIMU). Therefore, the weight settings of WLS are crucial to improve the measurement accuracy of RIMU. In this study, an optimal WLS weight estimation method based on the fruit fly optimization algorithm (FOA) is proposed. First, an optimal WLS weight criterion for an RIMU is established. This criterion determines the quality of the weight using the fused measurement data rather than the error analysis results, and it can therefore compensate for more errors than the traditional method. Second, a constrained multi-dimensional fruit fly optimization algorithm (CMFOA) is proposed to determine the optimal WLS weight according to the optimal weight criterion. The CMFOA extends the FOA from one dimension to multiple dimensions to match multiple weighting factors, and the concept of smell contamination is proposed to solve the constrained optimization problem for the aforementioned criterion. The CMFOA is then applied to estimate the optimal weight of WLS in an RIMU. The simulation and experimental results show that the proposed method is superior to the traditional methods used for the data fusion of RIMU.