A sensor data fusion algorithm based on suboptimal network powered deep learning

Abstract

To improve the performance of data fusion in wireless sensor networks, based on the suboptimal network powered deep learning model, a data fusion algorithm that combines a stacked autoencoder (SAE) and a clustering protocol is proposed. The fusion algorithm for sensor data is investigated by suboptimal network powered deep learning to analyse the fusion processing at the sensor measurement level and the fusion processing at the localization result level. This paper establishes a suboptimal network powered deep learning model and derives its weighted least squares solution algorithm based on this model, also, the best linear unbiased estimation method for fusion of vibration sensor and vision sensor localization results is adopted. First, to overcome the filter divergence caused by rounding error, reduce the computational complexity, ensure the non-negativity of the covariance, and improve the convergence speed of the filter, the target state volume after fusion using the proposed model is closer to the real state volume of the target compared to the state volume of each local sensor detection, verify the validity of the algorithm was increased.

Keywords

• Suboptimal networks
• Deep learning
• Sensor data
• Fusion algorithms