.png){kind=icon}
Return
.png){kind=icon}
.png){kind=icon}
.png){kind=icon}
.png){kind=icon}
.png){kind=icon}
Abstract
En 中文
Sensor data have dynamic characteristics over time and environment, which makes it difficult to effectively learn the deep correlation between multi-source sensor data when fusion is performed, and thus unable to adapt to the conflict characteristics of the data, resulting in poor fusion effect. Therefore, a dynamic fusion method based on improved RBF network algorithm is proposed. Based on the extended Kalman filter and particle filter, the multi-source sensor data are extracted and normalized to generate the processing format suitable for RBF neural network, so as to improve the accuracy and reliability of fusion. To accommodate the temporal and environmental variations in sensor data, we utilize the dynamic properties of Radial Basis Function (RBF) neural networks. In this model, input data is processed using a Gaussian activation function. Additionally, we incorporate a genetic algorithm to refine the RBF neural network. Through iterative crossover and mutation procedures, we search for the most effective network parameters within the global parameter space. This approach enhances the network's performance, enabling it to deeply comprehend the intricate relationships within multisensor data from diverse sources, and facilitating a comprehensive analysis of such data. Accordingly, the optimized RBF network is used to achieve dynamic fusion of multi-sensor data. Through experimental verification, the data integrated by the method is clearly displayed through the visualization page, which can provide a good basis for the subsequent process. Based on the data fusion results of the algorithm to improve the experimental smart home, the degree of intelligent perception, interaction smoothness, and stability are maintained at a high level, which proves that the method has a high fusion effect, and it can provide effective and reliable data support to promote the development of the field of intelligent management effectively.
Journal
IF:
4.2
Papers:
414
Citations:
1.1K
Organization
No organization information available
Cited Papers
No cited papers available