Research On Multi-sensor Data Fusion Based On Fuzzy Theory And Neural Networ

With the development of science and technology,the types of sensors are more abundant,and the performance is gradually improved,but only the use of a single sensor is far from forming a comprehensive and complete perceptual description of complex systems,so multi-sensor data fusion technology is needed to comprehensively process data to obtain more comprehensive,accurate and real results.Data fusion algorithm is the core of multi-sensor data fusion technology,and the application of fuzzy theory and neural network to multi-sensor data fusion has become a hot topic in recent years,but there are still some problems.In this dissertation,some problems existing in the application of fuzzy theory and neural network algorithms to data fusion systems are studied,and the main contents are divided into the following three aspects:（1）Multi-sensor data fusion research based on fuzzy theory.In this dissertation,the data fusion algorithm of similar sensors based on fuzzy support is studied,and the dynamic bending distance in the dynamic time regularization algorithm is introduced instead of the absolute distance in the fuzzy support function,and the reliability of the sensor itself is combined with the support degree as the weighted fusion weight,and the data fusion algorithm based on improved fuzzy support is proposed.Finally,the proposed method is tested with the fusion algorithm based on fuzzy support degree and the fusion algorithm based on arithmetic averaging method for multi-humidity sensor data fusion,and the comparison results are displayed in two experimental scenarios,which shows that the fusion accuracy and robustness of the proposed method are higher.（2）Research on multi-sensor data fusion based on neural network.In this dissertation,the sample data composed of multi-sensor data is analyzed by principal components and dimensionality reduction,and then the parameters of the BP neural network are optimized by fireworks algorithm（Fireworks Algorithm）,and a data fusion model based on the FWA-BP neural network is established.Finally,combined with the example of PM_2.5 concentration prediction,the proposed model is compared with the BP neural network data fusion model and the BP neural network data fusion model optimized by particle swarm algorithm,and the average absolute error and root mean square error are selected as the evaluation indicators,and the comprehensive analysis shows that the proposed model fusion accuracy is higher,which verifies the feasibility of the proposed model applied to multi-sensor data fusion.（3）Research on multi-sensor data fusion based on fuzzy theory and neural network.In this dissertation,the advantages of fuzzy theory and neural network combined for multi-sensor data fusion system are studied,and a data fusion model based on adaptive neural fuzzy inference system（Adaptive network-based fuzzy inference system）is proposed.The fuzzy C-means clustering process is optimized by subtractive clustering,and the optimal clustering result is determined by introducing the clustering effectiveness index,the initial structure of ANFIS is constructed based on the optimal clustering result,and the model is trained by using the hybrid algorithm to establish a data fusion model based on ANFIS.Finally,the proposed model is applied to the water quality rating evaluation problem,and compared with the ANFIS data fusion model based on meshing,the results show that the fused inference results of this dissertation are more accurate and algorithm complexity are lower,which verifies the effectiveness of the proposed model applied to multi-sensor data fusion.