交互系统设计中分布式多平台信息融合仿真

Distributed Multi-Platform Information Fusion Simulation in Interactive System Design

为了提高交互系统的用户满意度,提升系统服务性能,针对当前交互系统设计中分布式多平台信息融合方法中存在融合响应延时较长、融合值及测试精度较低等问题,提出基于模糊神经网络的交互系统设计中分布式多平台信息融合方法。分别采用多元判别分析法与局部切空间排列法提取各平台信息的线性特征与非线性特征,过滤掉冗余信息,消除干扰信息影响,获取有价值的特征信息。利用T-S模糊系统模型较强的自适应能力,及其自动学习更新、模糊隶属等特性,结合BP神经网络算法进行信息融合,并通过系数修正提高多平台信息融合结果的准确性。实验结果表明,与当前信息融合方法相比,所提方法的融合响应延时较低,信息融合值更高,测试精度受信息输出维度变化的影响较小,且信息识别能力及适应性较强。

In the current interactive system design, the fusion response delay is long, and the fusion value and test accuracy is low. In order to improve the user satisfaction and the service performance of interactive system, this paper presented a method to fuse the distributed multi-platform information in network interactive system design based on fuzzy neural network. Respectively, the multivariate discriminant analysis method and the local tangent space alignment method were used to extract the linear and nonlinear features of platform information, so as to filter out redundant information and eliminate the influence of interference information. Thus, valuable feature information could be obtained. Moreover, T-S fuzzy system model had better adaptive ability, automatic learning update, fuzzy membership and other characteristics. BP neural network algorithm was used to fuse information, so that the accuracy of multi-platform information fusion results was improved by coefficient correction. Simulation results show that, compared with current information fusion methods, the proposed method has lower fusion response delay and higher information fusion value. Meanwhile, the test accuracy is less affected by the change of information output dimension, and the information recognition ability and the adaptability are stronger.