基于HMM的家用机器人语音交互系统故障预测方法的研究

摘要

对家用机器人语音交互系统故障预测有利于提高系统的安全性、减少生产成本和提高维护保养的能力。根据隐马尔科夫模型的故障预测方法，利用基于语音芯片RSC4128的语音交互系统的故障预测问题，归纳和分析了语音交互系统的衰退过程，对于语音交互系统PWM输出信号、运动控制信号与串行通信信号进行监测，选择各个状态样本来训练HMM，然后利用得到的模型对待测信号进行观测，可以获得待测信号的似然概率，可以预测系统的当前状态，有利于家用机器人的语音交互系统的测试与发展。%The domestic robot voice interactive system failure prediction wil help improve the security of the system,reduce production costs and improve maintenance have the capacity.According to the fault prediction method hidden Markov model,the use of fault prediction voice chip RSC4128 interactive voice system based on summarized and analyzed the recession process voice interaction system,voice interactive system for PWM output signal,motion control signal and a serial communication signal monitoring,select each state sample to train HMM,and then use the resulting model to treat the measured signal observed can be obtained likelihood probability test signal,the current state of the system can be predicted,in favor of domestic robots voice interac-tive system testing and development.