Application Of Human-computer Speech Interaction In Intelligent Dispatching

With the vigorous development of the electric power industry in our country, a new generation of Smart Grid system is being gradually established. As a set of auxiliary decision-making of intelligent software system used for power grid dispatching, the system of electric power dispatching automation contains a large number of background computation. Timely, information can be processed and transferred to dispatchers, which provides effective guarantee for the safe operation of power system. However, the operation of intelligent dispatching system is still tedious at the present stage and the expanding power system will call for more monitor messages. Along with more information needed to monitor, the dispatchers are suffering more serious tasks. Accordingly, their ability to handle sudden accident weakens.In order to preferably meet the actual demand of power grid dispatching, this paper proposes a project of integrating the speech technology into the intelligent dispatching system, realizing information exchange between dispatchers and intelligent dispatching system by speech. As matters stand, the accuracy of speech recognition system, for dispatching terminology is low. In addition, the application research of speech technology in the field of power system is lack. To solve the above problems, this paper designs a speech interaction system, which is applied to intelligent dispatching. The process which is aimed at improving the work efficiency of human-computer interaction interface and power dispatching is paid attention to the following points. Firstly, dispatching system runs safely and stably. Secondly, dispatchers’daily work strength is lower. Finally, user-centered design principles are easy to realize.The Microsoft Speech SDK is a free software development kit for Microsoft Windows and provides services based on Component Object Model (COM), which makes underlying protocols out of the application layer. The Microsoft Speech platform is adapted to design the speech recognition and synthesis system. Microsoft Speech SDK is selected as the development tool. According to natural language of dispatchers and characteristics of power grid dispatching phrase, this paper compiles the corresponding grammar. In addition, dispatchers are scheduled for speech training based on frequently-used dispatching command.The last part is a series of experiments of getting started with the speech interaction system, which this thesis designs and implements, mainly for accuracy test of speech recognition system. In the process of experiments, the data is recorded in detail. In accordance with the correct strings of words, terminology and key words in the record results, the corresponding recognition accuracy is recorded. Recognition accuracy has improved significantly through training for professional vocabulary, and its average recognition rate of dispatching instruction is90.25%. Experimental results show showed that this system can greatly improve the recognition accuracy of dispatching special words and be more suitable for intelligent dispatching than the typical Chinese speech interaction systems.