语音识别技术在配电网工程建设中的应用

Application of speech recognition technology in distribution network engineering construction

为了实现配电网调度的智能语音交互,研究基于连续时序分类(CTC)和注意力机制的端到端语音识别技术,构建改进CTC-ATT语音识别模型,并利用循环神经网络自适应映射模型进行优化。实验结果表明,改进CTC-ATT语音识别模型对配电网调度指令和调度术语的识别正确率分别为92.45%和91.99%,能对配电网的调度指令术语进行高效精准地识别,帮助调度人员提升配电网调度的效率,对配电网工程的建设发展具有实用意义,为智能调度领域的发展提供了新的研究思路。

In order to realize the intelligent speech interaction of distribution network dispatching,the end-to-end speech recognition technology based on connectionist temporal classification(CTC) and attention mechanism is studied to construct the improved CTC-ATT speech recognition model,and the cyclic neural network adaptive mapping model is used for optimization.The experiment results show that the recognition accuracy of the improved CTC-ATT speech recognition model for distribution network dispatching instructions and dispatching terms is 92.45% and 91.99%,which can effectively and accurately recognize the dispatching instructions,help dispatchers improve the efficiency of distribution network dispatching,and have practical significance for the construction and development of distribution network engineering,which provides a new research idea for the development of intelligent scheduling field.