Page 72 - 《应用声学》2024年第6期
P. 72
1248 2024 年 11 月
图的特征学习,捕捉有效的上下文信息。相比于文 [8] 王素宁, 朱俊杰, 李志勇, 等. 基于 DTW 算法的电力调度语
献 [33] 模型、文献 [34] 模型和文献 [35] 模型,本文模 音识别研究和应用 [J]. 电力与能源, 2021, 42(1): 35–38, 64.
型参数量分别降低 75.38%、97.40、76.86%,CER 分 Wang Suning, Zhu Junjie, Li Zhiyong, et al. Research and
application of power dispatching speech recognition based
别降低36.50%、63.11%、39.83%,平均耗时分别降低 on DTW algorithm[J]. Electricity & Energy, 2021, 42(1):
58.63%、71.41%、44.67%。本文可为电力调度自动 35–38, 64.
化提供一定参考。但仍存在一些不足,如仅对声学 [9] 胡翔, 杨洋, 蒋长江, 等. 一种基于深度神经网络的电力系统
模型进行研究和识别模型性能还有待提高,后期将 调度控制语音识别模型 [J]. 电子器件, 2023, 46(1): 90–95.
Hu Xiang, Yang Yang, Jiang Changjiang, et al. A speech
结合一些模型提高识别能力。
recognition model for power system scheduling control
based on deep neural networks[J]. Chinese Journal of Elec-
tron Devices, 2023, 46(1): 90–95.
参 考 文 献
[10] Wang Z H, Gao F. Research on voice interaction model of
intelligent power dispatching based on DCGAN[J]. Nan-
[1] 顾晓东, 唐丹宏, 黄晓华. 基于深度学习的电网巡检图像缺陷
otechnology for Environmental Engineering, 2021, 6(3):
检测与识别 [J]. 电力系统保护与控制, 2021, 49(5): 91–97.
53.
Gu Xiaodong, Tang Danhong, Huang Xiaohua. Defect
[11] Zhang Q R, Zhai H T, Ma Y Y, et al. Enhanced-
detection and recognition of power grid inspection images
deep-residual-shrinkage-network-based voiceprint recogni-
based on deep learning[J]. Power System Protection and
tion in the electric industry[J]. Electronics, 2023, 12(14):
Control, 2021, 49(5): 91–97.
3017–3031.
[2] 夏玉果, 董天天, 丁晟. 基于轻量化深度迁移神经网络的电子
[12] 王泽霞, 陈革, 陈振中. 基于改进卷积神经网络的化纤丝饼表
元器件识别 [J]. 电子器件, 2023, 46(6): 1673–1679.
面缺陷识别 [J]. 纺织学报, 2020, 41(4): 39–44.
Xia Yuguo, Dong Tiantian, Ding Sheng. Electronic com-
Wang Zexia, Chen Ge, Chen Zhenzhong. Surface defect
ponent recognition based on lightweight deep transfer neu-
recognition of synthetic fiber cake based on improved con-
ral network[J]. Chinese Journal of Electron Devices, 2023,
volutional neural network[J]. Journal of Textiles, 2020,
46(6): 1673–1679.
41(4): 39–44.
[3] 周艳真, 查显煜, 兰健, 等. 基于数据增强和深度残差网络的
[13] 许洪强, 蔡宇, 万雄, 等. 电网调控大数据平台体系架构及关
电力系统暂态稳定预测 [J]. 中国电力, 2020, 53(1): 22–31.
键技术 [J]. 电网技术, 2021, 45(12): 4798–4807.
Zhou Yanzhen, Cha Xianyu, Lan Jian, et al. Transient
Xu Hongqiang, Cai Yu, Wan Xiong, et al. Architec-
stability prediction of power systems based on data aug-
ture and key technologies of power grid regulation big
mentation and deep residual networks[J]. Electric Power,
data platform[J]. Power System Technology, 2021, 45(12):
2020, 53(1): 22–31.
4798–4807.
[4] 赵涛, 张羿, 王永和, 等. 基于深度学习的人机语音交互平
[14] 邱志斌, 石大寨, 况燕军, 等. 基于深度迁移学习的输电线路
台 [J]. 信息系统工程, 2019, 12(1): 102–104.
涉鸟故障危害鸟种图像识别 [J]. 高电压技术, 2021, 47(11):
Zhao Tao, Zhang Yi, Wang Yonghe, et al. A human com-
puter speech interaction platform based on deep learn- 3785–3794.
ing[J]. China CIO News, 2019, 12(1): 102–104. Qiu Zhibin, Shi Dazhai, Kuang Yanjun, et al. Bird species
[5] 赵晴, 李庭瑞, 罗睿, 等. 基于双字典类标签语言模型的电力 image recognition based on deep transfer learning for bird
related faults in transmission lines[J]. High Voltage Tech-
调度语音识别 [J]. 电子测量技术, 2021, 44(13): 121–126.
Zhao Qing, Li Tingrui, Luo Rui, et al. Power dispatch nology, 2021, 47(11): 3785–3794.
speech recognition based on dual dictionary class label [15] 徐冬冬. 基于 Transformer 的普通话语声识别模型位置编码
language model[J]. Electronic Measurement Technology, 选择 [J]. 应用声学, 2021, 40(2): 194–199.
2021, 44(13): 121–126. Xu Dongdong. Selection of position encoding for man-
[6] 鄢发齐, 王春明, 窦建中, 等. 基于隐马尔可夫模型的电力调 darin phonetic recognition model based on transformer[J].
度语音识别研究 [J]. 武汉大学学报 (工学版), 2018, 51(10): Journal of Applied Acoustics, 2021, 40(2): 194–199.
920–923. [16] 颜宏文, 陈金鑫. 基于改进 YOLOv3 的绝缘子串定位与状态
Yan Faqi, Wang Chunming, Dou Jianzhong, et al. Re- 识别方法 [J]. 高电压技术, 2020, 46(2): 423–432.
search on power dispatching speech recognition based Yan Hongwen, Chen Jinxin. A method for insulator
on hidden Markov model[J]. Journal of Wuhan Univer- string positioning and state recognition based on im-
sity(Engineering Edition), 2018, 51(10): 920–923. proved YOLOv3[J]. High Voltage Technology, 2020, 46(2):
[7] 窦建中, 罗深增, 金勇, 等. 基于深度神经网络的电力调度语 423–432.
音识别研究及应用 [J]. 湖北电力, 2019, 43(3): 16–22. [17] 杨德举, 马良荔, 谭琳珊, 等. 基于门控卷积网络与 CTC
Dou Jianzhong, Luo Shenzeng, Jin Yong, et al. Re- 的端到端语音识别 [J]. 计算机工程与设计, 2020, 41(9):
search and application of speech recognition for power dis- 2650–2654.
patching based on deep neural networks[J]. Hubei Electric Yang Deju, Ma Liangli, Tan Linshan, et al. End to end
Power, 2019, 43(3): 16–22. speech recognition based on gated convolutional networks