基于特征提取和BP神经网络的音乐喷泉控制系统研究  

Research on Music Fountain Control System Based on Feature Extraction and BP Neural Network

作　　者：貟婧[1] YUN Jing(Xianyang Normal University,Xianyang Shaanxi 712000,China)

机构地区：[1]咸阳师范学院,陕西咸阳712000

出　　处：《自动化与仪器仪表》2024年第7期313-316,共4页Automation & Instrumentation

基　　金：校级《咸阳师范学院学术骨干》(XSYXSGG 2D2112)。

摘　　要：当前的音乐喷泉控制系统存在音乐特征识别单一,控制模式消耗大等问题,因此,研究利用特征提取与反向传播网络设计了具备智能化特性的音乐喷泉控制系统,并对其有效性进行了验证。实验结果表明,音乐特征提取中,以乐曲1为例,音乐的节拍维持在0.5~1之间;节奏变动强度维持在1.26~2.92之间;大三度维持在0.03~0.13之间,特征向量基本表达出来。另外,反向传播网络验证中,其对于平静情绪的识别率最高,为97.34%;将其与其他算法对比来看,对比算法A、B、C、D的最高值分别为83.35%、91.25%、87.22%和95.89%,均低于研究算法。在此基础上,研究设计的系统模型验证中较好地依据音乐的变化而变化。综合来看,设计的音乐喷泉控制系统具备实用性,能有效应用在实际的音乐喷泉中。The current music fountain control system has problems such as single music feature recognition and high control mode consumption.Therefore,a music fountain control system with intelligent characteristics was designed using feature extraction and backpropagation networks,and its effectiveness was verified.The experimental results show that in music feature extraction,taking Song 1 as an example,the beat of the music is maintained between 0.5 and 1;The intensity of rhythm variation is maintained between 1.26 and 2.92;The third degree is maintained between 0.03 and 0.13,and the feature vectors are basically expressed.In addition,in backpropagation network validation,its recognition rate for calm emotions is the highest,at 97.34%;Compared with other algorithms,the highest values of algorithms A,B,C,and D are 83.35%,91.25%,87.22%,and 95.89%,respectively,which are lower than the research algorithms.On this basis,the system model validation designed in the study is well adapted to changes in music.Overall,the designed music fountain control system has practicality and can be effectively applied in practical music fountains.

关 键 词：特征 BP网络 音乐喷泉控制系统 情绪 

分 类 号：TU986.43[建筑科学—城市规划与设计] TP273[自动化与计算机技术—检测技术与自动化装置]