| 摘 要: 动态效率优化神经网络算法(DEONN)的提出旨在提高振动能量收集设备的能量转换效率。DEONN利用深度学习技术,结合多层感知器架构,优化了发电机的关键组件(电枢、换向器、刷子、磁场及外壳)参数,提升了能量转换效率。开展实验实现该算法预测不同运行条件下的电机效率,具体为通过建立一个包含隐藏层的神经网络,输入转速、负载电阻和线圈数等特征,预测不同工况下的电机效率。实验结果表明,实测效率与预测效率具有高度一致性,预测效率为88.5%,验证了DEONN在预测发电机的转速、负载电阻和线圈数等关键性能参数方面的有效性。 |
| 关键词: 动态效率优化;神经网络;能量转换;发电机参数 |
|
中图分类号: TP391.41
文献标识码: A
|
| 基金项目: 国家自然科学基金(61871132). |
|
| Application of Dynamic Efficiency Optimization Neural Network Algorithm in Vibration Energy Collection |
|
QI Bin, ZHANG Qingbo
|
(School of Electronic Inf ormation, Zhejiang Business Technology Institue, Ningbo 315010, China)
qibingood@gmail.com; 305372820@qq.com
|
| Abstract: The proposal of Dynamic Efficiency Optimization Neural Network ( DEONN) algorithm aims to enhance energy conversion efficiency of vibration energy collection equipment. With deep learning technology and a multi-layer perceptron architecture, DEONN optimizes parameters of key components of the generator (armature, commutator, brush, magnetic field, and casing), improving energy conversion efficiency. Experiments are conducted to achieve the algorithm's prediction of motor efficiency under different operating conditions. Specifically, it is to establish a neural network containing hidden layers and input features such as rotational speed, load resistance, and coil count, to predict motor efficiency under different operating conditions. The experiment results show a high consistency between the actual efficiency and the predicted efficiency, with a predicted efficiency of 88.5% . This validates its effectiveness in predicting key performance parameters such as rotational speed, load resistance, and coil count. |
| Keywords: dynamic efficiency optimization; neural network; energy conversion; generator parameters |
