基于量子加权长短时记忆神经网络的状态退化趋势预测
|
摘要
提出基于量子加权长短时记忆神经网络(QWLSTMNN)的旋转机械状态退化趋势预测方法。首先采用小波包能量熵误差构建状态退化特征集,然后将该特征集输入QWLSTMNN以完成旋转机械状态退化趋势预测。在QWLSTMNN中,将输入层权值量子位扩展到隐层以获取额外的梯度信息;利用隐层权值量子位的反馈信息以获取输入序列的全部记忆,改善了原长短时记忆神经网络(LSTMNN)的非线性逼近能力和泛化性能,使所提出的状态退化趋势预测方法具有较高的预测精度;另外,采用新型的基于量子相移门和量子梯度下降法的学习算法以实现QWLSTMNN的网络量子参数(即权值量子位和活性值量子位)的快速更新,提高了网络收敛速度,使所提出的预测方法具有较高的计算效率。滚动轴承状态退化趋势预测实例验证了该方法的有效性。
A novel state degradation trend prediction method of rotating machinery is proposed based on quantum weighted long short-term memory neural network( QWLSTMNN). Firstly,the state degradation feature sets are constructed by using wavelet packet energy entropy error. Then these feature sets are input to QWLSTMNN to achievethe state degradation trend prediction of rotating machinery. In the proposed QWLSTMNN,the input-layer weight qubits are extended into the hidden-layer to gain extra gradient information,and the whole memory of input sequences can be obtained by using the feedback information of hidden-layer weight qubits in the hidden-layer.Therefore,the nonlinear approximation capability and generalization capacity oforiginallong short-term memory neural network( LSTMNN) is improved. And the higher state degradation trend prediction accuracy of the proposed method based on QWLSTMNN can be realized. Besides,a new learning algorithm based on quantum phase-shift gate and quantum gradient descent is presented to quickly update such network quantum parameters as weight qubits and activity qubits. Thus,the network convergence speed is improved,accordingly,the higher computational efficiency can be obtained for the proposed trend prediction method. Case study of state degradation trend prediction for rolling bearing demonstrates the effectiveness of the proposed method.
A novel state degradation trend prediction method of rotating machinery is proposed based on quantum weighted long short-term memory neural network( QWLSTMNN). Firstly,the state degradation feature sets are constructed by using wavelet packet energy entropy error. Then these feature sets are input to QWLSTMNN to achievethe state degradation trend prediction of rotating machinery. In the proposed QWLSTMNN,the input-layer weight qubits are extended into the hidden-layer to gain extra gradient information,and the whole memory of input sequences can be obtained by using the feedback information of hidden-layer weight qubits in the hidden-layer.Therefore,the nonlinear approximation capability and generalization capacity oforiginallong short-term memory neural network( LSTMNN) is improved. And the higher state degradation trend prediction accuracy of the proposed method based on QWLSTMNN can be realized. Besides,a new learning algorithm based on quantum phase-shift gate and quantum gradient descent is presented to quickly update such network quantum parameters as weight qubits and activity qubits. Thus,the network convergence speed is improved,accordingly,the higher computational efficiency can be obtained for the proposed trend prediction method. Case study of state degradation trend prediction for rolling bearing demonstrates the effectiveness of the proposed method.
引文
[1]景博,汤巍,黄以锋.故障预测与健康管理系统相关标准综述[J].电子测量与仪器学报,2014,28(12):1301-1307.JING B,TANG W,HUANG Y F.Summary of PHM system standards[J].Journal of Electronic Measurement and Instrumentation,2014,28(12):1301-1307.
[2]张焱,汤宝平,熊鹏.多尺度变异粒子群优化MKLSSVM的轴承寿命预测[J].仪器仪表学报,2016,37(11):2489-2496.ZHANG Y,TANG B P,XIONG P.Rolling element bearing life prediction based on multi-scale mutation particle swarm optimized multi-kernel least square support vector machine[J].Chinese Journal of Scientific Instrument,2016,37(11):2489-2496.
[3]HERP J,AMEZANI R,MPHAMMAD H,et al.Bayesian state prediction of wind turbine bearing failure[J].Renewable Energy,2018,116(2):164-172.
[4]SOUALHI A,RAZIK H,CLERC G.et al.Prognosis of bearing failures using hidden Markov models and the adaptive neuro-fuzzy inference system[J].Transactions on Industrial Electronics,2014,61(6):2864-2874.
[5]GUO L,LI N P,JIA F,LEIY G.et al.A recurrent neural network based health indicator for remaining useful life prediction of bearings[J].Neurocomputing,2017,240(3):98-109.
[6]HOCHREITER S,SCHMIDHUBER J.Long short-term memory[J].Neural Computation,1997,9(8):1735-1780.
[7]MA X L,TAO Z M,WANG Y H,et al.Long short-term memory neural network for traffic speed prediction using remote microwave sensor data[J].Transportation Research Part C:Emerging Technologies,2015,54(3):187-197.
[8]ESKANDARIA E,AHMADI A,GOMAR S.Effect of spike-timing-dependent plasticity on neural assembly computting[J].Neurocomputing,2016,191(3):107-116.
[9]YUKALOV VI,YUKALOV EP,SORNETTE D.Information processing by networks of quantum decisionmakers[J].Physica A:Statistical Mechanics and Its Applications,2018,492(4):747-766.
[10]KAK S C.On quantum neuralcomputing[J].Information Sciences,1995,83(3):143-160.
[11]CAO H X,CAO F L,WANG D H.Quantum artificial neural networks with applications[J].Information Sciences,2015,290(1):1-6.
[12]SILVA A J D,OLIVEIRA W R D.Comments on“quantum artificial neural networks with applications”[J].Information Sciences,2016,370-371(22):120-122.
[13]李鹏华,柴毅,熊庆宇.量子门Elman神经网络及其梯度扩展的量子反向传播学习算法[J].自动化学报,2013,39(9):1511-1522.LI P H,CHAI Y,XIONG Q Y.Quantum gate Elman neural network and its quantized extended gradient backpropagation training algorithm[J].Acta Automatica Sinica,2013,39(9):1511-1522.
[14]HU G S,ZHU F F,REN Z.Power quality disturbance identification using wavelet packet energy entropy and weighted support vector machines[J].Expert Systems with Applications,2008,35(1-2):143-149.
[15]LEE J,QIU H,YU G.Rexnord technical services,“bearing data set”,IMS,University of Cincinnati,NASA ames prognostics data repository[Z].NASA Ames,2007.
[2]张焱,汤宝平,熊鹏.多尺度变异粒子群优化MKLSSVM的轴承寿命预测[J].仪器仪表学报,2016,37(11):2489-2496.ZHANG Y,TANG B P,XIONG P.Rolling element bearing life prediction based on multi-scale mutation particle swarm optimized multi-kernel least square support vector machine[J].Chinese Journal of Scientific Instrument,2016,37(11):2489-2496.
[3]HERP J,AMEZANI R,MPHAMMAD H,et al.Bayesian state prediction of wind turbine bearing failure[J].Renewable Energy,2018,116(2):164-172.
[4]SOUALHI A,RAZIK H,CLERC G.et al.Prognosis of bearing failures using hidden Markov models and the adaptive neuro-fuzzy inference system[J].Transactions on Industrial Electronics,2014,61(6):2864-2874.
[5]GUO L,LI N P,JIA F,LEIY G.et al.A recurrent neural network based health indicator for remaining useful life prediction of bearings[J].Neurocomputing,2017,240(3):98-109.
[6]HOCHREITER S,SCHMIDHUBER J.Long short-term memory[J].Neural Computation,1997,9(8):1735-1780.
[7]MA X L,TAO Z M,WANG Y H,et al.Long short-term memory neural network for traffic speed prediction using remote microwave sensor data[J].Transportation Research Part C:Emerging Technologies,2015,54(3):187-197.
[8]ESKANDARIA E,AHMADI A,GOMAR S.Effect of spike-timing-dependent plasticity on neural assembly computting[J].Neurocomputing,2016,191(3):107-116.
[9]YUKALOV VI,YUKALOV EP,SORNETTE D.Information processing by networks of quantum decisionmakers[J].Physica A:Statistical Mechanics and Its Applications,2018,492(4):747-766.
[10]KAK S C.On quantum neuralcomputing[J].Information Sciences,1995,83(3):143-160.
[11]CAO H X,CAO F L,WANG D H.Quantum artificial neural networks with applications[J].Information Sciences,2015,290(1):1-6.
[12]SILVA A J D,OLIVEIRA W R D.Comments on“quantum artificial neural networks with applications”[J].Information Sciences,2016,370-371(22):120-122.
[13]李鹏华,柴毅,熊庆宇.量子门Elman神经网络及其梯度扩展的量子反向传播学习算法[J].自动化学报,2013,39(9):1511-1522.LI P H,CHAI Y,XIONG Q Y.Quantum gate Elman neural network and its quantized extended gradient backpropagation training algorithm[J].Acta Automatica Sinica,2013,39(9):1511-1522.
[14]HU G S,ZHU F F,REN Z.Power quality disturbance identification using wavelet packet energy entropy and weighted support vector machines[J].Expert Systems with Applications,2008,35(1-2):143-149.
[15]LEE J,QIU H,YU G.Rexnord technical services,“bearing data set”,IMS,University of Cincinnati,NASA ames prognostics data repository[Z].NASA Ames,2007.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |