Degradation Trend Consistency Study And RUL Prediction Of Relay Based On SIMO Millimeter-wave Radar

Relays as a kernel component are widely used in various control systems.Its reliability has a non-negligible effect on the efficiency and performance of the entire control system.Introducing the remaining useful life(RUL)estimation theory to realize the predictable maintenance,the Relay can be replaced before it fails to improve system performance and reduce maintenance costs.In order to solve the current problems and needs of relay RUL,this paper proposes a non-contact,non-destructive detection of relay contact displacement based on millimeter-wave radar.The auxiliary particle filter(APF)technology combined with the degradation model and convolutional neural network(CNN)model without a pooling layer is used to achieve relay RUL prediction.The main research contents of the paper are:First,single input multiple output(SIMO)antenna technology,multiple signal classification(MUSIC)algorithm,and adaptive beamforming technology are introduced into millimeter-wave radar to distinguish and detect the contact displacement of multiple relays.Second,the distortion degree(peak to peak value and skewness)of the relay contact displacement waveform is extracted from the raw data collected as a performance degradation indicator.A feature extraction method based on fourth-order polynomial fitting is used to reflect the consistent degradation trend of different relays.Then,two degradation models(exponential and stochastic process models)are defined.The APF technology is used to update the model parameters and degradation status according to the latest measured values,and the RUL prediction of the relay is preliminarily realized.In addition,considering the influence of ambient temperature and degradation trend consistency extraction on relay RUL prediction performance,a relay multi-dimensional degradation feature dataset is constructed.Finally,to retain more original degradation characteristics of relays,a CNN model without a pooling layer is built to realize the RUL prediction of relays further.This paper uses the relay control and radar measurement modules to build the relay accelerated degradation experimental platform,verifying the effectiveness of non-contact nondestructive testing of multi-relay contact displacement based on millimeter-wave radar.The effects of ambient temperature and degradation trend consistency extraction on relay RUL prediction performance are discussed and analyzed by designing several groups of comparative experiments.The experimental results show that the millimeter-wave radar can accurately detect the relay contact displacement in real-time.Considering the ambient temperature and introducing the degradation trend consistency extraction can improve the RUL prediction performance and accuracy to a certain extent.