Search

CN-122020232-A - Method for early warning faults of brush wheel shaft group of automatic pressure-regulating brushing machine in PCB wet process

CN122020232ACN 122020232 ACN122020232 ACN 122020232ACN-122020232-A

Abstract

The embodiment of the application provides a fault early warning method for a brush wheel shaft group of an automatic voltage-regulating brush machine in a PCB wet process, which comprises the following steps of S1, collecting fault signals of the brush wheel shaft group, wherein the fault signals comprise vibration signals, three-phase current signals of a driving motor, bearing temperature signals and brush pressure signals, S2, extracting features of the fault signals to obtain high-dimensional feature vectors, S3, inputting the high-dimensional feature vectors into a pre-trained deep learning diagnosis model to perform feature level fusion and state recognition, and outputting a comprehensive score representing the health state of the brush wheel shaft group by the diagnosis model, and S4, judging whether to send out early warning signals according to whether the comprehensive score exceeds a preset dynamic early warning threshold or not, wherein the dynamic early warning threshold can be automatically adjusted according to the real-time operation condition of equipment. The method improves the detection sensitivity of weak fault characteristics, ensures the early warning accuracy under different working conditions, reduces the phenomena of false alarm and missing alarm, and prolongs the service life of equipment.

Inventors

  • ZHAN YANSHENG

Assignees

  • 昆山市钰富鑫机电设备有限公司

Dates

Publication Date
20260512
Application Date
20251207

Claims (8)

  1. 1. A fault early warning method for a brush wheel shaft group of an automatic voltage-regulating brush machine in a PCB wet process is characterized by comprising the following steps: S1, collecting fault signals of a brush wheel shaft group, wherein the fault signals comprise vibration signals, three-phase current signals of a driving motor, bearing temperature signals and brush pressure signals; S2, extracting features of the fault signals to obtain high-dimensional feature vectors; S3, inputting the high-dimensional feature vector into a pre-trained deep learning diagnosis model for feature level fusion and state identification, and outputting a comprehensive score representing the health state of the brush wheel shaft group by the diagnosis model; and S4, judging whether to send out an early warning signal according to whether the comprehensive score exceeds a preset dynamic early warning threshold value, wherein the dynamic early warning threshold value can be automatically adjusted according to the real-time operation condition of the equipment.
  2. 2. The method according to claim 1, wherein the step S2 comprises: S201, carrying out envelope demodulation analysis on the vibration signal, and extracting frequency band energy related to bearing fault characteristic frequency; s202, performing fast Fourier transform on the current signal, and analyzing the amplitude change of specific harmonic components of the current signal; S203, calculating the temperature rise gradient of the temperature signal; s204, calculating the dynamic fluctuation variance of the pressure signal.
  3. 3. The method according to claim 2, wherein the step S201 includes: determining an optimal frequency band containing the bearing fault characteristic frequency through a resonance frequency band screening technology; Performing envelope demodulation on the signal in the optimal frequency band by using Hilbert transform to obtain an envelope spectrum; And calculating the frequency band energy value corresponding to the bearing fault characteristic frequency in the envelope spectrum.
  4. 4. The method of claim 2, wherein the diagnostic model includes a one-dimensional hole convolution and a modified covariance mutual-attention mechanism, wherein the hole factors of the one-dimensional hole convolution are configured to be dynamically adjusted according to the input high-dimensional feature vectors, wherein the modified covariance mutual-attention mechanism is configured to calculate covariance matrices between different modal features, obtain minimized covariances of the different modal features, and calculate different feature weighting coefficients according to the minimized covariances.
  5. 5. The method of claim 4, wherein the method for dynamically adjusting the dynamic early warning threshold comprises: Establishing a threshold adjustment model taking the running speed of the grinder and the current conductivity of the grinding slurry as input variables; the dynamic early warning threshold is adjusted based on the operating speed and the slurry conductivity.
  6. 6. The method according to claim 5, wherein the step S3 includes: S301, performing fault mode matching based on the extracted features of the diagnostic model, and outputting most probable fault types and positioning information; s302, generating the comprehensive score and a diagnosis report containing recommended checking positions and recommended maintenance measures according to the identified fault mode.
  7. 7. The method of claim 6, wherein continuously optimizing the diagnostic model after the diagnostic model is placed in service using an on-line transfer learning strategy comprises: when the working condition of the brush wheel shaft set changes due to the fact that a new PCB board type is replaced by a production line, fine adjustment is carried out on the diagnosis model by utilizing sample data under a small amount of new working conditions; And adjusting part of parameters in the diagnosis model to enable the diagnosis model to be quickly adapted to new operation conditions, and maintaining high-precision diagnosis capability.
  8. 8. The method of claim 7, further comprising predictive maintenance support: performing time sequence trend analysis on the comprehensive scores to obtain analysis results; And predicting the residual service life of the key components of the brush wheel shaft group based on the result, and providing a time window suggestion.

Description

Method for early warning faults of brush wheel shaft group of automatic pressure-regulating brushing machine in PCB wet process Technical Field The application relates to the field of operation and maintenance of brush wheel shaft groups of a brushing machine, in particular to a fault early warning method of a brush wheel shaft group of a brushing machine with automatic pressure regulation in a PCB wet process. Background The PCB wet process is a key link in the manufacturing process of the printed circuit board, wherein an automatic pressure-regulating brush grinder is used as core equipment for surface treatment, and the stable operation of a brush wheel shaft group directly relates to the product quality and the production efficiency. In an actual production environment, the brush wheel shaft group is in a high-speed, heavy-load and wet-type corrosive working condition for a long time, and mechanical parts of the brush wheel shaft group are easy to be damaged gradually, if potential faults cannot be found in time, batch product defects can be caused, equipment interlocking damage can be caused, and serious economic loss is caused. In the prior art, although a scheme for attempting to introduce vibration or a temperature sensor to monitor the state exists, the scheme is limited to simple alarm with a single parameter threshold, and the deep fusion and intelligent analysis of multi-dimensional operation information are lacked, so that the early warning accuracy is insufficient, and the phenomena of false alarm and missing alarm coexist. Especially for an automatic pressure-regulating brush grinder, the failure of a shaft group is usually progressive failure under the interaction of mechanical, electrical and technological parameters, the effective prediction of the residual service life of the automatic pressure-regulating brush grinder is difficult to realize by the existing method, and prospective decision support cannot be provided for planned maintenance, so that the realization of the actual predictive maintenance is challenged. Therefore, an innovative method capable of deeply fusing multi-source information and having early accurate diagnosis and prediction capabilities is urgently needed in the field, so that the reliability of equipment is improved, and the production quality and efficiency are guaranteed. Disclosure of Invention The application provides a fault early warning method for a brush wheel shaft group of an automatic pressure-regulating brush machine in a PCB wet process, which can realize accurate early warning of early faults of the brush wheel shaft group through the synergistic effect of multi-source signal fusion and an intelligent diagnosis model, effectively overcomes the limitation of traditional single signal monitoring, remarkably improves the detection sensitivity of weak fault characteristics, ensures the early warning accuracy under different working conditions by means of a self-adaptive threshold mechanism, greatly reduces the false alarm and missing alarm phenomena, prolongs the service life of equipment, and reduces the unplanned downtime, thereby effectively reducing the comprehensive operation and maintenance cost while improving the quality consistency of products. In a first aspect, a method for early warning faults of a brush wheel shaft set of an automatic voltage-regulating brush machine in a PCB wet process is provided, the method comprising: S1, collecting fault signals of a brush wheel shaft group, wherein the fault signals comprise vibration signals, three-phase current signals of a driving motor, bearing temperature signals and brush pressure signals; S2, extracting features of the fault signals to obtain high-dimensional feature vectors; S3, inputting the high-dimensional feature vector into a pre-trained deep learning diagnosis model for feature level fusion and state identification, and outputting a comprehensive score representing the health state of the brush wheel shaft group by the diagnosis model; and S4, judging whether to send out an early warning signal according to whether the comprehensive score exceeds a preset dynamic early warning threshold value, wherein the dynamic early warning threshold value can be automatically adjusted according to the real-time operation condition of the equipment. It should be understood that the vibration signal, the three-phase current signal of the driving motor, the bearing temperature signal and the brushing pressure signal form a complementary system capable of three-dimensionally sensing the brushing wheel shaft group from four dimensions of mechanical dynamics, electrical load, thermodynamic state and process execution effect. The vibration signal can sensitively capture early mechanical impact of parts such as a bearing, a gear and the like, the current signal reflects torque fluctuation and abnormal load of the whole transmission chain from a motor to a brush wheel, the temperature signal directly reveals f