Mechanical fault intelligent diagnosis method for implicit excitation adversarial training under small sample

Abstract

The invention discloses a mechanical fault intelligent diagnosis method for implicit excitation adversarial training under a small sample. The method comprises the steps of dividing one-dimensional signals generated by mechanical equipment under different working conditions into a training set and a test set; constructing an encoder model, a generator model and a discriminator model, and traininga set training model by generating an adversarial training mechanism and a mutual information maximization and feature matching strategy; and inputting the training set into a trained encoder to obtain a corresponding feature code, then constructing and training an intelligent diagnosis model by using the feature code, finally using the model for fault diagnosis of the test set, and evaluating a result. Through a generative adversarial training mechanism and a mutual information maximization and feature matching strategy, under the condition of small samples, information association of the samples and feature codes is established and strengthened, the most essential category feature information of data is mined and used for training of an intelligent diagnosis model, and the accuracy of diagnosis is improved. The generalization ability, the fault diagnosis accuracy and the fault diagnosis stability of the model can be effectively improved.

Description

technical field [0001] The invention relates to a mechanical equipment failure intelligent diagnosis technology, in particular to a mechanical failure intelligent diagnosis method for implicit incentive confrontation training under small samples. Background technique [0002] Fault diagnosis of mechanical equipment is of great significance to ensure the safe and economic operation of equipment and the safety of people's lives and property. Signal feature extraction technology is an important means to realize fault diagnosis. For mechanical equipment that has been working under complex and severe working conditions such as high temperature and variable load for a long time, it is difficult to establish a mechanical physical model that matches the actual working conditions with a fault diagnosis method based on signal processing technology, and it is highly dependent on expert experience and prior knowledge. Feature extraction and manual selection of knowledge limit its accura...

AI Technical Summary

Problems solved by technology

For mechanical equipment that has been working under complex and severe working conditions such as high temperature and variable load for a long time, it is difficult to establish a mechanical physical model that matches the actual working conditions with a fault diagnosis method based on signal processing technology, and it is highly dependent on expert experience and prior knowledge. Feature extraction and manual selection based on knowledge limit its accuracy and applicability in high-dimensional, multi-source and noise-containing fault diagnosis tasks, and it is difficult to meet the needs of future mechanical equipment fault diagnosis

[0003] Since the neural network has adaptive learning ability and strong nonlinear mapping characteristics, it can perform adaptive feature extraction and pattern recognition on complex information, which provides a new technical means for fault diagnosis and status monitoring. The knowledge processing method has shown great application potential in the field of equipment fault diagnosis. However, data-driven intelligent fault diagnosis models often rely on a large number of high-quality original data samples and label information, but in the actual production and operation of mechanical equipment However, the acquisition of fault signals is subject to many restrictions such as equipment installation, and the amount of data available for model training is very limited, which will inevitably lead to insufficient model training and over-fitting problems, making the detection accuracy of the fault diagnosis model low and general. Poor chemical ability

Images