Core Tip: In the aerospace industry, the performance of the hydraulic system directly affects the safety of the aircraft and the life of the passenger. The hydraulic pump is the power source of the hydraulic system, so the status monitoring of the hydraulic pump
In the aerospace industry, the performance of the hydraulic system directly affects the safety of the aircraft and the life of the passenger. The hydraulic pump is the power source of the hydraulic system, so it is especially important for the status monitoring and fault diagnosis of the hydraulic pump.
Bearing failure is one of the common failure modes of hydraulic pumps. Because the additional vibration caused by the bearing failure is weak relative to the natural vibration of the hydraulic pump, it is difficult to separate the fault information from the signal. So far, there is a lack of effective methods for troubleshooting faults in hydraulic pump bearings. This paper proposes feature extraction in frequency domain and frequency domain, aiming at solving the problem of bearing feature extraction and using integrated BP network to solve multi-fault diagnosis and identification and robustness problems.
1. Feature extraction of hydraulic pump bearing faults
For mechanical systems, if there is a fault, it will definitely cause additional vibration of the system. The vibration signal is a dynamic signal, which contains a wealth of information and is ideal for troubleshooting. However, if the additional vibration signal is overwhelmed by the inherent signal or external interference to the fault signal, how to extract the useful signal from the vibration signal is very important.
According to the tribology theory, when a damage occurs on the inner ring, the outer ring raceway and the roller of the bearing flow surface, the surface of the raceway is smoothly damaged, and whenever the roller rolls over the damage point, a vibration is generated. Assuming that the
bearing parts are rigid, regardless of the influence of contact deformation, the rollers are pure rolls along the raceway.
The Hilbert transform is used in signal analysis to find the envelope of the time domain signal to smooth the power spectrum to highlight fault information. Define the signal: the best envelope. The cepstrum envelope model essentially performs cepstrum analysis on the signal obtained from the sensor, and then extracts the cepstrum signal from the sensor, which double-emphasizes the fault information and provides the extraction of fault features with small signal-to-noise ratio. in accordance with.
2. Principle of integrated BP network for fault diagnosis
The organizational structure of the neural network is determined by the domain characteristics of the problem being solved. Due to the complexity of the fault diagnosis system, the application of neural networks to the design of obstacle diagnostic systems will be a problem of organization and learning of large-scale neural networks. In order to reduce the complexity of the work and reduce the learning time of the network, this paper decomposes the fault diagnosis knowledge set into several logically independent sub-sets, each sub-set is decomposed into several rule subsets, and then the network is organized according to the rule subset. Each rule subset is a logically independent mapping of sub-networks, and the associations between the subsets of rules are represented by the weight matrix of the sub-network. Each sub-network independently uses the BP learning algorithm to perform learning training separately. Since the decomposed sub-network is much smaller than the original network and the problem is localized, the training time is greatly reduced. The information processing capability of hydraulic pump bearing fault diagnosis using integrated BP network is derived from the nonlinear mechanism characteristics of neurons and BP algorithm.
3. Research on robustness of neural networks
The robustness of a neural network refers to the fault tolerance of a neural network to faults. As we all know, the human brain is fault-tolerant, and the damage of individual neurons in the brain does not seriously degrade its overall performance. This is because every concept in the brain is not only stored in one neuron but scattered in many nerves. Yuan and its connections. The brain can re-express the knowledge that is forgotten by the damage of a part of the neurons in the remaining neurons by learning again. Since the neural network is a simulation of the biological neuron network, the biggest feature of the neural network is the "associative memory" function, that is, the neural network can be combined with previous knowledge, in the case of partial information loss or partial information uncertainty. The remaining feature information makes a correct diagnosis.
