Home>Bearing knowledge>Hydraulic pump bearing fault diagnosis network law research

/*728*90 created on 2018/5/16*/ var cpro_id = "u3440131";

Hydraulic pump bearing fault diagnosis network law research

---

/*250*250 was created on 2017/12/25*/ var cpro_id = 'u3171089';

In the aviation industry; the operating function of the hydraulic system directly affects the safety of the aircraft and the life of the passenger; and the hydraulic pump is the power source of the hydraulic system; therefore, the monitoring of the condition of the hydraulic pump and the diagnosis of the disease are particularly important. The bearing fault is common in hydraulic pumps. One of the faulty forms; the additional oscillation caused by the bearing fault is related to the inherent oscillation of the hydraulic pump; therefore, it is difficult to separate the fault information from the signal. So far; the diagnosis of the hydraulic pump bearing fault is still short Very useful method. This paper proposes the characteristics of the frequency domain and the scrambling frequency domain. The pod 诮饩鲋岢 卣魈 卣魈 训 训 训 训 训 獠 . . . . . . . . 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络 网络
1. The extraction of the characteristics of the hydraulic pump bearing fault is related to the mechanical system; if there is a fault, it will definitely lead to additional oscillation of the system. The oscillating signal is a dynamic signal; it contains rich information; it is suitable for diagnosis of the disease. However, if it is an additional oscillating signal Because the inherent signal or external disturbances engulf the fault signal is very large; then how to extract the useful signal from the oscillating signal is very important.
According to the theory of conflict; when the inner ring of the bearing movable surface, the outer ring raceway and the roller show a damage; the smoothness of the raceway is damaged; every time the roller rolls over the damage point; an oscillation will occur. The part is a rigid body; the effect of the touch deformation is not considered; the roller is purely rolled along the raceway.
Hilbert is used to change the envelope of the time domain signal used in signal analysis; to reach the power spectrum and then to the outstanding information. The signal is: the best envelope. The cepstrum envelope model is the signal obtained from the sensor. Perform the cepstrum analysis; then carry out the enveloping of the cepstrum signal. The fierce 赝 赝 赝 收 收 收 收 收 收 收 收 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉 〉
2, the integration of BP network for the diagnosis of the principle of the diagnosis of the neural network arrangement is determined by the classification of the problem of the classification of the problem. Because of the complexity of the diagnosis system; the application of neural network in the depiction of the diagnosis system; will be the large planning nerve Network arrangement and learning problems. In order to reduce the complexity of homework; reduce the learning time of the network; this paper divides the common sense of diagnosis into several logically independent sub-sets; each sub-set is subdivided into several rule subsets; A subset of rules is used to arrange the network. Each rule subset is a logically independent sub-network mapping; the communication between the rules subsets; the sub-network weight matrix shows that each sub-network is independently used to use BP learning algorithms to separate Learning exercises. Because the differentiated sub-network is much smaller than the original network planning and the problem is somewhat complicated; then the practice time is greatly reduced. The information processing of the hydraulic pump bearing fault diagnosis using the integrated BP network can be derived from the neurons. Nonlinear mechanism characteristics and BP algorithm.
3. Research on the robustness of neural networks The robustness of neural networks refers to the fault tolerance of neural networks. It is well known that human brains are fault-tolerant; the damage of individual neurons in the brain does not make its overall function severe. Downgrade; this is because every concept in the brain is not only stored in one neuron; it is distributed among many neurons and their connections. The brain can learn again; the common sense that is forgotten by the damage of some neurons De novo expression in the remaining neurons. Because the neural network is an imitation of the biological neural network; so the biggest feature of the neural network is the "associative recall" function; that is, the neural network can be combined by the common sense of the past; Or some information is uncertain; use the remaining characteristic information to make a correct diagnosis.
Table 2 shows the success rate of correct diagnosis and identification of some of the six characteristic information of the bearing is incorrect or uncertain.
Table 1 Neural network robustness statistics table input characteristics uncertainty element diagnosis success rate a characteristic parameter uncertainty 100%
Two characteristic parameters are uncertain 94%
Three characteristic parameters are undefined 76%
Four characteristic parameters are uncertain 70%
Five characteristic parameters are undefined 20%
Six characteristic parameters are undefined 8%
It can be seen from Table 1 that the use of integrated neural network for diagnosis of defects can result in an appropriate high success rate (76% to 100%) in the case of losing a lot of information (nearly half of the characteristic parameters are uncertain). Neural networks have a strong talent.

---

Recommend to friends comments close window

Bearing related knowledge

Bearing scale selection of bearing device method and disassembly Analysis of NSK Skills: Classification and Function of Pre-tightening of Bearings for Cylindrical Bore Bearings

This article links to http://
Please indicate the bearing network http://

Previous:Analysis of the skills of the launching machine sliding bearing (3)
Next: The cause of damage to the roller bearing