Rotor Dynamic Balancing Based On Deep Learning

With the rapid development of China's industry,rotating machinery is widely used in related equipment such as aerospace,power generation,automobiles,ship propulsion and gas turbine machinery.Rotor unbalance,as a common fault in rotating machinery,has an extremely important impact on the healthy and stable operation of the rotor system.Can it effectively reduce the rotor unbalance and ensure the long-term safe and stable operation of the system has become a rotating machinery and equipment in various important fields An important guarantee for the key role.In the field of aero-engines,the higher thrust-to-weight ratio requirements make the rotor structure tend to be flexible,and the accompanying problem is that the vibration caused by the unbalanced amount continues to increase,endangering the safety and reliability of the engine.How to study the corresponding vibration characteristics in the complex closed structure of the bearing-rotor-casing,and accurately identify the unbalance of the rotor through the vibration measured at the bearing has become a problem of academic concern.Considering the above situation,this paper introduces a neural network into the process of identifying the rotor rotor imbalance for a certain type of aero-engine rotor imbalance,and studies the feasibility of neural network for rotor imbalance recognition.In this paper,a certain aero-engine model is established and different neural network training structures are established.First,the neural network is trained on the unbalance of a single rotor in the engine and the unbalance is identified;Second,the unevenness of the double rotor of the engine is carried out through an intermediate bearing measure the training of the neural network and the identification of the unbalance;Finally,calculate the bearing-to-bearing transfer function,train the neural network of the unbalance of the engine rotor with the casing and identify the unbalance.Through the artificial neural network training and network output analysis of the above content in this article,we can know that the use of artificial neural network for rotor imbalance identification is reasonable,and the use of simulated data for network training can achieve the identification of imbalance.