Rotor Assembly Optimization via Deep Learning Vibration Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for controlling and assembling high-speed rotary equipment, such as aircraft engines, fail to comprehensively consider key factors like coaxiality, unbalance, rigidity, and high-speed vibration response, leading to suboptimal performance and potential failures due to unaddressed rotation errors and geometrical parameters.
Innovation Solution
A deep learning regulation and control method that establishes models for coaxiality, unbalance, and vibration amplitude, using a Monte Carlo method to generate data and a BP neural network for prediction, optimizing the assembly of multiple stages of rotor/stator by accounting for various error factors and dynamic properties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If single objective optimization of coaxiality is performed solely, then the coaxiality model is simple, but the model does not consider rotation errors around X-axis and Y-axis, unbalance parameter, and rigidity parameter, leading to incomplete vibration control
Solution Approach 1:
The patent segments the vibration control problem into multiple independent optimization objectives: coaxiality optimization, rotation error correction (X and Y axes), unbalance parameter adjustment, and rigidity parameter optimization. Each segment is modeled and optimized separately but integrated together to achieve comprehensive vibration control, resolving the contradiction between model simplicity and control completeness.
Solution Approach 2:
The patent extends the optimization from a single coaxiality dimension to multiple dimensions by incorporating rotation errors around X-axis and Y-axis, unbalance parameters, and rigidity parameters. This multi-dimensional approach transforms the problem from scalar optimization to vector optimization, enabling comprehensive consideration of all key factors affecting vibration while maintaining systematic model structure.
2Measurement precision
If comprehensive models for coaxiality, unbalance, rigidity and vibration amplitude are established, then the measurement precision is improved, but the device complexity and calculation difficulty increase
Solution Approach 1:
The patent replaces complex mechanical measurement systems with a computational model-based approach. By establishing mathematical models for coaxiality, unbalance, rigidity, and vibration amplitude, the system uses calculation and simulation rather than purely mechanical measurement, thereby achieving high measurement precision while managing complexity through software rather than hardware complexity.
Solution Approach 2:
The patent transforms physical parameters (coaxiality, unbalance, rigidity) into measurable and adjustable model parameters. By changing the representation from physical dimensions to model parameters, the system achieves precise measurement and control while simplifying the complexity of direct physical measurement and adjustment mechanisms.
3Manufacturing precision
If multiple parameters (coaxiality, unbalance, rigidity, vibration amplitude) are optimized simultaneously, then the manufacturing precision is improved, but the assembly time and computational time increase
Solution Approach 1:
The patent performs preliminary optimization calculations and simulations before actual assembly. By pre-calculating the optimal parameters for coaxiality, unbalance, rigidity, and vibration amplitude, the system reduces the time required during actual assembly operations. The preliminary computational work prepares adjustment schemes in advance, minimizing on-site adjustment time and computational delays during assembly.
Data Source
AI summary
The present invention provides a deep learning regulation and control and assembly method and device for large-scale high-speed rotary equipment based on dynamic vibration response properties. The present invention starts from geometrical deviation of multiple stages of rotor/stator of an aircraft engine, amount of unbalance of rotor/stator, rigidity of rotor/stator and vibration amplitude of rotor/stator, considers the influence of the area of the assembly contact surface between two stages of rotors/stators, and sets the rotation speed of rotor/stator to be the climbing rotation speed to obtain vibration amplitude parameters. According to the calculation method of the coaxiality, amount of unbalance, rigidity and vibration amplitude of multiple stages of rotor/stator, an objective function taking assembly phases as variables is established, a Monte Carlo method is used to solve the objective function, and a probability density function is solved according to a drawn distribution function to obtain the probability relationship between the contact surface runout of the rotor/stator of the aircraft engine and the final coaxiality, amount of unbalance, rigidity and vibration amplitude of multiple stages of rotor/stator, thereby realizing assembly optimization and distribution of tolerances of multiple stages of rotor/stator.