CN-121997150-A - Intelligent helicopter rotor latent fault diagnosis method and system based on acoustic signals

CN121997150ACN 121997150 ACN121997150 ACN 121997150ACN-121997150-A

Abstract

The invention belongs to the technical field of helicopter fault diagnosis, and relates to an intelligent diagnosis method and system for potential faults of a helicopter rotor wing based on acoustic signals. The method comprises the steps of implanting characteristic parameters into typical fault characteristics to generate rotor wing models in various fault states, solving flow field parameters to obtain dynamic pressure load on the surface of a rotor wing, collecting sequence signals of sound pressure changing along with time through a virtual acoustic sensor array, extracting the fault characteristics, constructing a fault classification and discrimination model based on machine learning, and outputting the fault states and the fault types of the rotor wing. The method can comprehensively reflect the influence of rotor faults on acoustic radiation characteristics, effectively solve the problem that helicopter rotor fault samples are difficult to obtain through high-fidelity geometric modeling and fault implantation and CFD and CSD coupling strategies, and construct a fault diagnosis model by extracting multi-dimensional fault characteristics and combining a machine learning algorithm, has high fault diagnosis accuracy, and can realize automatic identification and type positioning of faults.

Inventors

YAO DAN
ZHANG ZHE
LIU ZEZHENG
PANG JIE
XIE JIAYU
NIE RUI

Assignees

中国民用航空飞行学院

Dates

Publication Date: 20260508
Application Date: 20260410

Claims (10)

1. The intelligent diagnosis method for potential faults of the helicopter rotor wing based on the acoustic signals is characterized by comprising the following steps: Constructing a three-dimensional geometric model of a helicopter rotor wing, implanting typical fault characteristics into characteristic parameters of the three-dimensional geometric model, and generating rotor wing models in various fault states; Simulating a dynamic rotation process of a rotor wing model in a fault state by adopting an overlapped grid method, and solving flow field parameters through a CFD and CSD coupling strategy to obtain a dynamic pressure load on the surface of a rotor wing blade; Mapping dynamic pressure load into an acoustic field, calculating far-field radiation noise based on an FW-H acoustic analogy equation, and acquiring a sequence signal of sound pressure changing along with time through a virtual acoustic sensor array; preprocessing the sequence signals, extracting fault characteristics by adopting a signal processing algorithm to obtain an acoustic characteristic matrix reflecting the running state of the rotor wing, and recording corresponding fault labels; Constructing a fault classification discrimination model based on machine learning, training the fault classification discrimination model by utilizing an acoustic feature matrix and a corresponding fault label, and establishing a mapping relation between acoustic features and a fault mode; and collecting real-time acoustic signals of the helicopter rotor to be diagnosed, inputting a trained fault classification and discrimination model, and outputting the fault state and the fault type of the rotor.
2. The intelligent diagnosis method for potential faults of a helicopter rotor based on acoustic signals according to claim 1, wherein the characteristic parameters comprise blade properties, blade rotation speed, blade diameter, blade chord length, blade base blade passing frequency, blade number and blade torsion distribution.
3. The intelligent diagnosis method for potential faults of a helicopter rotor based on acoustic signals according to claim 1 is characterized in that typical fault characteristics comprise blade cracks, weight unbalance and rotor blade icing, wherein the blade cracks are simulated by arranging notches with preset depth and preset width at the span of a blade, material parameter correction or rigidity degradation modeling is carried out on a notch area in a CSD model and used for reflecting structural dynamics effects of the blade cracks, and the weight unbalance is simulated by adding a mass block with mass of a set size to the tip of the blade.
4. The intelligent diagnosis method for potential faults of a helicopter rotor based on acoustic signals according to claim 1, wherein the flow field parameters are solved through a CFD and CSD coupling strategy to obtain dynamic pressure loads of the rotor blade surface, comprising: establishing a CFD solving model and a CSD calculating model; Calculating a CSD calculation model and obtaining initial deformation data of the rotor blade; inputting initial deformation data into a CFD solving model, and updating a blade grid; calculating flow field parameters by using the CFD solving model to obtain dynamic pressure load and returning to the CSD calculating model; And (5) iterating the calculation until the rotor blade structure deformation and the dynamic pressure load are converged.
5. The intelligent diagnosis method for potential faults of the helicopter rotor based on acoustic signals according to claim 1 is characterized in that a dynamic rotation process of a rotor model in a fault state is simulated by adopting an overlapped grid method, the method comprises the steps that the overlapped grid comprises a background grid and a component grid, the background grid is a fixed grid covering the whole calculation domain, the component grid is a movable grid wrapping rotor blades, the movable grid rotates together with the blades, the overlapped area of the component grid and the background grid is set to be a set ratio of the rotor diameter, and the grid interpolation is carried out in a linear interpolation mode.
6. The intelligent diagnosis method for potential faults of helicopter rotor wing based on acoustic signals as claimed in claim 4, wherein when flow field parameters are solved through CFD and CSD coupling strategies, a LES model is adopted to solve turbulence structures in core fluid, a RANS model is adopted to cover wall boundary layers, a control equation of the RANS model is in a differential form, a set time step meets the Nyquist sampling theorem and a dynamic grid time scale relation, and In order to achieve a density of the particles, In order to be able to achieve an average speed, In order to be able to achieve an average pressure, As a unit tensor of the number of units, As a result of the resultant of the volumetric forces, As a tensor for the average viscous stress, In order to be the reynolds stress tensor, Is the total energy of the filtering per unit mass, For filtering the heat flux, the RANS model control equation differential form is expressed as: ; ; 。
7. The intelligent diagnosis method for potential faults of the helicopter rotor based on the acoustic signals, which is disclosed by claim 1, is characterized in that the preprocessing of the sequence signals comprises the steps of denoising the sequence signals by adopting a wavelet soft threshold denoising method, normalizing the denoised sequence signals and segmenting the acoustic signals according to time.
8. The intelligent diagnosis method for potential faults of the helicopter rotor based on the acoustic signals is characterized in that a signal processing algorithm is adopted to extract fault characteristics to obtain an acoustic characteristic matrix reflecting the running state of the rotor, the method comprises the steps of converting a time domain signal into a frequency domain signal by utilizing fast Fourier transformation, analyzing the passing frequency and energy distribution of the rotor, filtering non-characteristic frequencies by utilizing wavelet packet transformation, extracting energy, entropy and peak factors of each frequency band as fault sensitive characteristic quantities, and converting sound wave data into the fault characteristic matrix.
9. The intelligent diagnosis method for potential faults of the helicopter rotor based on acoustic signals according to claim 1, wherein the machine learning model comprises one or more of a support vector machine, a random forest, a convolutional neural network and a cyclic neural network, and parameters of the machine learning model are adjusted through an Adam or SGD optimization algorithm.
10. The helicopter rotor latent fault intelligent diagnosis system based on the acoustic signals is characterized by comprising a geometric modeling and fault implantation unit, a coupling solving unit, an acoustic analysis unit, a feature extraction unit, a model construction and training unit and a fault diagnosis unit; The geometrical modeling and fault implantation unit is used for constructing a three-dimensional geometrical model of the helicopter rotor, implanting typical fault characteristics into characteristic parameters of the three-dimensional geometrical model, and generating rotor models in various fault states; the coupling solving unit is used for simulating the dynamic rotation process of the rotor wing model in the fault state by adopting an overlapped grid method, and solving flow field parameters through a CFD and CSD coupling strategy to obtain the dynamic pressure load on the surface of the rotor wing blade; The acoustic analysis unit is used for mapping dynamic pressure load into an acoustic field, calculating far-field radiation noise based on FW-H acoustic analogy equation, and acquiring a sequence signal of sound pressure changing along with time through the virtual acoustic sensor array; The feature extraction unit is used for preprocessing the sequence signals, extracting fault features by adopting a signal processing algorithm, obtaining an acoustic feature matrix reflecting the running state of the rotor wing, and recording corresponding fault labels; The model construction and training unit is used for constructing a fault classification discrimination model based on machine learning, training the fault classification discrimination model by utilizing the acoustic feature matrix and the corresponding fault labels, and establishing a mapping relation between acoustic features and fault modes; the fault diagnosis unit is used for collecting real-time acoustic signals of the helicopter rotor to be diagnosed, inputting a trained fault classification and discrimination model, and outputting the fault state and the fault type of the rotor.

Description

Intelligent helicopter rotor latent fault diagnosis method and system based on acoustic signals Technical Field The invention belongs to the technical field of helicopter fault diagnosis, and particularly relates to an intelligent diagnosis method and system for potential faults of a helicopter rotor wing based on acoustic signals. Background The helicopter is used as core equipment in the field of general aviation, and plays an irreplaceable role in the fields of emergency rescue, medical transportation, police patrol, agriculture and forestry operation, offshore petroleum service and the like by virtue of the unique vertical lifting and hovering capability. However, the helicopter is usually in a very bad running environment (such as strong wind, sand dust, high salt mist at sea, etc.), and has very complex mechanical structure, especially, when the rotor system is used as a lifting surface and a control surface of the helicopter, huge alternating pneumatic load, centrifugal force and coriolis force are born, and faults such as fatigue crack, unbalanced mass, loose connection, etc. are very easy to occur. Helicopter accident rates are statistically much higher than fixed wing aircraft, and rotor system failure is one of the main causes of catastrophic accidents. Traditional helicopter maintenance modes rely mainly on timing maintenance and post-maintenance. The timing maintenance has the contradiction of over maintenance or under maintenance, and the post maintenance often means that the accident has occurred. In order to meet the high-strength and high-reliability operation requirements of modern aviation, the transition to on-demand maintenance and predictive maintenance is needed. Currently, helicopter health and usage monitoring systems (HEALTH AND Usage Monitoring System, abbreviated as HUMS) are widely used, and mainly rely on vibration signal analysis. Although vibration diagnostic techniques are relatively mature, significant limitations are exposed in practical applications: (1) By adopting a contact measurement mode, the vibration sensor (accelerometer) must be physically installed on the measured component, and for a rotor blade rotating at a high speed, the installation of the sensor not only requires a complex slip ring primer or a wireless transmission system, but also can change the mass distribution and the aerodynamic shape of the blade, and even the sensor itself falls off, so that potential safety hazards are caused. (2) Vibration signals generated by faults can be received by a body sensor only through the transmission of multi-layer structures such as a blade, a hub, a main speed reducer, a body and the like, the paths are complex, high-frequency characteristic attenuation is serious, the interference of other vibration sources such as an engine and a transmission system is extremely easy, the signal-to-noise ratio is low, and the characteristic extraction is difficult. (3) The vibration sensor can only reflect the local state near the mounting point, has insufficient sensitivity to early local faults (such as micro cracks) of the large flexible rotor blade, and has limited space coverage capability; (4) The construction of the data-driven diagnostic model requires a large amount of sample data with fault labels, however, the risk of carrying out destructive fault tests (such as artificial crafted flight) on a real helicopter is extremely high, the cost is extremely high, and the problem that the data acquisition is difficult is caused by the existing diagnostic algorithm. The prior art is limited to vibration signal analysis, does not solve the fundamental problem of contact measurement, and ignores the great potential of acoustic signals in reflecting the aerodynamic characteristics and early structural changes of the rotor. The acoustic signal is used as a non-contact measurement means, contains rich global modal information, is extremely sensitive to aerodynamic shape changes (such as ice accumulation and deformation), is flexible in sensor deployment and low in cost, and is an important direction for health monitoring of the next generation of helicopters. Disclosure of Invention In order to solve the technical problems, the invention provides an intelligent diagnosis method and system for potential faults of a helicopter rotor wing based on acoustic signals. In a first aspect, the invention provides an intelligent diagnosis method for potential faults of a helicopter rotor wing based on acoustic signals, comprising the following steps: Constructing a three-dimensional geometric model of a helicopter rotor wing, implanting typical fault characteristics into characteristic parameters of the three-dimensional geometric model, and generating rotor wing models in various fault states; Simulating a dynamic rotation process of a rotor wing model in a fault state by adopting an overlapped grid method, and solving flow field parameters through a CFD and CSD coupling strategy to obtain a