An eVTOL electric drive power suite dynamic balance measurement and correction method and device

Abstract

The present application relates to a kind of eVTOL electric drive power set dynamic balance measurement and correction method and device, belong to aviation electric propulsion and motor manufacturing field.The present application includes acceleration vibration sensor, laser key phase sensor, real-time controller and test management software, and the influence coefficient calculated by finite element dynamics model is prior initial value to realize free test restart, known counterweight is regarded as virtual test weight in trimming, and influence coefficient matrix is corrected on-line by recursive least square algorithm using measured power frequency vibration vector, so that influence coefficient matrix changes from open loop calculation to closed loop convergence.For wide speed flexible rotor, FFT frequency domain analysis is used to extract power frequency vector to calculate phase angle;Real-time controller synchronously collects vibration and key signal under the same clock reference.The present application has balance precision of G2.5, and unbalance amount is reduced by 95.92% in one time trimming, supports single and double face correction and batch measurement.

Description

Technical Field

[0001] This invention relates to the field of aviation electric propulsion and motor manufacturing technology, specifically to a method and apparatus for measuring and correcting the dynamic balance of an eVTOL electric drive system. Background Technology

[0002] A crucial component of the low-altitude economy is low-altitude aircraft, especially electric vertical take-off and landing (eVTOL) aircraft. To meet the diverse needs of different user groups and flight routes, low-altitude aircraft primarily utilize three power types: multi-rotor, lift and cruise, and vector thrust. All of these power types require rotors as the aircraft's propulsion source, typically employing an electrically driven rotor system to drive the rotor's rotation.

[0003] Dynamic balancing of high-speed rotating rotors is a critical issue in both manufacturing and product use. For electrically driven propulsion systems used in aircraft, substandard dynamic balancing can significantly impact the stability and safety of the aircraft.

[0004] Currently, dynamic balancing measurements of rotors in conventional equipment such as motors, steam turbines, turbines, and pumps typically treat the rotor as a single rigid rotor body, with operating speeds far below the first-order bending critical speed. Vibration is measured on a low-speed dynamic balancing machine in a single or double-sided manner to calculate the unbalance mass and phase angle. However, the electric drive rotor of an eVTOL is a power assembly integrating a propeller, motor, and hub, and is a flexible rotor. During eVTOL flight conditions such as hovering, climbing, cruise, and descent, the operating speed will exceed the first-order or higher-order critical speeds, making it impossible to accurately test the rotor's dynamic balance and correct unbalance at a single steady state or specific speed.

[0005] In existing technologies, traditional dynamic balancing machines have limited functionality, only capable of testing low-speed steady-state rigid rotors and unable to test the dynamic balance of flexible rotors in electric drive assemblies. Traditional trial weight methods require multiple physical trials and cycles of speed-up-measurement-stop, which are prone to errors in trial weight installation angle, trial weight mass, and the risk of trial weight falling off at high speeds. The operation is cumbersome and inefficient.

[0006] In recent years, researchers have proposed virtual dynamic balancing methods without trial weights based on finite element models. Li Zhiwei et al. addressed the problem of multiple start-ups and shutdowns required for balancing multi-disc flexible rotors at multiple speeds and in multiple planes. They studied a virtual dynamic balancing method without trial weights based on dynamic response analysis, considering dual-speed, dual-plane dynamics. Using the finite element software Dyrobes, they established a finite element model similar to the dynamics of a multi-disc flexible rotor test bench. Through finite element calculations of the rotor system's vibration response at the equilibrium speed, they extracted the influence coefficient matrix between unbalanced excitation and vibration response at different speeds, and then calculated the balance correction. Although this method achieves trial-weight-free start-up, the influence coefficient matrix is ​​entirely dependent on the accuracy of the finite element model. Uncertainties exist in the bearing support characteristics, damping parameters, and connection stiffness of the actual rotor system, and model errors can lead to insufficient measurement accuracy.

[0007] Furthermore, when calculating the unbalanced phase angle using the traditional time-domain zero-crossing method, environmental noise can cause the peak value of the acceleration signal to be unstable, resulting in a large phase angle error, which makes it difficult to meet the requirements of high-precision dynamic balance measurement.

[0008] Based on this, the present invention is proposed. Summary of the Invention

[0009] To address the shortcomings of existing technologies, such as the limitations of the pure finite element method without trial weight in terms of accuracy (limited by model accuracy), the inability of the influence coefficient matrix to adaptively correct with measured data, and the low accuracy of the phase angle measurement in the traditional time-domain zero-crossing method, this invention aims to propose a method and apparatus for dynamic balance measurement, unbalance mass, and phase angle correction of the electric drive rotor of an eVTOL. By integrating the finite element prior model with online recursive correction based on measured data, the influence coefficient matrix achieves adaptive evolution from prior initial values ​​to measured convergence, maintaining dynamic balance measurement accuracy throughout the entire flight envelope.

[0010] The technical solution of the present invention is as follows:

[0011] This invention provides an eVTOL electric drive force sleeve dynamic balance measurement and correction device, comprising:

[0012] An acceleration vibration sensor, at least one, is used to output a time-domain vibration acceleration signal to measure the vibration acceleration value generated by the rotation of the power sleeve rotor;

[0013] A laser keying sensor is used to output keying pulse signals to determine the initial zero point of the unbalanced phase angle and to calculate the actual operating speed of the power sleeve rotor.

[0014] The real-time controller incorporates a finite element solver and an FFT frequency domain analysis module. The FFT frequency domain analysis module transforms the acquired time-domain vibration acceleration signal and key phase pulse signal using FFT to extract the power frequency (i.e., rotational frequency) vibration frequency vector corresponding to the current operating speed of the power sleeve rotor, which is then used as the measured power frequency vibration vector. The finite element solver establishes a finite element dynamic model based on the input finite element dynamic model parameters of the power sleeve rotor and calculates the a priori initial value of the influence coefficient matrix. The real-time controller uses the a priori initial value of the influence coefficient matrix as initial conditions, utilizes the change in the measured power frequency vibration vector during dynamic balancing measurement and balancing correction, treats the known balancing weight as a virtual trial weight, dynamically corrects the influence coefficient matrix online using a recursive least squares algorithm, and calculates the unbalanced mass and unbalanced phase angle based on the corrected influence coefficient matrix.

[0015] The test management software is used to input or import the finite element dynamic model parameters of the power sleeve rotor, call the finite element solver to generate the a priori initial value of the influence coefficient matrix, start the dynamic balancing test, and display the correction amount of the influence coefficient matrix and save the updated results after the balancing retest.

[0016] A sensor mounting assembly for mounting an acceleration vibration sensor and a laser key phase sensor near the rotor of the power sleeve being measured;

[0017] A counterweight is used to install a correction counterweight on the power sleeve rotor; the weight of the counterweight applied during the balancing and correction process is used as a known input for the recursive correction of the influence coefficient matrix.

[0018] The real-time controller also integrates the following modules:

[0019] (1) Multi-channel synchronous acquisition module: Under the drive of a single hardware clock reference, the signals of two acceleration vibration channels and one key phase channel are acquired synchronously to ensure that the sampling time of each channel is consistent;

[0020] (2) Unbalance and Balancing Calculation Module: Read the measured power frequency vibration vector output by the FFT frequency domain analysis module and the influence coefficient matrix output by the finite element solver, perform vector calculations to obtain the single- or double-sided unbalanced mass and unbalanced phase angle of the power sleeve rotor at the specified power frequency, and calculate the corrective counterweight mass and angle. Here, the power frequency is the aforementioned rotational frequency corresponding to the current measured rotational speed of the power sleeve rotor, representing the calculation state corresponding to the component extracted from the FFT spectrum that is equal to the rotational speed frequency.

[0021] Furthermore, there are two acceleration vibration sensors used to achieve dual-plane dynamic balance correction, and the acceleration vibration sensors are equipped with IEPE preamplifiers.

[0022] As a further embodiment of the present invention, the laser phase sensor is used in conjunction with a reflector, and the red light emitted by the laser phase sensor is perpendicular to the reflector; the red light emitted by the laser phase sensor is perpendicularly irradiated onto the surface of the reflector to form a light spot, and the reflectivity of the light spot to the reflector is close to 100%; the width of the reflector is the arc length corresponding to the central angle of the power sleeve rotor of 3°~5°.

[0023] It should be noted that the color contrast between the reflector and the surface of the power sleeve rotor should be large. Observation using the laser phase sensor shows that the reflectivity of the light spot on the reflector is close to 100%, while the reflectivity of other parts of the rotor is 10.9%~15.3%. Furthermore, the width of the reflector must be moderate. Too narrow a width will cause pulse signal frame loss at high rotor speeds; too wide a width will result in an excessively high duty cycle for the pulse signal, leading to pulse signal overlap at high speeds. In this invention, the width of the reflector is the arc length corresponding to the central angle of the power sleeve rotor of 3°~5°. This width range allows the duty cycle of the pulse signal output by the laser phase sensor to be maintained between 0.8% and 1.4%, ensuring that within the speed range of 120~6000 rpm, neither too narrow a reflector will cause high-speed frame loss, nor too wide a reflector will cause pulse overlap.

[0024] As a further aspect of the present invention, the distance between the measuring end face of the acceleration vibration sensor and the laser key phase sensor and the tangent of the rotor being measured is in the range of 10~35mm. Exceeding this range will result in a large deviation in the dynamic balance measurement results or damage to the sensors.

[0025] This invention also provides a method for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve, comprising the following steps:

[0026] Step 1: Input the finite element dynamic model parameters of the power sleeve rotor. The finite element solver built into the real-time controller calculates the prior initial value A(0) of the influence coefficient matrix based on the finite element dynamic model.

[0027] Step 2: Perform initial vibration measurement without adding test weight to obtain the initial measured power frequency vibration vector V0, and calculate the actual unbalance of the power sleeve rotor and the balancing scheme based on the current influence coefficient matrix A(k);

[0028] Step 3: Install the counterweights according to the balancing scheme and re-measure the vibration to obtain the measured power frequency vibration vector V1 after balancing;

[0029] Step 4: The real-time controller utilizes the measured change in the power frequency vibration vector before and after balancing. With known counterweight Using the prior initial values ​​of the influence coefficient matrix As the initial conditions, a recursive least squares algorithm is used. Online correction of the influence coefficient matrix;

[0030] in, Here is the gain matrix. This is the influence coefficient matrix after the k-th measurement / correction. It is the influence coefficient matrix after the (k+1)th measurement / correction. It is the known weight applied during the k-th measurement, where k is the iteration number, k=0,1,2,…; It is the kth measurement / correction , It is the kth measurement / correction ;

[0031] Step 5: Repeat steps 2 to 4 for batches of power sleeve rotors of the same model to gradually converge the influence coefficient matrix.

[0032] As a further aspect of the present invention, the calculation of the actual imbalance is achieved through... Solve for U, where U is the actual unbalance of the power sleeve rotor.

[0033] As a further aspect of the present invention, the actual unbalance (including a complex vector containing the phase) comprises the unbalance mass and the unbalance phase angle. The unbalance phase angle is calculated using frequency domain analysis. The time-domain vibration acceleration signal and the key phase pulse signal are converted into a frequency domain vector sequence through FFT transformation. The power frequency vector corresponding to the current measured speed of the power sleeve rotor is selected from the frequency domain vector sequence. ,in The power frequency phase angle of the time-domain vibration acceleration signal. The power frequency phase angle of the key phase pulse signal. This is the initial phase zero point.

[0034] It should be noted that for calculating the unbalanced phase angle, this invention abandons the traditional time-domain zero-crossing method and adopts a frequency-domain analysis method. The time-domain zero-crossing method simultaneously acquires acceleration and key phase pulse signals during testing, records the time difference between the peak point of the acceleration signal and the pulse signal, and the rotational speed, thus calculating the unbalanced phase angle. This method is simple and intuitive, but in actual testing, environmental noise can cause instability in the peak value of the acceleration signal, resulting in a large peak time error, making it difficult to accurately obtain the unbalanced phase angle. In this invention, the position where the laser key phase pulse sensor spot illuminates the reflector is taken as the phase zero point, and the time-domain acceleration signal before FFT is acquired starting from this phase zero point. Therefore, in the above formula... The value is always zero. Extensive experiments have shown that the unbalanced phase angle deviation calculated using the frequency domain analysis method described above is less than 0.5°, which is far lower than the 13.2° unbalanced phase angle deviation calculated using the time domain zero-crossing method.

[0035] As a further aspect of the present invention, the real-time controller uses a single hardware clock reference to achieve multi-channel synchronous acquisition. The ADC sampling clocks of the two acceleration vibration channels and one key phase channel are generated by uniformly dividing the same reference clock, and the sampling time deviation between channels is less than one sampling period. The key phase pulse signal is captured by a hardware latching circuit and timestamped under the same clock reference.

[0036] The key phase channel provides a reference marker for each revolution, while the acceleration vibration channel provides a continuous vibration waveform. By comparing the time difference between the vibration waveform and the key phase pulse, the phase angle of the imbalance point can be calculated.

[0037] As a further aspect of the present invention, the real-time controller dynamically allocates computing resources according to the working mode. FFT frequency domain analysis and synchronous acquisition are hard real-time tasks, occupying the highest priority; influence coefficient calibration is a medium-priority task; and finite element solution is a low-priority background task.

[0038] Wherein, sampling rate F s = 1000Hz, FFT points are set to 8192; therefore, the resolution is 0.122Hz, which balances signal sampling rate, FFT resolution, and computational efficiency of the real-time controller. The FFT process uses the Hanning window function, Hamming window function, or Force window function.

[0039] As a further aspect of the present invention, the test management software automatically determines whether the current imbalance is qualified based on the rotor mass of the power sleeve and the balance quality level requirements of the dynamic balance. The allowable imbalance calculation formula is as follows:

[0040] ;

[0041] In the formula, For allowable imbalance, unit M represents the mass of the power sleeve rotor, in kg. The unit for balanced quality level is mm / s; ω is the angular velocity of the power sleeve rotor, in rad / s.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] 1. This invention integrates finite element prior calculation with recursive least squares measured correction, so that the influence coefficient matrix starts from the prior initial value and gradually converges to the true value under the drive of measured data, thus breaking through the limitation that the accuracy of the pure finite element method depends on the accuracy of the model.

[0044] 2. This invention uses the influence coefficient matrix calculated by the finite element dynamics model as the a priori initial value to achieve a trial-free restart. In batch measurement, the influence coefficient matrix is ​​continuously corrected by the recursive least squares algorithm, forming a closed loop of measurement-balancing-correction-reuse. The larger the batch size, the higher the accuracy, which is highly compatible with the mass production process of eVTOL power kits.

[0045] 3. The eVTOL power kit's operating speeds cover flight conditions such as hovering, climbing, cruise, and descent. The rotor flexibility effect, bearing support characteristics, and thermal deformation drift with speed and operating conditions. This invention can use measured data before and after balancing to correct the influence coefficient matrix online at each speed, and store the convergence results at different speeds in the test management software in the form of a speed-influence coefficient mapping database. Here, the speed refers to the actual operating speed of the power kit rotor during the current dynamic balancing measurement, which is represented by the rotational angular velocity Ω in theoretical derivation and by the speed n in the interface and operation process of the test management software. During measurement, the corresponding influence coefficient matrix is ​​automatically called according to the current speed, which can maintain the accuracy of dynamic balancing measurement throughout the entire flight envelope.

[0046] 4. This invention uses frequency domain analysis to calculate the unbalanced phase angle, with a phase angle deviation of less than 0.5° and high phase angle measurement accuracy.

[0047] 5. The balancing accuracy of this invention can reach G2.5, which is high. The imbalance amount in a single balancing operation can be reduced by 95.92%. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the eVTOL electric drive force dynamic balance measurement and correction device described in this invention;

[0049] Figure 2 This is a schematic diagram of the installation of the acceleration vibration sensor, laser key phase sensor, and sensor mounting base described in this invention;

[0050] Figure 3 A physical image of a power sleeve rotor and its weight distribution points for a tested eVTOL;

[0051] Figure 4 This is a time-domain signal waveform diagram of the accelerometer and laser key phase sensor described in this invention;

[0052] Figure 5 The waveforms of the frequency domain amplitude spectrum and phase spectrum of the power sleeve rotor in its original state as described in this invention are shown.

[0053] Figure 6 This is a spectrum waveform diagram of the power sleeve rotor with virtual test weight described in this invention;

[0054] Figure 7This is the interface for calculating the prior initial value of the influence coefficient matrix in the test management software described in this invention.

[0055] Figure 8 This describes the data flow call relationship between the finite element solver and the FFT frequency domain analysis described in this invention.

[0056] Figure 9 This is the interface for dynamic balance measurement and vector decomposition of the test management software described in this invention;

[0057] Figure 10 Three balancing schemes for imbalance correction are provided for the test management software described in this invention;

[0058] Figure 11 The images show the vibration waveforms of the power sleeve rotor before (blue) and after (pink) correction, as described in this invention.

[0059] Figure 12 This is a front view of the actual real-time controller described in this invention;

[0060] Figure 13 This is a rear view of the actual real-time controller described in this invention. Detailed Implementation

[0061] The present invention will be described in detail below with reference to specific embodiments. These embodiments are merely some, not all, implementations of the present invention. All other implementations obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0062] Example 1

[0063] like Figure 1 , 2 As shown, the eVTOL electric drive force sleeve dynamic balance measurement includes:

[0064] An accelerometer vibration sensor with an IEPE preamplifier is used to measure the vibration acceleration generated during the rotation of a power sleeve rotor. To achieve dual-sided dynamic balance correction, two accelerometer vibration sensors are required for each power sleeve rotor. The sensor is designed to measure low-frequency, low-level vibrations. The built-in IEPE preamplifier allows for signal isolation and floating, and the sensitivity must reach 1000mV / g within ±5% of an ambient temperature of 25℃.

[0065] A laser phase-keying sensor is used to determine the initial zero point of the unbalanced phase angle. The laser phase-keying sensor works in conjunction with a reflector. The reflector is attached to the side of the power sleeve rotor, and the red light emitted by the laser phase-keying sensor is perpendicular to the reflector. The reflectivity of the light spot on the reflector is close to 100%. The width of the reflector is the arc length corresponding to a central angle of 3°~5° on the power sleeve rotor. When the power sleeve rotor rotates to the red light spot, the laser phase-keying sensor generates a pulse signal.

[0066] The acceleration vibration sensor, laser key phase sensor, and the tested eVTOL electric drive force sleeve of this invention need to be installed on a lifting platform with a cantilever and lifting assembly. The lifting platform allows the power sleeve to undergo dynamic balance testing at heights ranging from 1150 to 3080 mm above the ground. This design aims to simulate the real flight aerodynamic environment and implement graded safety control.

[0067] First, the height of the rotor blades above the ground directly affects the "ground effect." When the rotor is close to the ground, turbulent airflow causes distortion in thrust and vibration data. By adjusting the lifting platform, the free hovering condition can be simulated at a high position to obtain accurate dynamic balance data. At a low position, it can be used to simulate the specific installation height of the actual aircraft to avoid structural resonance.

[0068] Secondly, regarding testing safety, high-speed rotating dynamic balancing tests pose a risk of rotor debris splashing or breaking. The lifting platform supports a "low-level inspection, high-level operation" strategy: a low height facilitates safe maintenance, low-speed inching, and emergency intervention; when fully extended, it effectively isolates ground-rebound airflow and debris, providing a safe buffer space for sudden mechanical failures and greatly reducing potential hazards to laboratory personnel and equipment.

[0069] Simultaneously, the real-time controller synchronously acquires data using the same clock reference, ensuring that the starting point of the acquired effective vibration signal sequence is consistent with the pulse signal. Furthermore, the current rotational speed of the power sleeve rotor can be calculated using the number of pulses per unit time or the time interval between adjacent pulses.

[0070] The sensor mounting assembly includes a sensor mounting base 15, a bracket 10, and a cantilever 14. All components are connected by quick-release bolts. Each connection point is equipped with a locking screw, and a locking torque of not less than 5 N·m is applied. After locking, the overall sliding amount of the bracket is less than 0.05 mm. The measuring end face of the acceleration vibration sensor 16 and the laser phase sensor 17 is 10~35 mm away from the tangent of the measured power sleeve rotor.

[0071] In some embodiments, the structure of the lifting platform can be as follows: the support 10 is further provided with a numerical slide rail and a slider 13, the slider 13 and the slide rail are slidably connected, the slider 13 and the slide rail are stopped by a pneumatic brake 12, the cantilever 14 is fixedly connected to the slider 13, and the slider 13 is driven to rise and fall by the lift 11. Different flight altitudes are simulated by controlling the height of the sensor mounting base 15.

[0072] The real-time controller incorporates a rotor finite element dynamics model and an adaptive influence coefficient identification module. Based on the rotor's geometry, material parameters, and bearing support characteristics, a finite element dynamics model is established. This model calculates the vibration response at each correction plane under the rotational centrifugal force generated by a unit unbalance, generating initial values ​​for the influence coefficient matrix. Subsequently, the unbalanced mass and phase angle of single and double planes are calculated using a dynamic balancing algorithm. During balancing and retesting, the known balancing weight is treated as a virtual trial weight. The influence coefficient matrix is ​​corrected online using a recursive least squares algorithm based on the changes in vibration data before and after balancing. A physical diagram of the real-time controller is shown below. Figure 12 , 13 .

[0073] The real-time controller integrates a multi-channel synchronous acquisition module, an FFT frequency domain analysis module, a finite element solver, and an unbalance and balancing calculation module. Among these:

[0074] The multi-channel synchronous acquisition module synchronously acquires signals from two acceleration vibration channels and one key phase channel under the drive of a single hardware clock reference;

[0075] The FFT frequency domain analysis module converts the time-domain vibration signal and the key phase signal into a frequency domain vector sequence using the FFT (Fast Fourier Transform) algorithm, and extracts the power frequency vibration vector (amplitude and phase angle) corresponding to the current rotational speed of the power sleeve rotor.

[0076] The finite element solver establishes a finite element dynamic model based on the input parameters such as the rotor geometry, material parameters, bearing support stiffness, and damping, and calculates the prior initial values ​​of the influence coefficient matrix.

[0077] The unbalance and balancing calculation module reads the power frequency vibration vector and influence coefficient matrix, performs vector calculations, obtains the single- or double-sided unbalanced mass and unbalanced phase angle of the power sleeve rotor at the power frequency (the rotational frequency corresponding to the current actual speed of the power sleeve rotor), and calculates the corrective counterweight mass and angle.

[0078] The test management software is used to initiate the dynamic balancing calibration process. Users input or import finite element dynamic model parameters (including rotor geometry, material density and elastic modulus, bearing stiffness and damping coefficients, etc.) into the software interface. The test management software then calls the finite element solver built into the real-time controller to establish the finite element dynamic model and generate prior initial values ​​for the influence coefficient matrix. Subsequently, clicking on the dynamic balancing test performs unbalance measurements and balancing calculations.

[0079] like Figure 9 As shown, the dynamic balance measurement and vector decomposition interface of the test management software displays the power frequency vibration amplitude and phase angle measured by the two acceleration vibration sensors at the current speed in real time, and automatically calculates the actual unbalanced mass (e.g., 31.564g) and unbalanced phase angle (e.g., 160.19°) of the power sleeve rotor. At the same time, the location of the unbalanced point is intuitively marked in the form of a polar coordinate target diagram.

[0080] After balancing, the user clicks "retest and correction". The test management software reads the original time-domain signals before and after balancing, calls the real-time controller to complete the online correction of the influence coefficient matrix and displays the correction amount.

[0081] The real-time controller completes the unbalance measurement and balancing calculation. After balancing correction and retesting, the test management software displays the correction amount of the influence coefficient matrix and automatically saves and updates it. For power sleeve rotors of the same model and batch, the test management software automatically accumulates the correction data of each measurement, gradually improving the accuracy of the influence coefficient matrix. The test management software interface uses a target diagram corresponding to the actual power sleeve rotor angle (e.g., ...). Figure 10 It displays the convergence process of the imbalance point, the correction balancing point, and the influence coefficient matrix.

[0082] It also includes a balance, which is a desktop weighing instrument with an RS-232 serial port, a weighing range of 3kg, a resolution of 0.1g, and a minimum weighing weight of 1g.

[0083] It also includes a counterweight block, which comprises five sizes of pad weights: small pad 0.9g, medium pad 2.0g, large pad 3.1g, short screw 4.1g, and long screw 4.6g.

[0084] In this invention, unless otherwise specified, vibration signal and bond phase signal refer to the original time-domain signals (i.e., time-domain waveforms) acquired by the acceleration vibration sensor and the laser bond phase sensor, respectively, to distinguish them from the frequency-domain vector obtained after FFT transformation. Time-domain vibration signal and bond phase signal refer to the same two sets of physical quantities as vibration signal and bond phase signal. The term "time-domain" is used only to emphasize the original state of the signal before FFT transformation, in order to distinguish it from the output frequency-domain vector.

[0085] Example 2

[0086] This invention employs frequency domain analysis to calculate the unbalanced phase angle. The acceleration vibration signal and the key phase pulse signal are processed by FFT and converted into a frequency domain vector sequence. The power frequency vector corresponding to the rotor speed of the power sleeve is then selected from the frequency vector sequence.

[0087] like Figures 4-6 As shown, Figure 4 The waveforms of the time-domain signals synchronously acquired by the acceleration vibration sensor and the laser key phase sensor are shown, illustrating the measurement method of the acceleration signal peak and the time difference ΔT between the key phase pulse, which is relied upon by the traditional zero-crossing method in the time domain. The interference of environmental noise on the peak position can also be seen in the figure. After the time-domain vibration signal and the key phase pulse signal are transformed by FFT, they are converted into frequency domain vector sequences. The power frequency vector corresponding to the rotor speed of the power sleeve is selected from the frequency vector sequence.

[0088] like Figure 5 , 6 As shown, the power frequency is 12.451 Hz, and the initial measured vibration amplitude is 1.2066 m / s. 2 The phase angle was 159.46°; during the balancing and retesting, the vibration was measured after installing an 18.6g counterweight in the 0° direction, and the amplitude was 0.58918m / s. 2 The phase angle was 135.32°; during the balancing and retesting, a 18.6g counterweight was installed in the 180° direction before measuring the vibration, and the amplitude was 1.8886m / s². 2 The phase angle is 168.06°; the real-time controller calculates the unbalanced mass of the power sleeve rotor as 31.564g and the unbalanced phase angle as 160.19°, and uses the change in the measured power frequency vibration vector ΔV(k) before and after balancing to complete the online correction of the influence coefficient matrix.

[0089] In this embodiment, the user does not need to perform physical weight testing; they can directly input the finite element dynamic model parameters (outer diameter of the power sleeve rotor: 60mm, inner diameter: 20mm, length: 150mm, material density: 7800kg / m³) into the test management software. 3 Elastic modulus 2.1×10 11 Pa, Poisson's ratio 0.3, bearing 1 stiffness 5×10 6 N / m, damping 500 N·s / m, bearing 2 stiffness 5×10 6 (N / m, damping 500 N·s / m), the real-time controller automatically establishes a finite element dynamic model and generates prior initial values ​​for the influence coefficient matrix. Subsequently, only one initial vibration measurement without additional test weight is needed to calculate the actual unbalance of the power sleeve rotor and complete the balancing. The change ΔV(k) of the measured power frequency vibration vector after balancing is used by the real-time controller to correct the influence coefficient matrix online, and the corrected influence coefficient matrix is ​​used as the initial measurement value for the next power sleeve rotor of the same model.

[0090] Example 3

[0091] For the rotor-bearing system, the finite element method is used to discretize the power sleeve rotor along the axial direction into Timoshenko beam elements. Each beam element has two nodes, and each node has four degrees of freedom (two translations and two rotations). Considering the gyroscopic effect, moment of inertia, and lateral shear deformation of the power sleeve rotor, as well as the support stiffness and damping characteristics of the bearing, a dynamic mathematical model of the system is established. The differential equations of motion for the rotor-bearing system can be expressed as:

[0092] ;

[0093] In the formula, F(t) is the unbalanced excitation force vector (unit N); M is the system mass matrix (unit kg·m). 2 ), which is assembled from the uniform mass matrix of each beam element; q is the generalized displacement vector of the nodes, which includes the translation and angular displacement of each node; It is the velocity vector; Ω is the acceleration vector; C is the damping matrix, which includes the structural damping of the power sleeve rotor and the bearing oil film / electromagnetic damping (unit: N·s / m); Ω is the rotational angular velocity of the power sleeve rotor (rad / s). f is the rotation frequency (Hz); K is the gyroscope matrix, reflecting the Coriolis force effect when the power sleeve rotor rotates; K is the stiffness matrix, which includes the bending stiffness of the power sleeve rotor and the bearing support stiffness (unit N / m).

[0094] For the steady-state unbalanced response of the rotor at a rotational angular velocity Ω, the motion frequency domain equation of the rotor-bearing system is:

[0095] ;

[0096] F0 is the complex amplitude vector of the unbalanced force, and i is the imaginary unit. This represents the reset amplitude vector (i.e., the amplitude and phase of the steady-state response) of each node in the rotor-bearing system in the frequency domain.

[0097] Define the dynamic stiffness matrix Z(Ω) of the rotor-bearing system as:

[0098] ;

[0099] Then the frequency response function (FRF) matrix H(Ω) is the inverse of the dynamic stiffness matrix:

[0100] ;

[0101] The FRF matrix H(Ω) is a complex matrix whose elements This matrix represents the complex amplitude of the steady-state displacement response at degree u when a unit harmonic excitation force is applied at degree j. It comprehensively describes all the dynamic characteristics of the rotor-bearing system at a given rotational angular velocity Ω, including natural frequencies, mode shapes, damping characteristics, and coupling relationships between degrees of freedom.

[0102] The process of establishing the finite element dynamic model (FEM) of the power sleeve rotor is as follows:

[0103] The power sleeve rotor is discretized along the axial direction as follows: Each element is a Timoshenko beam element with two nodes and four degrees of freedom per node: lateral displacement (x, y) and rotation about the two lateral axes. , The degrees of freedom of the unit are arranged as follows:

[0104] ;

[0105] It is the nodal displacement vector of a single beam element;

[0106] x1 is the lateral translational displacement of the first beam element node in the X direction, in meters;

[0107] y1 is the lateral translational displacement in the Y direction of the first beam element node, in meters;

[0108] It is the angular displacement (section rotation) of the first beam element node about the X-axis, in rad;

[0109] It is the angular displacement (section rotation) of the first beam element node about the Y-axis, in rad;

[0110] x2 is the lateral translational displacement of the second beam element node in the X direction, in meters;

[0111] y2 is the lateral translational displacement of the second beam element node in the Y direction, in meters (m).

[0112] It is the angular displacement (section rotation) of the second beam element node about the X-axis, in rad;

[0113] It is the angular displacement (section rotation) of the second beam element node about the Y-axis, in rad;

[0114] Element mass matrix of each beam element Unit gyroscope matrix and element stiffness matrix All of these parameters can be derived from the shape function integrals of Timoshenko beam theory. The local matrices of each beam element are assembled according to the nodal degrees of freedom (superimposed onto the corresponding global positions) to obtain the global M, C, and G matrices. M The K-matrix. A finite element dynamics model with N nodes, where the dimensions of each global matrix are K. .

[0115] The bearing's support stiffness and damping are superimposed onto the global matrix as lumped parameters. For the bearing at node k, its stiffness matrix... and damping matrix The degrees of freedom of the corresponding k-node are superimposed on the global K and C matrices. On the sub-block. This modeling method can flexibly handle different bearing types (rolling bearings, sliding bearings, magnetic levitation bearings, etc.) and different support boundary conditions (rigid supports, elastic supports, free ends, etc.).

[0116] Calculation of unit unbalanced excitation force:

[0117] At the correction plane, define the unit unbalance U. unit for:

[0118] The amplitude of the centrifugal force generated by this unit unbalance at a rotational angular velocity Ω for: .

[0119] For two-plane dynamic balancing, the rotational centrifugal force generated by the unit unbalance needs to be applied at the first and second correction planes respectively, and two independent finite element analysis calculations need to be performed.

[0120] Example 4

[0121] This invention divides the acquisition of the influence coefficient matrix into two stages: Stage 1 is the prior initial value calculated by the finite element dynamic model theory (in this embodiment), and Stage 2 is the online recursive correction driven by subsequent measured data (see Embodiment 6). The role of the prior initial value is to provide an initial estimate of the influence coefficient matrix for the first dynamic balancing measurement under the condition of no physical trial weight data, so that this invention can achieve trial-free restart.

[0122] In this embodiment, the initial a priori value of the influence coefficient matrix calculated based on the finite element dynamics model is denoted as A(0) (distinct from A(k) after subsequent recursive least squares correction). This matrix is ​​theoretically calculated using the finite element dynamics model of the rotor-bearing system and is written into the shared memory area of ​​the real-time controller before the first measurement begins, serving as the initial basis for calculating the unit unbalance. After each subsequent balancing remeasurement, the real-time controller recursively corrects A(0) based on the measured data, gradually converging it to the true value.

[0123] Methods for calculating prior initial values ​​of influence coefficients in finite element dynamic models:

[0124] like Figure 7 As shown, in the prior initial value calculation interface of the influence coefficient matrix in the test management software, the user inputs or imports the rotor's geometric dimensions (outer diameter, inner diameter, length), material parameters (density, elastic modulus, Poisson's ratio), and finite element dynamic model parameters such as bearing support stiffness and damping coefficient. After clicking the "Calculate Influence Coefficients" button, the software calls the finite element solver built into the real-time controller to establish the finite element dynamic model, automatically generates the prior initial value A(0) of the influence coefficient matrix, and displays it on the interface. This interface also displays the amplitude and phase angle of each element of the 2×2 influence coefficient matrix in tabular form, making it easy for the user to confirm the rationality of the model calculation.

[0125] Apply a unit unbalance U at the correction plane unit The generated centrifugal force F unit,j Afterwards, the complex amplitude D of the displacement response at measuring point r of the acceleration vibration sensor i,j for:

[0126] ;

[0127] In the formula This represents the transfer function element in the FRF matrix H(Ω) corresponding to the displacement degree of freedom at the measuring point r of the accelerometer and the excitation force degree of freedom at the calibration plane. Accelerometers typically measure vibration acceleration, and the acceleration response has a complex amplitude. for:

[0128] ;

[0129] Define influence coefficient Let: The complex amplitude of the vibration acceleration response at the measuring point r of the acceleration vibration sensor caused by the unit unbalance (1 g·mm) at the calibration plane (m / s² / (g·mm)), that is:

[0130] ;

[0131] For two-dimensional dynamic equilibrium, the influence coefficient matrix A is a 2×2 complex matrix:

[0132] ;

[0133] The physical meaning of each influence coefficient is as follows: The unit unbalance at the first correction plane ( The first acceleration response of the vibration sensor caused by the phase. The acceleration response of the second acceleration vibration sensor caused by the unit imbalance at the first correction plane; The acceleration response of the first acceleration vibration sensor caused by the unit imbalance at the second correction plane; The acceleration response of the second acceleration vibration sensor caused by the unit imbalance at the second correction plane. Each All are complex numbers, and their modulus The amplitude sensitivity and phase angle of the vibration response. This indicates that the response lags behind the excitation in phase.

[0134] For single-plane dynamic balancing measurements, the influence coefficient degenerates into a scalar complex number.

[0135] Example 5

[0136] The process of solving for rotor imbalance using the influence coefficient matrix is ​​as follows:

[0137] When the power sleeve rotor operates at a rotational angular velocity Ω, the initial measured power frequency vibration vector is obtained from the acceleration vibration sensor. It is also the initial complex amplitude vector of the power frequency vibration acceleration:

[0138] ;

[0139] in , representing the initial vibration amplitude (m / s) at the first acceleration vibration sensor. 2 ) and phase angle (°); This represents the initial vibration amplitude (m / s) at the second acceleration vibration sensor. 2 ) and phase angle (°);

[0140] Based on the principle of linear superposition of influence coefficients, the actual unbalance quantity is measured. The following conditions must be met:

[0141] ;

[0142] Therefore, the actual unbalance U of the power sleeve rotor is obtained by solving the above system of linear equations:

[0143] ;

[0144] For a 2×2 influence coefficient matrix A, its inverse matrix is :

[0145] ;

[0146] In the formula, , where is the rank of matrix A.

[0147] The solution yields: , ,in Let g be the unbalanced mass at the first correction plane. Let g be the unbalanced mass at the second correction plane. , These represent the radii (mm) of the corresponding unbalanced masses. , These are the corresponding unbalanced phase angles (relative to the reference mark).

[0148] In the description of this invention, It is the kth measurement / correction , It is the kth measurement / correction k is the iteration number, k=0,1,2,…

[0149] Example 6

[0150] The correction mechanism and algorithm convergence of the influence coefficient matrix adaptive identification method built into the influence coefficient adaptive identification module are as follows:

[0151] Traditional experimental trial-and-error methods obtain the influence coefficient matrix through three physical experiments, while this invention treats the balancing and correction process itself as a virtual trial-and-error, eliminating the need for a dedicated trial-and-error operation. The correction mechanism of the fusion algorithm in this invention is as follows:

[0152] Step 1, Initial Measurement: Without adding a test weight, accelerate to a rotational angular velocity Ω, and measure the initial measured power frequency vibration vector V0;

[0153] Step 2, Balancing Correction: Install the known counterweight U on the correction plane according to the actual unbalance calculated in the initial calculation. t (Equivalent to a virtual trial weight), accelerate to the rotational angular velocity Ω, and re-measure the measured power frequency vibration vector V1 after balancing, which is also the complex amplitude vector of the power frequency vibration acceleration;

[0154] Step 3, Online Correction:

[0155] The online correction equation of this invention is as follows: the balancing correction is regarded as a virtual trial weight, and the facility is added with a known balancing weight. Afterwards, the change in vibration before and after balancing is: According to the principle of linear superposition The real-time controller calculates the correction amount of the influence coefficient matrix based on this. When multiple power sleeve rotors appear in a batch or during multiple alignments, the influence coefficient matrix is ​​updated recursively using the Recursive Least Squares (RLS) algorithm:

[0156] ;

[0157] in Here is the gain matrix. This is the influence coefficient matrix after the k-th measurement / correction. It is the change in the measured power frequency vibration vector before and after balancing. It is the known weight applied during the k-th measurement; the prior initial value of the influence coefficient matrix. Calculated by the finite element dynamics model, as the number of measurements k increases, the influence coefficient matrix gradually converges to the true value; its convergence is equivalent to that calculated by the finite element theory: under the ideal conditions of accurate finite element dynamics model parameters and no measurement error, the converged A is consistent with the result calculated by the experimental trial-and-retake method difference formula.

[0158] Compared with traditional single methods (Li Zhiwei et al.'s pure finite element method without trial and error), the algorithm of this invention has the following advantages:

[0159] 1) No-Test-Weight Startup: The influence coefficient matrix is ​​calculated theoretically using the finite element model as the initial a priori value. The first power-driven rotor can be directly measured without physical test weights, avoiding the three cycles of speed-up-measurement-stop, and eliminating the risks of test weight installation angle errors, test weight mass errors, and high-speed test weight detachment. Furthermore, since the influence coefficient matrix is ​​dynamically corrected from measured data, it overcomes the limitations of the pure finite element method, which is not dependent on actual measurements and whose accuracy is limited by the model.

[0160] 2) Accuracy Adaptive: The influence coefficient matrix is ​​corrected online by the real-time controller based on the measured data, gradually offsetting the modeling errors of the finite element model (uncertainties in geometric parameters, material parameters, bearing characteristics, boundary conditions, etc.), so that the measurement accuracy continuously improves with the number of iterations, breaking through the limitation of the pure finite element method's accuracy depending on the model.

[0161] 3) Batch self-learning: For batch testing of the same type of power sleeve rotor, the measured data of each power sleeve rotor is used to correct the influence coefficient matrix, forming a closed loop of measurement-balancing-correction-reuse. The larger the batch size, the more accurate the influence coefficient matrix, and the measurement efficiency and accuracy are improved simultaneously, taking into account the measured reliability of the traditional trial weight method and the high efficiency of the finite element method without trial weight.

[0162] 4) Operating Condition Adaptability of the Power Kit for eVTOL Electric Drives: The dynamic balancing measurement of the power kit for eVTOL electric drives differs fundamentally from that of conventional motor rotors. The eVTOL power kit rotor is a flexible rotor assembly integrating the propeller, motor, and hub. Its operating speed covers flight conditions such as hovering, climbing, cruise, and descent. The speed range is wide and continuously changes with the flight stage. The bending mode, bearing oil film characteristics, and thermal deformation caused by motor temperature rise of the power kit rotor are all different at different speeds, resulting in a significant drift of the influence coefficient matrix with speed and operating conditions. Traditional methods calibrate a fixed influence coefficient matrix once by trial weight at a single speed, which cannot cover the entire flight envelope and will produce large measurement errors under variable speed conditions. The adaptive identification method of the influence coefficient matrix of the present invention can correct the influence coefficient matrix online using measured vibration data before and after balancing under the rotational speed and operating conditions of each power kit rotor. The convergence results at different speeds are stored in the test management software in the form of a speed-influence coefficient mapping database. During measurement, the corresponding influence coefficient matrix is ​​automatically called according to the current speed, thereby achieving accurate measurement and correction of dynamic balance within the entire flight envelope. This effectively solves the problem of dynamic balance measurement of the power kit under multiple operating conditions and a wide speed range of eVTOL electric drive, while meeting its requirements for mass production, high consistency, fast cycle time and airworthiness.

[0163] Example 7

[0164] like Figure 3 As shown, the tested dynamic sleeve has 6 evenly distributed counterweights (see...). Figure 3 The arrows in the diagram correspond to 0° / 60° / 120° / 180° / 240° / 300° respectively. The test management software performed vector decomposition on the actual unbalance of 31.564g / 160.19° and provided three balancing methods (e.g., ...). Figure 10 As shown in the figure, the unbalanced mass after balancing is 1.287g, and the actual unbalanced amount is reduced by 95.92%.

[0165] like Figure 11 As shown, the comparison of the vibration waveforms of the power sleeve rotor before (blue curve) and after (pink curve) correction intuitively demonstrates the technical effect of the present invention: before correction, the vibration amplitude was large and the waveform was messy, while after correction, the vibration amplitude was significantly reduced and the waveform tended to be stable, indicating that one balancing operation effectively eliminated most of the imbalance, verifying the superiority of the present invention in terms of dynamic balance measurement accuracy and correction efficiency.

[0166] Example 8

[0167] The hardware architecture design of the real-time controller in this invention focuses on solving three key design problems, centered on the accuracy and real-time requirements of dynamic balancing measurement: first, the data flow and calling relationship between the finite element solver and the FFT frequency domain analysis module; second, the synchronous acquisition of vibration signals and bond phase signals using the same reference; and third, the allocation of computational resources in trial weight mode and finite element mode. The following sections will explain these three design points in detail:

[0168] 1) Data flow and calling relationship between the finite element solver and the FFT frequency domain analysis module

[0169] The data flow and calling relationship between the finite element solver and the FFT frequency domain analysis module are as follows: Figure 8 As shown: Vibration signal / key phase signal, multi-channel synchronous acquisition module, FFT frequency domain analysis module, power frequency vector (amplitude, phase angle), unbalance and balancing calculation module, balancing scheme;

[0170] Rotor finite element parameters (finite element dynamic model parameters), finite element solver, influence coefficient matrix A (a priori initial values), and unbalance and balancing calculation module.

[0171] The FFT frequency domain analysis module is responsible for transforming the synchronously acquired time-domain vibration signal and key phase signal into a frequency domain vector sequence distributed from low frequency to high frequency through FFT, and extracting the power frequency vector (frequency, amplitude, phase angle) corresponding to the operating speed of the power sleeve rotor for use by the unbalance and balancing calculation module. The finite element solver is responsible for assembling the system mass matrix, stiffness matrix and damping matrix according to the input power sleeve rotor geometric dimensions, material parameters, bearing support stiffness and damping parameters, obtaining the FRF matrix, and then calculating the a priori initial value of the influence coefficient matrix. The calculation results are written into the shared memory area of ​​the real-time controller for the unbalance and balancing calculation module to call through a unified interface.

[0172] The calling relationship between the two is as follows: the FFT frequency domain analysis module always runs alongside the real-time acquisition task; the finite element solver is only called in finite element mode and is a background pre-calculation task. When the test management software starts the dynamic balancing test, the real-time controller first checks whether the influence coefficient matrix in the shared memory area is ready: in trial weight mode, it is written after online calibration of the initial measurement, left trial weight, and right trial weight data; in finite element mode, it is written after pre-calculation by the finite element solver. The unbalance and balancing calculation module reads the influence coefficient matrix and the power frequency vector to complete the calculation of single-sided / double-sided unbalance mass and phase angle.

[0173] 2) Synchronous acquisition under the same clock reference

[0174] Synchronous acquisition of vibration signal and key phase signal is the key to ensuring the accuracy of unbalanced phase angle calculation. The real-time controller uses a single hardware clock reference to achieve multi-channel synchronous acquisition: (1) The real-time controller has a built-in high-stability clock source (such as a temperature-compensated crystal oscillator or a temperature-compensated crystal oscillator, with frequency stability better than (1) The sampling clocks of the two acceleration vibration channels and one key phase channel are all generated by the same reference clock, ensuring that the sampling time of each channel is strictly synchronized and the sampling time deviation between channels is less than one sampling period; (2) The key phase pulse is captured by the hardware latch circuit and timestamped under the same clock reference, sharing a time axis with the sampling time sequence of the vibration signal; (3) The controller takes the sampling sequence number corresponding to the key phase pulse as the time origin, and truncates and aligns the vibration signal sequence to ensure that the time starting point of the vibration data sequence participating in the FFT analysis is strictly consistent with the key phase pulse, thereby ensuring the measurement accuracy of the power frequency vector phase angle; (4) Since the time difference in the phase angle is measured by the same clock reference, its measurement error depends only on the frequency stability of the clock source, eliminating the phase error introduced by the relative drift between multiple clock sources.

[0175] 3) Differences in computational resource allocation between trial-and-error mode and finite element mode

[0176] The real-time controller dynamically allocates computing resources according to the working mode. The resource requirements of the two modes differ as follows: the trial weight mode requires three measurements in sequence: initial measurement, left trial weight, and right trial weight, and saves three sets of time-domain data and corresponding power frequency vectors, which places a greater demand on storage resources (data cache RAM); however, the influence coefficient matrix calibration only involves... Difference operations on complex matrices have low computational cost and low computational power requirements; the finite element method does not require saving multiple sets of trial data, so it has low storage resource requirements; however, the finite element solution requires assembling the system matrix and performing complex matrix inversion operations (such as inverting the FRF matrix to obtain the influence coefficient matrix), which involves intensive floating-point operations, is time-consuming, and has high computational power requirements.

[0177] Based on the above differences, the resource allocation strategy of the real-time controller is as follows: FFT frequency domain analysis and synchronous acquisition are hard real-time tasks, occupying the highest priority (such as the highest level interrupt) to ensure the real-time performance of sampling and spectrum analysis; influence coefficient calibration (trial mode) is a medium-priority task; finite element solution (finite element mode) is a low-priority background task, scheduled to be executed during non-real-time periods such as speed-up, speed-down, or shutdown intervals to avoid competing with real-time acquisition tasks for CPU resources; the controller adjusts memory partitioning according to the current mode, allocates a large-capacity data buffer area for storing three measurement data in trial mode, and allocates a large-capacity matrix operation work area for finite element solution in finite element mode to achieve optimal configuration of computing resources.

[0178] Furthermore, it should be understood that those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A device for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve, characterized in that, include: An acceleration vibration sensor, at least one, is used to output a time-domain vibration acceleration signal to measure the vibration acceleration value generated by the rotation of the power sleeve rotor; A laser keying sensor is used to output keying pulse signals to determine the initial zero point of the unbalanced phase angle and to calculate the actual operating speed of the power sleeve rotor. The real-time controller incorporates a finite element solver and an FFT frequency domain analysis module. The FFT frequency domain analysis module transforms the acquired time-domain vibration acceleration signal and key phase pulse signal using FFT to extract the power frequency vibration frequency vector corresponding to the current operating speed of the power sleeve rotor, which is then used as the measured power frequency vibration vector. The finite element solver establishes a finite element dynamic model based on the input finite element dynamic model parameters of the power sleeve rotor and calculates the a priori initial value of the influence coefficient matrix. The real-time controller uses the a priori initial value of the influence coefficient matrix as initial conditions, utilizes the change in the measured power frequency vibration vector during dynamic balancing measurement and balancing correction, treats the known balancing weight as a virtual trial weight, dynamically corrects the influence coefficient matrix online using a recursive least squares algorithm, and calculates the unbalanced mass and unbalanced phase angle based on the corrected influence coefficient matrix. The test management software is used to input or import the finite element dynamic model parameters of the power sleeve rotor, call the finite element solver to generate the a priori initial value of the influence coefficient matrix, start the dynamic balancing test, and display the correction amount of the influence coefficient matrix and save the updated results after the balancing retest. A sensor mounting assembly for mounting an acceleration vibration sensor and a laser key phase sensor near the rotor of the power sleeve being measured; Counterweights are used to install and correct counterweights on the power sleeve rotor; The weight applied by the counterweight during the balancing and correction process is used as a known input for the recursive correction of the influence coefficient matrix.

2. The eVTOL electric drive force sleeve dynamic balance measurement and correction device according to claim 1, characterized in that: The laser phase sensor is used in conjunction with a reflector. The red light emitted by the laser phase sensor is perpendicular to the reflector. The width of the reflector is the arc length corresponding to the central angle of the power sleeve rotor of 3°~5°.

3. The eVTOL electric drive force sleeve dynamic balance measurement and correction device according to claim 1, characterized in that: The distance between the measuring end face of the acceleration vibration sensor and the laser key phase sensor and the tangent of the rotor being measured ranges from 10 to 35 mm.

4. A method for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve, performed using the eVTOL electric drive force sleeve dynamic balance measurement and correction device as described in any one of claims 1 to 3, characterized in that, Includes the following steps: Step 1: Input the finite element dynamic model parameters of the power sleeve rotor. The finite element solver built into the real-time controller calculates the prior initial value A(0) of the influence coefficient matrix based on the finite element dynamic model. Step 2: Perform initial vibration measurement without adding test weight to obtain the initial measured power frequency vibration vector V0, and calculate the actual unbalance of the power sleeve rotor and the balancing scheme based on the current influence coefficient matrix A(k); Step 3: Install the counterweights according to the balancing scheme and re-measure the vibration to obtain the measured power frequency vibration vector V1 after balancing; Step 4: The real-time controller utilizes the measured change in the power frequency vibration vector before and after balancing. With known counterweight Using the prior initial values ​​of the influence coefficient matrix As the initial conditions, a recursive least squares algorithm is used. Online correction of the influence coefficient matrix; in, Here is the gain matrix. This is the influence coefficient matrix after the k-th measurement / correction. It is the influence coefficient matrix after the (k+1)th measurement / correction. It is the known weight applied during the k-th measurement, where k is the iteration number, k=0,1,2,…; It is the kth measurement / correction , It is the kth measurement / correction ; Step 5: Repeat steps 2 to 4 for batches of power sleeve rotors of the same model to gradually converge the influence coefficient matrix.

5. The method for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve according to claim 4, characterized in that: The calculation of the actual imbalance is achieved through... Solve for U, where U is the actual unbalance of the power sleeve rotor.

6. The method for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve according to claim 4, characterized in that: The actual imbalance includes the unbalanced mass and the unbalanced phase angle. The unbalanced phase angle is calculated using frequency domain analysis. The time-domain vibration acceleration signal and the key phase pulse signal are converted into a frequency domain vector sequence through FFT transformation. The power frequency vector corresponding to the current measured speed of the power sleeve rotor is selected from the frequency domain vector sequence. ,in The power frequency phase angle of the time-domain vibration acceleration signal. The power frequency phase angle of the key phase pulse signal. This is the initial phase zero point.

7. The method for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve according to claim 4, characterized in that: The real-time controller uses a single hardware clock reference to achieve multi-channel synchronous acquisition. The ADC sampling clocks of the two acceleration vibration channels and one key phase channel are generated by uniformly dividing the same reference clock, and the sampling time deviation between channels is less than one sampling period. The key phase pulse signal is captured by a hardware latching circuit and timestamped under the same clock reference.

8. The method for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve according to claim 4, characterized in that: The real-time controller dynamically allocates computing resources according to the working mode. FFT frequency domain analysis and synchronous acquisition are hard real-time tasks, occupying the highest priority; influence coefficient calibration is a medium priority task; and finite element solution is a low priority background task.

9. The method for measuring and correcting the dynamic balance of an eVTOL electric drive force sleeve according to claim 4, characterized in that: The test management software automatically determines whether the current unbalance is acceptable based on the rotor mass of the power sleeve and the balance quality level requirements of the dynamic balance. The allowable unbalance calculation formula is as follows: ； In the formula, For allowable imbalance, unit M represents the mass of the power sleeve rotor, in kg. The unit for balanced quality level is mm / s; ω is the angular velocity of the power sleeve rotor, in rad / s.

Images