Vibration reduction method and torque assembly

Abstract

The invention relates to a method for damping (10) vibrations (74) of a torque assembly (14) of a vehicle (24), comprising the following steps: providing (12) a torque assembly (14) having at least one torque component (18) for transmitting a drive torque (16) for driving the vehicle (24) forward and rotating at a rotational speed; acquiring (40) at least one vibration parameter (42) of an undesired, existing or imminent vibration (74) associated with the torque assembly (14); performing (60) a damping measure (62) for suppressing the vibration (74), which is dependent on the vibration parameter (42); wherein the at least one vibration parameter (42) is calculated by a trained neural network (44) based on evaluation information (46) associated with the vibration (74) as input data (48). Furthermore, the invention also relates to a torque assembly (14).

Description

Technical Field

[0001] Introduction This invention relates to a vibration reduction method according to the overarching concept of claim 1. Furthermore, this invention also relates to a torque assembly. Background Technology

[0002] DE 10 2004 032 150 A1 discloses a method for reducing clutch shudder vibration in a vehicle powertrain system. The shudder vibration can be reduced by generating a superimposed torque and adding it to the transmitted torque of the clutch. This superimposed torque is calculated based on the filtered variation curve of the vehicle acceleration. Summary of the Invention

[0003] The objective of this invention is to improve the speed and accuracy of vibration reduction.

[0004] At least one of the aforementioned tasks is solved by a vibration reduction method having the features described in claim 1. This allows for faster and more targeted vibration suppression, enabling vehicles to operate more comfortably and safely.

[0005] The vehicle may be a passenger car, a commercial vehicle, or a two-wheeled vehicle.

[0006] The torque assembly can be arranged in the vehicle's powertrain. The torque assembly can be an electric axle. The torque assembly can include an electric motor, an inverter, and / or a gearbox. The gearbox can be single-stage or multi-stage.

[0007] The driving torque can be provided by a driving element, particularly an electric motor. The torque assembly may include the driving element. The electric motor may include a stator and a rotor that rotates relative to the stator. The torque component may be a rotor shaft connected to the rotor, or a component connected to the rotor shaft in a torque-transmitting manner.

[0008] The vibrations may be undesirable because they can affect the safety, comfort, and / or reliability of the torque components and / or the vehicle.

[0009] The vibration can be torsional, rotational, and / or linear. The vibration can be generated at the torque assembly via the vehicle wheels. The vibration can be caused by road surface irregularities, such as potholes or speed bumps. The vibration can be at the mounting or suspension points of the torque assembly. The vibration can be torsional vibration generated during rotation that transmits drive torque. This torsional vibration can be triggered by changes in the drive torque.

[0010] The evaluation information may include time-series data. It may include time-varying curves of rotational speed, such as the rotor speed of a rotor, or the wheel speed, acceleration, angular acceleration, and / or torque of at least one wheel of the vehicle, especially drive torque. Fourier transforms and / or wavelet transforms for calculating vibration parameters based on the evaluation information can be omitted. The evaluation information may indicate impending vibrations in a predictive or causal manner. It may include information about road surface irregularities, such as the geographical location of these irregularities and / or the vehicle's geolocation. The evaluation information may be obtained through measurement; it may be information derived from measured values; it may be unprocessed or processed sensor data.

[0011] The vibration parameters may include, or are, the slope, frequency, amplitude, and / or phase in the time-domain signal. The vibration parameters may include, or are a vibration indicator, for indicating the presence of disturbed vibration. The vibration parameters may be acquired predictively and before the vibration occurs.

[0012] The neural network can be trained using training data. The training data may contain multiple sets of mappings between evaluation information and corresponding vibration parameters. The training data may contain labeled data. The training data may contain time-series data and labeled sections exhibiting disturbance vibrations, and may include the corresponding vibration parameters. The training data may contain measured data, particularly data recorded at least under the driving conditions of the vehicle or comparable vehicles, and / or simulated data.

[0013] The neural network may have been tested and / or validated. Validation can be performed using validation data. Testing can be performed using test data. Test data and / or validation data may contain multiple sets of mappings between evaluation information and corresponding vibration parameters.

[0014] The neural network can be a convolutional neural network (CNN). CNNs typically consist of multiple convolutional layers that automatically and adaptively extract features from the input data. Each convolutional layer applies a set of filters that slide across the input data to capture local features. Following the convolutional layers, the CNN may have downstream pooling layers to reduce the spatial size of the representations and help reduce the risk of overfitting, as well as one or more fully connected layers to perform classification or regression based on the extracted features.

[0015] The neural network can be a recurrent neural network (RNN). An RNN typically includes a recurrent layer that performs sequential data processing, in addition to at least one input layer and at least one output layer. An RNN can include a Long Short-Term Memory (LSTM) architecture and / or a Gated Recurrent Unit (GRU) architecture. Furthermore, one or more fully connected layers may be present to perform classification or regression based on the extracted features.

[0016] The neural network can have an encoder-decoder architecture, particularly an autoencoder or a Transformer incorporating an attention mechanism. The encoder processes the input data and transforms it into a compact internal representation state that summarizes key information from the input data. The encoder can contain recurrent neurons (such as RNNs, LSTMs, or GRUs) or convolutional neurons (such as CNNs). The decoder receives the internal state generated by the encoder and uses it to generate output data, i.e., at least one vibrational parameter. The decoder can also have recurrent or convolutional layers.

[0017] The neural network can be a deep neural network (DNN). Between at least one input layer and one output layer, the neural network can have multiple intermediate layers containing a large number of neurons.

[0018] Once the vibration parameters are calculated, vibration reduction measures can be implemented immediately. Therefore, it is possible to respond quickly to unwanted vibrations.

[0019] In a preferred embodiment of the invention, the vibration parameters are calculated by the neural network based on evaluation information contained in at most one vibration cycle of the vibration. This allows for faster calculation of the vibration parameters and faster vibration reduction.

[0020] In another preferred embodiment, the evaluation information is set to include at most a fraction of the vibration cycle. The evaluation information may include at most a first half, particularly a first third, of the vibration cycle.

[0021] In one particular embodiment, the vibration period is preferably the initial first vibration period of the vibration. Therefore, a rapid response can be made to the occurrence of the vibration.

[0022] In another preferred embodiment, the vibration reduction measure includes introducing a force-controlled torque into the torque component. The control torque can be introduced by a drive element that simultaneously provides drive torque, preferably by an electric motor.

[0023] In one particular embodiment, the control torque is preferably superimposed on the drive torque. The control torque may be applied additionally in addition to the drive torque, or it may be introduced in a manner that is added to the drive torque.

[0024] In another specific embodiment, the vibration reduction measure preferably relies on at least one damping variable calculated by the neural network or another trained neural network, using at least one vibration parameter as input data. The description of the neural network characteristics also applies to the other neural network. When using a neural network with evaluation information as input data to calculate the damping variable, the damping variable can be internally dependent on the vibration parameter. The vibration reduction measure can be implemented indirectly depending on the vibration parameter and directly depending on the damping variable.

[0025] In a preferred embodiment of the invention, the damping variable preferably includes a set value for the control torque. This set value can be dynamic. This allows the vibration reduction process to be adaptive.

[0026] In another advantageous embodiment, at least one vehicle's operating parameters and / or driving parameters are incorporated into the calculation of the damping variable. These operating parameters and / or driving parameters can be assigned as input data to the neural network or the other neural network used to calculate the damping variable.

[0027] The operating parameter can be the current drive torque.

[0028] The driving parameters may be the vehicle's speed and / or its position. The vehicle's position can be calculated using navigation data.

[0029] In addition, to solve at least one of the above tasks, a torque component having the features of claim 10 is also proposed. Attached Figure Description

[0030] Other advantages and advantageous embodiments of the present invention will become apparent from the accompanying drawings. The invention will now be described in detail with reference to the accompanying drawings. The drawings show: Figure 1 This invention discloses a vibration reduction method according to a specific embodiment. Figure 2 Evaluation information used as input data in a vibration reduction method according to a specific embodiment of the present invention; Figure 3 : A side view of a torque assembly according to a specific embodiment of the present invention; Figure 4 : A perspective view of a torque component according to another specific embodiment of the present invention. Detailed Implementation

[0031] Figure 1A vibration reduction method according to a specific embodiment of the present invention is shown. The vibration reduction method (10) for a torque assembly (14) is for a vehicle (24) and includes: providing (12) a torque assembly (14) having a torque component (18) for transmitting drive torque (16) to drive the vehicle (24) forward and rotating at a certain speed. The torque assembly (14) is, for example, an electric shaft (20) in the powertrain (22) of the vehicle (24). The torque assembly (14) includes an electric motor (26), an inverter (28) for electrically driving it, and a gearbox (30). The gearbox (30) is connected to a vehicle axle (32) in a torque-transmitting manner. The axle (32) has at least two vehicle wheels (34). The electric motor (26) includes a rotor (36), and the torque component (18) may be a rotor shaft (38) of the rotor (36) and is connected to the gearbox (30) in a torque-transmitting manner.

[0032] Using a trained neural network (44), at least one vibration parameter (42) of an undesirable, existing, or impending vibration associated with the torque component (14) is acquired (40) as input data (48), with evaluation information (46) associated with the vibration. The vibration may be torsional vibration at least at the torque component (18) and is applied to the torque component (18) for example, through vehicle wheels (34), axles (32), and gearbox (30) under the influence of road surface unevenness (58). The torsional vibration may cause undesirable noise and vibration in the torque component (14).

[0033] The vibration parameter (42) can be a vibration indicator (50) indicating the presence of vibration, or more specifically, a parameter of vibration, especially frequency. The evaluation information (46) can be time-series data (52), such as the rotor speed (54) of the rotor (36). This allows existing vibrations to be identified and evaluated. Alternatively, the vibration parameter (42) can be calculated by a neural network (44) relying on the vehicle location (56) and previously geolocated stored road surface irregularities (58) as evaluation information (46). This allows the vibration parameter (42) of an impending, yet-to-be-started vibration to be calculated.

[0034] Subsequently, vibration damping measures (62) for suppressing vibration are implemented (60), which depend on the vibration parameter (42). The vibration damping measures (62) preferably include a force-controlled torque (64) introduced by an electric motor (26), which is superimposed on the drive torque (16). The vibration damping measures (62) depend on at least one damping variable (68) calculated by a neural network (44) or another trained neural network (66) with the vibration parameter (42) as input data, thus depending on the vibration parameter (42) in an indirect manner. The damping variable (68) may include a set value of the control torque (64). The operating parameters (70) and / or driving parameters (72) of the vehicle (24) may also be included as input data when calculating the damping variable (68).

[0035] Figure 2 Evaluation information (46) used as input data in a vibration reduction method according to a specific embodiment of the present invention is shown. The evaluation information (46) is, for example, time series data (52) of the value A of the rotor speed (54) depending on time t. When the vehicle wheel is impacted (e.g., due to uneven road surface), the rotor speed (54) may begin to vibrate at time t0. Preferably, the vibration parameters are calculated and vibration reduction measures (62) are initiated when the vibration (74) initially occurs, especially at time t1. Therefore, the vibration (74) can be suppressed, and a suppressed rotor speed (76) curve is obtained compared to the case where the rotor speed (54) continues to oscillate due to no vibration reduction.

[0036] The vibration parameter (42) can be calculated using rotor speed information within at most the first vibration cycle (78) as evaluation information (46), preferably depending on the first third of the first vibration cycle (78).

[0037] Figure 3 A torque assembly (14) according to a specific embodiment of the present invention is shown. The torque assembly (14) is, for example, an electric axle (20) that provides drive torque (16) at an axle (32) of a vehicle (24). The axle (32) is mounted on a vehicle frame (82) via a vehicle suspension (80). The torque assembly (14) has a center of gravity (84) that is offset relative to the axle (32). Due to the force (86) caused by road surface unevenness (58) on the axle (32), combined with the offset center of gravity (84) of the torque assembly (14), the mounted torque assembly (14) will vibrate (74).

[0038] Figure 4A perspective view of a torque assembly (14) according to another specific embodiment of the present invention is shown. The torque assembly (14) is configured as an electric axle (20) and mounted on the axle (32) of a vehicle. The torque assembly (14) includes an electric motor (26), an inverter (28), and a gearbox (30).

[0039] List of reference numerals 10 Vibration Reduction Methods 12 provided 14 Torque Components 16 Drive torque 18 Torque Components 20 electric shafts 22 Powertrain System 24 vehicles 26 Electric motors 28 Inverters 30 Gearbox 32 axles 34 wheels 36 rotors 38 Rotor shaft 40 Collection 42 Vibration parameters 44 Neural Networks 46. ​​Assessment Information 48 Input Data 50 Vibration Indicator 52 Time Series Data 54 Rotor speed 56 Vehicle Location 58. Uneven road surface 60 Execution 62 Vibration Reduction Measures 64 Control torque 66 Neural Networks 68 Damping Variables 70 Operating Parameters 72 Driving Parameters 74 Vibrations 76 Rotor speed 78 Oscillation Period 80 suspension 82 frame 84 Center of gravity 86 Forces

Claims

1. A method for damping (10) vibration (74) of a torque assembly (14) of a vehicle (24), comprising the following steps: Provide (12) the torque assembly (14), the torque assembly having at least one torque component (18) for transmitting drive torque (16) to drive the vehicle (24) forward and rotating at a speed, acquire (40) at least one vibration parameter (42) regarding an existing or impending undesired vibration (74) associated with the torque assembly (14), implement (60) a vibration reduction measure (62) for suppressing the vibration (74), which depends on the vibration parameter (42), characterized in that the at least one vibration parameter (42) is calculated by a trained neural network (44) based on evaluation information (46) associated with the vibration (74) as input data (48).

2. The vibration reduction (10) method according to claim 1, characterized in that, The vibration parameter (42) is calculated by the neural network (44) based on the evaluation information (46) of the vibration period (78) containing at most one of the vibrations (74).

3. The vibration reduction (10) method according to claim 2, characterized in that, The evaluation information (46) may contain at most a fraction of the vibration period (78).

4. The vibration reduction (10) method according to claim 2 or 3, characterized in that, The vibration period (78) is the first vibration period (78) of the vibration (74).

5. The vibration reduction (10) method according to any one of the preceding claims, characterized in that, The vibration reduction measure (62) includes applying a control torque (64) to the torque component (18).

6. The vibration reduction (10) method according to claim 5, characterized in that, The control torque (64) is superimposed on the drive torque (16).

7. The vibration reduction (10) method according to any one of the preceding claims, characterized in that, The vibration reduction measure (62) is performed based on at least one damping variable (68) calculated by the neural network (44) or another trained neural network (66) with at least one vibration parameter (42) as input data (48).

8. The vibration reduction (10) method according to claim 5 or 6 in conjunction with claim 7, characterized in that, The damping variable (68) includes the set value of the control torque (64).

9. The vibration reduction (10) method according to claim 7 or 8, characterized in that, The calculation of the damping variable (68) incorporates the operating parameters (70) and / or driving parameters (72) of at least one vehicle (24).

10. A torque assembly (14) of a vehicle (24) having at least one torque component (18) for transmitting drive torque (16) to drive the vehicle (24) forward and rotating at a rotational speed, and which may experience undesirable vibrations (74), and is configured to operate by the vibration reduction (10) method according to any of the preceding claims.

Images