An article analyzing NVH of electric drive system

NVH is the abbreviation for sound, vibration and harshness. For electric passenger cars, it mainly manifests itself in high-frequency electromagnetic noise. The total noise level of electric vehicles is generally lower than that of fuel vehicles. However, its spectrum is more concentrated and mainly within the sensitive range of the human ear, making it easier for passengers to feel and complain. The basic relationship between the noise of these two types of vehicles is shown in the following two figures.

Figure 1 Differences in noise levels at different speeds

Figure 2 Relationship between the frequency of electric drive noise and the sensitive area of ​​the human ear

The reasons are generally the following 6 points:

1. No shielding effect: Electric vehicles generally do not have engines and turbochargers, etc., and the high-noise and wide-band sound sources make the electromagnetic noise lack shielding and more prominent;

2. Stronger torque impact. The motor torque response speed is significantly higher than that of fuel vehicles. Its instantaneous torque impact will put higher requirements on the strength and life of the transmission system, and aggravate NVH problems such as jitter and high-frequency vibration;

3. Mainly electromagnetic noise. It is mainly caused by electromagnetic harmonics of the motor body and PWM harmonics of the IGBT switch of the controller. The main noise frequency is relatively high and coincides with the sensitive range of the human ear;

4. Wider speed range. Unlike the 5-9 gearboxes commonly used in passenger cars, electric vehicles generally only have 1-2 gears. The wider speed adjustment range of the motor means a wider range of vibration excitation and resonance risks, making it impossible to avoid efforts to improve and avoid the dynamic stiffness of the shell structure;

5. Strong demand for lightweighting. Due to limited battery capacity, electric vehicles have a strong demand for weight reduction. An overly lightweight structure also leads to a reduction in rigidity, which increases the risk of resonance and noise amplification;

6. High integration brings additional noise. Integrated design is conducive to reducing weight and cost, but it may cause some parts to be re-excited by vibration and generate more noise.

The picture below shows the Nissan Leaf electric drive system. After the integrated design, a new set of resonant noise bands around 1700Hz were added.

Figure 3 Nissan Leaf electric drive integration noise changes

Where there is demand, there is a direction for effort. In order to achieve good NVH performance, major OEMs and electric drive component companies generally adopt the V-shaped R&D process shown in the figure below.

Figure 4 V-shaped flow chart for electric drive NVH performance development

In the above process, the overall NVH indicators of the next generation of models are generally determined based on the NVH performance indicators of existing and competing models, and are gradually split and transferred to the electric drive system. Then, the whole vehicle and bench tests are carried out on competing products and existing products to accumulate and understand the actual performance. If possible, mapping and reverse simulation analysis are carried out on competing products to try to understand more detailed design concepts, performance indicators, NVH advantages and disadvantages, etc.

Then define the overall structural design, motor electromagnetic design, reducer structure and NVH design, controller structure design and other solutions and performance of the new product, and gradually conduct test benchmarking and verification, and timely predict, discover, improve and optimize NVH problems until the NVH requirements of the whole vehicle or the optimal value at a reasonable cost are met.

During this period, methods and tools such as D-FEMA, P-FEMA, A3, and 8D may be used to help locate problems and improve product performance and quality.

When encountering NVH problems, due to the complexity of the structure and the coupling relationship between different components, it is generally impossible to simply and directly locate the problem. You can also try the black and white box test method to screen and locate it. As shown in the figure below.

Figure 5: Black and white box method for noise analysis

The prediction, reproduction and improvement of NVH problems of electric drives can be mainly carried out from three aspects: noise source, propagation path and receiver.

Generally, the focus is on optimizing the noise source. It is mainly caused by the electromagnetic pulsation harmonics of the motor stator and the gear meshing transmission error vibration, which are transmitted to the housing to generate radiated noise. The source of the problem is shown in the figure below.

Figure 6 Relationship between vibration excitation and noise

Transmission path optimization is generally achieved from the design and matching of the dynamic characteristics of the frame and suspension stiffness and damping. Since the high-frequency vibration component of the electric drive system is higher than that of conventional fuel vehicles, the simulation and experimental requirements for the dynamic characteristics of the suspension at 1000Hz and above are higher, and most similar experimental equipment cannot accurately measure such high-frequency performance, which may create a threshold for further NVH performance optimization.

If the suspension and frame cannot be significantly improved, the electric drive surface can also be optimized by wrapping the acoustic package. However, this will bring higher space, weight and cost, and sometimes the noise reduction effect is limited.

NVH optimization on the receiver side generally includes vehicle acoustic design and active noise reduction.

When developing detailed forward NVH performance, the process shown below is generally used.

Figure 7 Forward NVH development process

Generally, according to the load conditions, the dynamic excitation caused by gear meshing is calculated and coupled with the transmission system to obtain the shell surface acoustic radiation. At the same time, for the motor stator and rotor considering the control strategy, the harmonic components of the air gap magnetic field are solved, and the stator vibration and noise are calculated accordingly. The two are superimposed to obtain the overall shell dynamic characteristics, such as dynamic stiffness and noise map.

Based on the simulated and measured main noise orders, it may be possible to reversely locate the main noise position and possible causes and make adjustments.

Such as improving manufacturing and assembly accuracy, skewing poles and slots of stators and rotors, grooving on the surfaces of stators and rotors, matching stator coil winding slots, gear modal optimization, tooth surface contact spot and transmission error optimization, PWM control algorithm optimization, active shock absorption and active harmonic injection, and shell reinforcement rib optimization.

Reverse NVH performance development generally involves scanning a 3D model from the physical object and performing multiple prototype tests to determine the final design.

For the reducer, the main problem is the unstable pulsation of gear meshing, and the load on the bearing position requires simulation analysis. See the following two figures.

Figure 8 Bearing load simulation results

Experimental acquisition of bearing load. Due to the complex shape of the reducer and the small size of the local area, it is not easy to obtain it by pasting strain gauges. Generally, it is obtained indirectly through acceleration sensors and microphones.

If possible, it is recommended to use the photoelastic method for the experiment. Generally, with the help of polarized light and a transparent shell model, and reasonable loading, and then by observing the distribution of colored interference light patterns, the force transmission path and stress concentration degree of the shell can be indirectly obtained.

Figure 9 Load data of dynamic meshing of bearings

The figure below shows the influence of mechanical vibration of the rotating system on the bearing load.

Figure 10 Schematic diagram of the influence of vibration excitation on the rotor system

In the figure above, the bearing stiffness and damping data are key input parameters, which are generally obtained from the bearing supplier through experimental measurements and conversion from the deformation curves under different loads.

In order to ensure the NVH performance of the gear, it is necessary to accurately predict the deformation of the gear and the housing, the resulting gear meshing misalignment, and make the vibration fluctuations caused by the misalignment as smooth as possible, and then reasonably modify the tooth surface and design the housing stiffness.

When the motor is running in reverse charging mode, the reverse tooth surface of the gear will be fully loaded, which may increase the risk of gear whistling. However, the probability of occurrence and impact of this condition are low, and it can generally be downgraded for use.

Under the action of alternating loads, the reducer housing may resonate. Generally, modal analysis and frequency sweep vibration analysis methods are used to obtain the response degree of vibrations of different orders and summarize them in the figure below, which can be used to separate the noise source.

Figure 11 Reducer modal MPa diagram

The following two figures show the contribution of gear modal orders of different designs to the vibration response.

Figure 12 Relationship of reducer gear vibration response 1

Figure 13 Relationship of reducer gear vibration response 1

The NVH problem caused by resonance is usually more severe when the following conditions are met at the same time: