The reliability of the electric drive system is fundamental to the performance and longevity of an electric vehicle car. Unlike traditional internal combustion engine vehicles, the powertrain of an electric vehicle car is characterized by high transient torque response and regenerative braking capabilities. This results in more severe load fluctuations during real-world operation, particularly under frequent start-stop and acceleration-deceleration scenarios. These high transient loads significantly increase the risk of shock loading and fatigue damage to critical mechanical components such as shafts, gears, and bearings. Consequently, traditional durability test specifications, often derived from limited user surveys or specific driving conditions, may not adequately capture the actual load spectra and damage characteristics experienced by electric vehicle car systems under diverse Chinese user conditions. Therefore, investigating the load characteristics of electric drive systems based on actual usage patterns is a critical technical challenge for advancing the reliability engineering of new energy vehicles.
The advent of telematics and big data analytics offers a transformative solution. By leveraging extensive, real-world operational data from a large fleet of electric vehicle cars, we can obtain a comprehensive understanding of usage patterns across different regions, road conditions, and driving behaviors. This research is based on the analysis of over one million kilometers of driving data collected from more than 200 electric vehicle cars operating across seven Chinese cities, including Shanghai, Guiyang, Chongqing, Changchun, Qingdao, Xining, and Yancheng. The vehicles primarily consisted of high-usage models like ride-hailing and taxi services, ensuring the data reflects demanding operating conditions. The collected parameters included timestamp, GPS location, and vehicle speed at a 1 Hz sampling frequency.
The core technical challenge lies in transforming this vast amount of user speed data into the mechanical loads (torque and speed) acting on the electric drive system’s components. Direct measurement via CAN bus for such a large sample is impractical. Therefore, this study employs a vehicle longitudinal dynamics modeling approach to simulate these loads. The model calculates the required drive or brake torque based on the balance between tractive force and the sum of all motion resistances. The fundamental equation governing the longitudinal dynamics of an electric vehicle car is given by the force balance during driving:
$$F_t = F_f + F_w + F_i + F_j$$
Where \(F_t\) is the tractive force at the wheels (N). The resistance forces are:
- Rolling Resistance: \(F_f = mgf \cos \alpha\)
- Aerodynamic Drag: \(F_w = \frac{C_D A v^2}{21.15}\)
- Grade Resistance: \(F_i = mg \sin \alpha\)
- Acceleration Resistance: \(F_j = m \frac{dv}{dt}\)
Here, \(m\) is the vehicle mass (kg), \(g\) is gravitational acceleration, \(f\) is the rolling resistance coefficient, \(\alpha\) is the road grade angle, \(C_D\) is the drag coefficient, \(A\) is the frontal area (m²), \(v\) is the vehicle speed (km/h), and \(\frac{dv}{dt}\) is the acceleration (m/s²).
The motor torque \(T\) (N·m) and speed \(n\) (rpm) can then be derived from the tractive force and vehicle speed:
$$T = \frac{F_t \cdot r}{i_g \eta}, \quad n = \frac{v \cdot i_g}{0.377 \cdot r}$$
where \(r\) is the tire radius (m), \(i_g\) is the total gear ratio, and \(\eta\) is the drivetrain efficiency. The parameters for a typical SUV-style electric vehicle car used for model validation are summarized in Table 1.
| Parameter | Symbol | Value | Unit |
|---|---|---|---|
| Vehicle Mass | \(m\) | 2600 | kg |
| Tire Radius | \(r\) | 0.36 | m |
| Rolling Resistance Coefficients | \(f_0, f_1, f_4\) | 0.015, 0.0015, 0.0002 | – |
| Drag Coefficient | \(C_D\) | 0.24 | – |
| Frontal Area | \(A\) | 2.5 | m² |
| Drivetrain Efficiency | \(\eta\) | 0.95 | – |
| Total Gear Ratio | \(i_g\) | 10 | – |
The validity of this simulation approach was rigorously verified. Speed data from an instrumented test electric vehicle car was fed into the model, and the simulated motor torque output was compared against actual CAN bus measurements. The comparison was conducted in the time domain, frequency domain, and, most importantly, in the damage domain using rainflow cycle counting and Miner’s linear cumulative damage rule. The results showed a damage error of less than 10%, which is within an acceptable range for engineering applications, confirming the model’s reliability for large-scale load transformation.
To analyze operational patterns, the continuous driving data was segmented into distinct kinematic phases based on speed and acceleration thresholds, as defined in Table 2.
| Driving Phase | Speed Condition | Acceleration Condition |
|---|---|---|
| Acceleration | v > 5 km/h | a > 0.15 m/s² |
| Deceleration | v > 5 km/h | a < -0.15 m/s² |
| Constant Speed | v > 5 km/h | -0.15 ≤ a ≤ 0.15 m/s² |
| Idling | v ≤ 5 km/h | -0.15 ≤ a ≤ 0.15 m/s² |
A high-level analysis of the aggregated user data reveals the proportional contribution of each phase to total operating time, distance, and component damage. The results are summarized in Table 3. Acceleration phases account for the largest share of both time and distance, highlighting their prevalence in urban and suburban driving for an electric vehicle car. More critically, acceleration and deceleration phases are responsible for the overwhelming majority of fatigue damage to key components, underscoring that transient load cycles are the primary drivers of mechanical wear and failure.
| Driving Phase | Time Share | Distance Share | Approx. Damage Share (Shaft/Gear/Bearing) |
|---|---|---|---|
| Acceleration | 39% | 43% | ~65% / ~70% / ~60% |
| Deceleration | 27% | 28% | ~30% / ~15% / ~25% |
| Constant Speed | 21% | 28% | ~5% / ~15% / ~15% |
| Idling | 13% | 1% | ~0% / ~0% / ~0% |
Delving deeper, we performed a statistical characterization of each driving phase. The duration of idling events, for instance, was found to follow a lognormal distribution, with 95% of idling periods lasting less than 19 seconds. Constant speed phases are typically very short, with 95% lasting less than 7 seconds and covering distances under 250 meters at average speeds below 85 km/h. This pattern is indicative of the stop-and-go nature of traffic, even on open roads, for a typical electric vehicle car.
The acceleration phase is of particular interest. Key parameters like maximum acceleration, average acceleration, speed change (\(\Delta v\)), and duration were analyzed. The distributions for average acceleration and speed change during acceleration are lognormal. A joint distribution analysis of \(\Delta v\) versus average acceleration reveals that most acceleration events are of low intensity (\(\Delta v < 30\) km/h, average acceleration < 1 m/s²), representing gentle, frequent accelerations common in city traffic for an electric vehicle car. High \(\Delta v\) events tend to have lower average accelerations, suggesting users prefer smoother acceleration when building up speed over a longer period.
Similarly, for deceleration, the average deceleration follows a lognormal distribution. The joint distribution of \(\Delta v\) and average deceleration shows concentration in the region of \(\Delta v < 45\) km/h and average deceleration < 1.5 m/s². This indicates a prevalence of moderate braking behavior, which is favorable for efficient regenerative braking in an electric vehicle car.
To link kinematic states directly to component damage, we analyzed the damage contribution as a function of the average speed and average acceleration/deceleration of each segment. Different components exhibit distinct damage “hotspots” in this 2D domain, as conceptualized in Table 4. The analysis confirms that low-to-medium speed ranges (20-50 km/h) combined with moderate acceleration/deceleration levels are the most damaging operational regimes for the electric drive system of an electric vehicle car. This is precisely the scenario emblematic of congested urban and suburban driving.
| Component | Primary Damage Concentration Zone | Implication |
|---|---|---|
| Shaft | Avg. Speed: 20-30 km/h; Avg. Acceleration: 1-1.5 m/s² Avg. Speed: 20-40 km/h; Avg. Deceleration: -1.5 to -1 m/s² |
Damaged by torque transients during both acceleration and deceleration (e.g., frequent starts and stops). |
| Gear | Avg. Speed: 20-40 km/h; Avg. Acceleration: 1-1.5 m/s² | Primarily damaged during acceleration events where high torque is applied under load. |
| Bearing | Avg. Speed: 20-50 km/h; Avg. Acceleration: 0.5-1.5 m/s² | Exhibits a broader damage zone, affected by a wider range of speed and acceleration combinations. |
User behavior varies significantly by region. To compare fairly despite differences in total mileage, we introduced the metric of “Damage per Unit Distance” (\(D_{unit}\)) for each user’s drive/brake segments:
$$D_{unit} = \frac{D}{M}$$
where \(D\) is the total damage calculated via the appropriate model (shaft, gear, bearing) and \(M\) is the total distance traveled in the segment. A comparative analysis across cities showed marked disparities. For example, users in Yancheng and Changchun exhibited significantly higher average unit damage intensity across all components compared to users in Shanghai. This suggests more aggressive driving patterns (sharper acceleration and braking) in those regions, which directly translates to higher mechanical stress on the electric vehicle car’s drivetrain per kilometer driven.
A crucial outcome of this big-data study is the construction of a “User Target” load spectrum. By statistically extrapolating the acceleration/deceleration event frequency from the 200+ users to a full vehicle life (e.g., 300,000 km), we generated a representative load collective for the Chinese market. Comparing this User Target with established industry durability test standards (like the AKB profiles) reveals a significant gap, as summarized in Table 5. While existing standards are adequate for covering high-load, high-stress events, they underestimate the frequency of low-to-medium intensity load cycles that dominate real-world usage of an electric vehicle car in China. This discrepancy means that a component validated solely against such standards might be over-engineered for rare high loads but under-tested for the high-cycle, lower-load fatigue that characterizes daily operation.
| Aspect | User Target (Chinese Data) | Typical Existing Standards (e.g., AKB) |
|---|---|---|
| Focus | Derived from massive, real-world operation data of electric vehicle cars. | Often based on limited data, expert judgment, or specific driving routes. |
| Load Intensity Coverage | Extremely high frequency of low-to-medium intensity cycles. | Good coverage of high-intensity cycles; relatively lower frequency of medium/low cycles. |
| Representativeness for Chinese Users | High, as it directly reflects diverse regional behaviors and road conditions. | Potentially low, as they may not reflect the specific traffic patterns and driving habits prevalent in China. |
| Implication for Testing | Suggests a need for test profiles that emphasize high-cycle fatigue in the medium load range. | May lead to validation gaps for high-cycle fatigue failure modes. |
In conclusion, this research demonstrates a comprehensive methodology for leveraging user big data to understand the true drive-brake load characteristics of electric vehicle cars. By integrating telematics data with vehicle dynamics modeling, we successfully transformed user speed profiles into component-level mechanical loads. The analysis unequivocally identifies low-to-medium speed, frequent变速 (speed variation) conditions as the primary contributor to the fatigue damage of shafts, gears, and bearings in an electric vehicle car. Furthermore, the study quantifies significant regional variations in user behavior and, most importantly, highlights a substantive gap between the load spectra derived from actual Chinese user data and those embodied in current industry durability standards. These findings provide critical data-driven insights and a methodological framework to support the reliability-forward design, validation, and development of more representative durability test specifications for electric vehicle car powertrain systems, ultimately leading to more robust and longer-lasting vehicles.