Two key trends are driving the design of new vehicles around the world: software-defined vehicles (SDVs) and electric vehicles (EVs). Vehicles are undergoing a transformation from hardware-defined features and capabilities to software-defined, offering new opportunities for agile development, continual improvements and remote maintenance.
SDVs usher in a new development approach that enables vehicle improvements over time based on deeper access to real-time vehicle data, cloud processing for digital twins and machine-learning training, and vehicle over-the-air (OTA) updates to improve software and ML models. As part of this continuous integration and continuous deployment, model-based design tools are used to more efficiently develop software algorithms that will run in the vehicle.
Global carmakers are investing heavily in the electrification of their vehicles to help reduce greenhouse gases and offer EVs that will provide their customers with the excitement of EV driving but with acceptable driving range and access to charging stations. They have committed to transition their fleets from internal combustion engines (ICE) to EVs over the next decade, and the deployment of EVs is well underway. The combination of this shift to a more software-based and electric future is bringing key changes to the industry such as a software-defined EV motor.
The transformation to software-defined electric vehicles (Source: NXP Semiconductors)
What is a software-defined EV motor?
The EV motor will benefit from the industry’s move to the SDV approach that offers access to deep vehicle data for monitoring the performance and aging of the EV motor, along with more powerful automotive microcontrollers that can support new features over time and deployment of software upgrades through OTA updates. The software-defined EV motor becomes a dynamic product that evolves and improves over time by leveraging real-time, in-vehicle data and supports cloud development and feature-enhancement deployment.
A software-defined EV motor solution impacts all development stages of the EV motor control system. It can enable faster development cycles, enhance performance, monitor for maintenance needs and extend the lifetime of the system. The high-level lifecycle of a solution is shown in Figure 1.
Figure 1: High-level lifecycle of a software-defined EV motor solution (Source: NXP Semiconductors)
Design
The trend in motor control design is the use of high-level modelling tools like MathWorks’ MATLAB and Simulink, allowing designers to concentrate on using their key expertise on EV motor control and systems rather than programming. Modelling tools work at the algorithm level, where they can be optimized to improve efficiency and performance.
Modelling tools enable three key advantages: flexibility, speed and safety. Using a modelling tool allows algorithms to be tested and analyzed quickly in software without the need for hardware integration and evaluation. This offers both flexibility and speed in the design of an EV motor control module.
Developing at the algorithm level using a software model can be especially useful when analyzing strategies for smooth control of a motor. Once the model provides the desired features and performance, it can be directly synthesized into a program by the modelling tool. This approach supports safety applications because correct code is automatically generated by the tool that can be used in production systems. A nice benefit of model-based development is that cost can be reduced with less programming.
The only disadvantage of model-based development is that the code generated may be less efficient than code created by programmers. However, NXP offers an optimized Automotive Math and Motor Control Library (AMMCLib), providing essential building blocks that work with the Simulink code generation flow and improves the performance for key algorithm components.
Field-oriented control is a popular EV motor control algorithm that is supported by the combination of Simulink, Codegen and the AMCCLib to optimize performance on NXP processors. Also, the NXP S32K39 MCU and S32E real-time processor offer higher performance than traditional automotive MCUs for rapid and streamlined development. The continued improvement of silicon technology enables devices with faster clock speeds, more processor cores and more memory.
Figure 2: NXP model-based design toolbox for Mathworks MATLAB and Simulink modelling tools (Source: NXP Semiconductors)
Another way that software can influence the design of motor control systems is by replacing external hardware. While this strategy has been typically used to reduce the cost, it can provide flexibility if external hardware is replaced by a combination of hardware and software inside the MCU.
An example of this approach is the implementation of a safe software resolver that runs on the S32K39 MCU to determine the EV motor’s angular position and replaces costly external hardware. The software component provides greater flexibility and can support the addition of new features. It can also simplify control module designs because one ECU design can be used in different configurations that would normally require different ECUs. This can help decrease the number of ECUs that need to be developed and replaced in vehicles.
Maintain
With the availability of cellular connections to the vehicle and higher in-vehicle computing performance, vehicle health can be monitored and analyzed in-vehicle and remotely. This provides the opportunity to monitor the aging of vehicle components. A high-performance MCU like the NXP S32K39, in addition to performing motor control, can execute AI/ML processing to monitor systems in the vehicle and detect anomalies. Special hardware, such as DSP and ML processor cores, can be used to accelerate machine learning.
While monitoring parameters in the vehicle provides new capabilities, reporting the data back to the cloud enables new opportunities for data analysis. Processing in the cloud can run more advanced models and maintain “digital twins” of vehicles for a better estimate of preventative maintenance requirements. The data can be used to analyze vehicle data across vehicle fleets. In addition, lower-resolution “digital twins at the edge” can run in real time in the S32K39 MCU to help optimize the EV motor operation.
As an example, an OEM can use data from its fleet of electric vehicles to diagnose a problem with electric motors that are failing more than expected. Data from a single vehicle may not cover enough of the operating conditions or repetitions. By analyzing the data from the fleet of vehicles, the root cause of the problem could be identified. Fixing the root cause can reduce cost because the vehicles do not need to be serviced, offering a better customer experience and helping improve brand quality and loyalty.
Another feature is that while monitoring the vehicle, it is possible to measure key components to determine if higher performance is possible. A good example of this is measuring the characteristics of the power devices in the electric motor drive system to determine if they are capable of handling more power. Selling this feature offers a potential additional source of revenue to the vehicle manufacturer. It could be sold as a “one off” upgrade or for a monthly fee with continued monitoring.
While monitoring vehicles and predictive maintenance is an advantage for most vehicles, it is essential for emergency-service vehicles like police, fire and ambulances. These vehicles need to always be operational and not have any unexpected failures.
Extend
The lifetime of an EV could be extended by using collected vehicle data to drive digital-twin models in the cloud that can help improve algorithms and machine-learning models that run in the vehicle. Updates to the EV motor control software can be made with secure OTA updates. This can be done without the need to upgrade any hardware or service the vehicle.
In addition, the lifetime can be extended by adapting some vehicle features. This may be useful as parts age and are costly to replace, as slightly reduced performance and features may keep a customer content with their current vehicle. While this may restrict a vehicle manufacturer’s desire to regularly replace vehicles, it is better for the environment because resources are not needed for new vehicles.
The creation of the software-defined EV allows faster, more flexible and safe development, requiring the higher performance and features of new MCUs. The challenge for developers is to learn how to effectively use these new MCUs to leverage their features and performance.
More than silicon
The automotive industry’s move to SDVs and EVs are having major impacts across the vehicle, but one of the most important and exciting areas is the software-defined motor that is intelligently leveraging in-vehicle data and processing, along with the cloud, to improve over time using agile development. NXP has been investing in both SDVs and EVs with its S32 automotive processors that support end-to-end processing needs, including the central vehicle computer (S32G), EV propulsion control (S32E), EV zonal control (S32Z) and high-performance EV traction inverter control (S32K39), as well as other EV processing needs (S32K37).
However, silicon alone is just a foundation; it is not enough on its own. NXP has been working with industry partners, such as MathWorks for model-based design using MATLAB and Simulink and Amazon Web Services (AWS) for cloud services like AWS CodePipeline for code development, training and simulation and supporting OTA deployments to the EV motor controllers, to offer an end-to-end solution that can realize software-defined motors. This foundation is being built to offer a DevOps platform for software-defined EV motor innovations and accelerate customer developments.
Advertisement
