In automotive the biggest trend is the growth in hybrid electrical vehicles and electric vehicles as they act as a front liner of the eco-friendly transition of this century. This change in propulsion type opens the door to opportunities on the electronics that govern the hybrid car’s powertrain.
In a plug-in hybrid car (PHEV) the powertrain consists of two different physical engines that must be controlled at the same time. Usually the PHEV runs on the battery until a standard threshold of battery charge is reached, e.g. it runs on battery until the battery reaches 20% of its charge. This is done to give charging space in the battery to be recharged by other factors such as regenerative braking, and because it is not known when the battery will be plugged next. Once the 20% limit is reached the vehicle switches to the combustion engine. The combustion engine recharges the battery. However, when driven faster or when more power is needed from the car, this 20% limit is overruled; and the combustion engine is used to meet the specific demand at the time. This is suboptimal as the microcontroller (ECU) that steers the blending of these two engine -types does not have any historic or real-time information on when such demands will be happening.
To control these two modes of a PHEV a combination of multiple ECUs has to work closely with sensors. These microcontrollers have the potential to achieve significant savings in energy consumption. “Powertrain blending” is the smart process of determining the “right-mix” of using the power from the combustion engine and from the battery pack. Such smart blending is made possible by inferencing the required mix from different sensors - done by ECUs with low CPU-footprint. Smart, data-fed software can help to better decide when to draw power from which engine source.
This smart operation delivers more mileage with the same available power, i.e. a more efficient use of energy for different driving conditions,thereby reducing the operational costs. This can also be a smaller electric motor and lower battery pack power which help to reduce the production cost. Similar approaches can be applied to any hybrid powertrain, even non plug-in, although PHEV usually offers the largest potential for improvement.
Current mode of operation
Most Plug-in Hybrid Electric Vehicle (PHEV) implement powertrain operation in two modes: Charge Depleting mode (EV-mode) and Charge Sustaining mode (HEV-mode). This is often referred to as CD-CS operation. Charge-depleting or EV mode refers to a mode of vehicle operation that is dependent on the energy from the battery pack. Battery electric vehicles operate solely in this mode. Most plug-in hybrids operate in charge-depleting mode at start, and switch to charge-sustaining mode after the battery has reached its minimum state of charge (SOC) threshold, exhausting the vehicle’s pure electric range. When the battery charge drops below the threshold, the battery depletion is stopped and the vehicle controller adopts its blending strategy: A charge-depleting operating strategy in which the engine is used to supplement battery/ motor power. This prevents rapid battery depletion and maintains the state of charge.
Typically the PHEV behaves as an EV when the battery charge is high; and uses the engine when battery charge is low. This helps the car to operate as eco-friendly as possible until battery charge is available, which is sensible if the next charging stop is unknown. However deviations from this simple behavior can occur frequently e.g. when driving aggressively and the car needs a lot of power to cater.
Teraki and Widesense
Teraki’s smart data extraction combined with Widesense’s Model-based AI application provides a far better PHEV powertrain blending. This solution reduces energy consumption by using predictive insight about car behavior and future vehicle power demand. This insight is a significant progress from the current EV and HEV mode as it combines external and internal data by which it applies knowledge of environment; recorded mobility patterns and actual car/driver behaviour to dynamically mix the use of the two powertrain engines. The accurate prediction of vehicle power demand is derived based on the driving pattern, environment and feedback from sensors. The historical data from the vehicle is applied to learn mobility patterns on different routes. The traffic and weather forecasts are applied to predict the demand throughout the course of the trip; and high-resolution on-board telemetry predicts the instantaneous demand for the very near term. Widesense AI-powered model clusters all the information together to obtain the right mix ICE/battery usage and save 9% (!) in energy consumption. Teraki’s smart telemetry processing supplies the required high-resolution data to ‘fuel’ that insight.
With the help of larger variables such as weather and traffic, road segments and topography, driving behavior, vehicle status (incl. Battery SoC) and load we can predict and make insights that continuously improve with time for the vehicle’s optimal blending strategy that uses less energy than the CD-CS strategy that relies only on current load requirement from the car data.
As we slowly and surely progress into the future of connected vehicles, applied AI provides far better utilization of energy. For most AI-powered applications, it is essential to have rich sensor data. Teraki helps extract such high-resolution data from the car. This high-resolution sensor data helps the AI models work at a 30% higher accuracy rates. This ability to get that accurate information from sensors, enables the entire ICE/battery blending management and -hence - provides 9% energy efficiency improvement in the car.