Inverter Thermal Prediction Using Error Maps

The patent introduces an error map as an intermediary component that stores pre-calculated temperature differences between predicted and actual temperatures. This error map serves as a mediator to compensate for measurement errors in coolant temperature, especially when coolant flow is abnormal. The error map contains correction values that adjust the predicted temperature based on actual temperature sensor readings, thereby resolving the contradiction between being able to perform temperature prediction and ensuring its reliability under varying coolant flow conditions.

The patent implements a feedback mechanism by continuously updating the error map with actual temperature measurements from temperature sensors. The system compares predicted temperatures with actual measured temperatures, calculates the difference (error), and uses this feedback to refine future temperature predictions. This closed-loop feedback system ensures that even when coolant temperature measurements are inaccurate due to abnormal flow conditions, the temperature prediction remains reliable through continuous correction based on actual device temperature readings.

The patent implements preliminary action by using the error map to pre-compensate for temperature prediction errors before making current limiting decisions. The system calculates corrected temperature predictions by applying error compensation values from the error map, which contains pre-stored correction data based on historical temperature measurements. This preliminary correction ensures that current limiting actions are based on accurate temperature predictions, preventing both premature current reduction (which would harm driving performance) and late current limiting (which would cause overheating).

The patent replaces direct reliance on potentially inaccurate coolant temperature measurements with a corrected temperature prediction system that uses error compensation. Instead of mechanically trusting the coolant temperature sensor reading alone, the system substitutes this with a computationally corrected temperature value that accounts for measurement errors. This substitution allows for more accurate thermal management decisions, maintaining driving performance by avoiding unnecessary current limitation while ensuring overheating prevention through corrected temperature awareness.