Electric Drive Winding Temperature Estimation Using NTC Readings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electric drive systems, such as those in electric vehicles, face issues with thermal stress leading to phase unbalancing and potential damage due to lack of or malfunctioning winding temperature sensors, necessitating a method to predict and manage winding temperatures effectively.
Innovation Solution
A predictive model based on historical thermistor negative temperature coefficient (NTC) and winding temperatures is developed and used by a controller to estimate winding temperatures, allowing adjustments to operating parameters like switching frequency, DC voltage, and power factor to prevent damage and maintain system balance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If embedded winding temperature sensors are installed to monitor temperature, then winding temperature can be directly measured, but device complexity and cost increase
Solution Approach 1:
The patent uses NTC temperature sensors as intermediary devices placed in thermal communication with the windings through the stator core. These sensors indirectly measure winding temperature by detecting the temperature of the stator core, which is thermally coupled to the windings, thereby avoiding direct sensor installation in the windings while still achieving temperature monitoring
Solution Approach 2:
The patent replaces direct mechanical/physical temperature sensing (embedded sensors) with an indirect measurement approach using NTC thermistors and mathematical modeling. The system substitutes physical sensor installation with a combination of external temperature sensing and computational estimation to achieve the same measurement objective
2Device complexity
If NTC temperature sensors are used to estimate winding temperature, then device complexity is reduced, but measurement precision may be affected
Solution Approach 1:
The system implements a feedback mechanism where NTC sensor readings are continuously fed into a predictive model that has been trained on historical temperature data. The model adjusts its estimates based on the relationship between NTC readings and actual winding temperatures, improving measurement precision through iterative learning and adaptation
Solution Approach 2:
The patent transforms the temperature measurement problem by changing from direct winding temperature sensing to using NTC sensor readings as input parameters. A predictive model processes these parameter changes to estimate winding temperature, converting an indirect measurement into a precise estimation through mathematical transformation
3Measurement precision
If predictive model with historical data is used, then measurement precision is improved, but loss of time for model training occurs
Solution Approach 1:
The system performs preliminary model training during manufacturing or initial operation phases, storing the trained predictive model in memory before actual use. This preliminary action ensures that when the system enters production mode, temperature estimation can proceed without additional training time, as the model is already prepared and optimized
Solution Approach 2:
The patent separates the model training phase from the operational phase. Historical data is collected and used for training, then the trained model is stored and reused repeatedly. The training process is discarded after completion, while the recovered model continues to provide accurate temperature estimates without requiring ongoing training time
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables effective temperature management and prevention of damage in electric drive systems, even without embedded winding temperature sensors, by providing accurate estimates and adjustments to operating conditions, thus enhancing system reliability and longevity.
Implementation Method 1
retrieving, from memory and by the controller, predictive model indicating predictive winding temperatures associated with known thermistor negative temperature coefficient (NTC) temperatures
Data Source
AI summary
A method for operating an electric drive system is provided. For example, a controller retrieves, from memory, a predictive model indicating predictive winding temperatures associated with known negative temperature coefficient (NTC) temperatures, wherein the predictive model is based on historical NTC temperatures and historical winding temperatures, as well as one or more operating parameters. The controller receives current NTC information indicating one or more current NTC readings corresponding to one or more switching devices within the inverter. The controller determines estimated winding temperatures corresponding to the electric machine based on the predictive model and the one or more current NTC readings as well as current operating parameters. The controller provides instructions to adjust inputs to the electric machine based on the estimated winding temperatures.