Self-adaptive reduced order thermal modeling of underwater propulsion motors
To enhance the comprehensive performance of underwater propulsion motors,an adaptive re-duced-order model was developed to analyze the dynamic multimodal thermal field performance character-istics. This approach addresses the issues of high resource consumption and local optimum traps encoun-tered by existing algorithms in handling high-dimensional complex problems. In the time domain,an'e-quation-free' non-intrusive dynamic mode decomposition method was employed to extract the primary dy-namic modes from time series data,achieving dimensionality reduction and mode decomposition for pre-cise system behavior description and future state prediction. In the spatial domain,a combination of an expectation-improvement-based adaptive strategy and radial basis functions was used for parametric ap-proximation and order reduction. By evaluating the prediction uncertainty of each candidate point in the design space,the model quality was iteratively improved,balancing global exploration and local exploita-tion. This multi-step collaborative modeling framework enables accurate field solution predictions from limited large-scale simulation datasets. The model' s effectiveness and reliability in predicting temperature variations under both normal and abnormal conditions were validated through experimental systems,dem-onstrating high accuracy and stability.
high speed motordynamic thermal characteristicsreduced order modeldynamic mode decompositionself-adaptive samplingradial basis function
NETL
NSTL
万方数据
