Noise Adjustment Using Explainable Boosting Machine
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating and adjusting noise generated by devices, such as cars and mechanical products, are inefficient and fail to accurately assess how device features contribute to noise evaluation, leading to unpleasant sounds for humans.
Innovation Solution
The use of an explainable boosting machine (EBM) to determine the dependence of noise evaluation characteristic numbers on device features, allowing for precise identification of contributing factors and efficient noise adjustment by changing device features with significant influence on noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If linear regression approaches are used to evaluate noise based on psychoacoustic variables, then the evaluation process is simple, but the accuracy of noise evaluation characteristic number determination is insufficient
Solution Approach 1:
The patent replaces traditional linear regression statistical methods with an Explainable Boosting Machine (EBM) system that uses decision trees and machine learning algorithms. This substitution enables non-linear relationships between device features and noise evaluation to be captured, significantly improving measurement precision while maintaining interpretability through the explainable nature of EBM.
Solution Approach 2:
The patent transforms the evaluation approach by changing from fixed linear regression parameters to dynamic parameters derived from trained EBM models. The system learns optimal parameter configurations from training data, allowing the evaluation model to adapt to different device types and noise characteristics, thereby improving accuracy without excessive complexity.
2Loss of information
If traditional noise evaluation methods are used, then the process is straightforward, but the ability to identify specific device feature contributions to noise is insufficient
Solution Approach 1:
The patent segments the noise evaluation into distinct device feature contributions by using individual decision trees for each device feature. Each decision tree analyzes the relationship between a specific device feature and the noise evaluation characteristic number, enabling precise identification of which features contribute most to unwanted noise and in what direction they should be adjusted.
Solution Approach 2:
The system implements feedback by using the trained EBM model to predict how changes in device features will affect noise evaluation. This feedback mechanism guides the adjustment process by indicating which features should be modified and in what direction, significantly improving noise adjustment efficiency while reducing information loss about feature contributions.
3Manufacturing precision
If device features are adjusted without precise evaluation guidance, then adjustments can be made freely, but the effectiveness of noise optimization is reduced
Solution Approach 1:
The patent applies preliminary action by training the EBM model in advance using comprehensive training data that captures the relationships between device features and noise evaluations. This pre-trained model provides ready-to-use guidance for noise optimization, eliminating the need for time-consuming trial-and-error adjustments and enabling precise manufacturing decisions to be made quickly.
Data Source
AI summary
A method for adjusting a noise generated by a device. The method includes: ascertaining, for each device feature, a dependence of a contribution of the device feature to a noise evaluation characteristic number on the value of the device feature by training an EBM with combinations of values for the device features and, for each of the combinations, an associated value of the noise evaluation characteristic number, each of a plurality of decision trees of the EBM ascertaining a contribution to the noise evaluation characteristic number for a device that has the value for the device feature; ascertaining values of the device features for a device of which the noise is to be adjusted; ascertaining changes for one or more of the device features to the values of the device features for improving the noise evaluation characteristic number; and adjusting the device according to the ascertained changes.