MATERIAL SELECTION APPARATUS AND METHOD FOR VEHICLE PARTS USING A BRS (BUZZ, RATTLE, SQUEAK) NOISE PREDICTION ALGORITHM

Smart overview of the Invention

The patent application outlines a method and apparatus for selecting materials for vehicle parts using a Buzz, Rattle, Squeak (BRS) noise prediction algorithm. It leverages an optimized machine learning model to understand the correlation between different materials and environmental factors. The process involves gathering material properties, test conditions, and BRS noise data for model training. Using Bayesian optimization, a Gaussian process regression model is generated to predict an affective quality index for each material, helping in the selection of the optimal material.

Technical Background

Traditionally, vehicle performance was judged based on driving, braking, and steering capabilities. However, as these aspects have reached a certain standard, the focus has shifted to enhancing passenger comfort by reducing noise, vibration, and harshness (NVH). With the rise of eco-friendly vehicles like hybrid and electric cars, which operate more quietly than traditional vehicles, BRS noise has become a more prominent issue. These noises arise from complex interactions between materials and environmental factors, necessitating improvements in material selection.

Solution Description

The proposed solution addresses the limitations of existing methods by using a machine learning algorithm to predict BRS noise based on learned correlations between vehicle parts and environmental factors. The apparatus includes several components: an input unit for setting conditions, a data acquisition unit for collecting material properties and BRS noise data, and a controller to manage these components and select the optimal material. A Gaussian process regression model is employed to predict noise and affective quality indices, guiding the selection process.

Apparatus Components

Key components of the apparatus include:

• Input Unit: Receives external signals to set various test conditions.
• Physical Property Data Acquisition Unit: Measures properties of rubber materials using devices like surface roughness and tensile force measurement tools.
• BRS Noise Data Acquisition Unit: Collects noise data from friction tests between rubber and dissimilar materials, such as aluminum or carbon fiber reinforced plastics.
• Controller: Manages data acquisition and selects materials using the regression model.
• Display and Storage Units: Present results and store acquired and processed data.

Methodology

The method involves acquiring training data to develop a BRS noise prediction model, generating a Gaussian process regression model through Bayesian optimization, and extracting affective quality index values for different materials. Tests are conducted to measure material properties and friction noise data. Psychoacoustic characteristics, such as loudness and sharpness, are evaluated to determine an affective quality index. The model iteratively adjusts hyperparameters to optimize predictions, ultimately aiding in the selection of the most suitable material based on comprehensive indices.