Crash Test Data Compression Using Grouped PCA Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for compressing and storing vehicle crash test simulation data are inefficient, particularly when dealing with large numbers of simulations that require finer geometry approximations and higher precision, leading to increased memory requirements and suboptimal compression rates.
Innovation Solution
The method involves dividing the database into entity groups and performing principal component analysis separately on each group's observation matrices, allowing for lossy compression with adjustable accuracy, where similar coordinates are grouped and quantized, and error values are stored for reconstruction, enabling efficient storage and retrieval of simulation data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If data from a large number of test sequences are compressed using conventional methods, then compression rate is achieved, but accuracy is lost and precision cannot be adjusted
Solution Approach 1:
The database is divided into entity groups based on geometric similarity, and principal component analysis is performed separately on each group. This segmentation allows different precision levels to be applied to different groups, maintaining overall accuracy while reducing total data volume.
Solution Approach 2:
The method introduces adjustable precision parameters that allow users to specify the desired accuracy level for reconstruction. By changing these parameters, the system can optimize between compression rate and reconstruction accuracy according to specific application requirements.
2Measurement precision
If finer geometry approximation and higher number of simulation results are archived, then measurement precision is improved, but memory requirements increase
Solution Approach 1:
By segmenting the data into entity groups and applying principal component analysis to each group separately, the method achieves efficient compression that can store a large number of high-precision simulation results without proportionally increasing memory requirements.
Solution Approach 2:
The adjustable precision parameters enable the system to optimize storage requirements while maintaining the necessary geometry approximation precision, allowing more simulation results to be archived in available memory.
3Volume of stationary object
If conventional compression methods are used, then data storage is reduced, but compression rate is insufficient for future requirements
Solution Approach 1:
The division into entity groups enables more effective application of principal component analysis, achieving higher compression rates by capturing the specific variation patterns within each group rather than treating all data uniformly.
Solution Approach 2:
By introducing adjustable precision parameters, the system can optimize the compression rate according to future requirements, allowing the same system to adapt to increasing storage efficiency needs without redesign.
Data Source
Figure 1~2b
Figure 3~4
Figure 5~6b
AI summary
A method for compressing observations of a multitude of test runs, in particular simulation runs, at one or more observation points, especially time points, wherein the observations comprise instance coordinates of instances of one or more grids, each instance coordinate being assigned to an observation point, in particular a time point, of a test run, comprising the following procedural steps: dividing instances of the grid into several instance groups and generating a multitude of group-related data sets from the observations, wherein the instance coordinates of the instances of a common instance group are summarized from the observations in group-related data sets; extracting the instance coordinates of the group-related data set for each coordinate type separately and summarizing the coordinate-type-related and group-related instance coordinates thus extracted in an observation matrix.Determine for each coordinate-type and group-related observation matrix: - a basis matrix whose columns form at least a subset of a basis, in particular an orthonormal basis, of the respective observation matrix; - a coefficient matrix, as a weighting of the columns of the basis matrix, in particular by performing a complete or partial principal component analysis separately for each of the coordinate-type and group-related observation matrices; determine a working dimension and reduce the basis matrix to a reduced basis matrix with a number of columns corresponding to the working dimension; and reduce the coefficient matrix to a reduced coefficient matrix with a number of rows corresponding to the working dimension; and store the reduced basis matrix and the reduced coefficient matrix.