Bosch Develops a Single Platform for Automotive Test Data Analysis and Visualization – Noticeperiod.com (1)

1

Welcome
2
Bosch Develops a Single Platform for Automotive
Test Data Analysis and Visualization
Noticeperiod.com
Industries Automotive Application Areas Data
analytics, Test and measurement Capabilities
Data analysis, Parallel computing, Desktop and
web deployment Products Used MATLAB, MATLAB
Compiler, Parallel Computing Toolbox, Signal
Processing Toolbox Country India
3
Bosch Develops a Single Platform for Automotive
Test Data Analysis and Visualization Bosch, the
worlds largest independent parts supplier to the
automotive industry, must meet various quality
requirements for its products. Using test
benches, components undergo functional,
endurance, and other testing. To enable
engineers to rapidly and accurately interpret
test data from measurement devices, test
benches, and vehicles, Bosch developed ENValyzer
(Engineering Test Data Visualizer and Analyzer),
a MATLAB based tool for analyzing and
visualizing measurement data. ENValyzer has
reduced the time engineers spend on data analysis
and visualization, improved accuracy, and
enabled enhanced visualization of test bench
results, says Sharath SL, project manager at
Bosch India. MATLAB saved us months of
development time because it had the data
importing, signal processing, statistical
analysis, and visualization capabilities we
needed built in.
Challenge Testing is one of the most critical
phases of the engineering product development
life cycle, and it demands enormous time and
effort. At Bosch, products are subjected to
various kinds of testing. Engineers must create
test scenarios within the limitations of
spreadsheets and other data post processing
tools. The resulting measurement data
4
comes in many formats, which are dictated by data
acquisition software, test bench manufacturers,
and other acquisition techniques. Bosch
engineers recognized several drawbacks with using
disparate tools. First, the in- house tools
required ongoing maintenance. Second, the amount
of data the teams needed to process was growing
beyond the limits of the tools. Third, the
results the tools produced were not accurate
enough to enable the engineers to precisely
determine the quality of the component under
test in many cases, the data post processing
software could not be enhanced with new features.
Fourth, using the tools to configure and analyze
the data involved numerous manual steps. Bosch
wanted to develop and deploy a single platform
for accurately analyzing and visualizing large
amounts of engineering test data from a range of
automotive systems. MATLAB enabled us to speed
the development of ENValyzer, a customizable,
easy-to- use tool for analyzing, visualizing,
and interpreting engineering data in a wide
variety of formats. Now, our engineers can
validate components faster and more accurately
than was possible with spreadsheets and
third-party tools. Sharath SL,
Bosch Solution The Bosch engineering tools team
used MATLAB to develop ENValyzer, a tool that
simplifies analysis without compromising
integrity and helps engineers arrive at better
decisions. The team used the object-oriented
programming capabilities of the MATLAB language
to simplify ongoing maintenance tasks, including
the creation of more than 250 class definition
files for the complete application. ENValyzer
uses MATLAB capabilities to read and write
measurement data in multiple formats from test
benches, vehicles, and acquisition
systems. Using MATLAB and MATLAB toolboxes, the
team added several general-purpose analysis
capabilities to ENValyzer, including functions
for regression analysis, curve fitting,
filtering, spectral analysis, data smoothing, and
principal component analysis (PCA) calculations.
They also developed MATLAB functions for
domain-specific analysis. To enable engineers
to visualize analysis results in ENValyzer, the
team used MATLAB to add single, secondary,
matrix plot, and multiple axis views. The team
added support for generating analysis and
visualization reports in PDF, HTML, and
Microsoft PowerPoint formats that show results
in tables and plots. Users can create and
customize reporting templates for various domains.
5

• Once the team had created this initial version of
  ENValyzer, they used Signal Processing Toolbox
  functions to add more advanced features for
  performing Fourier analysis, removing noise with
  Chebyshev and Butterworth filters, and applying
  Savitzky- Golay smoothing filters.
• They used these new features to automate
  processes frequently performed by test and
  validation engineers in specific domains. For the
  steering group, for example, they added an
  ENValyzer function that performs filtering,
  smoothing, and other signal processing
  operations on steering angle, torque, and other
  measurement channels to automatically assess the
  quality of steering gears.
• The team added support for running calculations
  on multiple processor cores simultaneously using
  Parallel Computing Toolbox, enabling engineers
  to analyze several data files at once.
• Using MATLAB Compiler, they created a standalone
  version of ENValyzer that can be used by test
  engineers who do not have MATLAB installed.
• Bosch is currently using ENValyzer in production,
  and now offers the tool commercially to other
  companies via the MathWorks Connections Program
  and through other marketing forums. Bosch
  engineers in India, Germany, and North American
  use ENValyzer to evaluate common rail system and
  steering system data and to validate steering
  gear and fuel-level sensors.
• Results
• Validation times reduced by an average of 4050.
  MATLAB enabled us to automate analysis steps in
  ENValyzer and to analyze several data files at
  once on multiple cores, says Sharath. With
  these improvements, our engineers have reduced
  validation cycle times by 4050.
• Three to four months of development time saved.
  A key requirement for ENValyzer was the ability
  to import data from binary files in various
  formats, says
• Sathvik Tarikere Sathyanarayana, senior engineer
  at Bosch India. We found it easy to understand
  the data structure and write the code in MATLAB,
  and that helped us
• complete development in about six months. It
  would probably have taken us three or four
  months longer if we had used another environment
  for developing the post
• processing software.

6
Learn more about Bosch www.bosch.com Team
Noticeperiod.com
7
Thank You!!