This training module is the foundational class of the training series Analyzing Noisy Data From Physical Tests & Transient FEA. The module covers the key aspects of DPS (Digital Signal Processing) fundamentals, explained with terminology and examples applicable to Mechanical Engineers working in the field of linear and nonlinear structural mechanics, or similar. The concepts taught in the training apply to data obtained from physical testing, numerical simulation models, or a combination of both.
Course Objectives & Benefits
- Learn all about the evils of aliasing and how to collect digital data properly, avoiding aliasing.
- Perform spectrum analysis to interpret frequency content via DFS (so-called FFT) and other approaches.
- Filter data with confidence to remove high frequency noise or low frequency drifting.
- Learn why and how to properly upsample or decimate data.
- Learn about filter-induced distortions, including how to identify them and techniques to avoid them.
- Experience “pushing the same buttons” in different software can yield very different results (and the reasons this occurs).
- Get general-purpose strategies for applying DSP to FEA results & physical test data.
Course Outline
- Introduction & Motivation
- Explore cases from a large range of examples that demonstrate the critical need to handle noisy/messy data properly.
- DSP Fundamentals
- Time-Domain & Frequency-Domain
- Understanding frequency content
- Filtering
- General concept of digital filters (lowpass, highpass, and more)
- Impulse response and filter signatures
- Filter-induced distortions and minimizing their effects
- Data Collection From Physical Testing & Requesting Simulation Output
- Sampling, resampling, and avoiding aliasing
- Antialias filtering
- DSP requires constant sampling
- DSP Using Various Software (Your Results May Vary!)
- Learn why different software packages can yield strange spectrum and filtering results.
- Strategies For Applying DSP in the Real World
- General-purpose work-flows to process and explore frequency-rich data.
- Conclusions
- Workshops
- The class provides several fully-solved examples including their raw data files that the attendees can explore to further solidify their learnings.
Course Duration
- Four sessions, 3 hours each (12 hours total)
Prerequisites
- No previous training in digital signal processing is needed. Previous exposure to working with noisy/messy data is beneficial, but not required.
Training Delivery Options
- Public Web Training via Zoom
- Request a customized course for your company/organization.
- Can be delivered LIVE over the web or in-person at your site.
- You can customize content, including having use utilize one or more of YOUR datasets in workshop examples.
- Click here to request a quote.
