I teach this course on signal processing and spectral analysis—specifically tailored for structural engineers (not electrical engineers, mind you!)—once every year at the University of California, San Diego, as part of the Structural Engineering Department curriculum. Both undergraduate and graduate students are welcome to enroll.

The course is divided into three main parts:

1. Signals as Standalone Quantities (Part 1): We begin by studying analog and discrete signals, both in their natural form and in the frequency domain.

2. Machine Learning for System Identification (Part 2): In this section, we explore how to learn system behavior—linear or nonlinear—using machine learning techniques. These systems may be "grey-box" (with some prior knowledge) or "black-box" (with no prior information). We also introduce basic concepts in feature reduction and feature analysis. 

3. Linear Time-Invariant (LTI) Systems (Part 3): Here, we focus on systems that transform input signals into output signals via a transfer function. We examine the linear dynamics of structural systems that are linear, time-invariant, causal, and stable, by starting with single-degree-of-freedom models and extending to multiple-degree-of-freedom systems subjected to dynamic loading. This section also covers concepts like filtering and windowing. I refer to these as "white-box systems," where the governing laws are known—such as the linear second-order differential equations used in solid mechanics.

Numerous real-world examples are included as part of the assignments to reinforce learning (not shared here). This course requires at least basic coding skills. I have found application of these ideas into wide ranging fields spanning from engineering, data science, to finance.

​

Please note that these lecture notes are a work in progress. You may (will) encounter typos or minor (or major) errors, and I do my best to update and correct them as I discover them (mostly when I teach). I hope you find these notes useful.