The `structural equations' are what I would call systems of equations, which would be nothing out-of-the-way in modeling with differential equations. This is an explanation of how to set up and use such systems when you are beginning with statistics, specifically, linear relationships with error terms. There are lists and cookbook examples of how to map out relations and equations and get from there, eventually, to testable hypotheses about your actual system.
It's not written in the mathematical style, that is, it isn't deductive; it builds up from examples and rules-of-best-practice instead, with standards of how to draw the flowcharts that make them a notation for how you're thinking about the system (curved vs straight lines, for instance. There's a whiff of the drafting template in this, updated to drag-and-drop software).I think it's longer than a deductive exposition would need to be, but on the other hand the examples are useful for a different case of mind.
Nonlinear relationships are dealt with, but not very thickly; there are separate works on nonlinear structural equations.
Find in a Library: Structural Equation Modeling and Natural SystemsSo wrote clew in Science.