For example, the heat equation is ∂u/∂t−α∇2u=0, but what would a solution to this be?
A solution to this equation is a function $u(\vec{x},t)$ that describes the temperature of the material in question at different locations, and how it evolves over time.
We say this is a solution to the equation, because out of all possible functions $u(\vec{x},t)$, we have to find the one that satisfies the equation given (the heat equation).
The wiki article talks about solving the equation with the Fourier series. So essentially, it's not always possible to solve differential equations in the same way we can solve simple linear or quadratic equations, so instead we have to approximate them through various methods?
The issue is more that there are an infinite number of possible solutions. For example, in the one-dimensional version of the problem, any function of the form
$u(x,t)=A\exp(-\lambda\alpha{}t)\left(\sin(\sqrt{\lambda}x) + B \cos(\sqrt{\lambda}x)\right)$
will be found to satisfy the heat equation itself.
So we need some additional information to narrow down the solution to one that fits our physical situation. These are the boundary conditions and the initial conditions.
For example, we might have a situation where the left end of a bar (of length L) is connected to a ideal heatsink at 100 C, and the right end of the bar is connected to an ideal heatsink at 0 C. Then our boundary conditions are
$u(0, t) = 100$
$u(L, t) = 0$
And we might have an initial condition from the fact that before the bar was connected to those heat sinks it was heated to a uniform temperature of 25 C.
$u(x, 0) = 25$.
We use Fourier analysis to find the combination of all the possible solutions that satisfy these conditions.