1) There are two possible interpretations of “minimal generating set”. Unfortunately, your conjecture is false under either one.
Interpretation 1. $\{a_1,\ldots,a_n\}$ is a minimal generating set for $G$ if $G=\langle a_1,\ldots,a_n\rangle$, but for any proper subset $X\subsetneq \{a_1,\ldots,a_n\}$, $G\neq\langle X\rangle$.
That is, the set generates, but no proper subset does.
For a counterexample here, take $G=C_{12}=\langle x\rangle$ be cyclic of order $12$, and take $a_1=x^2$, $a_2=x^3$. Then $\{a_1,a_2\}$, and no proper subset of $\{a_1,a_2\}$ generates. The order of $a_1$ is $6$, the order of $a_2$ is $4$, but $x^6$ can be expressed as either $(x^2)^3(x^3)^0$, or as $(x^2)^0(x^3)^2$, so the representation is not unique.
Interpretation 2. $G=\langle a_1,\ldots,a_n\rangle$, and if $X$ is any set with $\langle X\rangle = G$, then $|X|\geq n$; that is, $\{a_1,\ldots,a_n\}$ generates, and no set of cardinality strictly smaller than $n$ generates.
This time take $G=C_2\times C_4$, with $x$ a generator of order $2$ and $y$ a generator of order $4$ with $\langle x\rangle\cap\langle y\rangle =\{e\}$. Now let $a_1=xy$ and $a_2=y$. This is a generating set of smallest possible order; $|a_1|=|a_2|=4$. But $y^2$ can be expressed both as $(a_1)^2(a_2)^0$ and as $(a_1)^0(a_2)^2$. So the representation is not unique either.
(The second example also works for the first interpretation, but the failure in the first interpretation is much more radical, since you can have really big “minimal generating sets” even for groups that can be generated by very few elements.)