Let $\mathbb K$ be $\mathbb R$ or $\mathbb C$. Let $(X, \mathcal M, \mu)$ be a measure space and let $F$ be a Banach space over $\mathbb K$. I would like to define an integral of a function $f:X \rightarrow F$ that satisfies suitable conditions. My method is motivated by Bourbaki's Integration. They assume that $X$ is a locally compact Hausdorff space and $\mu$ is a Radon measure. So mine is sort of a generalization of their definition of the integral.
If $g: X \rightarrow [0, \infty]$ is a non-negative extended real-valued function, we denote by $\int^* g d\mu$ the upper integral of $g$(for the definition of the upper integral, see Lebesgue's monotone convergence theorem for upper integrals).
We denote by $\mathcal F(X, F)$ the set of functions $X\rightarrow F$. It is a vector space over $\mathbb K$. If $f\in \mathcal F(X, F)$, then we denote $N_1(f) = \int^* |f(x)| d\mu(x)$. We denote by $\mathcal F^1(X, F)$ the set $\{f\in \mathcal F(X, F): N_1(f) \lt \infty\}$. Then $\mathcal F^1(X, F)$ is a vector subspace of $\mathcal F(X, F)$. It is easy to see that $N_1$ is a seminorm on $\mathcal F^1(X, F)$. It can be proved by using Lebesgue monotone convergence theorem for upper integrals(Lebesgue's monotone convergence theorem for upper integrals) that $\mathcal F^1(X, F)$ is complete with the seminorm $N_1$(you just mimic the proof of the corresponding theorem in Bourbaki's Integration).
A function $f: X\rightarrow F$ is called a simple function if $f(X)$ is a finite set and $f^{-1}(a) \in \mathcal M$ for all $a\in f(X)$. $f$ is said to have a finite support if $\mu(f^{-1}(a)) \lt \infty$ for all $a\in f(X) - \{0\}$. Let $\mathcal S$ be the set of simple functions of finite support. It is a subspace of $\mathcal F^1(X, F)$. We denote by $\mathcal L^1(X, F)$ the closure of $\mathcal S$ in $\mathcal F^1(X, F)$ with respect to the seminorm $N_1$.
If $f\in \mathcal S$, we can define $\int f d\mu \in F$ in the obvious way. Since $|\int f d\mu| \le \int |f(x)| d\mu(x) = N_1(f)$, the map $\psi: \mathcal S\rightarrow F$ defined by $\psi(f) = \int f d\mu$ is a continuous linear map. Since $\mathcal S$ is dense in $\mathcal L^1(X, F)$ and $F$ is a Banach space, $\psi$ can be extended to a unique continuous linear map $\phi: \mathcal L^1(X, F)\rightarrow F$. We call an element $f \in \mathcal L^1(X, F)$ integrable and call $\phi(f)$ its integral. We denote $\phi(f)$ by $\int f d\mu$ or $\int f(x) d\mu(x)$.
Now I would like to find some criteria to determine when a function $X \rightarrow F$ is integrable. A function $f:X\rightarrow F$ is said to be $\mathcal B(F)$-measurable if $f^{-1}(U) \in \mathcal M$ for all open subsets $U$ of $F$, where $\mathcal B(F)$ is the $\sigma$-algebra generated by all open subsets of $F$.
A function $f:X\rightarrow F$ is said to be Bochner measurable if it satisfies the following condition.
There exists a sequence of simple functions $f_n,n=1,2,\cdots$ defined on $X$ such that $f(x) = \text{lim}_{n\rightarrow\infty} f_n(x)$ almost everywhere on $X$.
It can be proved that an integrable function is Bochner measurable(you just mimic the Bourbaki's proof of the corresponding fact).
Conversely suppose $f$ is a Bochner measurable function. There exists a null set $N$ and a sequence of simple functions $f_n, n = 1, 2, \cdots$ such that $f(x) = \text{lim}_{n\rightarrow \infty} f_n(x)$ for all $x\in X - N$. Since the pointwise limit of a sequence of $\mathcal B(F)$-measurable functions is $\mathcal B(F)$-measurable(see Limit of measurable functions is measurable?), if we redefine $f$ as $f(x) = 0$ for $x \in N$, then $f$ is $\mathcal B(F)$-measurable. Hence the function $x \rightarrow |f(x)|$ is measurable. If $\int |f(x)| d\mu \lt \infty$, it seems to me that $f$ is integrable in our sense.
How do you prove this if it is correct?