Since you are not responding, I'll only offer a very simple concept that accepts both DC and AC. It provides a very simple answer to your question:
The max input is 0~24Vdc and -24V~24Vac. [...]
The voltage that I want to bias is regulated by the voltage divider so
the range of AC voltage coming to Arduino is between -3.3V and +3.3V.
I want to get the negative part of this input to positive, but I
really don't know a proper method to do this.
The following is mostly just a bridge rectifier, except that instead of diodes it uses BJTs (which can get a little closer to ideal than simple diodes may.)
Keep in mind that this does NOT take into account any reading of the datasheet information regarding your ADC. In fact, I'm almost certain that the following conceptual idea won't work well with your ADC. The source impedance needed by your ADC is probably much lower. (So it should be buffered or else the resistor values should be substantially reduced in magnitude.) Also, given the large resistor values it may be okay only up to \$600\:\text{Hz}\$ but not much beyond that (unless you get specialized BJTs or else change out the resistor values for smaller values.)
The only reason I'm offering it is because it's simple to understand why it works (it's just a bridge and a voltage divider.)
simulate this circuit – Schematic created using CircuitLab
There's also no opto-isolation here. Also, note that although I've labeled the leads as AC 1 and AC 2, they also work with DC input and you don't have to worry about polarity. That said, the above also won't work at all between \$-650\:\text{mV}\$ and \$+650\:\text{mV}\$ and will have some significant error until you reach a magnitude somewhat greater than a volt, or so. See this opamp full wave precision rectifier example for something a little better.
I didn't proceed to get anything that works about this problem, so I'm
not adding any circuits here. If there's any information missing
please let me know.
I can't tell you what's missing. Only you know what you want to achieve. But some suggestions might include the allowable input impedance (resistance and capacitance) of your voltmeter, if you will be wanting different ranges supported, the accuracy and precision of each range, frequency range and responses, and the operating temperature range over which all of the above is to be met, to name some thoughts.
The point of this exercise is to get you to think closely about and then discuss more details of what you hope to achieve. I'll delete this answer the moment you've provided better detail and others have decided to write a good answer for you.
