<p>In Vehicular Ad-hoc Networks (VANETs), effective information exchange is essential. However, limited bandwidth restricts transmission capacity. Under heavy traffic, the channel load increases as the number of vehicles grows, which can lead to channel congestion. In particular, network congestion can severely impact network performance when it reaches a certain level. Therefore, it is important to control channel congestion to provide reliable messages to vehicles. We propose a Multi-parameter Congestion Control and Avoidance Protocol to reduce network congestion in dense situations. This scheme is divided into two aspects: real-time congestion control and congestion prevention. We focus on using vehicle density and combine various transmission parameters to design the scheme in this paper. The congestion control aspect first reduces the vehicle collision problem by adjusting the vehicle speed. Secondly, the transmission power is adjusted according to the real-time local density of the vehicle. The scheme selects a suitable relay node after adjusting the power so that the loss node can receive the message without failure. Finally, the data transmission rate and the beacon generation rate are changed based on the Channel Busy Ratio (CBR), therefore reducing channel congestion. In congestion prevention, the state of the node at the next moment is forecasted by online learning. We flexibly adjust the beacon generation rate according to the node state to reduce the amount of data in the network. The scheme keeps the delay below 70ms, maintains the data arrival rate above 88%, and the channel occupancy between 0.5 and 0.8. It successfully relieves congestion, ensures reliable data transmission.</p>