Speed prediction device and method thereof

A speed prediction method comprising:

• selecting a prediction section of a road for which to predict a speed of a vehicle and rear sections located behind the prediction section in a traveling direction of the vehicle;
• receiving driving information including speeds of vehicles passing the prediction section and the rear sections, receiving time information including one or more pieces of information on time, date, day of a week, and month at a time of driving of vehicles, and receiving section information including a length and a speed limit of each of the prediction section and the rear sections;
• generating processing information including an average speed obtained by giving a higher weight on the prediction section than weights on rear sections and giving a higher weight on a rear section closer to the prediction section than a weight on a rear section farther to the prediction section and calculating an average using the weights and the driving information;
• generating indicator information representing a degree of road congestion by calculating actual transit times of the vehicles for each of the prediction section and the rear sections by dividing the lengths of the prediction section and the rear sections respectively by speed values of the vehicles included in the driving information, calculating transit time limits of the prediction section and the rear sections by dividing the lengths of the prediction section and the rear sections by the respective speed limits, and dividing the actual transit time by the transit time limit;
• and
• generating a predicted speed by calculating the processing information, the time information, and the indicator information and inputting the calculated information including a real-time speed of the vehicle traveling on the road into a prediction model trained to predict a future speed of the vehicle;
• and
• transmitting information on the predicted speed to a route search server that generates shortest-time route information by analyzing the predicted speed and transmits the shortest-time route information to a navigation device for a user to drive the vehicle according to the shortest-time route information
• wherein the prediction model is a deep neural network (DNN) model.