A Neural Network Based Combined Forecasting Method for Bus Passenger Flow by Period

Abstract

The invention discloses a neural network-based time-period combination prediction method for bus passenger flow, which includes four steps: collecting basic relevant information, determining optimal input data, determining a combination prediction algorithm, and determining the performance of the prediction algorithm. This method fully considers the applicability of different algorithms in different situations, assigns the weight value of each time period to the two neural network algorithms in different time periods, and combines the prediction advantages of different single algorithms in different time periods to obtain the optimal prediction results in each time period. It makes up for the shortcomings of a single algorithm with large limitations, improves the prediction accuracy, prediction stability and generalization of the model, thereby improving the overall performance of the model, and making the bus passenger flow prediction more reliable and accurate. In addition, the present invention screens data through a single algorithm, and guarantees the prediction performance of the combined algorithm to the greatest extent under the condition that the amount of information in the input data is sufficient.

Description

technical field [0001] The invention relates to the technical field of bus passenger flow forecasting by time period, in particular to a combined forecasting method of bus passenger flow by time period based on a neural network. Background technique [0002] In recent years, in the face of the deteriorating traffic conditions, many cities have adopted measures such as restricting purchases, licensing, and vigorously developing public transportation. [0003] People pay more and more attention to the development of public transportation. The large carrying capacity of public transportation can greatly improve the efficiency of urban road utilization, and the routes and frequency of departures are basically fixed. rights and operational efficiency will be greatly improved, the development of public transport laws will help ease urban congestion, and the public transport system is constantly improving. For example, the rapid bus BRT system has appeared. The emergence of BRT ha...

AI Technical Summary

Problems solved by technology

[0009] In order to solve the above problems, the present invention provides a time-series variable weight coefficient combination algorithm based on a neural network-based time-series variable weighting coefficient combination algorithm with better generalization and more stable prediction performance based on neural networks. The advantages in different time periods make up for the relatively large defect of a single algorithm, and improve the overall forecasting performance of passenger flow forecasting. The method includes the steps of collecting basic relevant information, determining the optimal input data, determining the combined forecasting model, and determining the performance of the forecasting model, which are carried out sequentially.

Images