Road traffic flow prediction method

Abstract

The invention discloses a road traffic flow prediction method. The method comprises the following steps: (1), collecting road traffic flow; (2), processing and cleaning data; (3), performing chaotic initialization on population particle parameters to be optimized; (4), adopting the optimized parameters for LSVVM sample training; (5), outputting prediction results after accuracy requirements are met. With the adoption of the road traffic flow prediction method based on a least square support vector machine optimized with an improved particle swarm algorithm, global optimal performance of particles is greatly improved by a chaotic ergodicity and variable weight combined model, so that parameters of a support vector machine model are relatively optimized, accuracy and speed of the predictionmodel are improved to a certain extent, the new data generalization prediction error of the model is lower, and generalization performance is improved.

Description

technical field [0001] The invention relates to a road traffic flow prediction method, which belongs to the technical field of road traffic flow prediction. Background technique [0002] At present, there are many methods for short-term traffic flow prediction, and the predecessors have done a lot of fruitful work: using the fruit fly algorithm to optimize the least squares support vector machine for traffic prediction, but the fruit fly algorithm is prone to "premature" and convergence Strange circle with low accuracy; use the improved BP neural network to predict the short-term traffic flow of a street, and the BP neural network adds momentum to the system to prevent the system from "slipping" the local solution, thereby avoiding system oscillation , but it still does not solve the problem that may fall into local optimum; based on the hybrid model of non-parametric regression and support vector machine regression, using the K nearest neighbor search mechanism to reconstru...

AI Technical Summary

Problems solved by technology

[0002] At present, there are many methods for short-term traffic flow prediction, and the predecessors have done a lot of fruitful work: using the fruit fly algorithm to optimize the least squares support vector machine for traffic prediction, but the fruit fly algorithm is prone to "premature" and convergence Strange circle with low accuracy; use the improved BP neural network to predict the short-term traffic flow of a street, and the BP neural network adds momentum to the system to prevent the system from "slipping" the local solution, thereby avoiding system oscillation , but it still does not solve the problem that may fall into local optimum; based on the hybrid model of non-parametric regression and support vector machine regression, using the K nearest neighbor search mechanism to reconstruct the state similar to the current traffic flow time series, support vector machine parameters and K The parameters adopt the grid division method, and it is difficult to find the global optimal solution or satisfactory solution; the wavelet neural network model is used to predict and analyze the road traffic flow, which combines the strong nonlinear ability of the neural network and the characteristics of wavelet local analysis. Frequency and window decompose the signal in the frequency domain and time domain, but its generalization ability is not high

Images