Multi-element time sequence prediction method

Abstract

The present invention provides a kind of multivariate time series forecasting method, this method comprises: Step 1, based on local regression seasonality and trend decomposition algorithm, decompose multivariate time series seasonally into trend sequence, cyclic sequence and irregular sequence; Step 2, Use linear or nonlinear regression algorithm to predict the trend sequence to obtain the predicted value of the trend sequence; step 3, based on the combination model, predict the cyclic sequence that introduces external factors and historical variables respectively to obtain the initial predicted value of the cyclic sequence, using the feedforward neural network The initial predicted value is fused to obtain the predicted value of the cyclic sequence; step 4, the predicted value of the trend sequence and the predicted value of the cyclic sequence are summed to obtain the predicted value of the multivariate time series. The invention solves the problem of prediction accuracy of multivariate time series, can more reasonably analyze the characteristics of multivariate time series and the influence of external factors, and obtain more accurate time series prediction results.

Description

Technical field [0001] The present invention relates to the technical field of multivariate time series forecasting, and in particular to a forecasting method based on a combined model and seasonal time series analysis of a multivariate time series. External influencing factors are introduced to model the time series, so as to improve forecast accuracy and reduce forecast errors. purpose. Background technique [0002] The time series analysis method originated in 1927 when the mathematician Yule proposed to establish an AR-Autoregressive model to predict the law of market changes. At present, time series forecasting has become the basis for decision-making in many fields, such as energy, finance, economy, and agriculture. [0003] Take the energy field as an example. In the energy field, energy load data can be viewed as a time series. Energy load forecasting, including power load forecasting, heat load forecasting, etc., can predict future demand based on historical energy consu...

AI Technical Summary

Problems solved by technology

Since a single machine learning model may only reflect a certain aspect of the original data, it can no longer meet the current energy load forecasting needs. For example, the GMM-Gaussian Mixture Model can introduce the impact of external factors on the load, but It is difficult to reflect the influence of historical load; ARIMA-Autoregressive Integrated Moving Average Model (ARIMA-Autoregressive Integrated Moving Average Model) is easy to reflect the influence of historical load, but it cannot introduce external factors, so it is necessary to combine multiple machine learning models, each model It can provide raw data information from different angles, and each model is connected and supplemented to achieve the purpose of accurately predicting energy load

Images