A PM2.5 prediction and early warning method and system based on nonlinear theory

Abstract

The invention discloses a PM2.5 prediction and early warning method and system based on nonlinear theory, a model training step and a model prediction step; the PM2.5 concentration time-series data is divided into two groups, which are respectively used as a training time-series data set and a test time-series Training set: S-level wavelet decomposition is performed on the data of the training time series data set, time-frequency analysis is performed, the one-dimensional information is expanded into high-dimensional information, the implicit information of PM2.5 historical data is extracted, and the training time series index data set is obtained ; Construct a prediction model; Train the prediction model; For the test time series training set, perform MLRC-LSSVR model prediction, perform variance analysis on the model prediction results, and obtain the upper bound value of the confidence interval as the final prediction result. The invention can provide adjustable parameters of the model, and adapt to the prediction and early warning work of PM2.5 concentration in different regions by changing the adjustable parameters.

Description

technical field [0001] The invention relates to the field of air quality prediction and early warning, in particular to a PM2.5 prediction and early warning method and system based on nonlinear theory. Background technique [0002] The main component of smog is PM2.5, which is a colloidal mixture of particles with a particle size of less than 2.5 μm. The influencing factors of PM2.5 are complex, and its concentration changes present nonlinear characteristics. [0003] At present, there are mainly two types of air pollutant concentration prediction methods: statistical models and deterministic models. Among them, the statistical model is generally based on historical data to establish a correlation model between air quality and influencing factors. Its advantage is that it has relatively low requirements for input data, but its prediction accuracy is low, it is difficult to reflect regional air quality, and it cannot determine the causes and sources of pollution. The numeri...

AI Technical Summary

Problems solved by technology

The single-step forecast works well, but when performing multi-step forecasts, each step of the forecast requires the output of the last forecast as input. In this iterative process, the last forecast result will affect the forecast result at the next time point , the error will gradually accumulate until the end, and the prediction effect will gradually weaken

[0006] In summary, in the prior art, there is still a lack of an effective solution to the problem of PM2.5 prediction

Images