A short-term data flow prediction method based on long short-term memory network model

Abstract

The invention discloses a short-term data flow prediction method based on a long-short-term memory network model. The steps are: first obtain a plurality of training samples in observation points, then extract the characteristics of the training samples, and perform training on the training samples according to the characteristics of the training samples. Classification, respectively, the two types of data flow value change trend are severe and gentle, or the change trend is rising and falling; all training samples are used to train the LSTM model, and the main model after training is obtained, and then the two types are used respectively. The training samples train the main model respectively to obtain the first type sub-model and the second type sub-model respectively. Obtain the test samples of observation points, classify the test samples through the classifier, and then input the test samples into the first type of sub-model or the second type of sub-model according to the classification results, and use the first type of sub-model or the second type of sub-model to predict Quantitative flow value at the next time point of the outgoing observation point. The method of the invention improves the accuracy of short-term data flow prediction.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition and artificial intelligence, and in particular relates to a short-term data flow prediction method based on a long-short-term memory network model. Background technique [0002] With the continuous and stable development of the world economy, services in many countries are data flows, such as network load flow and traffic flow, which represent the characteristics of their services, and forecasting is the best way to optimize these services, such as forecasting Internet load The data provides more suitable resource scheduling for the scheduling of the next moment; for the traffic flow prediction, it can optimize the allocation of traffic resources. [0003] At present, many methods for short-term data flow prediction use a single model for prediction. However, data flow is a nonlinear and random signal, and it is difficult to distinguish and separate multiple mixed components. Therefor...

AI Technical Summary

Problems solved by technology

[0003] At present, many methods for short-term data flow prediction use a single model to make predictions. However, data flow is a nonlinear and random signal, and it is difficult to distinguish and separate multiple mixed components.

Therefore, there is a bottleneck in the prediction effect of a single model. When a model has a good prediction effect on the data flow in the case of congestion, the prediction effect on the data flow in the unblocked situation often needs to be improved. Therefore, the existing short-term data flow prediction method There is a defect that the prediction accuracy is not high

Images