24th Conference on Practical Aspects of and Solutions for Software Engineering (KKIO@FedCSIS'23)

1 / 2023-05-14 21:17:47

Research on temperature prediction model based on fusion machine learning

RNN,BP neural network

In order to ensure the temperature stability of

the integrated op amp circuit and the measurement instru-

ment circuit, the ambient temperature prediction of the circuit

operation is critical to the circuit protection. The models of

BP neural network, RNN algorithm and LSTM algorithm are

constructed separately, and are conbined based on the stacking

ensemble algorithm, which highlights the nonlinear relationships.

The proposed forecasting frame integrates the advantages of

multiple models and combines with the regression forecasting

and time series forecasting. The reliable data is obtained through

focused crawler technology and data preprocessing. This paper

employs the normal distribution to eliminate abnormal datas.

The experimental results show that the accuracy of the ensemble

model is 93.2%, which is improved compared with the single

model.