Practice and Application of LSTM in Temperature Prediction of HVAC System

The optimization of HVAC system is an important part of building energy conservations. As a complex nonlinear system with large inertia and large lag, the traditional HVAC system adjusts temperature based on the current indoor temperature, so that the control does not involve future changes. Therefore, accurately predicting the building indoor temperature will help more stable control and provide a reference for system tuning. Air handling unit dataset, which is established by the US Department of Energy for experiments, is used in this article, and the long short-term memory (LSTM) model is also used to predict the temperature. The MAE of predictions of the final model is about 0.077 °F, and the RMSE is about 0.107 °F. The experimental results show that the LSTM can efficiently learn the trends and features of the data, and higher prediction accuracy than the traditional back propagation neural network (BP) and multi-layer perceptron (MLP).

[1]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[2]  Geoffrey E. Hinton,et al.  Speech recognition with deep recurrent neural networks , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[3]  Qian Xu,et al.  MS-LSTM: A multi-scale LSTM model for BGP anomaly detection , 2016, 2016 IEEE 24th International Conference on Network Protocols (ICNP).

[4]  Hui He,et al.  Short-Term Non-Residential Load Forecasting Based on Multiple Sequences LSTM Recurrent Neural Network , 2018, IEEE Access.

[5]  H. D. McGeorge,et al.  Heating, ventilation and air conditioning , 1999 .

[6]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[7]  Zhang Xu.  Analysis of Air Conditioning Load Prediction by Modified Seasonal Exponential Smoothing Model , 2005 .

[8]  Frédo Durand,et al.  The visual microphone , 2014, ACM Trans. Graph..

[9]  Andrej Kitanovski,et al.  Neural network, ARX, and extreme learning machine models for the short-term prediction of temperature in buildings , 2019, Building Simulation.

[10]  Xiaojuan Liu,et al.  Uncertainty Analysis of Weather Forecast Data for Cooling Load Forecasting Based on the Monte Carlo Method , 2018, Energies.

[11]  Duncan S. Callaway,et al.  Using smart meter data to estimate demand response potential, with application to solar energy integration , 2014 .

[12]  Richard M. Schwartz,et al.  Fast and Robust Neural Network Joint Models for Statistical Machine Translation , 2014, ACL.

[13]  Ivan Medennikov,et al.  LSTM-Based Language Models for Spontaneous Speech Recognition , 2016, SPECOM.