The purpose of this exercise is to automate a temperature control process with an LSTM network. The LSTM network is trained from a PID (Proportional Integral Derivative) controller or a Model Predictive Controller (MPC). LSTM (Long Short Term Memory) networks are a special type of RNN (Recurrent Neural Network) that is structured to remember and predict based on long-term dependencies that are trained with time-series data. An LSTM repeating module has four interacting components.

The LSTM is trained (parameters adjusted) with an input window of prior data and minimized difference between the predicted and next measured value. Sequential methods predict just one next value based on the window of prior data. In this case, the error between the set point and measured value is the feature and the heater value is the output label. Proportional Integral Derivative (PID) control automatically adjusts a control output based on the difference between a set point (SP) and a measured process variable (PV).

Objective: Train and deploy an LSTM network to adjust the heater (Q) to regulate the TCLab temperature to a requested set point.

ML for Engineers: https://apmonitor.com/pds
LSTM Case Study: https://apmonitor.com/pds/index.php/Main/LSTMAutomation

source