Data-Driven Adaptive Control Technology for Sewage Treatment Based on Testbed

This study investigates the sewage treatment process and the production of oxygen, using the input of sewage as the re-search object. The input and output data of the oxygen generator were collected, the input data of waste water and output data of processed water, and appropriate control variables and controlled quantities were selected. A BP feedback neural network model was trained and identified to represent the generation process. A model-free adaptive control algorithm based on multi-input and multi-output tight-format dynamic linearization was then applied. The Pseudo Jacobi matrix estimation of the S-Function function and the estimation function of the controlled quantity were written, and relevant parameters were adjusted to achieve data-driven control simulation of the treatment process.