SMPL: Simulated Industrial Manufacturing and Process Control Learning Environments

Abstract

Traditional biological and pharmaceutical manufacturing plants are controlled by human workers or pre-defined thresholds. Modernized factories have advanced process control algorithms such as model predictive control (MPC). However, there is little exploration of applying deep reinforcement learning to control manufacturing plants. One of the reasons is the lack of high fidelity simulations and standard APIs for benchmarking. To bridge this gap, we develop an easy-to-use library that includes five high-fidelity simulation environments: BeerFMTEnv, ReactorEnv, AtropineEnv, PenSimEnv and mAbEnv, which cover a wide range of manufacturing processes. We build these environments on published dynamics models. Furthermore, we benchmark online and offline, model-based and model-free reinforcement learning algorithms for comparisons of follow-up research. † † Official documentation:https://smpl-env.readthedocs.io/en/latest/index.html Official implementation: https://github.com/smpl-env/smpl Code of experiments: https://github.com/smpl-env/smpl-experiments Preprint. Under review.

Sensor Selection and State Estimation of Continuous mAb Production Processes

“…The diagram presented in Figure 1 illustrates the upstream system of a continuous mAb production process, which is a modified version from [9]. The system primarily consists of a well-stirred bioreactor and a microfiltration unit/cell retention device.…”

“…The remaining content of the stream is allowed to pass on into the buffer tank with a flow rate denoted as F 2 . A cooling jacket is used around the bioreactor to regulate the temperature of the bioreactor components [9]. Finally, the stream from the buffer tank, which contains the product, metabolites, and any toxic by-products, is directed downstream for product purification.…”

“…Finally, the stream from the buffer tank, which contains the product, metabolites, and any toxic by-products, is directed downstream for product purification. The mathematical model of the upstream process is based on the work by [9,34,35].…”

“…The upstream process entails antibody production within a perfusion bioreactor, where a continuous supply of nutrients is introduced. The downstream process involves purifying the produced antibodies through a number of chromatography steps [9].…”

“…They represent the conversion of viable cells and total cells within the culture [9], where X is the cell concentration in cells/L, µ is the specific growth rate in min −1 , and µ d is the specific death rate in min −1 . The subscripts v and t signify viable and total cells, and the subscript 1 represents the bioreactor throughout the paper.…”