On-line monitoring of pharmaceutical production processes by means of near infrared spectrum with hidden Markov model

Abstract

In recent years, a few pharmaceutical companies have started to use on-line monitoring for quality assurance. The state-of-the-art online monitoring method is based on near infrared (NIR) spectrum. Because the pharmaceutical production process is a dynamic process, the NIR method used alone may not work very well. This paper introduces a new method for on-line monitoring of pharmaceutical production process. It consists of several steps: First, it uses partial least squares (PLS) to extract the features from NIR spectrum. Next, it uses hidden Markov model (HMM) to model the production process. Based on two sets of powder blending experiments, it is found that the new method gives more precise information to help us control the process. It is believed that the proposed method is effective and can be used for various pharmaceutical production processes.