Pichia Pastoris Fermentation Control Using Multi-Model Predictive Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Pichia pastoris fermentation process is a highly nonlinear and strongly coupled system, making it difficult to establish an accurate mathematical model, and traditional control methods like PID and dynamic model-based control fail to achieve effective control results due to strong non-linearity and parameter variations.
Innovation Solution
A multi-model predictive control method using fuzzy C-means clustering, least squares support vector machines, and improved particle swarm optimization to establish optimal sub-prediction models, followed by designing a corresponding model predictive controller and constructing a multi-model fusion controller for improved control inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional PID control or dynamic model-based control is used, then the control system is simple to implement, but the control effect is poor due to strong non-linearity and parameter variations
Solution Approach 1:
The patent divides the nonlinear fermentation process into multiple local linear regions using fuzzy C-means clustering. Each cluster represents a specific operating region with its own simplified linear model, making the control system manageable while capturing the overall nonlinear behavior through piecewise linear approximation.
Solution Approach 2:
The patent implements dynamic model switching based on real-time fermentation state. The fuzzy inference system dynamically determines which local model to use at each moment, allowing the control system to adapt to changing process conditions while maintaining computational simplicity through localized linear models.
2Measurement precision
If a single global model is used for the fermentation process, then the model structure is simple, but the prediction accuracy is insufficient due to strong non-linearity and time-varying characteristics
Solution Approach 1:
The patent segments the global process model into multiple local linear models, each valid in a specific operating region identified by fuzzy clustering. This segmentation allows accurate prediction within each region while keeping individual model structures simple and computationally tractable.
Solution Approach 2:
The patent changes the model parameters (which local model to use and its weight) based on the current fermentation state. The fuzzy inference system adjusts model parameters dynamically, switching between different local models to maintain high prediction accuracy across varying operating conditions without requiring a single complex global model.
3Adaptability or versatility
If model parameters are fixed, then the control system is stable and easy to implement, but the adaptive ability is poor due to strong time-varying characteristics of the fermentation process
Solution Approach 1:
The patent makes the control system dynamic by implementing real-time model switching based on fermentation state. The fuzzy inference system continuously evaluates current conditions and adjusts which local model is active, enabling the system to adapt to time-varying characteristics while maintaining implementation simplicity through localized linear models.
Solution Approach 2:
The patent incorporates feedback through the fuzzy inference system that continuously monitors fermentation state and adjusts model selection accordingly. This feedback mechanism enables adaptive behavior by using real-time process information to determine the appropriate local model, improving adaptability without requiring complex adaptive algorithms.
Data Source
AI summary
Disclosed is a multi-model predictive control method for a Pichia pastoris fermentation process, including: dividing prior data into m training sample clusters by using a fuzzy C-means algorithm (FCM); obtaining, for each sample cluster, a corresponding prediction model by using a least squares support vector machine (LSSVM) and an improved particle swarm optimization method (IPSO); then, designing a corresponding predictive controller; and finally, calculating a deviation between an output of an object and an output of each sub-prediction model at each sampling time to establish a multi-model fusion predictive controller. According to the method, the adaptive ability of the model is improved and an actual state of a nonlinear system is described more accurately.