Environmental Factor Control Model for Real-Time Microbe Cultivation

The patent creates a digital replica of the human operator's decision-making process by training an AI model on historical operation data. The model learns from the operator's expertise and replicates their control decisions, enabling automated real-time control that mirrors human judgment without requiring continuous human intervention.

The patent replaces the mechanical human operator system with an automated AI-based control system. The AI model processes sensor data and generates control commands automatically, substituting the human-in-the-loop mechanism with an autonomous system that can respond in real-time while maintaining the quality of expert decision-making.

The patent transforms discontinuous training data into continuous form by filling missing values through interpolation and temporal aggregation. This ensures the AI model receives uninterrupted sequences of environmental data, enabling it to learn continuous patterns of environmental changes while maintaining the simplicity of working with originally discrete sensor measurements.

The patent introduces data preprocessing techniques as intermediaries between the raw discontinuous sensor data and the AI model. These preprocessing steps (interpolation, aggregation, normalization) act as mediators that convert discontinuous data into continuous form, allowing the model to learn temporal patterns without requiring changes to the underlying sensor infrastructure.