1. IoT technologies and their application in fermentation process monitoring
The Internet of Things (IoT) is a paradigm that enables seamless interconnection of devices, providing remote monitoring and control capabilities. IoT primarily relies on Radio Frequency Identification (RFID) technology and is underpinned by advanced sensor technologies that offer the ability to measure and understand the physical world. As the IoT ecosystem rapidly evolves, its scope has expanded from basic sensing and RFID data collection to comprehensive development across transmission and application layers.
The IoT framework consists of three core layers:
- Perception Layer: The foundation of IoT, incorporating essential sensors, RFID, QR codes, cameras, and more.
- Transmission Layer: Responsible for transferring data collected from the perception layer using technologies such as MODBUS, CAN bus, LoRa, NB-IoT, Wi-Fi, and 5G.
- Application Layer: Provides a wide range of industry applications, including smart homes, smart grids, environmental monitoring, and autonomous driving.
In microbial fermentation processes, environmental monitoring and optimization of dynamic parameters are essential for process control. Real-time monitoring of fermentation parameters is therefore of critical importance. With the continuous advancement of IoT sensing capabilities and transmission technologies, large-scale data monitoring throughout fermentation is now achievable. IoT technologies are increasingly contributing to traditional fermentation industries by enabling smart sensing, mobile app integrations, and remote process management, ultimately reducing batch variation and improving product quality.
2. Integration of multi-source and heterogeneous sensor data
Modern fermentation plants deploy diverse hardware systems from different manufacturers, including fermenter systems, online sensors, and process monitoring devices. These heterogeneous systems often lack interoperability due to inconsistent interfaces and protocols.
Fieldbus technologies serve as the backbone of industrial automation communication networks, linking devices on the plant floor to centralized control rooms. Common protocols include EtherCAT, CAN bus, Profibus, and Modbus. Among these, Modbus is widely adopted, operating on a master/slave architecture where the master device issues data requests or commands, and slave devices respond accordingly, enabling bidirectional communication.
To address the challenge of integrating varied hardware, the OLE for Process Control (OPC) standard was developed. OPC leverages Component Object Model (COM) and Distributed COM (DCOM) technologies to encapsulate device drivers and communication programs into OPC servers. These servers abstract the complexities of underlying physical devices, unify protocols such as Modbus, and provide standardized data exchange interfaces for client applications, ensuring seamless data integration across the fermentation process.
3. Data cleaning and preprocessing in fermentation processes
Data integrity is essential for building accurate, intelligent models for fermentation control. However, missing data is common in fermentation processes due to factors such as sensor failures, storage errors, manual omissions, or equipment malfunctions.
Typical types of missing data in fermentation include:
- Online Data Loss: Often caused by sensor or storage malfunctions.
- Offline Data Loss: May occur due to missed sampling or testing errors.
- Sparse Offline Data: Compared to online data, offline measurements are less frequent, making integration with continuous data streams challenging.
Standard methods for handling missing data include:
- Deletion: Removing incomplete records. This method is only suitable when missing values are minimal to avoid compromising data integrity.
- Imputation: Filling in missing values using techniques such as:
- Mean Imputation
- Hot Deck Imputation
- Expectation-Maximization (EM) Algorithms
Effective data preprocessing ensures more robust and reliable models for process optimization.
4. Characteristics and visualization of fermentation process data
Fermentation data serves as a crucial carrier of microbial metabolic information. Effective data visualization techniques empower engineers to better interpret process dynamics, enabling timely adjustments and control strategies. Visualization can also help identify critical points for process optimization, driving breakthroughs in fermentation efficiency.
To visualize fermentation data effectively, it’s important to understand its structure and classifications:
- Online Data: Continuously captured via sensors.
- Offline Data: Periodic manual sampling and laboratory measurements.
Through mathematical calculations, these raw data sources can be transformed into derived variables with clear meanings, typically categorized as:
- Process Variables: Related to input/output and environmental conditions.
- Reactor Variables: Reflecting physical states within the fermentation vessel.
- Cell Physiological Variables: Indicating the health and metabolic state of the microbial culture.
Each variable group offers unique insights and can be analyzed to explain process variations, providing valuable support for scientific decision-making and process optimization through advanced visualization dashboards and analytics.
