Nuclear energy can make an important contribution to the low-carbon energy supply for Industry 4.0, while Industry 4.0 can reform this industry in return. As a typical and complex man-machine-network integration system, various faults, insufficient automation and stressed human operators limit the further popularization of nuclear power plants (NPPs). At the same time, these issues can be addressed with artificial intelligence (AI) technologies. 

The development of advanced information technologies, such as the Internet of Things, Cloud Computation, and Artificial Intelligence (AI), enables the realization of the industry 4.0 vision, where 4.0 represents the fourth industrial revolution to attain an advanced level of automation. Nuclear energy can contribute more to the low-carbon energy supply for Industry 4.0, while Indus- try 4.0 can also reform this industry in return. The nuclear power system is a typical man-machine-network integration system, and its research and development, construction, operation, and other aspects have shown sufficient complexity. 

There are three major barriers and risks for current nuclear power plants (NPPs). They are: 

  • Firstly, as a complex system, various faults and failures can occur in instruments, equipment, or processes of an NPP, and these errors can significantly impact NPP's performance and security. 
  • Secondly, another concern lies in the insufficient automation level of NPP management. Although NPPs have initially been digitized after decades of development, most NPPs still use many traditional operation and control methods, which reduce operating efficiency and increase the risk of accidents. 
  • Due to the above reasons, human operators in NPPs are under great pressure because of high control requirements. Human factors engineering is important and has gained lots of attention for the design of NPPs. 

AI plays a significant role in eradicating these barriers and developing a top-level design for future NPPs. 

Current AI-based applications in NPPs 

Nuclear fuel is the core material of NPPs, which is related to the safe and reliable operation of NPPs. To ensure the quality of fuel composition and prevent possible accidents with nuclear fuel, it is necessary to standardize and systematically manage nuclear fuel. AI technology has been used to manage and process nuclear fuel efficiently, and some simulation experiments have been carried out. 

With the digitalization of NPPs, more and more nuclear data are generated. How to deal with them to better manage and maintain NPPs is particularly important. AI techniques such as neural networks have been widely used in nuclear data processing and life assessment of NPPs. How- ever, there are some problems, such as difficulty in quantifying nuclear data, low diagnostic efficiency and accuracy, single life evaluation index, and the influence of manpower on NPPs. 

Autonomous control is a kind of symbol of the realization of high-level automation. It could ease the issue of insufficient automation a great deal if applied. It is believed that less human involvement and intervention would bring less possible error and better efficiency, which would lead to improved economics for the operation of NPPs. 

Fault detection and diagnosis (FDD) have always been an important research field concerning an NPP's safety. Currently, NPPs suffer from over-intervention of human operators and bad accuracy and efficiency in FDD. With the development of AI and other related technologies, more and more methods are being applied in FDD. 

Apart from the physical plant-centered technologies, AI also plays a significant role in improving human operator-centered technologies. Human-machine interaction (HMI) is a typical scenario in NPP as human operators would receive collected data from sensors or devices of the plant and make decisions based on that, making it extremely important in the nuclear power industry. A well-designed HMI in the control room could help reduce operator errors and ensure the security of NPP. 

Conclusion 

There are also still many other problems remaining to be solved when applying AI in NPPs. For instance, the interpretability of the model is poor. Numerous AI models are "Black Box" with end-to-end architecture, making it hard to explain or understand these models. Similarly, the generalization of the model needs to be further enhanced. Many current AI models are designed especially for scenarios like human face recognition, while we want the models to be capable of generalizing to similar scenarios. 

Sources of Article

Want to publish your content?

Publish an article and share your insights to the world.

Get Published Icon
ALSO EXPLORE