Research on Adaptive Heartbeat Mechanism for Nuclear Instrumentation and Control Communication Networks

In nuclear power instrumentation and control systems, the heartbeat mechanism is a commonly used method to maintain the effectiveness of long-term TCP protocol communication connections. However, traditional heartbeat mechanisms employ a fixed heartbeat interval, making it difficult to adapt to the dynamically changing network conditions of nuclear power plants under different operational scenarios. This may result in ineffective maintenance of communication connections, thereby affecting the real-time performance and reliability of communication. To address this issue, considering the composition characteristics of nuclear power instrumentation and control equipment, an adaptive heartbeat mechanism based on a robust linear regression model is proposed. First, the median absolute deviation (MAD) of residuals is used to modify the IGG weight function. Then, the improved IGG weight function is incorporated into the normal equations of the least squares method to establish a robust linear regression model-the MAD-IGG-LS model. During the experiments, the model's prediction performance and computational resource consumption were initially validated. Subsequently, varying degrees of random disturbances were introduced into the training set to compare the traditional linear regression model, the IGG-weighted linear regression model, and the MAD-IGG-LS model. The experimental results demonstrate that the MAD-IGG-LS model exhibits strong predictive capability and low computational resource consumption. Moreover, in terms of robustness, it shows significant advantages over both the traditional least squares linear regression model and the least squares linear regression model with only the IGG weight function introduced.