Performance Assessment of Liquid Cooling Plates Using Time-Delay Correlated Evidence Under Dynamic Conditions

Performance evaluation of liquid cooling plates is crucial for ensuring the reliable and stable operation of electronic devices. The evidential reasoning (ER) rule, an effective method for handling uncertain information, has been widely applied in this context. Traditional ER methods, however, face limitations in liquid cooling plate performance evaluation due to dynamic input variations and the challenges of integrating multisource, time-delay correlated indicators. In order to overcome these challenges, this article develops an ER rule with time-delay correlation (ER-TDC). A directed graph model of the liquid cooling system is first constructed to calculate the theoretical outlet flow rates and temperatures. Residuals are then obtained by comparing these with measured values, ensuring that the data used for performance evaluation remain unaffected by condition fluctuations. In order to enhance fusion stability and reduce uncertainty in evaluation results, a fusion sequence determination method that accounts for time delays is introduced. Additionally, the traditional ER rule is improved to refine the calculation of evidence parameters, incorporating the relative total dependence coefficient (RTDCD) as a discount factor to facilitate the effective fusion of time-delay correlated information sources, thereby generating belief distributions and expected utilities. Finally, some experiments on liquid cooling plates are performed to validate the effectiveness and feasibility of the proposed method. Experimental results show that the proposed method exhibits robust and practical performance under dynamic conditions and provides an efficient and reliable solution for performance evaluation of complex engineering systems.