An absorber modeling method for optimal control of absorption refrigeration system operation

This article proposes a modeling method for the SPSO-WNN neural network in absorbers used in absorption refrigeration systems. The method establishes the input-output structure of the absorber model by analyzing the absorber’s working principles. A WNN neural network with a single implied layer is selected as the internal structure of the model, and the number of nodes in the implied layer is determined using the root mean square error and a trial-and-error method of decision coefficients. To address the limitations of the WNN neural network and improve the model’s prediction accuracy, the article uses Morlet wavelet basis functions and a simple particle swarm algorithm to improve the determination of the traditional neural network’s activation function in the hidden layer and the parameters of the neural network’s weights, scaling factor, and translation factor. The results of experimental simulations show that the absorber SPSO-WNN model established in this paper has less error than the traditional WNN neural network model, with the root mean square error of the two outputs being 1.73% and 2.36%, respectively. The model established by this method is simple and effective and can be used as an equation constraint in the optimal operation control strategy of absorption refrigeration systems.