Research on Fast Evaluation Algorithm Based on Hopfield Neural Network

Abstract

The fast evaluation algorithm based on Hopfield neural network (HNN) was presented to solve problems of multiple influencing factors, complex evaluation mechanism, inaccurate evaluation results and slow evaluation process in water quality evaluation. In storage of evaluation criteria information, this algorithm performed orthogonal and symmetric processing of information to guarantee accurate storage of information and steady operation of the network. After storage of the evaluation criteria, the connection weights of the network were determined and the actual measured data input into the HNN model, where such data were evaluated in a fast manner. In addition, the stability of this algorithm was verified, proving that this fast learning algorithm could ensure steady operation of HNN, which converged to the minimum. The fast learning algorithm avoided modular calculations currently applied in the existing algorithms, to greatly reduce the computation. As a result, only one iterative calculation would be needed to obtain the correct water quality evaluation results, which significantly enhanced the speed of water quality evaluation. At the same time, this algorithm accurately saved the evaluation criteria to ensure correctness of the final evaluation results. Finally, this algorithm was applied in surface water environmental quality evaluation and eutrophic water quality evaluation and compared with other algorithms, leading to the conclusion that this algorithm could evaluate water quality of different kinds in a fast and accurate manner.