MapReduce-Based BP Neural Network Classification of Aquaculture Water Quality

As information technology develops and prevails across the globe, the informatization and data processing efficiency in the aquaculture field are exerting increasing impact on the intelligent conversion of this field. The aquaculture water quality indicators were analyzed by a feedforward error back propagation algorithm (BP neural network) with strong nonlinear mapping capacity, and the complicated nonlinear relations among the parameters of the aquaculture environment were solved. An aquaculture element analysis model was proposed, the Johnson attribute reduction algorithm based on the discernibility matrix was used to optimize the traditional algorithm, and the network convergence speed was increased under a given accuracy. The MapReduece distributed programming model was then used to perform parallel design of the BP neural network algorithm to meet the needs of massive data processing in aquaculture platforms. Also, case studies were performed to analyze the aquaculture element model and the parallel learning algorithm, and the big data framework design and data analysis method are integrated to develop an efficient, fault-tolerant aquaculture data management, mining, visualization big data system.