论文部分内容阅读
由于海洋生态系统的高度复杂性和非线性,利用新兴的水信息学技术,包括模糊模式识别、遗传算法、人工神经网络等构建渤海湾叶绿素a预测模型。以渤海湾实测水质数据为依据,利用遗传算法的全局搜索能力优化BP神经网络的初始权值和阈值,从而避免陷入局部最优解,构建GANN叶绿素a预测模型。模型预测的均方根误差为3.81μg/L,仿真效果好于传统的BP网络模型。为进一步提高预测精度,用模糊模式识别方法遴选出与测试样本较匹配的训练样本,输入到GANN模型中进行训练。模型预测的均方根误差为1.45μg/L,预测效果有较大的提高。研究表明,所提出的改进模型应用在渤海湾叶绿素a的预测中是合理、可行的,具有较高的精度。
Due to the high complexity and nonlinearity of marine ecosystem, the chlorophyll a prediction model in the Bohai Bay is constructed by using emerging water informatics techniques, including fuzzy pattern recognition, genetic algorithms and artificial neural networks. Based on the measured water quality data in Bohai Bay, the initial weights and thresholds of BP neural network were optimized by using the global search ability of genetic algorithm to avoid falling into the local optimal solution and construct the GANN chlorophyll a prediction model. The root mean square error of the model prediction is 3.81μg / L, the simulation effect is better than the traditional BP network model. In order to further improve the prediction accuracy, the fuzzy pattern recognition method is used to select training samples that match with the test samples and input to the GANN model for training. The root mean square error of the model prediction is 1.45μg / L, the prediction effect is greatly improved. The research shows that it is reasonable and feasible to apply the improved model to the prediction of chlorophyll a in Bohai Bay with high precision.