圣莱安德罗上游水库的藻类水华导致东湾市政公用事业区的圣莱安德罗上游水处理厂出现味觉和气味问题。几次升级，包括安装颗粒活性炭过滤介质和中间臭氧，都不足以应对原水供应中的土工蛋白水平。1997年底，工作人员开始试验电子气味识别系统。该设备使用一个32维聚合传感器，该传感器与运行在个人计算机上的人工神经网络（ANN）相耦合。人工神经网络经过训练后，可以区分不同级别的气味强度，但不能直接测量浓度。 本文介绍了该设备，并给出了测试和网络培训的结果。包括表格，数字。

Algae blooms in the Upper San Leandro Reservoir have led to taste and odorproblems at East Bay Municipal Utility District's Upper San Leandro Water Treatment Plant. Several upgrades, including installation of granular activated carbon filter media and intermediate ozone were insufficient to cope with the geosmin levels in the raw water supply. In late 1997, staff began experimenting with an electronic odor identification system. This device uses a 32-dimensional polymerized sensor coupled to an artificial neural network (ANN) running on a personal computer that, once trained, can discriminate between different levels of odor intensity, but cannot directly measure concentration. This paper describes the device and presents the results of testing and network training efforts. Includes table, figures.