This paper presents a cluster water quality parameter collection and cloud analysis system, which is based on a NEROGM (1,1) to predict water quality accurately with non-equidistant stochastic oscillation sequences. Taking into account the non-equidistant characteristics of original data for water quality, a non-equidistant transformation is given to make a non-equidistant sequence become an equidistant sequence. Combining with the prediction capability of the traditional ROGM (1,1) (stochastic oscillation sequence Grey model) for equidistant stochastic oscillation sequence, a NEROGM (1,1) (non-equidistant ROGM (1,1)) is proposed to predict water quality. The water quality parameters can be collected centrally by Modbus to model NEROGM (1,1)s for different water quality parameters. The prediction data obtained by NEROGM (1,1) can be transmitted to cloud analysis system by 4G network. With the water quality parameters in Jilin province different areas, a cluster water quality parameter collection and cloud analysis system is carried out, and the effectiveness and feasibility is demonstrated. The cluster water quality parameter collection and cloud analysis system will provide guiding significance for water environment management. stochastic oscillation sequence.