Abstract: Coal mine drainage water is unpolluted groundwater with excellent water quality and potential to be developed into high-quality water resources. This study takes the drainage water of a coal mine in Inner Mongolia as the object. Through the analysis of Taste index and health index,it is found that the Taste index of water quality is low and the health index is high. In order to solve the problem that the Taste index and health index of produced water quality can not be considered in the traditional process,the hybrid nanofiltration reverse osmosis system is used to improve the traditional process. By comparing NE40,NE70 and NF270 nanofiltration membranes,the primary and secondary nanofiltration membranes are selected. The primary and secondary nanofiltration systems are used to improve the Taste index and health index of water quality respectively. The reverse osmosis produced water is used to control the total dissolved solids concentration of water quality in a reasonable range,and the total dissolved solids concentration of concentrated brine in the mixed desalination system is controlled below 1 000 mg/L. The recovery rates of primary nanofiltration,secondary nanofiltration and reverse osmosis systems are designed based on the prediction formula of water ion concentration.The results show that the NF270 membrane produced water quality of the primary nanofiltration system is better.The prediction formula shows that with the increase of recovery rate,the Taste index of produced water continues to decline and the health index continues to increase. It is more appropriate to adopt 70% recovery rate for primary nanofiltration. By comparing the predicted value of water ion concentration with the actual value,it is found that the prediction deviation is less than 10%,NE40 membrane is better for the secondary nanofiltration system. The prediction formula shows that with the increase of the recovery rate of the secondary nanofiltration system,the Taste index and the concentration of total dissolved solids in produced water accelerate to decline,and the health index first increases at a constant speed and then decreases sharply. The concentration of dissolved total solids in concentrated water first decreases and then increases. The higher the reverse osmosis recovery rate,the more the critical point of change moves forward. When the recovery rates of secondary nanofiltration and reverse osmosis systems are 50% and 75% respectively,the predicted value of produced water quality meets the requirements of healthy and delicious water,and the predicted value of dissolved total solids concentration in produced water is suitable to be 224.01 mg/L,The concentration of dissolved total solids in concentrated water shall be controlled below the standard line to 987.85 mg/L. After measuring the concentrated water quality of the secondary nanofiltration system and using the salt balance,the real mixed produced water quality is obtained when the recovery rates of the secondary nanofiltration and reverse osmosis systems are 50% and 75% respectively,in which the Taste index is 2.23,the health index is 8.88,and the total dissolved solids is 238.4 mg/L,which is close to the predicted produced water quality,and the produced water quality is significantly improved compared with the raw water. The results show that it is reliable to design and adjust the recovery rate of each system of mixed desalination process based on the prediction formula. The economic analysis shows that the operation cost of the mixed desalination system is only increased by 7% compared with the traditional process,the risk of inorganic scaling is small,and the added value of produced water quality is increased. Therefore,the system has potential economic benefits,The process applicability analysis shows that the system has the application potential to deal with different raw water quality. In addition,this study is based on the analysis method of taste index and health index,which lacks a comprehensive evaluation scheme. It is necessary to further study and improve the evaluation model.