Efficient and accurate monitoring and diagnosis are necessary for reliable and safe operation of the lithium battery system. During repetitive charging and discharging, the operating voltage increases gradually. This paper develops a progressive fault diagnosis based on box plot and expert system on battery knowledge to detect anomalies before battery pack completely fails. The method is to detect an abnormal value of the change rate of the cell voltage extreme difference at the SOC=50% during the discharge of the battery pack. Since the method detected the faulty battery before the second-use battery is completely rejected, it realizes the early warning of the progressive battery failure, and improves the reliability of the battery system.