Semiconductor chip test system based on machine learning

The invention relates to the technical field of semiconductor chip testing, in particular to a semiconductor chip testing system based on machine learning. The chip testing system comprises a chip testing unit, the chip testing unit is used for testing chips one by one and generating chip testing data, the output end of the chip testing unit is connected with a data analysis unit, and the data analysis unit is used for judging whether the chip testing data are qualified or not and generating judgment information. The output end of the data analysis unit is connected with a chip data prediction unit, and the chip data prediction unit is used for predicting whether a subsequent test chip is qualified or not according to the fluctuation frequency report. According to the invention, the method abandons the random extraction of the to-be-tested chip for detection in the prior art, locks the range threshold value of the to-be-tested chip according to the abnormal fluctuation condition of the data of the tested chip