The invention relates to a vacuum concentration estimation method based on a self-adaptive expert system. A system comprises a parameter setting module, a measurement module, a data calculation module, a process point sampling module, a process point filtering module, an expert database module, a real-time calculation module, an output module and an integrated control system. The method comprisesthe following steps: (1) parameter consulting, (2) data measuring, (3) data calculating, (4) process point sampling, (5) process point filtering, (6) expert database updating, (7) real-time calculating and outputting and (8) external controlling. Multi-dimensional storage, automatic optimization and dynamic modeling can be realized, and the method has relatively high adaptive capacity.