| 摘要 |
PURPOSE:To calculate dynamic characteristics from normal operation data by selecting input/output variables and simplifying input/output relation, and lightening an arithmetic load and also decreasing an error in identification due to almost noncorrelative input. CONSTITUTION:Data 1 on the entrance composition of raw material, data 2 on the feed amount of the raw material, data 3 on the recycled amount, data 5 on the amount of a refrigerant, and data 6 on the amount of a catalyst are converted by an A/D converter 2 into digital values, which are inputted to a data processor 3. The data processor 3 inputs three outputs supplied from a polymerization reaction vessel 1 through an A/D converter 4, i.e. data 4 on polymerization vessel pressure, data 7 on exit composition, and data 8 on polymerization vessel temperature in addition to said five input data, and performs specific processing to calculate process dynamic characteristics. In this case, a correlating method is used as a means of extracting only variables having large causal relation, and input/output variables are selected on physical or chemical inference basis to select 3, 5, and 6 as inputs and 7 as an output, thereby simplifying input/output relation. |
