|
Real-time monitoring of continuous automated production 1. ntroduction
to production In the process of production engineering in the petrochemical industry, the production process parameters determine the quality of the product. Adjusting the process parameters can change the product quality value. The production process parameters are closely related to the product quality. For a long time, a large number of production process parameters and product quality data have been stored in their respective databases. 2. Data collection of process parameters
and product quality values
Figure 1: fishbone diagram of
production process control Description: 1. Product quality affected by 5 process parameters in the figure 2. 1. t2, t3, t4 and t5 are the time that the material acts on a certain part of the material when the material flows in the pipeline. t6 is the time when the product quality analysis is completed. 2. 2. The red line segment is a section of material flowing in the pipeline. During the production process, the material (the red part in Figure 1) flows in the pipeline. When it flows to the process parameter 1, record the process parameter value v1 at t1, and when it flows to the process parameter 2, record the time t2 The process parameter value v2 is collected, so as to collect the process parameter values (v1, v2, v3, v4 and v5) at 5 different times; when t6 after a period of time, this part of the production is completed (the red part in Figure 1) , the product quality analysis result value (r) is also collected into its own database. Of course, the properties of the raw materials will also affect the quality of the products, so the property v0 of the raw materials must be collected, so that the complete process parameters and quality values (v0, v1, v2, v3, v4, v5 and r) are collected, that is, a Complete sample. In the long-term production, many such
data samples were collected from the database, in which the process parameter
values (v0, v1, v2, v3, v4 and v5) were stored
in the production real-time database system, and the
quality values were stored in the product quality
database system. 3. Data mining construction method
and real-time measurement quality We carry out data mining on the production real-time database system and product quality database system, and call out the process parameter values and quality values (as shown in Figure 2). Figure 2: Block diagram of the product quality system for
process parameter estimation 3. 1. Data matching: Because the process parameter data and product quality data are stored in different database systems, it is necessary to find the process parameter data and product quality data according to the data collection time, and match them into a complete sample. Lots of samples. 3. 2. Data screening: After the data is matched, some data will appear outliers or do not conform to actual laws for some reason, so these data should be eliminated. 3. 3. Weight configuration: 2. 3. 1. The process parameters have different effects on the quality value, so different weights should be given to the process parameters according to the actual influence. 3. 3. 2 Among each process parameter, some are temperature, some are pressure, and some are flow, so their order of magnitude (value range) is also different, which affects the quality value during calculation. It is small, but the actual situation is not necessarily the case, and the corresponding weights should be given to different process parameters to eliminate the undue influence of the order of magnitude on the quality calculation. 4. Create algorithm: comprehensively use metrology, statistics and other methods to establish the relationship algorithm between process parameters and quality. 5. Calculation by algorithm: When a set of real-time process parameters is collected according to the method of process parameter collection, the established method can be used to calculate the real-time product quality value. 3. Real time monitoring of product quality of production device |
