Page 237 - Mechanical Engineers' Handbook (Volume 2)
P. 237

226   Data Acquisition and Display Systems

                          can facilitate research, for example, on what conditions lead to the best yield of a particular
                          product.
                             The ease of performing this operation is a particular advantage of relational databases.
                          There are some disadvantages, including overhead due to the access methods, extra space
                          requirements due to the creation of indexes, and costs and complexity associated with the
                          database management system. Indexes may add as much or more than 100% to the size of
                          a database. Old data must be managed and removed as with any other storage system. This
                          typically is via an automated program, since the structure is not as simple as just looking
                          for the file creation date of a file.


           5.4  Using Third-Party Data Acquisition Systems
                          When data storage is fairly simple, it is not hard to store data in the above-mentioned
                          methods, but when one is using sophisticated methods of data compression, it is recom-
                          mended that systems that have robust implementations of those methods be used rather than
                          attempting to reinvent the wheel. They can be quite complex to implement reliably. Transfer
                          of data to those systems can occur through a variety of methods, including creation of files
                          that are captured by the other systems, insertion of data into standard interfaces such as OPC
                          or message buses, or insertion into database tables which are monitored by the other systems
                          (often ODBC links). Third-party systems often have software development kit (SDK) inter-
                          faces that allow the engineer with some programming skills to store data directly into the
                          system.
                             Process historians are optimized for storing time-based data. A technique used to provide
                          some relational capability to the data is the following:
                             • Store related data at exactly the same time stamp (time stored in the database for
                               when the data elements were collected).
                             • Treat data stored with the same time stamp as being part of the same record.
                             • Select a set of these ‘‘records’’ based on a time range.
                             • Search a variable for some attribute value (e.g., having some value or range of values).
                             • Provide to the display system the values of some related variable in the same record
                               having the same time stamp as the desired attribute variables.
                          This is functionally the same as performing a relational database query on a set of records
                          in a table, with criteria based on values in some columns.


           6  DATA DISPLAY AND REPORTING

                          There are a variety of ways to reference and display data acquired from a sensor and stored
                          in suitable media. The current value can be inspected, values can be stored for inspection
                          later, values can be trended, alarm conditions can be detected and reported, or some output
                          back to the process can be performed.


           6.1  Current-Value Inspection
                          Often, one wants to see the data as they are being collected. This can be of critical importance
                          in experiments which are hard or costly to repeat, allowing the researcher to react to situa-
                          tions as they occur. As it is collected, each data item will be called the current value for that
   232   233   234   235   236   237   238   239   240   241   242