Redundant features

Figure 1.7 An initial structure for the manufacture of benzene from toluene and hydrogen incorporating some redundant features.

The approach to heat exchanger network design discussed so far is based on the creation of an irreducible structure. No redundant features were included. Of course, when the network is optimized, some of the features might be removed by the optimization. The scope for the optimization to remove features results from the assumptions made during the creation of the initial structure. However, no attempt was made to deliberately include redundant features. 

An alternative approach is to create a reducible structure that deliberately includes redundant features and then subject this to optimization. Redundant features are then removed by the optimization. 

Figure 1.9 A superstructure for the manufacture of benzene from toluene and hydrogen incorporating some redundant features. (From Kocis GR and Grossman IE, 1988, Comp Chem Eng, 13 797, reproduced by permission.).

In Chapter 1, two alternative ways were discussed that can be used to develop the structure of a flowsheet. In the first way, an irreducible structure is built by successively adding new features if these can be justified technically and economically. The second way to develop the structure of a flowsheet is to first create a superstructure. This superstructure involves redundant features but includes the structural options that should be considered. This superstructure is then subjected to optimization. The optimization varies the settings of the process parameters (e.g. temperature, flowrate) and also optimizes the structural features. Thus to adopt this approach, both structural and parameter optimization must be carried out. So far, the discussion of optimization has been restricted to parameter optimization. Consider now how structural optimization can be carried out. 

In many circumstances, the supplied system will be based on a standard COTS product and include additional features that are superfluous in the intended context. These features cannot normally be disabled because they are integral to the COTS product. Such redundant features should be included in the Functional Specification, noted as superfluous and, if possible, rendered inaccessible to users within the implemented computer system. Standard features that support compliance, such as audit trails for electronic records, should be used even if not defined within the URS. In such circumstances it may be necessary to make additional design allowances for the inclusion of these features (e.g., for audit trail functionality, extra storage capacity may be required). Standard documentation for COTS products can be referenced by the Fnnctional Specification, if available for inspection, rather than reproduced. Care must be taken to refer to the correct version of COTS documentation and to keep cross-references up to date following any system upgrades. 

Installed spare capacity, including fail-safe redundancy features Future expansion with any imphcations for reduced performance Plant external I/O signals (type, format, range, accuracy, timing)... 

LIMS must be backed up on a regular basis to maintain the security of the database. The regulatory inspectors will not accept that data within the database has been lost due to the failure of the server or other incidents (e.g., fire or flood). In order to prevent losses the pharmaceutical manufacturer is responsible for implementing a reliable, robust, and documented backup regime. The frequency of backups must be assessed as part of the GxP assessment process as this will be determined by the frequency at which data will be entered into the database and the redundancy features in the server, and should take into account the risks of data loss. Automation of the backup regime is acceptable provided that it is validated normally as part of the operational testing phase of the project. 

Description of redundancy features (disk mirroring, RAID devices) Establishment of initial standard client... 

Installation qualification Change control reports Parameter change control records Descriptions of hardware redundancy features... 

Life is redundant. There are many redundant features incorporated into living things. This makes life very robust. Because of redundant structures and processes, living things will... 

The best correction to these problems is not always obvious. In the first example of the heat exchanger, a pressure sensor with redundancy features could be installed between the primary and secondary loops such that when the positive pressure differential falls below a stipulated value (well above the point at which the pressures are reversed), operations should be shut down and the secondary loop emptied. A better solution would be for the heat exchanger to incorporate an intermediate loop so that the secondary and primary sides could not be directly coupled. 

No discussion of crew responses to system failure is complete without considering the Fail Safe design concept. A system is fail safe if, in the event of a failure, the system or component automatically reverts to one of a small set of states known to be safe and thereafter operates in a highly restricted mode. This may involve complete loss of functionality, or reverting to back-up/redundant features. See AMC25.1309 (Amend 17, para 6b) and Kritzinger (2006, Chapter 7) for more information on this topic. 

Fuel cells offer operational advantages such as electrical energy control, quick ramp rate, remote and unattended operation, and high reliability because of inherent redundancy feature. 

The purpose of feature selection in chemical data processing is twofold firstly it can help us to find the most influential features for the target function that we are interested in secondly it can help us to delete the irrelevant, redundant features or the noisy features, in order to improve the quality of data set suitable for data processing. 

Although the deletion of redundant features can improve the generalization ability of resultant mathematical model, in many cases, the information of the deleted features can be recovered by using these deleted features as a part of output of learning machine in mathematical modeling. This novel method is called multitask learning. This method will also be briefly described in this chapter. 

After feature selection, there are usually many features not selected, and the information hidden in these discarded features shall be lost. In recent years, a new concept named multitask learning (MTL) has been proposed to reuse the redundant features [18 19]. The basic idea of multitask learning is to use the selected features as the input feature set, and combine the target values with some of the discarded features as the target output. It has proved that MTL can improve the accuracy of prediction of the base learning method. 

For the quadratic and nonlinear data sets, SVM-BFS can detect the irrelevant feature at the first step, while it cannot detect the irrelevant one for the linear data set, we think the reason is that the nonlinear redundant features can cause larger error than the random feature using the linear SVM. 

From the above experiments, we can see that the feature selection using support vector machine can efficiently eliminate the irrelevant feature and the redundant features. In the case of nonlinear data set, some features eliminated are not the functional but the original ones. Since the functional features can be also used to form a complete independent feature subset. It can be concluded that the result is also satisfactory. 

The optimal inspection period T can be obtained graphically by plotting the expected downtime D(T) against the inspection period T. The optimal period will be such that D(T) is minimised or alternatively, such that the availability is maximised. The result obtained from the described method only reflects the availability of the component analysed and does not account for any redundancy features incorporated within the system. 

---