Control limit knowledge-based systems

Wu and Joseph [35] incorporated fuzzy logic into a knowledge-based control system for control of composite curing. They used the fuzzy logic to interpret the sensors and adjust the amount of control reaction on a simulated process. Even though the limitations of the simulator used did not allow full evaluation of the advantages of this system, it did show that the controller could react to material and process variations and improve the process plan. 

Nonanalytic control can be described as cloning the biological control [87,88]. The basic mechanisms for implementation of such algorithms are rule-based controllers (RBCs) [ 89]. RBC were applied for assistive systems to restore locomotion with handcrafted [82,85,90,91], and lately automatically determined knowledge base [92-97] with limited success. An RBC is a system implementing If-Then production rules, where If part describes the sensory and motor state of the system, while Then part of a rule defines the corresponding motor activity to follow. 

The SRK framework is used with the cognitive work analysis approach to conduct the worker competencies phase. Vicente (1999) presented a worker competencies analysis of the Dual Reservoir System Simulation (DURESS) II process control microworld. Here, the worker competencies are expressed in terms of the skill-, rule-, and knowledge-based behaviors required. Most of the demands associated with industrial process control are cognitive rather than perceptual-motor in nature, and only limited tasks require skill-based behavior. Vicente suggested that operators should have basic perceptual skills and should be able to recognize and keep track of the labels for the various components within DURESS II. He also suggested that operators should be able to directly interact with displays rather than have to... 

A fundamental objective of a computer system applied to automate a pharmaceutical GMP operation is to ensure the quality attributes of the drug product are upheld throughout the manufacturing process. It is therefore important that quality-critical parameters are determined and approved early in the validation life cycle. The exercise should be undertaken to a written procedure with base information from the master product/production record file examined and quality-critical parameter values and limits documented and approved for the process and its operation. In addition, the process and instrument diagrams (P IDs) should be reviewed to confirm the measurement and control components that have a direct impact on the quality-critical parameters and data. This exercise should be carried out by an assessment team made up of user representatives with detailed knowledge of both the computer system application and process, and with responsibility for product quality, system operational use, maintenance, and project implementation. This exercise may be conducted as part of an initial hazard and operability study (HAZOP) and needs to confirm the quality-related critical parameters for use in (or referenced by) the computer control system URS. 

A review of the current state of scientific and technical knowledge on evaluation of the pre-crash phase set the starting point for this thesis (Chap. 2). Safety evaluation can be conducted at different levels (e.g., component-, system-, vehicle-based or with focus on the overall benefits in traffic). The method of choice depends on the level of evaluation and the underlying research question. Functions of active safety rely on sensors which perceive information from their environment and are thus subjeet to uneertainty. Besides possible technical limitations, the prediction of future movements of all involved participants contributes to this inherent uncertainty. As a eonsequenee, systems subject to uncertainties will not work perfectly in the sense of reliability. False-positive activations, e.g., due to misinterpretation of information or technical limitations, will occur with consequences on acceptance and controllability of the system. With an increasing number of false-positive activations, acceptance by the driver will decrease. In case of severe interventions in traffic, such as high velocity reductions and sharp decelerations, false-positive activations become a matter of controllability for the driver and the surrounding traffic and can ultimately have a negative impact on safety. 

---