E.S. are found to be suitable for those problems for which solution steps are not clearly defined. The action taken depends not only on the present values of data but on the outcome of previous decisions, historical data and trends.
1. Control and Monitoring:
Control and monitoring systems usually perform sequential tasks and are closely integrated. Monitoring is required for effective control and control is usually the ultimate objective of monitoring. The function of control/monitoring is the continuous interpretation of signals and actuation of alarms when intervention is needed. Expert systems have been developed to perform control and monitoring for computer system, nuclear power plants and medicine.
For example, PRANTS, assists managers in controlling the manufacture and distribution of Digital equipment of computer system. It monitors the progress of techniques in the implementation of assembly plans in accordance with customer’s orders, diagnoses problems, predicts shortages or surpluses of materials and prescribes solutions. REACTOR helps reactor operators diagnose and treat nuclear reactor accidents by monitoring instrument readings on a real-time basis.
ADVERTISEMENTS:
Limitations:
The requirements and conditions for monitoring and control vary according to the process monitored and are time and situation dependent.
2. Debugging:
Debugging is performed to identify remedies for malfunctions in a physical system such as a machine. The search space for remedies is usually limited, therefore debugging can admirably be performed by expert systems. Debugging usually accompanies diagnosis, which is the best understood by application area. The debugging task is usually employed for chemistry, computer systems, electronics, geology and medicine.
For example, TQMSTUNE (Chemistry) is used for fine tuning triple quadruple mass spectrometer.
ADVERTISEMENTS:
BLUE BOX (medicine). Treats various from of depressions.
Limitations:
Single fault is usually assumed with an appropriate correction prescribed: multiple, sequential or intermittent faults cannot be corrected.
3. Design:
The function of design is to establish specification to configure objects which satisfy particular requirements and constraints. Design expert systems are developed mostly for chemistry (organic molecules), computer systems, electronics (VLSI circuits) and military (map labeling) purposes.
ADVERTISEMENTS:
Limitations:
Problem scope may be so large that assessment of the consequences of design diagnoses is precluded. Sub-problems may be consequently portioned and sub-problem interdependency is difficult to be resolved.
4. Diagnoses:
Expert systems specialised in diagnosis troubleshoot faults in a system based on interpretation of data. This task is best understood and most highly developed of all expert system applications. They have been developed to locate defects in computers and electronic gadgets to diagnose faults in telephone networks and nuclear power plants, to diagnose nuclear reactor accidents, to locate problems related to drilling fluids used in drilling operations and to diagnose lung diseases.
Limitations:
ADVERTISEMENTS:
Limitations of expert systems specialised in diagnosis are similar to that of debugging. These systems tend to be based on single fault assumption and may not consider sequential, combined or intermittent faults. Many times, the faults originate with diagnostic equipment rather than with the physical system. The complexity of the physical system often precludes a full understanding of the system.
5. Instruction:
Instructional expert systems train students by diagnosing, debugging and correcting student behaviour or knowledge in a particular field. These systems develop a basis of the students acquired knowledge in a particular field, diagnose his/her deficiencies and prescribe training drills to correct the deficiencies. The instructional systems are used in expert systems in electronics (for example, the use of CAD system), plant operation, for example, to train operators of a steam propulsion part and medicine, for example, to instruct methods of anesthesia management.
Limitations:
One limitation with instructional expert systems lies in the complexity of interaction between teachers and students and the learning power of students.
6. Interpretation:
ADVERTISEMENTS:
Expert systems for interpretation perform analysis of sensor data to determine its meaning. The systems deal with real data measured from sensing instruments in the form of data stream, wave forms or pictures because of the real environment, this data may be flawed by noise, may be incomplete unreliable or erroneous.
These are developed for chemistry for example, to infer compounds molecular structure from mass spectral data, nuclear power plants for example, in accidents from reactor data, geology for example, to interpret dip-meter logs, medicine/or example, to interpret scanning densitometer data for inflammatory conditions and the military, for example, to interpret SONAR sensor data, detecting and identifying ocean vessels.
Limitation:
Interpretation may be based on partial or conflicting data from various sensors. These systems also interpret real-time processing of data.
7. Planning:
Expert systems specialised in devising plans and programs to achieve given goals, develop an entire course of action and occasionally reject a portion of a plan after implementation because it violates constraints. These systems are used in chemistry for example, to develop plans for assembling and testing the computer system on order, electronics/or example, to synthesise integrated circuits of nMos cells and military operations to develop plans for attacking enemy air fields.
Limitations:
The possible explosion of system complexity due to problem size and the consequences of action may effect planning.
8. Prediction:
Expert systems specialised in prediction and forecast the future from models or programs of the past and present which mirror real-world activity. Predictive expert systems combine traditional models programs with knowledge about the processes which originated them, to form the basis of predictions.
These systems can be used in military Applications, for example, to forecast when and where a major armed conflict will next occur, agricultural economic applications for example, to fore tall damage in agriculture due to black cutworm and economic applications to predict effects of a change in economic policy.
I and W expert systems assist an intelligence analyst in forecasting here and when an armed conflict will occur by analysing incoming intelligence reports on troop locations, activity and movements. PLANT was used to foretell the black cut worn damage to corn.
Expert System Research/Application in India:
The author has developed some expert systems. One of them was a knowledge- based expert system developed to avoid irregular heating up or cooling down and the associated thermal distortion of the turbine rotor and the casing (stator) of the 210 MW steam turbine. This expert system was developed in BASIC which has a powerful control structure.
The other expert system was a rule-based expert system developed in PROLOG which supports specifications and synthesis of computer communication protocol in the form of the communicating finite state machine. Bose and Dutta Majumder of Indian statistical Institute, Kolkata have developed two expert system.
CAN-SCAN-an expert system for use by a woman alone or in consultation with her general physician, for early detection, self-examination and counseling service in the domain of breast carcinoma (cancer) which has been commercialized and is available in the Indian and German markets.
ONCO-AID-an expert system which assists an oncologist (cancer specialist) for the diagnosis, treatment and planning and protocol management of breast cancer patients. Unlike CAN-SCAN (which has been commercialized) ONCO-AID is a more knowledge-intensive project.
An expert system for Breast cancer screening and Diagnosing ECSU/KBCS Technical Report. Dr. S.C. Patodi and his colleagues of the applied mechanics department, MS. Baroda University, Baroda, Gujarat have developed knowledge base expert systems, using various expert system tools for concrete technology, R.C.C. design and steel design.