Artificial Intelligence And Fuzzy Logic Controller In A Plc System Computer Science Essay

Artificial Intelligence And Fuzzy Logic Controller In A Plc System Computer Science Essay In this research paper, I am trying to analyse more about advanced Programmable logic controllers.Today it is hard to imagine a industry without PLC and other automatic controllers. As the production becomes more and more efficient, the controller work faster and the system become more complex. Artificial intelligence (AI) techniques reduce the complexity and they are used through PLC-based process control system. The working of artificial intelligence consist of -diagnostic, knowledge, expert and the structure of an AI system. Function such as AI fault diagnostics in process help in controlling and successfully predicting the outcomes based on resident knowledge. Here i will be researching more about the applications of PLCs such as use of PLC s with fuzzy logic. Basic fuzzy logic and also its fundamental concepts will be analysed. Use of the fuzzy logic controller in practical applications include providing real time logical control systems. In the end i will be concluding on how advanced PLC are more efficient than the conventional PLCs. In industry use of automatic controllers is increased, the use of PLCs. The programmable logic controllers are based upon the on/off logic, in PLCs we use normally closed or normally open switch, and these switch can turn on or turn off the devices. PLCs contain a small processing unit, memory, input and output interfaces. But the PLCs are not able to represent the all data of the process and they are not able to take action to remove the faults. But with the help of the Artificial Intelligence we can make the system to make the decision on faults. Artificial Intelligence is the branch of computer science. In AI we use the data from the process to solve the many faults in the industry. All the information used by the AI is gathered from the person working in the plant/machine and log book also provide information about fault. The information about the fault-what kind of fault that was and how that was solved. Once all the information we gather from the process, all information is sto red in the memory of the PLC or elsewhere will be used by the PLC to solve the complex problems of the process. . II. ARTIFICIAL INTELLIGENCE IN A PLC A. Three type of AI system The classification of AI system is very difficult because they are used in the many applications but however we can classify in three types: 1.diagnostic 2.knowledge 3.expert.All three type of AI system have similar characteristics. The system become more and more sophisticated as the size of the data base increases and the extent how the process data is used. Diagnostic AI system. This type of system is the fault detecting systems. They detect the fault in the application and they do not solve the problem. For example if the temperature of a tank is decreased the diagnostic system can diagnose the fault by reading the thermocouple values. These system use the knowledge to reach on a fault conclusion, these type of system are used in the applications that use a small data base and knowledge. Knowledge AI system An knowledge AI system is the enhanced diagnostic system. These type of systems are able to detect the fault and process behaviours based on the knowledge and they are also able to take the decisions concerning the process and/or the possible cause of a fault. Expert AI system This type of system comes on the first position in AI applications. Expert systems are more capabilities than the knowledge system. The expert system provides a further capability for examining process data with the help of statistical analysis and the system predict outcomes of the process that are based on present process assessments. The outcome calculation may be a decision and with the help of that decision process maintain the output in spite of a fault detection. The knowledge used in the expert AI systems are more complex than in the other AI systems; these type of system generate more feedback information. The expert systems also require more refined software programming to make decision, since their decision trees involve more options and attributes. The implementation of the expert systems is only done by with the help of extra programming and they also need more hardware. The system use the transducers to make the decision in the process and the total number of transducer used in this system is more than the other system. Programmable logic controller use the AI system, it will need two or more than two processer to make the all programming for the system. PLC system require more speed to operate in real time, the system should be fast. The system has large data to operate in the real time due to large data system also need large memory to store that data . B. Artificial Intelligence System Architecture The block diagram (Figure 1 ) shows the basic architecture of an AI system. It has three primary elements: 1 Global database, 2 Knowledge database, 3 Inference engine. The block diagram show that expert block first, that block provide the knowledge to the AI system and the knowledge is received from a person who know about the plant/process, how the machine perform their operation. The expert sends the all information (about system maintenance, faults) to the knowledge engineer. The process of transmitting the knowledge and gathering data is known as knowledge acquisition. C:UsersPARVEENDesktopimg3.jpg Figure.1 Artificial Intelligence system Architecture Global Database Global database contain the information about the process and the system, how to control them. The information contained by the global database is about the input and output data flow from the process. The global database is the storage area, the information about the process stored. The data stored in global database can be used any time to make the AI decision to control the process. PLC have memories to store the data and the Global database resides in the memory of the system that makes the system to take the AI decisions. We can also use the AI system with computer and the Global database will be in the hard disk of the computer. Knowledge Database The knowledge database store the information as the global database store about the process and the all information is supplied from the expert. It also contains information about the faults, process, causes of the problems and their solutions as well. Moreover, all the rules that help to make the decision are also stored in the knowledge database. The diagnostic system has knowledge database and that is less complex than the knowledge system. The knowledge system is less complicated than the expert system. It stored in the system memory. Inference Engine All the AI system has inference engine. All the decisions are made in the inference engine. Inference engine use the knowledge database to make the decision about the process and after that inference engine execute the rules in the process. It also uses the historical data of the process to make the real time decision. PLC system contain the central processing unit, CPU perform all the operation for the system and the inference engine may be inside the CPU or it may not be inside data that depends upon the diagnostic, knowledge, expert. C. Knowledge representation In the knowledge representation all the AI strategies are organised and the knowledge engineer represent the input of the expert. The knowledge database is used for the storage of the representation. The knowledge from the expert is changed in the form of rules (IF and THEN/ELSE) and we call it rule-based knowledge representation. It make the system capable to take action and decision. A PLC system is used with AI, all the control strategies are executed by software programs. Whenever a fault is detected by the system and at that time system makes a decision, inference engine also use the knowledge representation. The decision will be in the form of software. D. Rule-based knowledge representation It uses the knowledge from the expert and make the decision with the help of that knowledge. The rules contain two parts, first part antecedent (IF something happens) and the second part consequent (THEN take this action). All the rules are made for the process and they can be complex. A simple rule-based System may make a simple diagnostic rule, such as: IF the temperature of a tank is less than the set point, THEN turn on the heater. A more complex diagnostic formula may contain rules that further depend on a more complex diagnostic formula and they involve the rules that are depend on Parent rules: IF case 1, THEN  ¾Ã¢â€š ¬Ã‚  ELSE nothing â‚ ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã‚ ½ IF case 2, THEN  ¾Ã¢â€š ¬Ã‚  ELSE something â‚ ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã‚ ½ IF case 3  ¾Ã¢â€š ¬Ã‚  THEN nothing â‚ ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã‚ · â‚ ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã‚ · The decision tree makes the system capable to take the decisions. The figure 2 shows how the decision tree works to get a decision on the given process data. C:UsersPARVEENDesktopimg4.jpg â‚ ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Ã¢â€š ¬Ã‚  Figure 2 â‚ ¬Ã‚  Ã¢â€š ¬Ã‚  Decision Tree E. Knowledge inference This is the method used to draw conclusion by gathering the data. When the system execute the main control strategy at that time the inference of knowledge takes place in the inference engine. The knowledge inference is also takes place in knowledge database when computation of rule is going on. In the small control system the knowledge inference takes place on local basis. But in case of large systems knowledge inference takes place in the hierarchical system. To design a AI based PLC we need hardware, the need of hardware depends on the involvement of the AI. In all AI systems some common methods of rules are used for the implementation of knowledge inference. Methods are: 1. Forward Chaining, 2.Backward Chaining. Forward chaining-This method is used to find out the outcomes of a given data and receives the information from the global database. Forward Chaining is done by two methods: depth first, breadth first search. Backward chaining-This is also similar to the forward chaining. Basically it is used to find the antecedents. F. Basic Architecture of an AI based PLC Large and complex distributed control systems are made by the combination of small systems. They can communicate with each other either directly or with the help of local area network. The AI is added in the large systems, global database, knowledge database and knowledge inference from these is distributed all over the system. These large systems are made by the combination of small system and all the local system has their own local data base and knowledge database. The PLCs in the diagram shows they perform inference engine computations. In the large systems the supervisory PLC use the all subsystem and their local database to make a complex decision. The main computer we call it blackboard hold the all information from the small units. The main computer applies all the complex AI solution. C:UsersPARVEENDesktopimg5.jpg Figure 3 Architecture of an AI based PLC III. FUZZY LOGIC IN A PLC In industrial automation Programmable Logic Controller combine the simplicity, and reliability. Fuzzy logic is a part of AI which deals with reasoning used to imitate human decision making and thinking in machines. The reasoning is transformed in algorithms. These algorithms are used when the data cannot be converted in the binary form. The output of the process is the input for the fuzzy controller. Fuzzy logic performs three main actions. First is fuzzification, in this action the data received at the input is converted in the fuzzy form. Second is fuzzy processing, which involves the transformation of the input data according to IF†¦THEN rules formed by the user at the time of design and programming of the fuzzy control systems. After finishing the fuzzy processing (rule-processing stage) the fuzzy controller reach an outcome. Third is defuzzification process, this is final step of the fuzzy controller. In this step the final output data is converted into the real output data and after that the data is sent to the process with the help of output interface. The fuzzy logic controller is placed in the PLC rack in this case the controller does not have a direct contact with the process, the fuzzy logic controller will send the defuzzification data in the PLC memory location and PLC send that data to the process by the interface module. In the most of the fuzzy logic controller have their independent interface ports and they are also connected to the PLC with the help of the plug. The fuzzy controller can communicate with the process through the PLCs input/out ports. PLC can be interfaced with the intelligent fuzzy controllers. Interface of Fuzzy logic with PLC A Fuzzy logic controllers input interface can read the data from the 8 devices and it can transmit the data to 4 output devices with the help of the output interface. This interface is able to perform 128 rules, each rule can have maximum IF conditions and the action will be in the form of two THEN. The fuzzy logic controller has capability to perform all its computations in only in 6 msec if fuzzy logic unit works separately of the processer, as a result it providing fast functioning of fuzzy logic control. Fuzzy Logic and I/O Communication In below given Table 1, the Fuzzy Logic Unit (FLU) uses the programmable controllers memory to store the control parameters and fuzzy logic controller uses 10 words or registers. The position of the FLU module in the rack tells about the registers addresses. Assuming that the position of the FLU module takes the addresses 110 through 119, the use of the addresses by the FLC module as follows: The first four bits (0-3) the first word is (word 110) and its first four bits(0-3) enclose, in BCD, the FLU module uses as the number of inputs. 15 number bit turns on the fuzzy processing of this word. The second word (word 111) specifies that the location of the input data stored in the PLCs memory. It tells the starting register address. TABLE I inputs: bits 0-3 of word 110 specify the number of inputs to be read (8 max) (e.g., I = 8) Word 111: starting address where input data is located (length of I) (e.g., address = 120) Outputs: bits 0-3 of word 112 specify the number of outputs to be written (4 max) (e.g., O = 4) Word 113: starting address where output data is located (length of O) (e.g., address = 130) Word 114: used for flags and settings Words 115-119: available as working word addresses 3) As the first word, the third word is (word112) and the first four bits enclose the outputs in BCD. 4) The fourth word (word 113) store the address where the output data is stored, the output data is obtained by the fuzzy logic computations. Because fuzzy logic controller work with the other I/O interfaces, their input/output data must be send to the I/O modules working with them. Figure 4 shows how the memory addresses (words) used by the Fuzzy Logic Controller and it also shows the location of the input and output data according to the input/output devices. C:UsersPARVEENDesktopimg2.jpg Figure 4 The working of the Fuzzy Logic Unit works with I/O interfaces We can also use the block transfer instruction to transfer the data between FLU and input/output interfaces (Figure 5). C:UsersPARVEENDesktopimg1.gif Figure 5.Block Transfer of instructions IV CONCLUSION When we apply artificial techniques to a system, we need to add hardware as well as software to in the system. The program that system needed is depending upon the fault in the system, the fault detection is complex then the program will be more complex. We design a system that also has intelligence; this is possible by adding the data from the process. The data should be about the process regarding the last time fault and what type of fault that was, how that was solved and when was the last maintenance performed. The addition of artificial intelligence and fuzzy logic controller in a PLC make the system faster and the system will be able to take decision about the process. The system will be better than the conventional PLCs. . . V REFRENCES [1] Bikash Pal, Balarko Chaudhuri, Robust control in power systems, Spinger, 2005 . [2] Fuzzy Logic Toolbox Users Guide, The MathWorks, Inc.,2008. [3] PLC-5 Programmable Controlers, Rockwell Automation, USA, 2007. [4] C. P. Chuang, X. Lan, J. C. Chen, A systematic procedure for designing state combination circuits in PLCs, Journal of Industrial Technology, 1999;15(3):2-5. [5] S. Manesis, K. Akantziotis, Automated synthesis of ladder automation circuits based on state-diagrams, Advances in Engineering Software, 2005;36:225-233. [6] A. Rullan, Programmable logic controllers versus Personal computers for process control, Computers and Industrial Engineering, 1997;33:421-424. [7] J. Jang, P. H. Koo, S. Y. Nof, Application of design and control tools in a multirobot cell, Computers and Industrial Engineering, 1997;32:89-100. [8] P. Klingstam, P. Gullander, Overview of simulation Tools for computer-aided production engineering, Computers in Industry, 1999;38:173-186. [9] L. A. Bryan, E. A. Bryan, Programmable Controllers, Theory and Implementation, An industrial text company publication, U.S.A. 1997, p. 785.