The production process intelligence that is based on expert system and nerve network is decision-making system

  • Time:
  • Click:122
  • source:SHERBURNE CNC Machining
Much target optimizes the treatment course that the summary offers a kind to be based on expert system and union of nerve network photograph intelligence is decision-making method, built the information switching equipment between expert system and nerve network to make, the intelligence that used object-oriented method to design turning to machine process much target to optimize is decision-making system. Keyword: Proposes An Intelligent Decision System In Manufacturing Process Based On Expert System And Neural Net of Abstract   of   of   of Fu Yonghui of   of Peng Jingxiong of   of Chen Tongjian of   of Intelligent Decision System In Manufacturing Processbased On Expert System And Neural NetPeng Guan of network of nerve of system of production process expert.

The Date Change Technology Of Neural Net And Expert System Is Developed.

A Machine Tool Optimization System, which Runs In Microsoft Windows Was Built Based On Object-oriented Software Technology.

Key Words:mForeword of Anufacturing Process; Expert System; Neural Net1 loss of loss of a finished cost of ideal machining process demand, percent defective, tool, the sources of energy as far as possible low, and quality of productivity, production precision, surface as far as possible tall. But as a result of the complexity that machines a process, system nonlinear parameter changes and cause with treatment when denaturation, make traditional numeric calculation method and experiment plan to satisfy treatment process much target to optimize decision-making requirement hard legally, because this intelligence is decision-making,be production process the inevitable trend of decision-making development. The turning process intelligence that the article opens a gender through building is decision-making system, descriptive expert system and the intelligence that nerve network combines are decision-making method, offerred a viewpoint from overall situation to realize the new approach that treatment parameter function optimizes. The 2 turning process intelligence that are based on expert system and nerve network are decision-making 2.

Production process of 1 system model is decision-making all sorts of tie conditions that producing a course namely, below the limitation that waits like precision of durability of limitation of machine tool power, torque, cutting tool, treatment, make through choosing the cutting tool parameter, treatment parameter such as cutting dosage all sorts of optimizing the target if finished cost, productivity and profit margin get as far as possible optimize [1] . Model of Ⅰ of decision-making system MTOS- is like the turning process intelligence that we build the graph is shown 1 times, its joint action through expert system and nerve network will achieve the optimum solution of production process. Production process is decision-making the much target that is a model optimizes a problem, use the method that optimizes simple goal of translate into of much target problem the problem to undertake begging solving, allow to if linear increases authority law, good point,choose different method law and wait by division, its are main what the difference just depends on evaluating function is different. Use expert system to construct evaluation function, the extraction that optimizes parameter individually certainly is worth limits, will optimize variable join to rise individually with nerve network and undertake optimizing computation. The system is moving platform with Windows, use development of Microsoft Visual C++ . Graph 1 intelligence is decision-making systematic MTOS- Ⅰ 2.

The system of expert system expert in 2 decision-making systems includes knowledge base, database, formulary library and inferential plane. Knowledge base collect all sorts of knowledge that choose cutting dosage and experience, basically involve requirement of computational plan option, tie of affirmatory, correction coefficient and other parameter choose wait for content. Data of the standard data that database memory has place of choice cutting dosage to need, computational constant, test. The experience such as computation of force of the cutting rate that formulary library memory has all sorts of treatment processes, cutting is formulary. The knowledge of expert system basically originates " cutting dosage is compendious manual " [2] the experience knowledge that reachs an expert. Inferential machine is comprised by a group of programs, harmonious and control, whole system, and the environment with current foundation, transfer the material of knowledge base, formulary library and database, choose best parameter. In this system, designed parameter choice and system of tie judgement expert respectively, can mix according to the type of different machine tool of the input different treatment process, whether does the machine tool that judges some model satisfy torque of the power that machines a need, main shaft, choose proper cutting tool point of view, the treatment parameter that affirmatory need optimizes and make choice of take value scope, build evaluation function. 2.

3 nerve network is optimized implement nerve network reviews from organization, self-study with its and collateral calculative ability, make its show powerful dominant position in optimizing the operation that seek solution. The system chooses model of Markov nerve network to be optimize implement. The main characteristic of Markov network is, it does not need function of tectonic to nerve network energy, build a network easily to build a model according to different treatment process, and because its seek solution,the direction that algorithm can drop to case numerical value not only advances, and the way that allows to rise to case numerical value below certain circumstance is ongoing, be helpful for reaching overall situation best [3] . The parameter that treatment process needs each times to optimize forms an unit of Markov nerve network, every unit and other unit two-way join. For example external for round cutting, the horn before definition variable includes the lathe tool of durability of deepness of feed, cutting, cutting tool, cutting tool, tool cutting edge angle, deputy slant radius of circular arc of horn, point of a knife in all 8 variable, design the nerve network that has 8 unit, make each unit correspondence of nerve network needs optimized variable at, stipulate the 2nd the first unit and corresponding feed, unit is corresponding cutting deepness... . When nerve network moves, each unit calculates respective limits extraction a cost according to the current estimation of all sorts of parameter, the moving regulation that presses Markov nerve network next changes the current condition of the network, arrive when network thermal drop a certain when booking a value, each unit condition is direct and corresponding a group of optimized parameter. The unit of nerve network can solve the need trends increase and decrease of the problem according to begging, according to different treatment process trends reframes, because of this nerve network optimize a process not to rely on specific treatment target. The information exchange of 3 experts system and nerve network makes process intelligence the parameter that decision-making system uses expert system to need to optimize certainly, the nerve that decides nerve network from this yuan amount. When nerve network optimizes computation, also need to call expert system to come to what optimize parameter certainly to take a cost. The effective union of expert system and nerve network and the premise that work in coordination depend on mutual the information exchange between. The inquiry that if the graph is shown 2 times,we designed - the interface that translates type data to deliver a technology to regard data as exchange. When the system begins to move, pass standard interface to undertake inquiring to the treatment operand of make choice of by nerve network share first, what this object report goes out to need from personal place is variable a number and every variable span, next nerve network builds network unit according to the result of inquiry, when network unit content produces change, the current condition of reoccupy network transfers the interpreter function that machines an object as parameter, what this function basis reports the numeric changeover of an each unit is corresponding variable formerly is actual and numeric, next the opinion that the evaluation function that nerve network calls this boy or girl friend has machining parameter. Pass this kind of mechanism, nerve network share can operate depart to leave with specific treatment come, what it does not need to know to be being optimized currently when the job is what treatment operation, also do not need to know the actual physics meaning of each working unit. Expert system and exchange of nerve network news basically include: ① machines the report function of the operation through calling the report function of the machine tool to call some secondhand, those who obtain nerve network need is variable a number and respective span. ② basis inquires network of result initialization nerve. ③ transfers interpreter function and computation evaluates the value of function. Graph the control software that 2 nerve network and expert system information exchange 4 systems is designed 4.

1 machine tool kind and machine kind of machine tool type that differs to suit and different treatment process, the intelligence that designed open sex is decision-making system. Structure of this system software uses object-oriented design method, all lathe by same base kind derive comes out. This lathe base kind the mutual feature that defined lathe and the interface that when undertaking optimizing computation, need, certain member function is in base kind in it is simple theory function, if computation evaluates function, data,the interpreter waits. From lathe base kind can derive gives all sorts of lathe kind, the subclass that these derive give can be some is planted specific lathe, be like C620-1, also can be some kind of lathe base kind, those who be like lathe of all numerical control base kind. In derive kind in complete two works according to the feature of specific type machine tool: (1) joins new member function and data; (2) is right base kind simple theory function undertakes redefine. Want to state only in the main program base kind the object can call any derive kind object, on software structure implementation main program and other part have nothing to do, other program module can accept a system directly in. In each specific targets and data side, used program and the technology with separate data, make same kind the object is common share of a program. For example, the distinction between all sorts of different type of center lathe is power, rotate speed, feed and precision only normally differ somewhat, because this defined a center lathe in the program kind, this file photograph combines kind of text that describes lathe data by unified format with to be able to state some is planted the lathe of specific type, because this is in the system,add, it is very easy to decrease lathe to become. Machine kind also be by same a treatment radical kind derive comes out, derive kind right base kind perfect and machine tool kind similar, be like derive outside round turning kind, need to be opposite in base kind the basic parameter of round turning outside the definition on the foundation, include man-hour of material of material of diameter of total mental allowance of turning length, workpiece, workpiece, workpiece, cutting tool and parameter, turning to wait. 4.

2 systems control system of software     to design the software interface that has Windows color, include advocate all sorts of icon that use in window interface, menu, program and pushbutton, dialog box. Through friendly systematic interface the choice operates type, input the basic data and treatment parameter, trends that optimizes target and implementation to optimize a process certainly shows etc. Use the message drive means below MFC framework, the response that software interior provided all sorts of pair of informations when moving function, if the graph is shown 3 times,the system runs technological process. The person that use guides automatically in the system below, found operand of lathe, certain treatment, input relevant parameter, the system can be outputted automatically optimize a result. Graph the intelligence that 5 conclusion study 3 systems moving flow to make a process to coming true originally is decision-making undertook beneficial exploration. Method of decision-making system design is in the treatment process intelligence that advances to intelligence of process of MTOS- Ⅰ turning gets test and verify on decision-making system design and run test and verify through emulating systematic effectiveness and optimize a gender [4] . Devise a method according to what the article puts forward, also can expand systematic function the other machining process such as milling. CNC Milling CNC Machining