Expert system and performance prediction of the ho

Heat treatment expert system and performance prediction

1. Technical content and technical key

expert system as a new technology. It is still in a period of continuous development. Therefore, the structure of expert system does not have a fixed model. According to the current development, it is generally believed that the core of expert system mainly includes knowledge representation and reasoning mechanism, and heat treatment expert system is no exception

due to the particularity of the field of materials and heat treatment, the heat treatment expert system has its own characteristics. In terms of knowledge representation, the conventional data used in heat treatment, including material brand, part and product name, workpiece type and size, process specification, chemical composition, tensile strength, impact toughness, hardness, hardenability, phase transformation dynamics data, are generally expressed in numerical form, so the heat treatment expert system usually adopts relational database system to save knowledge, and uses database technology to realize data management and control. On this basis, the knowledge of heat treatment domain and heat treatment expert knowledge are inserted to realize the knowledge representation of expert system

in terms of reasoning and decision-making, the calculation of experience and theoretical formula is the main clue, supplemented by logical reasoning, to realize the function of decision-making. In the decision-making process, it is crucial to get the intermediate results according to the data input by the user and the known facts, which is also the "heart" of the whole system. In heat treatment expert system, this part is called data export system. Taking a comprehensive view of the current heat treatment expert system, its data export mechanism is no more than the following two ways. One is based on the calculation of phase transition dynamics. Typical systems in this regard are stamp system and PPS system. The other is based on the calculation of permeability. AC3 system and SSH system are typical in this regard. The data export system is further introduced below

the expert system based on the calculation of phase transition dynamics has three basic equations used in its data export system:

L) heat conduction differential equation. Using the two-dimensional transient heat conduction equation, its form is as follows:

here, T represents time, T represents temperature, R is the position coordinate, QV is the phase change latent heat, ρ Is density, CP stands for specific heat capacity, λ Represents thermal conductivity, for flat plate β＝ 0, for cylinder β＝ 1。

in the specific calculation, for simple shapes such as flat plates and cylinders, the difference numerical method is used to approximate the solution

the electromechanical adopted by electronic universal has changed from DC servo electromechanical in the early stage to exchange servo electromechanical now. 2) change the dynamic differential equations. Avrami equation is generally used to calculate the amount of structural transformation. Here is a variation form of Avrami equation

, where y represents the amount of transformation, and K, B1, B2, B3 are parameters related to temperature, composition and grain size

3) equations describing the relationship between structure and performance. The generalized linear mixing rate is used to calculate the properties, that is, a certain property of steel PJ is the integral sum of the properties of each constituent phase

, where x[t (R, t), yj] is the weight function, and YJ is the volume fraction of the constituent phase

expert system based on Hardenability calculation. The basic equation used by the data export system is as follows:

1) hardenability calculation. The hardenability calculation formula of boron containing steel and non boron steel is different

for non boron steel, di = AF × CF (4)

for boron containing steel Di = AF × CF × F (5)

where Di is the ideal critical diameter and AF is the alloy multiplier, that is, the product of multipliers of other alloy elements except carbon and boron; CF is the multiplier of carbon and grain size, and BF is the multiplier of boron. Where

af = fmnfsifnifcrfmofcufv (6)

here, FMN is the multiplier of manganese element, FSI is the multiplier of manganese element, and the rest are similar

2) calculation of tissue transformation. First, according to the following formula, the critical cooling rate VCI of various structures is calculated according to the chemical composition of steel and the austenitization parameters

υ ci＝g(P α, C. Si, Mn,...) (7)

Where, υ Ci is the critical cooling rate to obtain various structures, I = 1 corresponds to the critical cooling rate to obtain 100% of the volume fraction of martensite, I = 2 corresponds to the critical cooling rate to obtain 90% of the volume fraction of martensite + 10% of the volume fraction of bainite, and so on; P α Is the austenitizing parameter, C is the mass fraction of carbon, Si is the mass fraction of silicon, the rest is the same

determined by the actual cooling speed of the workpiece υ Calculate the volume fraction of various tissues. as υ ci＜ υ＜υ Ci+1, then

fm=am/（ υ ci+1- υ ci)( υ-υ ci)+BM

fB=AB/( υ ci+1- υ ci)( υ-υ ci)+BB (8)

fFP=AP/( υ ci+1- υ ci)( υ-υ CI) + BP

where FM, FB, and FFP are the volume fractions of martensite, bainite, and ferrite pearlite respectively, and am, AB, AP, BM, BB, and BP are constants related to the critical cooling rate

3) equations describing the relationship between structure and performance. The linear mixing rate is used to calculate the properties, that is, a certain property of steel PJ is the weighted average of the properties of each constituent phase

, where PJM is the property of martensite, pjb is the property of bainite, and pjfp is the property of ferrite pearlite

it should be pointed out that the expert system based on the calculation of phase transformation dynamics and the expert system based on the calculation of hardenability are not completely different. They penetrate each other in many aspects of technology and have similar functions

the expert system with the data export system as the core, its functions include the following aspects:

1) organization and performance prediction. The microstructure and properties of the workpiece after heat treatment are calculated and predicted according to the partial composition of the steel chemical dynamometer and the heat treatment parameters

2) process change analysis and optimization. Analyze the influence of the change of process parameters on the heat treatment results, and then analyze the heat treatment defects and optimize the process parameters

3) auxiliary design of heat treatment process. The computer aided design of heat treatment process is carried out by using the decision-making function of expert system

4) real time monitoring of the process. Multi time carbon potential control of carburizing process, optimization control of carburizing layer depth and carbon concentration distribution

5) select materials according to the geometric dimensions and performance requirements of parts. Select an appropriate brand for the workpiece to be heat treated. The hardenability of this brand is enough to ensure that the performance at the specified point on the workpiece section meets the use requirements when quenching under given conditions

an example is given below to illustrate the function of the expert system. It is assumed that the best steel should be selected for the 45mm diameter national shaft in the project. For production reasons, quenching oil (flow rate lm/s, quenching intensity H = 0.4cm-l) is used as the cooling medium. After quenching and tempering, in order to make the workpiece have better fatigue performance, the hardness after tempering at the 3/4 radius of the workpiece section is required to be 35hrc. After inputting the above objectives and constraints, the system first converts the constraints (tempering hardness 35hrc at the 3/4 radius of the workpiece section) through calculation, and finally determines that this requirement is equivalent to the hardness of 45.3hrc at 11mm away from the water-cooled end on the end quenched sample. Taking this as the constraint condition, the decision-making search of the sheep horn hammer was carried out, and the steel grades of GB 40crni, GB 4omnvb and GB 42CrMo were preliminarily obtained. Further comparing the hardenability bands of the three steel grades, the system determines that GB 42CrMo can well meet the above design requirements

in addition to the SSH structural steel hardenability material selection and process optimization system of Beijing Electromechanical Research Institute, some domestic universities and factories have also carried out a lot of research and development work on heat treatment expert system. For example, the carburizing process control system developed by Shanghai Jiaotong University takes the mathematical model describing the carburizing process as the knowledge representation, and takes the computer calculation results as the basis for judgment to realize the prediction of the concentration distribution and hardness distribution of the carburized layer; A new carbon potential control method is proposed to keep the whole process in an optimal state. Beijing University of Aeronautics and Astronautics has developed an auxiliary decision-making system for aerospace material heat treatment process, which combines part CAD with heat treatment theory, takes the material, process and standard database as the knowledge representation, and takes it as the basis for reasoning and decision-making, so as to realize the formulation of process flow and professional knowledge consultation. In addition, some factories are also carrying out the research and development of computer-aided heat treatment process, including the generation and management of process cards, the formulation of heat treatment process for special parts, etc

2. advantages, disadvantages and scope of use

at present, expert system has been applied in industrial production. For example, some automobile companies have widely used heat treatment expert systems in production for process analysis and formulation, existing production cycle optimization, auxiliary material selection, component design and other work, and achieved good results. Carburization process control systems have also been used in many factories in China. It can be said that expert system technology has been widely recognized by engineers and technicians. Heat treatment expert system plays an important role in reducing production costs, shortening production cycle and improving product quality, and a consensus has been reached on this point

however, it should also be noted that the utilization rate of expert systems in the design and industrial sectors is not very high. There are two reasons for this. On the one hand, we need to strengthen the work of promoting the use of expert systems; On the other hand, it is necessary to develop new and better methods so that the expert system can help complete the tasks of process design more directly and effectively

technically, the current heat treatment expert system is based on statistical data and empirical knowledge, and the accuracy and reliability of its results need to be further improved. In terms of use, when expert systems are applied to industrial processes and design work, they also include the traditional trial and error method (tral and egror) to varying degrees. In addition, for the major key parts related to the development of national economy in industrial production, as well as the basic parts widely used in industry, the internal stress and residual deformation produced in the heat treatment process not only affect the performance and quality, but also affect the subsequent assembly accuracy and processing cost. At present, this factor has not been considered in the heat treatment expert system, which cannot be said to be a defect of the current expert system

at present, the heat treatment expert system is mainly used in the heat treatment process of carbon steel and low alloy steel and related design activities, so it is necessary to broaden the application scope of the expert system. In addition, the decision-making process of the current expert system requires a lot of manual intervention, and further work is needed in the processing of decision results and human transformation to improve the traditional industrial machine interface, so as to make the expert system more convenient for engineers and technicians