JP2952464B2 - Expert system for casting plan support - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic casting method for planning a runner model, a weir model, and a model of a product part, that is, a main mold layout, which is buried in a sand mold in which casting is performed. On a casting plan support expert system.

[0002]

2. Description of the Related Art FIG. 22 schematically shows a conventional casting. The upper mold 1 and the lower mold 3 are both sand molds. That is, the upper die 1
Alternatively, a model 9 to be a product is buried in sand on the plates 5 and 7 serving as a platform for forming the lower mold 3 (FIG. (A)). At this time, the upper mold 1 and the lower mold 3 are assembled together with the model 13 for guiding the molten metal, which is a molten metal, at the part of the models 9 and 11 to be the products (FIG. 2B). . Then, the upper mold 1 and the lower mold 3 are separated (FIG. (C)), and the models 9, 11, and 13 are removed (FIG. (D)) to obtain a space 15 (FIG. (E)). . This space 1
The hot water is introduced into 5 and solidified, the sand mold is broken and the product is taken out.

In the conventional example of FIG. 22, the hot water is guided after the model is removed, but in other conventional examples, the model is formed of styrene foam, and the hot water is guided without removing the model. Styrofoam can be evaporated by the heat of hot water.

[0004] In the case where a space portion is designed inside a product, a core is used to form a space portion. Here, the concept of a main model, which is a model 9 as a product, includes: The core and the baseboard supporting the core are included.

Next, as shown in FIG. 23, a model 13 for guiding hot water includes, for example, a portion of a vertical gate 13A serving as a spout of hot water, and a horizontal runner continuous with the gate 13A. There is a portion 13B, a riser 13C which is continuous with the runner 13B and becomes a final solidification portion, and a portion of a weir 13D where the hot water is introduced into a portion which is continuous with the riser 13C and becomes a product.

[0006] How many of these models, in what shape,
The layout that is placed in consideration of which position, in which direction, and in which direction is the top and bottom has a significant effect on the quality and cost of the product, and depends on the knowledge (including know-how and even intuition) of a skilled technician. There were many things to do.

[0007] On the other hand, attempts have been made to support a casting method by using a computer. However, this conventional attempt is based on theoretical equations such as fluid dynamics (for example, Bernoulli's equation).
Based on this, the shape and dimensions of the main mold, weir, runner, etc. were to be calculated.

[0008]

However, the knowledge of skilled casting engineers (including know-how and even intuition) is required for casting methods.
Therefore, it takes a very long time to train newly skilled technicians, and some of them cannot meet the needs of diversified casting methods.

[0009] Computer support for calculating the shapes and dimensions of the main mold, weir, runner, etc., based on the theoretical formulas such as fluid dynamics is only partial, and the knowledge (skills, know-how,
In addition, it was difficult to provide overall support at the same level (including intuition).

The above problems have been described with respect to a casting technique using a sand mold. Similar problems also exist in a casting technique using a mold which is a mold for resin molding and a casting technique for die casting for precision casting of aluminum. I do.

The present invention has been made in order to solve the above problems, and a casting method for planning a layout of a runner model, a weir model, a main mold, and the like is based on the knowledge of a skilled engineer. It is an object of the present invention to provide a casting plan support expert system which can automatically perform the casting at a high level.

[0012]

In order to achieve the above object, according to the present invention, a hot water such as a molten metal is provided in a mold by a runner portion, a feeder portion, a weir portion, and the like. Before the product part is guided in order and casting is performed, a runner model, feeder model, weir model, and product part model, i.e., the layout of the main mold, are placed in the mold. This is a casting plan support expert system that automatically performs a casting plan to be performed, a means for inputting and registering a main mold shape and the number of main molds, and a knowledge of a skilled technician from a casting plan that has been used in the past in advance. The shape of the weir and the runner,
Means for classifying and registering a plurality of types of knowledge relating to position and dimensions, and, among a plurality of knowledge contents possessed by each of the classified knowledge, which knowledge content to use is sequentially determined for each knowledge Means, and a runner, a feeder, a weir, and a layout of the main mold are simulated in accordance with the main form, the number of main molds, and the knowledge content determined from the plurality of knowledges. Means for checking whether the contents are satisfactory or not. If the contents are not satisfactory, return to the input of the number of the main molds and repeat the procedure. It is a casting plan support expert system which is a feature.

According to a second aspect of the present invention, hot water such as molten metal is
Before the casting is performed in the mold in the order of the runner part, the part of the feeder, the part of the weir, and the part to be the product, the model of the runner placed in the mold, the model of the feeder, It is a casting plan support expert system that automatically performs a casting plan for planning a model of a weir and a model of a part to be a product, that is, a layout of a main mold. A means for inputting and registering a certain production constraint, and acquiring in advance the knowledge of a skilled technician from a casting plan that has a track record in the past, selecting the shape of the weir, selecting the weir shape, selecting the position and number of weirs Weir position selection knowledge, runner shape selection knowledge to select the shape of the runner, gate design knowledge to determine the dimensions of the weir and runner, and layout of the main mold, weir, riser, runner Five types of layout coordinate determination knowledge for determining coordinates Means for classifying and registering the knowledge, and a plurality of knowledge cards constituting each of the classified knowledge, wherein each knowledge is used depending on which knowledge card is adopted, and the classified knowledge is used. And the attribute value that satisfies the attribute. The attribute and the attribute value represent a constraint condition that is a condition as to whether or not the knowledge card can be adopted, and the constraint condition is satisfied. The decision card that is determined when the item is decided is expressed and the decision item is registered when the item is adopted, and whether the constraint condition is satisfied is determined by the master shape, the production constraint, and the registered condition. By determining whether the decision item of the knowledge card matches the constraint condition of the knowledge card,
According to the means for making a judgment for each knowledge card in order, and the main form, weir, feeder, and hot water according to the main form, the production constraints, and the decision items determined in each of the five knowledge cards, which are input first. Means for simulating the layout of the road and checking whether it does not protrude from the plate corresponding to the flat area of the sand mold, and if it protrudes from the plate, return to the input of the initial production constraint and follow the procedure This is a casting plan support expert system, characterized in that the procedure is repeated and the procedure is terminated when it does not protrude from the plate.

[0014]

According to the first aspect of the present invention, the system acquires in advance the knowledge of a skilled engineer from a casting method that has been used in the past,
Weirs and runners are classified and registered as a plurality of types of knowledge on shape, position, and dimensions. Then, first, the main mold shape and the number of main molds are inputted and registered. Next, based on the input main mold shape and the number of main molds,
Among a plurality of knowledge contents of each classified knowledge,
Decide which knowledge content to use. This decision is made for each knowledge in turn.

The runner, the weir, and the layout of the main mold are simulated in accordance with the main shape, the number of main molds, and the contents of knowledge determined from a plurality of knowledges. If the layout is not satisfactory, the procedure returns to the input of the number of the main molds, and the procedure is repeated. If so, end the procedure.

According to a second aspect of the present invention, the system acquires in advance the knowledge of a skilled technician from a casting method that has been used in the past, and selects the shape of the weir to select the shape of the weir, and selects the position and number of the weir. Weir position selection knowledge, runner shape selection knowledge to select the shape of the runner, gate design design knowledge to determine the dimensions of the weir and runner,
The information is classified and registered into five types of knowledge including layout coordinate determination knowledge for determining coordinates for laying out the main mold, weir, and runner.

First, a master shape and a production constraint which is a constraint on a production policy and production equipment are inputted and registered.

Next, based on the master shape and the production constraints, each knowledge is used depending on which of the plurality of knowledge cards constituting each knowledge is adopted.
Whether a certain knowledge card can be adopted depends on whether or not the constraint condition of the knowledge card is satisfied. Whether or not the constraint condition is satisfied is determined based on whether or not the master shape, the production constraint, and the decision items of the already registered knowledge card match the constraint condition of the knowledge card. When the knowledge card is adopted, the decision items of the knowledge card are registered. Judgment, adoption, and registration are performed for each knowledge card in order.

The runner, weir, and main mold layout are simulated in accordance with the main items, the production constraints, and the determination items determined in each of the five knowledge cards. Check whether it protrudes from the plate corresponding to the flat area. In the case of protruding from the plate, the procedure returns to the input of the first production constraint, and the procedure is repeated.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to FIGS.

"System Overview" FIG. 1 shows an overall overview of the system in this embodiment. This system uses an inference module (Inferenc) that automatically generates layout data of models (models of runners, models of weirs, master models, etc.).
e Module) 21 and its layout data, and a layout module (Layout Mod) that executes layout on a plate (see 5 in FIG. 22).
ule) 23.

An inference module 21 for automatically generating layout data is used to determine the quality and production efficiency based on a knowledge base created based on the technical knowledge, know-how, intuition, empirical rules, etc. of a skilled technician relating to past casting plans. Based on the production policies and restrictions on production equipment that have been considered, the best model plan (those that determine the shape, size and number of each model) and model layout (those that determine the arrangement of each model) are inferred. It is output as graphic information (for displaying on a display screen or the like) and layout instruction information (for example, actual numerical values for layout).

In the layout module 23 for executing the layout, first, the generated model plan and layout information are converted into spatial layout information of the model by the control device 25. Next, the layout position obtained by the spatial layout information is marked by an automatic layout machine 27 including an NC processing machine such as a machining center, or the automatic layout machine 27 including a layout dedicated machine.
To automatically lay out the model on the plate. Thereafter, casting sand is put on the plate to complete a sand mold.

Next, the configuration of the inference module 21 will be described in more detail.

(1) Data input unit (CAD) 29 The function of this data input unit 29 is composed of the shape and dimensions of the main mold and the production constraints (number of main molds, quality, etc.). Is a function for inputting
The user 31 interactively inputs necessary information (such as the shape and dimensions of the main mold and production constraints) for the request items graphically presented on the display screen.

(2) Inference mechanism (Inference E)
ngin) 33 The function of the inference mechanism 33 is based on information such as the master shape input by the user 31, taking into account the similarly input production constraints, and making full use of the knowledge of the knowledge base unit 35. It is a function to infer layout information of a plan and a model.

(3) Knowledge base section (Knowledge)
Base) 35 A knowledge card (later including conceptual and numerical data on technical knowledge, know-how, intuition, empirical rules, and the like regarding the casting method) cultivated over many years by a skilled engineer, which can be handled by a computer. Will be described in detail later) and registered.

(4) Database section (Date Bas)
e) 37 This database unit 37 relates to information on the size, shape, and number of objects (hereinafter referred to as “runners, risers, gates, etc.”) to be modeled, and dimensions and shapes of plates to be laid out. This is a database in which information is registered.

(5) Knowledge input correction mechanism (Knowleg)
dge Editor) 39 In the knowledge base unit 35 and the database unit 37, it is necessary to acquire and add the latest knowledge data, or to change / delete data with a low use record. The function of the knowledge input correction mechanism 39 supports acquisition, addition, change, and deletion of these data by the maintenance engineer 41.

Note that the inference module 21 is provided with an engineering workstation (NEWS 5000) 4
3, the data input unit 29, the inference mechanism 33, the knowledge base unit 35, and the database unit 37 are graphical information such as a Prolog language which is a symbol processing language, and output results and the like are created by Motif based on X-Window. ing.

"Input of master shape" The products handled by the system according to the present embodiment are mainly cylindrical shapes such as hydrants, fittings, fittings, and valves made of bronze casting (although more It is of course possible to handle products with complex shapes.) In particular, the product cast in this embodiment has the shape of FIG. Strictly speaking, the main shape includes not only a model of a part to be a product, but also a core and a baseboard supporting the core. FIG. 15 shows a specific main mold shape in this embodiment.

The main shape is expressed by a combination of six types of basic elements as shown in FIG. A specific example of this combination is shown in FIG. 17 for FIG. In the case of a main type with only a solid part, the outer shape is defined. In the case of a main mold including a hollow part, one basic element is defined by a combination of an outer shape and an inner shape, and one main mold is defined by a combination of a plurality of basic elements. In addition to specifying the geometric dimensions, the quality such as pressure resistance and the specification of the machined part after casting (Machinin
g) is a data expression (described later) that is added to each basic element.

In the casting method, by expressing the shape of the main mold in this way, information such as a portion requiring machining, a maximum wall thickness, and an element to which a weir is to be attached can be concretely and easily expressed. In general CAD, line and curve information (geometric information)
Can only be entered, and the system cannot determine what they mean. In this system, when defining a shape, such information can be expressed easily and concretely, and the system can automatically judge the meaning of the input information and obtain the necessary information, so that efficient Inference becomes possible.

"Input of Production Constraint" As shown in FIG.
Production constraints include general attributes and shape attributes on production policies or owned equipment. The contents of the production constraint are changed when it is impossible to generate a casting plan through a predetermined procedure, and the generation of the casting plan is made possible. By making this change, it is also possible to perform a casting simulation using the cost and the like as evaluation values.

As specific examples of the general attribute, a main type number,
There are the number of cores, the number of baseboards, the quality (pressure resistance), the presence or absence of machining such as thick finishing, the presence or absence of core blowing, and the machining reference plane.

Here, the number of the main molds is the number of main molds arranged on the plate. The productivity increases as the number of the main molds increases. The defect rate rises. Reducing the number can reduce the reject rate, but reduces productivity. Thus, there is an optimal number.

The skirting board is a portion that supports the core. The core is for forming a hollow portion of the product. The core is made separately from the product in advance, and is inserted into the product in a sand mold so that the portion becomes hollow after casting. This core is also made of sand and can be broken down to get a hollow part.

For example, as shown in FIG. 18, the rough shape of a product or a core (FIG. 18A) or a shape obtained by further subdividing the shape of a cylindrical main mold as shown in FIG. There is.

"Knowledge Classification" The knowledge of a skilled engineer is
Classified into one.

(1) Weir shape selection knowledge Knowledge for determining a weir shape by using the main shape, the number of main dies, the thickness, the baseboard shape, and the like as constraints.

In the present embodiment, the weir shape to be handled is the three weir shown in FIG.
Into three forms. These three weirs are called the main weir, the Bali weir,
It is called a ring weir.

The weir (FIG. 1A) is used in the case of a product shape in which shrinkage cavities are very likely to occur, or when it is impossible to attach a weir to the baseboard. Attach directly to the outer diameter of the product. The cross section of the product must be polygonal or semicircular.

The condition of the burr (FIG. 2B) is that the product has a thin plate shape. The relationship between the product thickness T at the weir position and the burr thickness t is t = T. Must be thinner than the thickness of the main weir (FIG. 1A).

The ring weir (FIG. 2C) is used when the skirting board length is 1.5 times or more the skirting board diameter.

(2) Weir position selection knowledge In order to determine the number of weirs and the mounting position of the weirs on the main die, using the main mold shape, skirting board shape, core shape (with or without blow-in mandrel) as constraints, etc. Knowledge of.

(3) Runner shape selection knowledge Knowledge for determining a runner shape using a weir shape, the number of main molds, and the like as constraints.

The runner types to be handled are I type and H type.

(4) Gate design knowledge Knowledge for determining the size of the gate, weir, runner, riser, and gate receiving bottom, with the weir shape, runner shape, main mold shape, etc. as constraints.

(5) Layout coordinate determination knowledge Knowledge for determining the layout coordinates of each object on the plate under the constraints such as the number of the main mold, the shape of the main mold, the weir shape, the riser, and the runner shape.

As shown in FIG. 3, when these five pieces of knowledge are used to determine the use of a certain piece of knowledge, they refer to another piece of knowledge (the direction of the arrow) to form a network. The layout is optimized in consideration of the entire constraints.

In the above (1) to (3), only the shape and number of weirs and runners, the mounting position on the main mold, and the like are determined. The dimensions are determined only in (4). You.
The reason why dimensions are not determined in (1) to (3) is that
This is because the dimensions of the weir and runner objects are determined by referring to the dimensions of other objects when the dimensions are determined. That is, since the dimensions of the other objects are not determined when the dimensions of the object are determined, there is a possibility that there is no data to be referred to and the determination cannot be performed. Furthermore, it is more efficient (shortening the inference time) to finally determine the dimensions collectively with the gate system design knowledge of (4).

[Acquisition of Knowledge] In this system, knowledge acquisition for extracting knowledge from a skilled engineer was performed through an interview. Generally, knowledge on production technology includes quantitative knowledge that can be expressed by specific numerical values and qualitative knowledge based on experience and intuition. The knowledge in casting plan work is also a combination of qualitative and quantitative knowledge and is complex. Also,
Regarding qualitative knowledge, it is unconscious and includes the ambiguity and subjectivity of each engineer, and even an expert may not be able to express it clearly. These factors make it very difficult to organize knowledge, known as the "knowledge acquisition bottleneck,
The success of an expert system depends on this cleanup.

When the work of a skilled engineer is analyzed in detail, it is often the case that past results are referred to in order to solve a new case. Therefore, in the present invention, attention has been paid to layout data that has a track record of past casting plans when acquiring knowledge. This proven data is the result of trial and error,
The shapes and dimensions are specifically expressed by numerical values. In addition, this figure takes into account quality, cost, workability, yield, and other factors. Therefore, it is considered that this data includes all the knowledge necessary for the casting method, such as the experience and know-how of a skilled person.

"Construction of Knowledge" When constructing an expert system, it is very important how to construct the acquired knowledge. In other words, how efficiently knowledge can be accumulated, searched, and changed greatly affects reliability and operability in this type of system.

In the present system, first, the knowledge on the casting method possessed by the skilled technician is represented by a format sheet called a knowledge card and registered in the knowledge base unit (FIG. 1). Here, weir shape selection knowledge, weir position selection knowledge, runner shape selection knowledge,
FIGS. 8 to 13 show examples of format sheets of gate system design knowledge and layout coordinate legal knowledge, respectively.

This knowledge card has three main contents, for example, as shown in FIG. The first part is a name (Name) of the knowledge card, a priority of the knowledge card (described later), and an item name for controlling execution of inference in a header portion of the card. Item names that control the execution of inference include Attribu
te: attribute, Value: attribute value, Procedure
e: Additional procedure, Note: There are special notes on quality.

The second part is a constraint condition (Constraint cond) that determines whether or not a knowledge card is adopted.
), which are expressed by the specific contents of the attributes, attribute values, additional procedures, and special notes on quality.

The third part is an item (Decision) determined when the knowledge card is adopted.

These are arranged in a frame type structure. Further, an additional procedure (Procedur) for determining from which object database the attribute and its value are obtained and which object database is to be registered (Create)
e) is described.

This frame type structure has been proposed by Minsky as a framework for modeling the process of human memory and cognition. In order to search a database, it is more efficient to organize information according to certain criteria. A group of such information (for example, a group of attributes and attribute values) is called a frame.

There is another point that has been devised. In other words, it is not only an expression of a knowledge card with attributes and attribute values, but also constraints and decision items are expressed by attributes and attribute values, and additional procedures are added and extended. That is it.

Then, in order to search for a plurality of knowledge cards and to check whether or not the knowledge card described therein can be adopted, the constraint condition and the known data (such as the previously entered master shape and It checks the production constraints and the consistency with the already adopted knowledge card attribute value), and if there is no problem, adopts the knowledge card. This check is performed by pattern matching.

As a result of this check, if it is determined that the constraint condition of each attribute value of the second part is satisfied, each attribute value of the decision item of the third part is determined as a result, and the corresponding object database (FIG. 1) 37)).

For example, as shown in FIG. 8, it is possible to incorporate constraints such as productivity (productivity) and pressure resistance (pressure proof) into special notes on quality.

Since the knowledge is expressed by the frame type structure in this manner, the maintainability such as addition and deletion of the knowledge is improved.

As a method of determining coordinates in the knowledge card for determining layout coordinates shown in FIG. 13, a plate size based on each attribute value of an object obtained from the experience of a skilled technician and the like, and a layout interval and coordinates of each object are determined in advance. Register it in the layout database. Next, a total number of layout databases matching the object attribute values obtained by the inference are searched, and the layout coordinates of each object described therein are adopted. And
When this knowledge card is adopted as shown in FIG. 13, this means that coordinate values based on the layout pattern of 6-20 (FIG. 13) are determined.

It should be noted that the number of knowledge cards created for this knowledge acquisition is 110.

[Object Data] The object data in the object database of the present system is composed of the following six items.

(1) Principal object The data consists of data such as the shape, size, number of the main mold, and the presence or absence of processing.

(2) Weir Object The weir object consists of data such as the shape of the weir, the number of the weir, and the mounting position on the main mold.

(3) Runner object The runner object is composed of data such as a runner shape and dimensions.

(4) Sluice bar object Consists of data such as dimensions.

(5) Gating Bottom Object Consists of data such as dimensions.

(6) Feeder object It consists of data such as shape and size.

These are arranged in a frame type structure ((2) to (6) are shown in FIG. 6).

[Inference Procedure] FIG. 2 shows an inference procedure in the present system. Inference is made up of two components: "generation of an object (a plan object such as a weir or a runner)" and "object layout". First, in "generation of object", the main shape of the main shape to be laid out on the plate and the attribute values of all objects are determined. Note that the attributes of each object are as shown in FIG. 6 (the upper two columns in the figure) and include a shape, a dimensional shape, and a number. Specific attribute values (two upper right columns in the figure) consisting of predetermined values are satisfied for each of these attributes.

Next, in the “object layout”, a past actual layout example database that satisfies the attribute values of all the objects is searched using the attribute values of all the objects as constraint conditions, and the coordinate values of the object layout (lower right in FIG. Column). At this time, the physical layout possibility of all objects in a finite space (for example, on a plate (see 5 in FIG. 22)) is checked. Modify the constraints and generate the required objects again. The check of the possibility of the layout may include not only the check of whether or not the layout is out of the finite space but also various other checks. For example, the ratio of product weight or product volume to the total weight or volume of the mold, S
/ M ratio (SandMetal Ratio) check, quality check, and production efficiency check can be performed. In this way, it is possible to devise a method of casting a stable product. Such a check of quality and production efficiency is performed based on the contents of the constraint conditions of each object.

[Resolution of Conflict] The present system can generate a plurality of layout-possible plans for one main model. This is because, after searching the total number of applicable knowledge cards during inference, all appropriate knowledge is adopted, and the next object is generated in a tree structure from these objects. However, considering the efficiency of the inference, unless some restrictions are applied during the generation of the plan, a combination explosion occurs and the practicality of the system is reduced.

Therefore, in the present system, as shown in FIG. 9, in order to resolve a conflict in knowledge application, a plurality of knowledge cards which may determine a plurality of attributes based on the same constraint conditions are used in the past for the number of times of use. The priority (Priority) is described in the knowledge card according to.

[Specific Example of Inference Procedure] A specific inference procedure is shown according to the numbers (S1 to S50) in FIGS.

S1 The user defines a main shape and constraints (by key input and mouse operation in the data input section).
Also, the priority of the knowledge sheet used at the time of executing the system is defined.

S2 The system registers (stores) the information of S1 in the main object database (hard disk) of the database unit.

S3 Weir shape selection is performed in the following steps S4 to S7.

S4 The system searches the knowledge card database (registered in the hard disk) of the knowledge base unit for a knowledge card related to the selection of the weir shape.

S5 The constraint conditions described in the card (here, as shown in FIG. 8, the product shape, the core shape, the number of main molds, and the maximum thickness) are the main mold shapes registered by the user in S2. (Pattern matching) to see if they match the production constraints. This is repeated for the number of constraints.

S6 If NO (the constraint condition was not satisfied in S5), S4 and S5 are repeated until they match.

S7 If yes (satisfies the constraint in S5), the attribute and attribute value described in the decision item of the pertinent knowledge card are registered in the weir object database of the database unit.

S8 Weir position selection is performed in the following steps from S9 to S12.

S9 The system searches the knowledge card database (registered in the hard disk) for a knowledge card related to the selection of the weir position.

S10: The constraints described in the card (here, as shown in FIG. 10, weir shape, main mold shape, core length, presence / absence of machining) are registered by the user in S2. It is checked (pattern matching) whether or not it matches the determination item of the knowledge card of the weir shape selection registered in S7, the mold shape, the production constraint. This is repeated for the number of constraints.

In this system, as in S10, when determining the attribute and attribute value of a certain object, the object attribute and attribute value that have already been inferred (that is, the determination items of the knowledge card are registered) are used. This is to form a network of knowledge that it is necessary to effectively use other knowledge when deciding whether certain knowledge can be applied based on the experience of casting method work. In this way, the inference of each object is dynamically (data of decision items registered every moment changes) and chained (decision items registered in the own knowledge card determine whether to adopt the next knowledge card). It is the feature of this system to repeat to the end.

If S11 is not satisfied (the S10 constraint condition is not satisfied), S9 and S10 are repeated until they match.

S12 If yes (all constraints are satisfied in S10), the attribute and attribute value described in the decision item of the relevant knowledge card are registered in the weir object database.

S13 The runner shape is selected from S14 to S1.
Step 7 is performed.

S14 The system searches the knowledge card database (registered in the hard disk) for a knowledge card related to the selection of the runner shape.

S15 Restrictions described in the card (here, weir shape, product shape, number of main molds, maximum diameter of main mold, direction of main mold to runner as shown in FIG. 11)
Is the main shape, production constraint and S7 registered by the user in S2.
It is checked (pattern matching) whether or not it matches the decision item of the knowledge card of the weir shape registration registered in step S12 and the decision item of the knowledge card of the weir position selection registered in step S12. This is repeated for the number of constraints.

If S16 is not satisfied (the constraint condition was not satisfied in S15), S14 and S15 are repeated until they match.

S17 If yes (all constraints are satisfied in S15), the attribute and attribute value described in the decision item of the relevant knowledge card are registered in the runner object database of the database unit 37 (FIG. 1). .

S18 The design of the sprue bar is described below from S19 to S
Step 22 is performed.

S19 The system searches the knowledge card database (registered in the heart disk) for a knowledge card relating to the design of the sprue stick.

S20: The constraints described on the card are the master shape, the production constraints registered by the user in S2, and the constraints.
Decision item of knowledge card of weir shape selection registered in 7, S12
It is checked (pattern matching) whether or not it matches the decision item of the knowledge card of weir position selection registered at step S17 and the decision item of the knowledge card of runner shape registration registered at step S17. However, in the case of the sprue bar, the constraint condition is loose, and there is a possibility that a plurality of knowledge cards can be adopted from the same constraint condition. Therefore, the above checks are substantially performed in the order of priority.

S21 If not (the constraint condition was not satisfied in S20), a knowledge card having a priority obtained by adding 1 to the priority number is checked.

S22 If yes (satisfies the constraint in S20), the attribute and attribute value described in the decision item of the pertinent knowledge card are registered in the gate entry object database.

S23 The gate bottom (the part to which the lower end of the gate bar is connected and further connected to the runner) is designed in the following steps S24 to S27.

S24 The system searches the knowledge card database (registered on the hard disk) for a knowledge card related to the gate bottom design.

S25 The constraints described on the card are the master shape, the production constraints registered by the user in S2, and the constraints.
Decision item of knowledge card of weir shape selection registered in 7, S12
Item of knowledge card for selection of weir position registered in step S17
Then, it is checked (pattern matching) whether or not the item matches the decision item of the knowledge card for selecting the runner shape registered in step S22 and the decision item of the knowledge card for selecting the sprue stick registered in step S22. However, in the case of the gate, the constraint condition is loose, and there is a possibility that a plurality of knowledge cards can be adopted from the same constraint condition. Therefore, the above checks are substantially performed in the order of priority.

If S26 is not satisfied (the constraint condition was not satisfied in S25), S24 and S25 are repeated until they match.

S27 If yes (satisfies the constraint in S25), the attribute and attribute value described in the determined item of the pertinent knowledge card are registered in the gate bottom object database.

S28 The design of the runner is described below from S29 to S3.
Perform the procedure in step 2.

S29 The system searches the knowledge card database (registered in the hard disk) for a knowledge card relating to runner design.

S30: The constraint conditions (here, pressure resistance, main die diameter, weir shape, runner shape) described in the card are as follows:
It is checked whether the user agrees with the master shape registered in S2, the production constraint, the decision item of the weir shape selection knowledge card registered in S7, and the decision item of the runner shape selection knowledge card registered in S17 ( Pattern matching). This is repeated for the number of constraints.

If S31 is not satisfied (S30 constraint condition is not satisfied), S29 and S30 are repeated until they match.

S32 If yes (all constraints are satisfied in S30), the attribute and attribute value described in the decision item of the pertinent knowledge card are registered in the runner object database.

S33 The design of the weir is described below from S34 to S37.
The procedure is as follows.

S34 The system searches the knowledge card database (registered in the hard disk) for knowledge cards relating to weir design.

S35 The constraint conditions (here, pressure resistance, product weight, wall thickness, and weir shape) described in the card are the shape of the main mold registered by the user in S2, the production constraints, and the weir shape registered in S7. It is checked (pattern matching) whether or not the selected item matches the determined item of the knowledge card. This is repeated for the number of constraints.

If S36 is not satisfied (the constraint condition was not satisfied in S35), S34 and S35 are repeated until they match.

S37 If yes (all constraints are satisfied in S35), the attribute and attribute value described in the decision item of the pertinent knowledge card are registered in the weir object database.

S38: The design of the riser is as follows from S39 to S4
Perform the procedure in step 2.

S39: The system searches the knowledge card database (registered on the hard disk) for a knowledge card relating to the design of the feeder.

S40: The constraints described in the card (here, the maximum diameter of the weir shape and the main shape as shown in FIG. 12) are the main shape and the main shape registered by the user in S2.
It is checked (pattern matching) whether or not the production constraint matches the decision item of the weir-shaped selection knowledge card registered in S7. This is repeated for the number of constraints.

If S41 is not satisfied (the constraint condition was not satisfied in S40), S39 and S40 are repeated until they match.

S42 If yes (all constraints are satisfied in S40), the attribute and attribute value described in the decision item of the pertinent knowledge card are registered in the hot water object database.

S43 The layout coordinates are determined in the following S4.
4 to S47.

S44 The system searches the knowledge card database (registered on the hard disk) for knowledge cards related to layout coordinate determination.

S45: The constraints described on the card (here, as shown in FIG.
The main mold shape and the weir shape) are the main mold shape registered by the user in S2, the production restriction and the decision items of the weir shape selection knowledge card registered in S7, and the runway shape selection knowledge card decision items registered in S17. Is checked (pattern matching). This is repeated for the number of constraints.

If S46 is not satisfied (the constraint condition was not satisfied in S45), S44 and S45 are repeated until they match.

S47 If yes (satisfies all constraints in S45), place on the plate based on the layout coordinates of all objects described in the decision item of the pertinent knowledge card, and check if it protrudes from the plate I do.

S48 If not (protruding), return to S1 and repeat the previous procedure again.

S49 If yes (it fits on the plate), this layout is registered as satisfactory in the layout coordinate object database.

S50 The attributes and attribute values of all object databases are determined and the process ends.

[Execution Result] The execution result of the system is shown in FIG.
Shown in The number of knowledge cards used was 76, and only one example was generated this time. FIG. 21 shows an example of a casting method by a skilled engineer. Both results agree very well.

The time required for a skilled engineer to create a casting plan was about 12 hours in total, including the time required for various literature checks, various calculations, past case searches, drawing, and the like. In this system, the total time from data input to graphic output is about 30 minutes (inference only: 15 seconds).
12 could be made more efficient.

In the above embodiments, the casting technique using a sand mold has been described. In the other embodiments, the casting technique using a mold which is a mold for resin molding and the casting technique for die casting for precision casting of aluminum are described. It is possible to carry out the invention.

[0135]

As described above, claim 1 or claim 2
According to the invention, the system acquires in advance the knowledge of a skilled technician from a casting method that has been proven in the past, and can use this knowledge, so that the advanced level according to the knowledge of the skilled technician Then, the casting method can be automatically performed. For this reason, the casting plan does not need to depend on the knowledge, know-how, and intuition of each skilled engineer, and it is not necessary to train a new skilled engineer. Therefore, it is possible to quickly respond to the needs for diversifying casting plans.

[0136] The use of such knowledge of a skilled engineer is performed by classifying this knowledge into a plurality of types of knowledge regarding the shape, position, and dimensions of the weir and the runner, and registering the knowledge. The determination of which knowledge content to use from among the knowledge content was made possible by sequentially determining each knowledge.

According to the second aspect of the present invention, the knowledge of the skilled technician is further divided into five types of knowledge of weir shape selection knowledge, weir position selection knowledge, runner type selection knowledge, gate design design knowledge, and layout coordinate determination knowledge. The expert system can be practically applied to the casting method.

Also, separately from the weir shape selection knowledge, the weir position selection knowledge, and the runner type selection knowledge, by classifying the gate system design knowledge, the weir and the number of runners are determined and then the weir is determined for the first time. Since the dimensions of the gate and the gate are determined, the inference time by the system of the present invention is reduced, and the operation of the system becomes more efficient.

[Brief description of the drawings]

FIG. 1 is an overall schematic diagram of a system according to an embodiment of the present invention.

FIG. 2 is a schematic flow chart showing the operation of the inference mechanism of FIG.

FIG. 3 is a diagram showing a network formed by five kinds of knowledge used when the inference mechanism of FIG. 1 performs inference.

FIG. 4 is a first half of a diagram illustrating a specific inference procedure according to the embodiment;

FIG. 5 is a latter half of a diagram showing a specific inference procedure of the present embodiment.

FIG. 6 is a diagram showing object data registered in the database unit of FIG. 6;

FIG. 7 is a diagram showing a shape of a weir in the object data of FIG. 6;

FIG. 8 is a diagram showing a knowledge card of weir shape selection knowledge among the five knowledges of FIG.

9 is a diagram for explaining the meaning of the priority of the weir shape selection knowledge in the knowledge card in FIG. 8;

10 is a diagram showing a knowledge card of weir position selection knowledge among the five knowledges of FIG.

11 is a diagram showing a knowledge card of runner shape selection knowledge among the five knowledges of FIG. 5;

12 is a diagram showing a knowledge card of feeder design knowledge, which is one of the gate design knowledges out of the five knowledges in FIG.

13 is a diagram illustrating a knowledge card of layout coordinate determination knowledge among the five knowledges of FIG. 5;

FIGS. 14A and 14B are views showing a product according to the present embodiment, wherein FIG. 14A is a side view and FIG. 14B is a front view.

FIG. 15 is a side view of a main mold which is a model of a part to be the product of FIG. 14;

FIG. 16 is a diagram showing six basic elements for expressing the shape of the main mold.

17 is a diagram showing a combination of the basic elements shown in FIG. 16 for expressing the main form of FIG.

FIG. 18 is a diagram illustrating an example of a shape attribute constituting a production constraint input in the present embodiment.

FIG. 19 is a diagram illustrating an example of a production constraint input in the present embodiment.

FIG. 20 is a diagram illustrating an execution result of the present embodiment.

21 is a diagram showing an example of a casting plan created by a skilled engineer for comparison with FIG.

FIG. 22 is a diagram showing an example of a schematic process of a conventional casting.

FIG. 23 is a view showing an example of an arrangement of a main die, a weir, a runner, a sprue bar and the like in a conventional casting.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Upper die 3 Lower die 5, 7 Plate 9, 11 Model of product part 13 Model for leading hot water 13 Gate stick 13B Runner 13C Feeder 13D Weir

Claims (2)

(57) [Claims] 1. A hot water is provided in a mold at a runner portion, a feeder portion,
Before the casting is performed by being guided in the order of the weir portion and the product portion, the model of the runner, the model of the feeder, the model of the weir, and the model of the product portion, i.e. This is a casting plan support expert system that automatically performs a casting plan for planning a die layout, and a means for inputting and registering a main mold shape and the number of main dies. Means for acquiring the knowledge of a skilled technician, classifying and registering the weirs and runners into a plurality of types of knowledge relating to shape, position, and dimensions; and a plurality of knowledge contents of each of the classified knowledge Of which
A means for deciding which knowledge content is to be used for each knowledge in order, and a runner and a hot water bath according to the initially input master shape, the number of master dies, and the knowledge content determined by a plurality of knowledge. ,
Means for simulating the layout of the weir and the main mold, and checking whether the layout is satisfactory.If the layout is not satisfactory, return to the initial number of main molds and repeat the procedure. A casting plan support expert system characterized by the fact that the procedure is completed when it is satisfactory. 2. A method according to claim 2, wherein the hot water is a part of a runner, a part of a riser in the mold,
Before the casting is performed by being guided in the order of the weir portion and the product portion, the model of the runner, the model of the feeder, the model of the weir, and the model of the product portion, i.e. A casting plan support expert system that automatically performs a casting plan for planning a mold layout, a means for inputting and registering a main mold shape, and a production constraint that is a constraint on a production policy and production equipment, Obtain the knowledge of skilled technicians from the casting methods that have been used in the past, and select the shape of the weir to select the shape of the weir, select the position and number of weirs, select the weir position, and select the shape of the runner Five types consisting of road shape selection knowledge, gate design design knowledge for determining the dimensions of the weirs and runners, and layout coordinate determination knowledge for determining the coordinates for laying out the main form, weirs, risers, and runners Classify into knowledge and register And a plurality of knowledge cards constituting each of the classified knowledge, wherein each knowledge is used depending on which knowledge card is adopted, and an attribute determined by the type of the classified knowledge, and the attribute is satisfied. The attribute and the attribute value express a constraint condition as a condition as to whether or not the knowledge card can be adopted, and a decision item to be determined when the constraint condition is satisfied is defined by the attribute value and the attribute value. The knowledge card in which a decision item is registered when it is expressed and adopted, and whether or not the constraint conditions are satisfied are determined by the master shape, production constraints,
A means for making a determination for each knowledge card in order by determining whether or not a decision item of the already registered knowledge card matches the constraint condition of the knowledge card; , Production constraints, and the decision items determined in each knowledge card of the five knowledge,
Means for simulating the layout of the main mold, weir, runner, and feeder, and checking whether or not it protrudes from the plate corresponding to the flat area of the sand mold. A casting plan support expert system characterized by returning to the input of the first production constraint, repeating the procedure, and terminating the procedure without protruding from the plate.