JPH03138758A - Commodity specification determining and estimating device - Google Patents

Abstract

PURPOSE:To enable a customer to estimate the cost and the period corresponding to specifications with a high precision by providing an estimating means which estimates the cost and the period required for production of a commodity based on commodity specifications determined by a commodity specification determining means. CONSTITUTION:The fundamental structure of the commodity is stored as structure description data in a data base; and when the customer inputs the form and the color of a presented sample as a appearance specification request, appearance specification request recognition results are connected to fundamental structure data to easily determine specifications of the commodity having the appearance desired by the customer. When commodity specifications according with the customer's will are determined, the cost and period required for production are estimated based on commodity specification data and data where the structure method added to structure description data describing the structure of the commodity is described. Thus, the cost of production and the appointed date of delivery of the commodity suiting customer's taste are detected.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention inputs and determines a customer's required specifications for a product as changeable, and calculates the cost required for manufacturing the product based on the determined product specifications. , relates to a product specification determination estimation device for estimating a period, in particular, from the appearance specifications of the product inputted by showing a sample similar to the appearance of the requested product, and furthermore the functional specifications inputted separately, The present invention relates to a product specification determination and quotation device that determines the specifications required by a customer for a product.

[Prior art] Until now, this type of device was disclosed in Japanese Patent Application Laid-open No. 63-120.
As described in Publication No. 68, it is equipped with a file containing a still image of a product catalog and information about the product, and when a customer specifies the type of product, the image and product information of that product are displayed. Once the quotation conditions are entered, a quotation will be output.

[Problems to be Solved by the Invention] However, the above-mentioned prior art can only be applied to estimates for products whose specifications have been determined. On the other hand, recently, customer needs for products have become more diverse, and customers are now ordering products that suit their tastes.
This is becoming more and more common. In such cases, since the specifications of each product are different,
It is necessary to clearly input the specifications and estimate the cost and period required for manufacturing according to those specifications, but up until now, such estimates have not been taken into account. It becomes.

The purpose of the present invention is to enable customers to clearly input product specifications according to their own preferences using samples, and also to determine the cost and period required to manufacture the product according to the specifications. To provide a product specification determination and estimation device that is capable of highly accurate estimation.

[Means for solving the problem] The above purpose is to
This is achieved by providing a product specification determining means for inputting and determining the product specifications in step 7, and an estimating means for estimating the cost and period required for manufacturing the product based on the product specifications determined by the means. More specifically, the product specification determining means includes at least a database that stores structure description data that describes the structure of various products, a plurality of sensors that manually determine the appearance of samples that indicate the appearance specifications for the product, and the sensors. a means for analyzing the data obtained by the method and recognizing the appearance of the sample; a means for storing the recognition result; and a means for comparing the recognition result data of the sample with the structural description data and converting the recognition result into structural elements of the product. means for associating data; means for displaying recognition results based on display display data generated from the recognition result data; means for accepting transformation instructions for the recognition result data; and recognition result data based on the transformation instructions. In addition, as an estimation means, a database that stores structure description data describing the structure of the product to which data describing the manufacturing method is added;
The method includes means for storing the determined product specifications for the product, and means for predicting data on the cost and period required to manufacture the product according to the product specifications using the structural description data and the determined required specifications for the product. It is achieved by doing. In particular, the three-dimensional shape and color of a sample as a product specification are configured to be recognized by analyzing data from a plurality of sensors.

[Function] The basic structure of the product is stored in the database as structural description data, but the appearance desired by the customer can be achieved by inputting the shape and color of the sample presented by the customer as appearance specification requirements. By linking the basic structure data with the appearance specification requirement recognition results, the specifications of a product having the appearance desired by the customer can be easily determined.

With the appearance specifications input in this way displayed on the display, if the customer inputs transformation instructions as necessary, the pre-recognition result data will be corrected based on this, and the sample It is possible to input different specifications.

By the way, the customer's transformation instructions are not only detailed instructions for each part of the product, but also general instructions for the entire product. If it is converted, the appearance specifications can be easily modified even by a batch transformation instruction. However, batch transformation instructions are often intuitive and therefore tend to be vaguely expressed, so if the transformation result is different from what the customer intended, use a graphic editor to modify the transformation result and use the modified value. By changing the parameters used in the mechanism that converts collective transformation instructions into individual instructions using , it becomes possible to learn the conversion parameters to match the customer's intentions.

Once the product specifications that match the customer's intentions have been determined as described above, the data describing the manufacturing method added to the product specification data and the structure description data describing the structure of the product can be used to determine the manufacturing method. It is possible to estimate the cost and period required.

When making an estimate, the accuracy of the estimate can be further improved by referring to data showing the status of the production line and the relationship between past estimate data and the actual cost and period required for manufacturing. be.

[Example] The present invention will be explained below with reference to FIGS. 1 to 12.

First, the product specification determination and estimation device according to the present invention will be described. FIG. 1 shows the configuration of an example of the device relating to product specification determination.

In this case, the database 1 stores in advance structural description data that describes the structure of various products, and the modeling processing unit 2 receives product types and attributes input based on requests from customers. , the data corresponding to the type and attributes are searched from the database 1, and by combining them, data describing the structure of the product specified by the customer is generated and stored in the product class object base 3. It has become. The image data of the sample appearance as a required specification for the product is detected by the sensor 4, and the image data is analyzed by the recognition section 5, so that the sample appearance can be recognized. At this time, in order to assist recognition in the perception 5, the prediction processing unit 6 predicts the appearance of the sample (prediction model) from the product class object base 3 and stores it in the prediction model base 7. In perception 5, recognition is performed using the predictive model, and the recognition results are stored in recognition object base 8. The structure interpretation unit 9 compares the sample recognition results stored in the recognition object base 8 with the prediction model base 7 and the data in the product class object base 3, and matches the recognition result data to the structural elements of the product. , the results are stored in the product instance object base 10. The graphic processing unit 12 generates display data from the data in the product instance object base 10, thereby
By displaying the appearance and functional specifications of the product based on the sample presented by the customer on the display 13, the customer can confirm the specifications. If the customer wants to change the displayed specifications, a change instruction is input from the command input section 14, and the specification determination module 11 changes the data in the product instance object base 10 based on this change instruction. The results are displayed so that the customer can confirm the changed specifications. This process of changing specifications is repeated until the customer is satisfied.

FIG. 2 shows the configuration of an example of a sensor as a main component thereof. In this example, the three-dimensional shape and color of the sample are input, and the sample is placed on a transparent net 44 attached to the inside of the frame of the stage 4 to hold it. The three-dimensional shape and color of the sample placed in this way can be detected by sensors from its surroundings. Specifically, for the position where the sample is placed, six color cameras 421, 422, 423, 424, 425, 426 are installed above, below, left and right, and in front and behind the sample.
are located as sensors, and slit projectors 43 are located at the upper right, upper left, lower right, and lower left, respectively.
1,432, 433, and 434 are arranged to irradiate the sample with slit light. those 6
The mutual positions of the color cameras on the stand have been checked in advance using a calibration pattern, and the positional relationship with the slit projector has also been checked in advance. Therefore, the three-dimensional position of the sample portion irradiated with the slit light can be detected. Furthermore, the three-dimensional shape and color of the sample can be input as one from the color image of the sample portion irradiated with the slit light.

With the sensor configuration as described above, the shape and color of the sample placed on the transparent net 44 are inputted from all around.

FIG. 3 shows the configuration of an example of the specification determination module 11 according to the present invention. The specification determination module 11 is for changing the appearance specifications of the product based on transformation instructions for the appearance of the product, and the types of transformation instructions include:
For example, for a model airplane, it is not just about each structural element, such as making the main wing longer, but also about the whole thing.
It is now also acceptable to refer to the whole, such as "long and thin." As shown in the figure, the specification determination module 11 is composed of three hierarchical units consisting of an input layer 111, an intermediate layer 112, and a product parameter layer 113, a command analysis section 110, and a learning mechanism 114. It has become a thing. Among these, the product parameter layer 113 is the product instance object base 10
The appearance specifications of the product can be changed by modifying the data in the product instance object base 10 based on the output from the product parameter layer 113. Further, the input layer 111 exists for each type of transformation instruction, and an example of the transformation instruction command will be explained in FIG. 4.

As shown in the figure, various types of words are prepared for overall shape modifiers, shape and color expressions, impression expressions, value expressions, environmental expressions, function expressions, and degree modifiers. Here, for example, assuming that a transformation instruction such as "long and thin" is input, this transformation instruction is decoded by the command analysis unit 110, and a predetermined numerical value is output only to the input crab unit 1111 corresponding to "long and thin". It looks like this. Based on this value, the input crab unit 1111 outputs a function value, but the function value from each input crab unit is multiplied by the intermediate layer weighting coefficient corresponding to the input crab unit and the intermediate unit, and the sum is the output to that intermediate unit. It is designed to be given as a value f1. That is, the output value f1m to a certain intermediate unit (m is the unit number for that intermediate unit) is given by a sigmoid function as shown below, using the unit number of the input unit as a variable.

f 1m=Σ((input crab unit output) 1×(middle layer weighting coefficient) 1.) ・・・・・・・・・(1)
Each intermediate unit is given fl++ as described above, but based on this, each intermediate unit also outputs a certain function value, which is then given to the product parameter unit.

The input/output relationship between each intermediate unit and each product parameter is similar to the input/output relationship between each input crab and each intermediate unit. That is, the output value f to a certain product parameter unit. . (n is the unit number for that product parameter unit) is the unit number m of the intermediate unit
is given as a variable as follows.

f. =Σ((intermediate unit output).×(product parameter layer weighting coefficient), .) ・・・・・・・・・(2) Now, f6. , a certain function value is obtained from each product parameter unit, and data in the product instance object base 10 is changed based on these function values. Through such processing, product parameters can be converted based on the modification instruction command. In addition, the explanation in the above example is for the case where there is a crab knit corresponding to "slender", but if
If there is no corresponding incoming crab knit, but instead there are incoming crab knits corresponding to "thin" and "long", each of these incoming crab knits is given a predetermined numerical value by the command analysis unit 110. Each incoming crab is given a numerical value, such as a numerical value equivalent to 1/2.

FIG. 5 shows an example of each unit in the input layer 111. As shown in the diagram, in addition to instructions for deforming each structural element, such as ``lengthen the main wing'' and ``lengthen the horizontal stabilizer,''
Since there are deformation instructions for the entire structure, such as ``elongate and elongate,'' it is possible to convert a collective deformation instruction into a deformation instruction for each structural element. In addition, there are also irikannits that correspond to sensory terms such as "smartly" and "heavily," so it is possible to convert sensory deformation instructions into deformation instructions for each structural element. Fast”
1. Depending on the Ikaninit that corresponds to a term indicating a function, such as ``sturdy'', it is also possible to convert a functional instruction into a modification instruction for each structural element.

As shown above, in order to convert transformation instructions into transformation instructions for recognition result data, it is necessary to appropriately set conversion parameters such as intermediate layer weighting coefficients and product parameter layer weighting coefficients. The method for determining is shown in Figure 6. In this case, first, the conversion parameters are initialized appropriately (process 601).

As the initial setting method, if appropriate transformation parameters can be predicted by some method, that method may be used, but if they cannot be predicted, random numbers may be used. Next, if some transformation instruction is input (process 602
), in response to this transformation instruction, performs transformation using the set transformation parameters, outputs the amount of modification to the data in the product instance object base 10 from the product parameter layer 113 (603), and performs the modification based on this amount of modification. The product specifications are displayed on the display 13 (process 604). From this display, it can be determined whether the modified result matches the intended transformation instruction (process 605), but if the modified result is the intended one, the series of processes ends. However, from an epithelial point of view, the initial transformation parameters are not properly defined, so it is easy to predict that they will not match the intended transformation instructions. In such a case, the transformation parameters can be modified (process 607) by inputting the intended transformation result as the desired transformation value (process 606). For example, in the case of a transformation instruction regarding appearance, the shape of the converted image is modified using the graphic editor function of the graphic processing unit 12 to match the intention; Enter the value of the function you
The difference between this intended result and the result converted using the original conversion parameters is determined, and the conversion parameters are corrected so as to reduce this difference. That is, by inputting the processing result for the transformation instruction and the intention of the person who gave the transformation instruction, and modifying the transformation parameters to reduce the difference between them, the intended content of the transformation instruction is automatically learned. It is something that This makes it possible to input and understand the intent of not only functional and performance requirements that can be clearly specified numerically, but also sensory specification requirements that are difficult to clearly define.

Next, in perception 5, the order in which data is taken in from a plurality of cameras and range finders connected as sensors and the samples are recognized will be explained. When it comes to samples, there are many types ranging from simple to complex, so a large number of sensors are required to recognize complex samples.

Acquiring and recognizing data from a large number of sensors will require a lot of time, but if the sample is simple, processing data from a small number of sensors may be sufficient. is clear. therefore,
In order to recognize a sample in the minimum necessary time, it is effective to determine the number of sensors to be used depending on the shape of the sample. In perception 5, the predicted appearance of the sample can be known by referring to the prediction model base 7, and an example of the procedure for using the sensor using this is shown in FIG. First, when there are six cameras as shown in FIG. 2, the symmetry of the sample is checked, and if it is symmetrical, only the cameras necessary for recognizing one side are used.

For example, in the case of a symmetrical sample, top, bottom, front,
After that, it is sufficient to use only the five cameras on the right side to sequentially capture images, and the camera on the left side is not necessary. Furthermore, if the sample is a polyhedron and all edges are detected by two or more cameras, the three-dimensional shape can be recognized by the stereo method using only the cameras.
There is no need to use a range finder. In cases other than the above, a three-dimensional shape cannot be recognized with a camera alone, so a range finder must be used. Even in this case, the portion where the three-dimensional shape needs to be recognized by the range finder can be limited, so it is only necessary to measure the three-dimensional shape of that portion. Furthermore, if the sample has a two-dimensional shape with a similar thickness, the shape can be recognized using only the upper and lower cameras.In this way, the relationship between the characteristics of the predicted shape and the necessary sensors can be determined based on geometry. It is determined by a mathematical calculation.

Next, a method for predicting the function of a product whose appearance has been determined based on the input sample will be described. FIG. 8 shows an example of data describing the structure of the product. This data includes a slot in which the structure of the product is described, a procedure for predicting the function of the product, and further points to sample recognition result data or correction data thereof. Therefore, predicting the function of a product based on the recognition result data can be realized by starting a procedure for predicting the function of the product using the recognition result data as a parameter.

Furthermore, when displaying recognition results, it is possible to accept instructions for the display range, direction, and magnification of the recognition results, and to display the results based on these instructions. This is an effective function when you want to change the recognition results. That is, on a two-dimensional display, only one side of the recognized sample is displayed, and the entire appearance cannot be seen from only one screen. In addition, even when attempting to change the appearance specifications of a product, it may not be possible to display the changed part on one screen, so such a display method is not only necessary, but also If it is approved, the recognition results can be checked and changed quickly. FIG. 9 shows an example of a configuration that enables such display. Here, the dial 121 is for inputting the angle to be rotated around the coordinate origin of the product shape description in the structure description data, and if the specified angle is input from this, it is indicated by the structure description data. By rotating the position of the sample recognition result data or its modified data, the recognition results can be displayed in a display direction that can be changed arbitrarily. In addition, by moving the stick 122 vertically and horizontally, the coordinate origin of the product shape description in the structure description data can be moved vertically and horizontally with respect to the screen coordinates, that is, in the vertical and horizontal directions. This is for scrolling the image on the stick 1.
23 itself can be moved up and down to change the display magnification of the recognition results. The graphic processing unit 12 performs display control based on the instructions from these human-powered devices, so that images related to the display range, display direction, and display magnification after the recognition result has been changed are displayed on the display 13. This is what is meant to be done.

By the way, if you examine the sample recognition results from various directions as described above, you may find parts that are not recognized or parts that are incorrectly recognized. It is necessary to perform the recognition of that part again to obtain a correct recognition result, but the method for obtaining a correct recognition result will be described. FIG. 1O shows the procedure of the method. In this case, first, the recognition result of the sample is checked by the means shown in FIG. 9 (process 101). As a result, if an unrecognized portion or a misrecognized portion is found (process 102), the screen is set to the screen on which that portion is displayed. In this screen setting, first, the range related to re-recognition on the screen is specified using the mouse 124 (process 103), and with this specification, the range related to re-recognition can be obtained by the coordinates describing the product shape in the structure description data. This is the result. If this coordinate position is given to perception 5 (process 104), perception 5 will perform recognition using an appropriate sensor for the position range as a recognition target (process 105). More specifically, in perception 5, a recognition target is detected using a sensor or recognition algorithm different from the previously stored sensor type and recognition algorithm type used to initially recognize the position range. The position range of the location is recognized. Through the series of processes described above, a recognition result different from the previous one is obtained and displayed, but if this recognition result is correct, the process ends. If the recognition result is incorrect, the above process is repeated until a correct recognition result is obtained.

Now, the specifications of the product can be determined as described above, but after this, the cost required to manufacture the product according to the specifications,
The period is now estimated. To explain this estimation method, the cost and period will be almost constant if the standard specifications for the product are almost the same, but for products that are ordered using the product specification determination and quotation system and that include many requests from customers. Since the specifications differ for each item, the cost and time required for manufacturing also differ for each item. In order to estimate the cost and period required for manufacturing, as shown in Figure 11, data that describes the manufacturing method in the slot of the product structure description data and the estimated cost and period required for manufacturing are required. Procedures to do so are required. The structure description data also points to data indicating the required specifications for the product and data indicating the status of the manufacturing line, such as sample recognition result data or its correction data, so these data should also be referred to. Therefore, an estimate can be made. In other words, the cost and time required to manufacture a product according to the required specifications are determined based on the required specification data for the product, such as manufacturing method description data, production line status data, and recognition result data, as parameters. Estimates are made by invoking procedures to predict costs and periods.

Costs and periods are estimated as described above, but here we will explain how to improve the estimation accuracy.
FIG. 12 shows an example of structural description data about products for improving estimate prediction accuracy. In this case, in addition to the data shown in Figure 11, the prediction processing parameter change procedure, prediction processing parameters, past prediction data, and data on the actual cost and period required for manufacturing are added. . Among these, in the prediction processing parameter change procedure, the prediction processing parameters are changed based on the difference between the past prediction data and the actual cost and period data required for manufacturing. As a result, the prediction processing parameters are calculated based on the prediction data and the actual manufacturing costs.
The data will be corrected so that it is closer to the period data.

Note that the predicted processing parameters are referred to in the procedure for predicting the cost and period required for manufacturing.

[Effects of the Invention] As explained above, according to the present invention, it is possible to clearly input the specifications required by the customer for the product, and in particular, it is possible to input the external shape, which is difficult to specify, based on a sample. Therefore, the appearance specifications can be easily input by changing it as necessary (particularly related to claim 2.10).

At this time, the deformation instructions are not only geometric deformation instructions for local areas, but also batch deformation instructions and sensory expressions for the entire shape, so it is extremely easy to recognize the appearance from the sample. The shape can be changed. Moreover, in the conversion process for such transformation instructions, the customer's intention can be learned.
Even ambiguous sensory instructions can be converted gradually and accurately (particularly related to claims 2.12 to 15).

Furthermore, after the product specifications are input in the manner described above, the cost and period required to manufacture the product can be estimated with high accuracy for the individual specification product based on the product specifications. (particularly related to claims 3, 16 and 17).

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the configuration of an example of a product specification human-powered device according to the present invention, FIG. 2 is a diagram showing the configuration of an example of a sensor as one component of the device, and FIG. FIG. 4 is a diagram showing an example of the configuration of a specification determination module as one component. FIG. 4 is a diagram showing an example of various transformation instruction commands when changing the appearance of a product. FIG. FIG. 6 is a diagram showing an example of an input layer unit as one component, FIG. 6 is a diagram showing a method for determining conversion parameters for converting a deformation instruction into a deformation instruction for recognition result data, and FIG. 7 is a diagram showing a sensor usage procedure. Figure 8 showing an example of
The figure shows an example of product structure description data that makes it possible to predict the functions of a product, FIG. 9 shows an example of a configuration in which the display of recognition results is changed in various ways, and FIG. Figure 11 shows an example of product structure description data for estimating the cost and period required to manufacture a product. Figure 12 shows how to improve the estimation accuracy. FIG. 3 is a diagram showing an example of product structure description data for use in the product structure. l...Database, 2...Modeling processing unit, 3.
... Manufacturing class object base, 4... Sensor,
5... Perception, 6... Prediction processing unit, 7...
- Prediction model base, 8... Recognition object base, 9... Structure interpretation unit, IO... Product instance object base, 11... Specification determination module, 1
2... Graphic processing unit, 13... Display, 14... Command input unit, 41... Stage, 4
21-426...Color camera, 431-434...
・Slit floodlight, 44...Transparent net, 110...
・Command interpretation unit, 111... Input layer, 112...
Middle layer, 113...Product parameter layer, 114...
Learning mechanism, 121...Dial, 122, 123...
・Stick, 124...Mouse

Claims (1)

[Scope of Claims] 1. Input customer's required specifications for the product as changeable.
A product specification determining and estimating device comprising: product specification determining means for determining product specifications; and estimating means for estimating the cost and period required for manufacturing the product based on the product specifications determined by the means. 2. As a product specification determination means, at least a database that stores structure description data describing the structure of various products, a plurality of sensors that input the appearance of a sample indicating the appearance specifications for the product, and data obtained by the sensors means for analyzing and recognizing the external appearance of the sample, means for storing the recognition result, and means for comparing the recognition result data of the sample with the structural description data to correspond the recognition result data to the structural elements of the product. a means for displaying a recognition result based on display display data generated from the recognition result data; a means for receiving a transformation instruction for the recognition result data; and a means for correcting and re-displaying the recognition result data based on the transformation instruction. 2. The product specification determining and estimating device according to claim 1, further comprising means for determining and estimating product specifications. 3. As an estimation means, a database that stores structure description data that describes the structure of the product to which data that describes the manufacturing method is added, a means that stores determined product specifications for the product, and the structure described above. 3. The method according to claim 1, further comprising means for predicting data on cost and period required for manufacturing a product according to the product specifications using descriptive data and determined required specifications for the product. Product specification determination and quotation device. 4. Claims 2 and 3, further comprising a stage for holding the sample on a transparent net in association with a plurality of sensors that input the appearance of the sample indicating the appearance specifications for the product.
The product specification determination and estimation device described in any of the above. 5. The product specification determining means includes means for accepting input of product types and attributes, and searching for data describing structures corresponding to the types and attributes from structure description data describing structures of various products; The product specification determination and estimation device according to any one of claims 2 to 4, further comprising: generating means. 6. Regarding the means of analyzing data obtained by multiple sensors and recognizing the appearance of a sample, data useful for recognition can be obtained by predicting the data to be input to each sensor from the structural description data of the product. The product specification input device according to any one of claims 2 to 5, further comprising means for sequentially inputting data from a sensor. 7. Means for adding data describing functions to structural description data describing the structure of the product, and predicting functions using sample recognition result data associated with each structural element or modified data of the data. The product specification input device according to any one of claims 2 to 6, comprising:. 8. In relation to the means for displaying recognition results based on display data generated from recognition result data of the sample, means for receiving instructions regarding the display range, direction, and magnification of the recognition results of the sample; The product specification input device according to any one of claims 2 to 7, further comprising means for displaying based on the information. 9. A means for instructing re-recognition of a specific part of the displayed recognition result, a means for re-recognizing the part using another method different from the previous recognition, and a means for displaying the re-recognized result. The product specification input device according to any one of claims 2 to 8. 10. In connection with a plurality of sensors that input the three-dimensional shape and color of the sample, means for analyzing data from the sensor and recognizing the three-dimensional shape and color of the sample is provided. 9. The product specification determination and estimation device according to any one of 9. 11. The plurality of sensors include color cameras placed above, below, left, right, front, and rear of the sample;
11. The product specification determining and estimating device according to claim 10, comprising slit projectors arranged at the upper left, lower right, and lower left. 12. Means for receiving transformation instructions for sample recognition result data for each structural element or for the entire product at once, and for converting the collective transformation instructions into transformation instructions for each structural element. Claim 2 comprising:
12. The product specification determination and estimation device according to any one of 12 to 12. 13. The product specification determining and estimating device according to claim 12, further comprising means for receiving an instruction indicating a function as a collective modification instruction and converting it into a modification instruction for each structural element. 14. The product specification determining and estimating device according to claim 12, further comprising means for receiving an instruction in sensory terms as a collective modification instruction and converting it into a modification instruction for each structural element. 15. The product specification determining and estimating device according to any one of claims 12 to 14, further comprising means for generating a conversion parameter for converting a modification instruction into a modification instruction for recognition result data by example learning. 16. In relation to the means for predicting data on the cost and period required for manufacturing a product, there is provided a means for predicting the cost and period using data indicating the status of the production line.
A product specification determination and estimation device according to any one of claims 3 to 16. 17. When predicting the cost and period required to manufacture a product from data describing the structure of the sample, data indicating the product specifications for the product, and data indicating the status of the production line, it is necessary to use past data and 17. The product specification determining and estimating device according to claim 3, further comprising means for changing parameters used for prediction using data showing a relationship between costs and periods actually required for manufacturing.