JP2009230169A - Parameter determination support device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To determine a proper operation parameter set by executing dimension reduction in a short time even when there are a large number of attributes of a precondition. <P>SOLUTION: This parameter determination support device includes: a precondition space construction part 2 constructing a precondition space that is a base group of a space of the precondition; and a precondition space part 3 converting coordinates of a recommendation precondition into coordinates on the precondition space. A recommendation parameter presentation part 4 calculates similarity between a plurality of case records and the recommendation precondition by use of the coordinates on the precondition space in the plurality of case records recorded in a case accumulation part 1 and the coordinates of the recommendation precondition converted into the coordinates on the precondition space, and determines the recommended operation parameter set by use of the similarity. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a parameter determination support apparatus that determines an optimal operation parameter set under the presented preconditions.

For example, Patent Document 1 below discloses a parameter determination support apparatus that determines a recommended operation parameter by similarity search after performing dimension reduction by principal component analysis for a precondition attribute group. .
As a prerequisite dimension reduction method, there is a technique in which a vector stored in a database is selected at random, and a dimension is reduced by an orthogonalized base vector (see, for example, Patent Document 2).

Japanese Patent Laid-Open No. 5-40852 (paragraph numbers [0055] to [0071], FIG. 3) Japanese Patent Laying-Open No. 2005-50197 (paragraph numbers [0011] to [0014])

Since the conventional parameter determination support device is configured as described above, the principal component analysis reduces the dimension of the prerequisite attribute group, but in the principal component analysis, the eigenvalues of the matrix whose size is the number of attributes It is necessary to carry out the solution process. Therefore, when the number of attributes is large, there is a problem that requires a lot of time.
In addition, since the principal component direction is expressed by a weighted sum of a large number of attributes, there is a problem that it is difficult to interpret the meaning.

  The present invention has been made to solve the above-described problems. Even when the number of prerequisite attributes is large, dimension reduction can be performed in a short time to determine an appropriate operation parameter set. An object is to obtain a parameter determination support device.

  The parameter determination support apparatus according to the present invention uses the preconditions and result evaluations in a plurality of case records recorded in the case record recording means to construct a precondition space that is a base group of the precondition space. A precondition space constructing means and a coordinate conversion means for transforming the coordinates of the recommended precondition that is the precondition to be recommended into coordinates on the precondition space constructed by the precondition space constructing means, and a recommended operation parameter determining means Using the coordinates in the precondition space in the plurality of case records recorded in the case record recording means and the coordinates of the recommended preconditions converted by the coordinate conversion means, the similarity between the plurality of case records and the recommended preconditions And the recommended operation parameter set is determined from the operation parameter sets in multiple case records using the similarity. Those were Unishi.

  According to this invention, using the preconditions and result evaluation in a plurality of case records recorded in the case record recording means, a precondition space construction that builds a premise space that is a base group of the precondition space And a coordinate conversion means for converting the coordinates of the recommended precondition, which is the precondition for the recommended object, into coordinates on the precondition space constructed by the precondition space construction means, and the recommended operation parameter determination means records the case record Using the coordinates on the precondition space in the plurality of case records recorded in the means and the coordinates of the recommended preconditions converted by the coordinate conversion means, the similarity between the plurality of case records and the recommended preconditions is calculated, Since it was configured to determine the recommended operating parameter set from among the operating parameter sets in multiple case records using the similarity, Even when the condition number of attributes of mass in a short time by carrying out the dimensionality reduction, there is an effect that it is possible to determine the proper operating parameters set.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a parameter determination support apparatus according to Embodiment 1 of the present invention. In the figure, the case accumulation unit 1 is a precondition that the state of the surroundings of the target apparatus before the operation is expressed, A plurality of case records including an operation parameter set including a plurality of operation parameters for setting the operation content of the target device and a result evaluation expressing the surrounding state of the target device after the start of the operation are recorded. The case storage unit 1 constitutes case record recording means.
The precondition space construction unit 2 uses the preconditions and result evaluations in the plurality of case records recorded in the case storage unit 1 to perform a process of constructing a precondition space that is a base group of the precondition space. . The precondition space construction unit 2 constitutes a precondition space construction means.

The precondition space conversion unit 3 performs a process of converting the coordinates of the recommended preconditions, which are recommended preconditions designated by the user, into coordinates on the precondition space constructed by the precondition space construction unit 2. The precondition space conversion unit 3 constitutes coordinate conversion means.
The recommended parameter presentation unit 4 uses the coordinates on the precondition space in the plurality of case records recorded in the case storage unit 1 and the coordinates of the recommended preconditions converted by the precondition space conversion unit 3 to generate a plurality of cases. The similarity between the record and the recommended precondition is calculated, and using the similarity, a recommended operation parameter set is determined from among the operation parameter sets in the plurality of case records, and a process of presenting it to the user is performed. The recommended parameter presentation unit 4 constitutes recommended operation parameter determination means.

Next, the operation will be described.
The case storage unit 1 records a plurality of case records including a precondition, an operation parameter set, and a result evaluation. In response to requests from the precondition space construction unit 2 and the recommended parameter presentation unit 4, the corresponding case record Is provided to the precondition space construction unit 2 and the recommended parameter presentation unit 4.

Here, FIG. 2 is an explanatory diagram illustrating the types of case records recorded in the case storage unit 1.
The simulated case record in FIG. 2A is a case record obtained when a simulator of the target device is available, and facilitates the accumulation of the case records by cooperating with the simulator. .
As shown in FIG. 2A, the attribute of the simulated case record is composed of three types: a precondition, an operation parameter set, and a result evaluation.

The prerequisite attributes include controllable attributes and non-controllable attributes.
For example, in the case of a simulator having a movable sensor, the controllable attribute includes the position of the sensor and the distance and range that can be detected when the sensor is pointed.
The uncontrollable attributes include, for example, the outside air temperature, humidity, and wind speed.
The attribute of the operation parameter set is a controllable attribute and is a recommended attribute. For example, in the case of a sensor system, there are switching between a mode in which many are found and a mode in which the found items are not lost.
The attribute of the result evaluation includes information indicating the state of the simulator after the operation is started and information obtained by counting the state within a certain time. For example, the number found per unit time or the average time kept per object.

FIG. 3 is an explanatory diagram illustrating a partial configuration example of the parameter determination support device when the result evaluation is collected from the simulator of the target device and the case record is generated.
In the figure, the environment setting unit 11 gives an optional precondition to the simulation unit 13 which is a simulator of the target device, as an environmental information attribute (a situation before the simulation unit 13 operates, Information that describes the situation, and information that can actually be measured is applicable).
The operation parameter setting unit 12 sets an arbitrary operation parameter set to the simulation unit 13 that is a simulator of the target device, and performs processing for instructing the operation start of the simulation unit 13.
The environment setting unit 11 and the operation parameter setting unit 12 constitute a simulator activation unit.

The simulation unit 13 is a simulator of the target device that starts an operation corresponding to the operation parameter set set by the operation parameter setting unit 12 under the preconditions set by the environment setting unit 11.
The result evaluation unit 14 evaluates the state around the simulation unit 13 after the simulation unit 13 starts its operation, or the total of the state in time, and the result evaluation and set by the environment setting unit 11 A case record (see FIG. 2A) including the set preconditions and the operation parameter set set by the operation parameter setting unit 12 is generated, and the case record is output to the case storage unit 1 in FIG. Perform the process. The result evaluation unit 14 constitutes a case record generation unit.

Next, the actual case record in FIG. 2B is a case record obtained when the operation state of the target device can be measured, and it is easy to accumulate the case records by cooperating with the actual device. It aims to make it easier.
As shown in FIG. 2B, the attribute of the actual case record is composed of three types of environment information (prerequisite), operation parameter set, and result information (result evaluation).
The attribute of the environment information is information describing the surrounding situation before the target device operates and the status of the target device itself, and information that can be measured corresponds to the precondition.
The attribute of the result information is a state around the target device after the target device starts operating, or the state is totaled within the time, and corresponds to measurable information.

FIG. 4 is an explanatory diagram showing a partial configuration example of the parameter determination support device when collecting the result evaluation from the target device (actual device) and generating a case record.
In the figure, the environment measuring unit 21 performs a process of measuring environment information indicating a situation around the function unit 23 before the function unit 23 as a target device starts operation.
The operation parameter setting unit 22 sets an arbitrary operation parameter set for the function unit 23, sets environment information measured by the environment measurement unit 21 as a precondition, and instructs the function unit 23 to start operation. To implement.
The environment measuring unit 21 and the operation parameter setting unit 22 constitute a target device activation unit.

The function unit 23 is a target device that starts an operation corresponding to the operation parameter set set by the operation parameter setting unit 22 under the preconditions set by the operation parameter setting unit 22.
The result measurement unit 24 performs a process of measuring the state around the functional unit 23 after the functional unit 23 starts operation, or the total of the state in time.
The result evaluation unit 25 evaluates the measurement result of the result measurement unit 24, and the case record including the result evaluation and the precondition and the operation parameter set set by the operation parameter setting unit 22 (see FIG. 2B) Is generated, and the case record is output to the case storage unit 1 of FIG.
The result measuring unit 24 and the result evaluating unit 25 constitute case record generating means.

Next, the case totalization record in FIG. 2 (c) is created by totaling a simulated case record in FIG. 2 (a) or a subset of the actual case record in FIG. 2 (b). By adding a case record or a subset of actual case records to the case record recorded in the case storage unit 1 of FIG. 1, it becomes easy to remove the uncontrollable attribute from consideration.
The attributes of the case summary record are composed of a precondition summary, an operation parameter set, and a result evaluation summary.
For example, if only the controllable attributes are counted among the preconditions of the simulated case record, it is possible to create a case count record having only the controllable attributes as the precondition aggregation attributes.
Specifically, for example, if there is a random number that simulates a disturbance in the precondition, or if the weather attribute is included, only the random number or the weather is different, but other preconditions and operation parameter sets Case records having the same attribute value are regarded as subsets, and the average and variance of the attribute values of the result evaluation are calculated in the subsets to be the attributes of the result evaluation aggregation.

The precondition space construction unit 2 collects a plurality of case records recorded in the case storage unit 1, and uses the preconditions and the result evaluations included in the plurality of case records, and the basis group of the precondition space Construct a precondition space that is
Hereinafter, the processing content of the precondition space construction unit 2 will be specifically described. FIG. 5 is a flowchart showing the processing contents of the precondition space construction unit 2.

When the precondition space construction unit 2 receives an instruction to select a precondition or an instruction to quantify the precondition attribute from the user, the precondition space construction unit 2 selects a selection target from among a plurality of case records recorded in the case storage unit 1. The preconditions are collected (from the preconditions to be selected, the preconditions whose attributes can be quantified are collected), and the attributes of the preconditions are quantified (step ST1).
However, it is possible to set so as not to accept the user's instruction, and when the user's instruction is not accepted or when there is no user's instruction, the attribute has already been quantified from the case storage unit 1 Try to collect prerequisites.
At this time, all the preconditions whose attributes have already been digitized may be collected, or the preconditions whose attributes have already been digitized may be selected at random and collected.

When the precondition space construction unit 2 receives a result evaluation selection instruction or a threshold instruction from the user, the precondition space construction unit 2 collects a result evaluation of a selection target from among a plurality of case records recorded in the case storage unit 1 ( Step ST2).
However, it is possible to set so as not to accept the user's instruction, and when the user's instruction is not accepted or when there is no user's instruction, all result evaluations are collected from the case storage unit 1. To.
In such a case, the threshold value is determined from the statistics of the attribute of the collected result evaluation. For example, the average value or median value of the attributes of all results evaluation is determined as the threshold value.
When the result evaluation attribute is composed of a finite number of values, one threshold value may be determined for each individual result evaluation instead of a single threshold value as a whole.

When the precondition space construction unit 2 collects the preconditions and the result evaluation from the case storage unit 1 as described above, the precondition space construction unit 2 selects one precondition from a plurality of preconditions and also evaluates the plurality of results. One result evaluation is selected from the list.
When the precondition space construction unit 2 selects one precondition and one result evaluation, the precondition space construction unit 2 gives the precondition and the result evaluation to the SVM application module (module that outputs the discriminant normal vector), thereby the SVM application module. A discriminant normal vector is acquired from (step ST3).
This process is repeated for the number of collected result evaluations, and discriminant normal vectors are acquired for the number of collected result evaluations.

Here, the SVM application module calculates a discrimination hyperplane that divides the space spanned by the attribute group of the precondition into two according to the result evaluation index.
For calculation of the discrimination hyperplane, a mechanism of a support vector machine (SVM), which is a known technique, can be used.
The discriminant hyperplane obtained by this mechanism is represented by the discriminant normal vector, the direction of the discriminant normal vector represents the normal direction of the discriminant hyperplane, and the length of the discriminant normal vector is the width of the margin between the two divisions. It represents.

When the precondition space construction unit 2 acquires the discriminant normal vector, the precondition space construction unit 2 accepts an instruction for specified performance from the user.
However, it is possible to set so as not to accept the user's instruction. When the user's instruction is not accepted or when there is no user's instruction, for example, a value exceeding 50% is set as the system default value. Set as specified performance.

The precondition space construction unit 2 measures the discrimination performance of a plurality of discriminant normal vectors, and if the discriminative performance is equal to or higher than the prescribed performance, the precondition space construction unit 2 adds the discriminant normal vector to the base group of the precondition space. A condition space is constructed (step ST4).
Here, the instruction of the specified basis number is accepted from the user, and when the basis number of the basis group in the space of the precondition does not reach the specified base number, the process returns to the precondition selection process of step ST1. You may try to add a base.

FIG. 6 is an explanatory diagram showing a mechanism of base generation of the precondition space.
In the example of FIG. 6, three example records x1, x2, and x3 are assumed, and are assumed to exist in a two-dimensional precondition space.
Further, the result evaluation indexes are assumed to be two, f and g, and the attribute values of the result evaluation in the three case records x1, x2, and x3 are f (x1), f (x2), f (x3), It is expressed by g (x1), g (x2), and g (x3).

As described above, each threshold value of the attribute of the result evaluation (or each set of threshold values) is determined by a user instruction or the like, and as shown in FIGS. 6A and 6B, three case records x1 , X2, x3 result evaluation indexes f, g are divided into two with the threshold as a boundary.
That is, in the result evaluation index f, as shown in FIG. 6A, the attribute value is divided into f (x1), f (x2), and f (x3). As shown in (b), it is assumed that the attribute value is divided into g (x1), g (x3), and g (x2).

When SVM is applied to the division of the result evaluation index f, a discrimination line indicated by a straight line in FIG. 6A and a normal vector F having a unit length of the discrimination line indicated by an arrow are obtained.
Similarly, when SVM is applied to the division of the index g of the result evaluation, a discrimination line indicated by a straight line in FIG. 6B and a normal vector G having a unit length of the discrimination line indicated by an arrow are obtained. It is done.
These two normal vectors F and G become new base vectors F and G in the precondition space, and the coordinates thereof are obtained by inner products of the base vectors F and G and the case records x1, x2, and x3.
In FIG. 6C, the inner product of the basis vector F of the index f of the result evaluation and each of the case records x1, x2, and x3 is F · x1 = −1, F · x2 = −1, and F · x3 = + 1. In this example, the inner product of the basis vector G of the result evaluation index g and each of the case records x1, x2, and x3 is G · x1 = −1, G · x2 = + 1, and G · x3 = −1. .
In FIG. 6C, a case record x4 is a recommended case record newly designated by the user, and the precondition included in the case record is a recommended precondition (recommended precondition). Suppose that

When the precondition space construction unit 2 constructs the precondition space, the precondition space conversion unit 3 converts the coordinates of the recommended preconditions included in the case record x4 designated by the user into coordinates on the precondition space. .
FIG. 6C shows an example in which the inner product of the base vector F and the case record x4 is converted to F · x4 = −2, and the inner product of the base vector G and the case record x4 is converted to G · x4 = −1. ing.

When the precondition space conversion unit 3 converts the coordinates of the recommended preconditions included in the case record x4 into coordinates on the precondition space, the recommended parameter presentation unit 4 converts the case record x1 recorded in the case storage unit 1 , X2, x3, and the recommended record of the case record x1, x2, x3 using the precondition space coordinates included in the case record x4 converted by the precondition space conversion unit 3 The similarity of the precondition is calculated, and using the similarity, a recommended operation parameter set is determined from the operation parameter sets included in the case records x1, x2, and x3 and presented to the user.
Hereinafter, the processing content of the recommended parameter presentation part 4 is demonstrated concretely. FIG. 7 is a flowchart showing the processing contents of the recommended parameter presentation unit 4.

For example, the recommended parameter presentation unit 4 calculates the Euclidean distance as the similarity between the preconditions included in the case records x1, x2, and x3 and the recommended preconditions included in the case record x4.
When the recommended parameter presenting unit 4 receives an instruction of the similarity reference value from the user, the recommendation parameter presentation unit 4 selects a case code having a similarity (Euclidean distance) equal to or greater than the similarity reference value from the case records x1, x2, and x3 (step ST11). .
Alternatively, when an instruction of the number of similarities N (for example, N = 2) is received from the user, the top N case codes with the highest degree of similarity are selected from the case records x1, x2, and x3.

In the example of FIG. 6C, the precondition with the highest similarity (Euclidean distance) is the precondition included in the case record x2, and the precondition with the next highest similarity (Euclidean distance) is the case. The precondition included in the record x1 and the precondition having the lowest similarity (Euclidean distance) is determined to be the precondition included in the case record x3.
Therefore, for example, when N = 2 is designated as the number of similarities, the case records x2 and x1 are selected, and the case record x3 is excluded.
It is possible to set not to accept an instruction such as a similar reference value from the user. When the user's instruction is not accepted or when there is no user's instruction, it is recorded in the case storage unit 1. All the case records x1, x2, and x3 may be selected.
Alternatively, a default similarity criterion value may be set in advance, and case records that are equal to or higher than the similarity criterion value may be selected. Alternatively, a default number of similarities may be set in advance, and case records may be selected in descending order of similarity by the number of similarities.

When the recommended parameter presenting unit 4 selects a case code as described above, the recommended parameter presenting unit 4 accepts an instruction of a weighting vector for a result evaluation attribute group from the user.
The recommended parameter presentation unit 4 calculates the inner product of the result evaluation attribute and the weighting vector included in each selected case code as a unified result evaluation value, and adds the unified result evaluation value to the case record (step) ST12).
However, it is possible to set so that the instruction of the weighting vector is not accepted from the user. When the instruction of the user is not accepted or when there is no instruction of the user, a default value is set in advance, for example, The unification result evaluation value may be calculated with all the weights set to “1”.

In the example of FIG. 6C, if all the weights are set to “1”, the case record x2 and the case record x3 both have a unified result evaluation value “0”, and the case record x1 has a unified result evaluation value “−2”. Is calculated.
Since the case record x3 is excluded in the process of step ST11, the unification result evaluation value of the case records x1 and x2 is calculated, and the unification result evaluation value is added to the case records x1 and x2.

When the recommended parameter presenting unit 4 receives an instruction on the number of selections from the user, the recommended parameter presenting unit 4 selects case records corresponding to the number of selections in descending order of the unification result evaluation value, and recommends an operation parameter set included in the case records. The parameter set is presented to the user (step ST13).
For example, if the selection number is “1”, the case record x2 having the highest unification result evaluation value is selected. If the selection number is “2”, the case record x2 having the highest unification result evaluation value and the case record are selected. x1 is selected.
It is possible to set so that the instruction of the number of selections is not accepted from the user. When the user's instructions are not accepted or when there is no instruction of the user, the default selection number (for example, recorded Half of the case records) may be set in advance.

Here, the processing content of the recommended parameter presentation unit 4 is shown for what is shown in FIG. 7, but it may be that shown in FIG.
That is, the recommended parameter presenting unit 4 selects a case code having a similarity (Euclidean distance) equal to or greater than a similarity reference value or a top N case codes having the highest similarity (step S11 in FIG. 7) ( Step ST21).
When the recommended parameter presenting unit 4 receives an instruction for the number of selections from the user, the recommended parameter presentation unit 4 selects case records for the number of selections in descending order of similarity (step ST22).

When the recommended parameter presenting unit 4 receives a weight vector instruction for the operation parameter set from the user, the recommended parameter presentation unit 4 adds the corresponding weight in the weight vector to the operation parameter set included in each selected case code.
The recommended parameter presentation unit 4 determines the operation parameter set having the maximum weight as the recommended operation parameter set, and presents the recommended operation parameter set to the user (step ST23).

Here, the processing content of the recommended parameter presentation unit 4 is shown in FIGS. 7 to 8, but it may be that shown in FIG. 9.
That is, when the recommended parameter presentation unit 4 receives an instruction of a weighting vector for the attribute group and the degree of similarity of the result evaluation from the user, for each of the case records x1, x2, and x3 recorded in the case storage unit 1, the recommended parameter presentation unit 4 Is calculated as a unified result evaluation value, and the unified result evaluation value is added to the case record (step ST31).
When the recommended parameter presenting unit 4 accepts an instruction of the number of selections from the user, as in the process of step ST13 in FIG. 7, the recommended parameter presentation unit 4 selects case records corresponding to the number of selections in descending order of the unification result evaluation value. Is presented to the user as a recommended operation parameter set (step ST32).

Here, the processing content of the recommended parameter presentation unit 4 is shown in FIGS. 7 to 9, but may be that shown in FIG. 10.
That is, when the recommended parameter presenting unit 4 receives an instruction for the number of selections from the user, the recommended parameter presentation unit 4 selects as many case records as the number of selections in descending order of the result evaluation value and the similarity (step ST41).
When the recommended parameter presentation unit 4 receives an instruction of the weighting vector for the operation parameter set from the user, the recommended parameter presentation unit 4 adds the corresponding weight in the weighting vector to the operation parameter set included in each selected case code, The operation parameter set having the maximum weight is determined as a recommended operation parameter set, and the recommended operation parameter set is presented to the user (step ST42).

As is apparent from the above, according to the first embodiment, the precondition space base group is obtained by using the preconditions and result evaluations in the plurality of case records recorded in the case storage unit 1. Precondition space construction section 2 for constructing a precondition space, and a precondition space section for converting recommended precondition coordinates, which are recommended preconditions, into coordinates on the precondition space constructed by the precondition space construction section 2 3, and the recommended parameter presentation unit 4 uses the coordinates on the precondition space in the plurality of case records recorded in the case storage unit 1 and the coordinates of the recommended preconditions converted by the precondition space unit 3. , Calculate the similarity between multiple case records and recommended prerequisites, and use that similarity to determine the recommended operational parameter set from the operational parameter sets in multiple case records Since it is configured, an effect that even when the number of attributes prerequisite of mass, to implement the dimensionality reduction in a short time, it is possible to determine the proper operating parameters set.
Also, according to the first embodiment, the precondition space is determined using the precondition attribute group and the result evaluation attribute group, so that the precondition space can be constructed at high speed, and the precondition space Since the base is related to the attribute of the result evaluation, an effect of improving the ability to explain the coordinates after the base conversion can be obtained.

  According to the first embodiment, the environment setting unit 11 that sets arbitrary preconditions in the simulation unit 13 and the operation parameter setting that sets the operation parameter set in the simulation unit 13 and instructs the simulation unit 13 to start operation. The result of expressing the surrounding state of the simulation unit 13 after the simulation unit 13 and the simulation unit 13 are started to operate, the preconditions set by the environment setting unit 11, and the operation set by the operation parameter setting unit 12 A case record including a parameter set is generated and a result evaluation unit 14 that outputs the case record to the case storage unit 1 in FIG. 1 is provided. There is an effect that it is possible to facilitate the accumulation of records.

  In addition, according to the first embodiment, the environment measurement unit 21 that measures environment information indicating the situation around the function unit 23 before the function unit 23 starts operation, and the operation parameter set are set in the function unit 23. In addition, the environment information measured by the environment measurement unit 21 is set in the function unit 23 as a precondition, the operation parameter setting unit 22 instructing the operation start of the function unit 23, and the function unit 23 after the operation starts. The result measurement unit 24 that measures the surroundings of the function unit 23, the measurement result of the result measurement unit 24 is evaluated, the result evaluation, and the preconditions and the operation parameter set set by the operation parameter setting unit 22 A case record is generated, and the result evaluation unit 25 that outputs the case record to the case storage unit 1 is provided, so that the actual device is linked to reflect the reality. An effect which is easy to use an example record.

  According to the first embodiment, a subset of the simulated case record shown in FIG. 2A or the actual case record shown in FIG. 2B is totaled to generate the case total record shown in FIG. Since it did in this way, the effect of being able to give the stable result by removing from the influence of disturbances, such as a weather, is considered, for example.

  According to the first embodiment, the result evaluation in a plurality of case records recorded in the case storage unit 1 is regarded as a binary attribute using a predetermined threshold, and the precondition in the case record The discriminant hyperplane that discriminates binary values and the margin of the discriminant hyperplane are calculated, and the discriminant normal vector consisting of the normal direction and the margin of the discriminant hyperplane is calculated, and the performance of the discriminant normal vector exceeds the specified performance. If so, since the precondition space is constructed by adding the discriminant normal vector to the base group of the precondition space, there is an effect of speeding up the construction of the precondition space. In addition, since the direction and unit of the base of the precondition space are related to the attribute of the result evaluation, there is an effect that the ability to explain the coordinates after the base conversion is improved.

BRIEF DESCRIPTION OF THE DRAWINGS It is a block diagram which shows the parameter determination assistance apparatus by Embodiment 1 of this invention. It is explanatory drawing which illustrates the kind of case record currently recorded on the case storage part 1. FIG. It is explanatory drawing which shows the example of a partial structure of the parameter determination assistance apparatus in the case of collecting result evaluation from the simulator of an object apparatus, and producing | generating a case record. It is explanatory drawing which shows the example of a partial structure of the parameter determination assistance apparatus in the case of collecting result evaluation from a target apparatus (real apparatus) and producing | generating a case record. It is a flowchart which shows the processing content of the precondition space construction part. It is explanatory drawing which shows the mechanism of the base generation of precondition space. It is a flowchart which shows the processing content of the recommendation parameter presentation part. It is a flowchart which shows the processing content of the recommendation parameter presentation part. It is a flowchart which shows the processing content of the recommendation parameter presentation part. It is a flowchart which shows the processing content of the recommendation parameter presentation part.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Case storage part (Case record recording means), 2 Prerequisite space construction part (Precondition space construction means), 3 Precondition space conversion part (Coordinate transformation means), 4 Recommended parameter presentation part (Recommended operation parameter determination means), DESCRIPTION OF SYMBOLS 11 Environment setting part (simulator starting means), 12 Operation parameter setting part (simulator starting means), 13 Simulating part, 14, 25 Result evaluation part (case record generating means), 21 Environment measuring part (target apparatus starting means) , 22 Operation parameter setting unit (target device activation unit), 23 function unit, 24 result measurement unit (case record generation unit).

Claims (5)

  Preconditions in which surrounding conditions of the target device before starting the operation are expressed, an operation parameter set including a plurality of operation parameters for setting the operation content of the target device, and the surroundings of the target device after starting the operation A case record recording means for recording a plurality of case records including a result evaluation in which the state of the condition is expressed, and a precondition and a result evaluation in the plurality of case records recorded in the case record recording means. , A precondition space construction means for constructing a precondition space that is a base group of the above precondition space, and a precondition space constructed by the above precondition space construction means with the coordinates of the recommended precondition that is a precondition to be recommended Coordinate conversion means for converting to upper coordinates, and coordinates in the precondition space in the plurality of case records recorded in the case record recording means Using the coordinates of the recommended preconditions converted by the coordinate conversion means, the similarity between the plurality of case records and the recommended preconditions is calculated, and the operation parameters in the plurality of case records are calculated using the similarities. A parameter determination support device comprising recommended operation parameter determination means for determining a recommended operation parameter set from the set.   An arbitrary precondition and an operation parameter set are set in the simulator of the target device, and a simulator starting means for operating the simulator and a state around the simulator after the simulator starts operating are expressed. And a case record generating means for collecting the above-mentioned preconditions and operation parameter sets and generating a case record including the preconditions, operation parameter sets and result evaluations, and collecting case records 2. The parameter determination support apparatus according to claim 1, wherein the means records the case record generated by the case record generation means.   Arbitrary operation parameter set is set in the target device, target device starting means for operating the target device, information indicating the state before and after the start of operation in the target device and the operation parameter set are collected, and the operation start Claim information is provided with information indicating the state before and after and a case record generating means for generating a case record from the operation parameter set, and the case record recording means records the case record generated by the case record generating means Item 1. The parameter determination assisting device according to Item 1.   4. The parameter determination assisting apparatus according to claim 2, wherein the case record recording means records a subset of the case records generated by the case record generating means.   The precondition space construction means regards the result evaluation in the plurality of case records recorded in the case record recording means as a binary attribute using a predetermined threshold, and for each precondition in the case record, the 2 The discriminant hyperplane for discriminating the value and the margin of the discriminant hyperplane are calculated, the discriminant normal vector composed of the normal direction of the discriminant hyperplane and the margin is calculated, and the performance of the discriminant normal vector is the specified performance. The parameter determination support apparatus according to claim 1, wherein the precondition space is constructed by adding the discriminant normal vector to the base group of the precondition space.

Images