WO2013179577A1

WO2013179577A1 - Multipurpose optimization system, information analysis system, multipurpose optimization method, and program

Info

Publication number: WO2013179577A1
Application number: PCT/JP2013/002898
Authority: WO
Inventors: 白木　孝; 友人安藤; 圭介梅津
Original assignee: 日本電気株式会社
Priority date: 2012-05-31
Filing date: 2013-04-30
Publication date: 2013-12-05

Abstract

A multipurpose optimization system that displays multiple evaluation items and multiple solution results for multipurpose optimization, wherein the multipurpose optimization system is equipped with a display method determination unit. When solutions are displayed with each of multiple evaluation items associated with one axis, this display method determination unit reflects the importance of or the restriction conditions for each evaluation item, that is, the parameters used in calculating a solution, in the display of the axes in the graph with which the solutions are displayed, or in the display of the background of the axes, or in the display of each solution.

Description

Multi-objective optimization system, information analysis system, multi-objective optimization method and program

The present invention relates to a multipurpose optimization system, an information analysis system, a multipurpose optimization method, and a multipurpose optimization program for displaying a plurality of solution results together with a plurality of evaluation items and performing multipurpose optimization and information analysis.

OR (Operations Research) used for making business decisions displays analysis information for optimization. Many of these display methods enumerate the best solution and the elements that yield it for one objective function. This is a method based on the premise that parameters of an objective function in optimization are determined in advance.

However, the objective function parameters for multi-objective optimization are not always set correctly with respect to the criteria of the judge. In order to correct such inappropriate parameter settings, there is a method in which a candidate selects a solution while confirming the trade-off of multiple evaluation items by presenting multiple candidate solutions during optimization. May be. There are also tools for this.

One of the analysis information display methods for such multi-objective optimization is Parallel Coordinate Plot (parallel coordinate display) described in Non-Patent Document 1. The most common example of parallel coordinate display is a method of displaying a plurality of evaluation items on a plurality of parallel vertical axes and expressing the numerical values of the evaluation items on each axis. The solutions in the objective function are then displayed as a set of lines that cross their axes.

In addition, as an example of a method for solving inadequate parameter setting, Patent Document 1 is called semi-automatic in which a judge makes a solution by making a correction based on a solution presented by optimization. A method is described.

JP 2009-181195 A

However, as described in Non-Patent Document 1, a method of simply assigning a plurality of evaluation items to a plurality of parallel vertical axes and expressing a numerical value of each evaluation item on each axis is obtained by dividing each evaluation item. Although it is easy to understand because it is displayed, the degree of influence on the objective function of each axis is difficult to understand. For this reason, there are many cases where it is not sufficient to compare the solutions sensuously. Further, there is a problem that the parameters of the objective function for optimization cannot be easily corrected if the algorithm and the degree of influence of various evaluation items are not understood after understanding the algorithm of the objective function.

For example, if you want to optimize a service provision system, the evaluation items may include cost, short-term sales expectations, long-term sales expectations, customer satisfaction, service provider satisfaction, etc. . When the weight to be assigned to each of the plurality of evaluation items is set as a parameter, the optimum value of the weight of such an evaluation item may vary depending on the evaluator and cannot be easily set. Therefore, trial and error is often performed to interactively adjust each weight after comparing candidate solutions for improving each evaluation item. However, in order for a judge to adjust such parameters, it is difficult to adjust the parameters because not only the business contents of the system to be optimized but also the tools and algorithms must be familiar.

In addition, in order to avoid insufficient parameter setting, each element of the solution (a unit for assigning which work to whom is taken as an example of the scheduling problem) as described in Patent Document 1 is directly improved. Learning with the results is one way. However, in the method described in Patent Document 1, although each element of the solution can be improved, each evaluation item (scheduling problem is taken as an example, cost, short-term sales expectation value, long-term sales (Expected value of revenue, customer satisfaction, service provider satisfaction, etc.) In order to understand the optimal solution and the change in the value of each evaluation item due to changes in individual parameters such as weight, Need knowledge of tools and algorithms. Therefore, even if it is desired to adjust each parameter interactively and perform trial and error, the adjustment is difficult unless the judge is familiar with the contents of the system work and is familiar with the tools and algorithms.

In addition, the method described in Patent Document 1 has a problem that the burden on a judge who sets each parameter is high. When the number of evaluation items is large, the operation load related to the parameter change is further increased.

Suppose, for example, that the judge is not a person who is knowledgeably suitable for the work of correcting each element. In this case, it is difficult for the determiner to operate with objective-based indexes such as cost and customer satisfaction instead of each element, and to compare the optimal solutions of the respective cases. Furthermore, such a judge cannot grasp the achievement status of the purpose of the solution, nor can it perceive the change in the optimum solution due to the change in the weight of each evaluation item. This is the first problem.

The second problem is that if the number of elements of the solution to be operated for improvement is large, the work load is large. When the load is large, for example, the judge wants to compare the solutions that arise when the parameters of each evaluation item are modified in various ways. May be limited.

Accordingly, the present invention provides a multi-objective optimization system, an information analysis system, a multi-objective optimization method, and a multi-objective optimization that can easily display the values of a plurality of evaluation items that become a plurality of options in order to obtain a solution desired by a judge The purpose is to provide a computerized program. To display the values of multiple evaluation items in an easy-to-understand manner, more specifically, when displaying each solution that combines the values of multiple evaluation items, visually confirm the achievement status of each solution for each purpose. It means to make it easier to analyze the superiority or inferiority of each solution.

In addition, the present invention is a multi-objective optimization system capable of operating parameters for optimization with an objective-based index and capturing changes in the achievement status of the objectives of each solution due to changes in parameters, An object is to provide an information analysis system, a multi-objective optimization method, and a multi-objective optimization program.

A multi-objective optimization system according to the present invention is a multi-objective optimization system that displays a plurality of solution results together with a plurality of evaluation items in multi-objective optimization, and displays a solution by associating each of the plurality of evaluation items with one axis. At the time, the display method decision means to reflect the importance or constraint condition of each evaluation item, which is a parameter used for solution calculation, in the axis, background of the graph or display of each solution in the graph for displaying the solution It is characterized by.

An information analysis system according to the present invention is an information analysis system that analyzes information by displaying a plurality of solution results together with a plurality of evaluation items, and each of the plurality of evaluation items is associated with one axis to solve a solution. When displaying, the display method decision means that reflects the importance or constraint condition of each evaluation item, which is the parameter used for solution calculation, in the axis, background of the graph or display of each solution in the graph for displaying the solution It is characterized by that.

The multi-objective optimization method according to the present invention is a multi-objective optimization method for performing multi-objective optimization by displaying a plurality of solution results together with a plurality of evaluation items, each of which corresponds to one axis. When displaying a solution, the axis used to display the solution, the background of the axis, or the display of each solution is the parameter used for calculating the solution and the importance or constraint condition of each evaluation item is reflected. To do.

The multi-objective optimization program according to the present invention is a multi-objective optimization program that performs multi-objective optimization by displaying a plurality of solution results together with a plurality of evaluation items, and each of the plurality of evaluation items is arranged on one axis. Processing to reflect the importance or constraints of each evaluation item on the axis, background of the axis, or display of each solution, and the parameters used for calculating the solution when displaying the solution in association with each other Is executed.

The first effect is that even a judge who has little knowledge about the work of correcting each element can easily understand the features of each solution and try the solutions by balancing various objective-based indicators.

The second effect can reduce the amount of work to be operated for improvement. This is because the balance of the goal-based index can be adjusted by the user operation without correcting each element of the solution.

It is a block diagram which shows the structural example of the information analysis system of 1st Embodiment. It is a block diagram which shows the structural example of a display method determination part. It is a flowchart which shows an example of operation | movement of the information analysis system of 1st Embodiment. It is explanatory drawing which shows the example which reflected the improvement direction of each evaluation item on the direction of the axis | shaft. It is explanatory drawing which shows the example which reflected the weight of each evaluation item on the length of the axis | shaft. It is explanatory drawing which shows the example which reflected the weight of each evaluation item on the space | interval of an axis | shaft. It is explanatory drawing which shows the example which folds and displays an axis | shaft according to the optimal value of each evaluation item. It is explanatory drawing which shows the other example which folds and displays an axis | shaft according to the optimal value of each evaluation item. It is explanatory drawing which shows the other example which folds and displays an axis | shaft according to the optimal value of each evaluation item. It is explanatory drawing which shows the example which changes the density of the color of the drawn line of each solution according to the fitness with respect to an objective function. It is explanatory drawing which shows the example which reflected the range of absolute restrictions in each evaluation item, and the condition adaptability in the color range and color density of the background color of an axis | shaft. It is explanatory drawing which shows the other example of the display which reflected the absolute constraint condition in each evaluation item on the background of an axis | shaft. It is explanatory drawing which shows the other example of the display which reflected the absolute constraint condition in each evaluation item on the background of an axis | shaft. It is explanatory drawing which shows the other example of the display which reflected the absolute constraint condition in each evaluation item on the background of an axis | shaft. It is explanatory drawing which shows the other example of a solution display. It is explanatory drawing which shows the other example of a solution display. It is explanatory drawing which shows the example which displays the upper limit of a constraint condition of each evaluation item, a minimum, and an optimal value on an axis | shaft. It is explanatory drawing which shows the example reflected and reflected on the width | variety of the background figure of an axis | shaft in the range of the absolute constraint conditions in each evaluation item, and condition adaptability. It is a block diagram which shows the structural example of the information analysis system of 2nd Embodiment. It is a flowchart which shows an example of operation | movement of the information analysis system of 2nd Embodiment. It is explanatory drawing which shows the example of parameter adjustment. It is explanatory drawing which shows the example of parameter adjustment. It is explanatory drawing which shows the example of parameter adjustment. It is explanatory drawing which shows the example of parameter adjustment. It is explanatory drawing which shows the example of parameter adjustment. It is a block diagram which shows the outline | summary of this invention. It is a block diagram which shows the other structural example of the multi-objective optimization system of this invention.

Embodiment 1. FIG.
Hereinafter, embodiments of the present invention will be described with reference to the drawings. The information analysis system of the first embodiment accepts an objective-based index such as cost and customer satisfaction from a user as a parameter reflected in an objective function. For example, the information analysis system may accept the importance for each index as a parameter reflected in the objective function, and set it as the weight of each evaluation item in the objective function. Note that the information analysis system may accept a constraint condition as a parameter in addition to the importance level and reflect it in the objective function. The information analysis system of the present embodiment reflects the received parameter value not only in the objective function but also in the optimization calculation solution display. That is, the information analysis system of the present embodiment also visualizes the contents of parameters in the goal-based index when displaying the solution.

The information analysis system may reflect, for example, the parameters used for the optimization calculation in the length of the axis line, the thickness of the axis line, the interval between the axes, and the direction of the axis in the parallel coordinate display. In addition, the information analysis system may reflect the parameters used for the optimization calculation by displaying the axes in a folded manner or by displaying the optimum values of the respective axes. The information analysis system also reflects the parameters used for the optimization calculation in the color of each solution in the parallel coordinate display, the line type, etc., the background color of the axis and its density, and displays the upper and lower limits on the axis. Alternatively, the degree of fitness of each axis with respect to the objective function may be reflected by displaying the figure in accordance with the axis.

More specifically, for example, when displaying each solution, the information analysis system displays the importance of each index, the length of the axis line of the evaluation item related to the index, the thickness of the axis line, the axis It may be reflected in the interval. Moreover, the information analysis system may reflect the directionality (improvement direction) of the values for the optimization of each evaluation item in the direction of the axis, for example. In addition, the information analysis system may reflect, for example, the optimum value or constraint condition of each evaluation item in the axis wrap display.

Also, the information analysis system may reflect, for example, the degree of suitability of each value in each evaluation item to the objective function in the axis background display. When reflecting in the axis background display, for example, the information analysis system may generate a graphic reflecting the degree of fitness for the objective function on each axis in the width with respect to the axis, and display the graphic on the axis background. . In addition, the information analysis system may display the degree of conformity to the objective function on each axis as a background of the axis by reflecting it in the color and its density. Note that the method of reflecting the fitness to the objective function on each axis is not limited to the method of displaying the axis background. For example, the information analysis system may display a line that satisfies the constraint condition in a polygonal line, or reflect the degree of fitness in the display color of points in the evaluation item in each solution.

In addition, the information analysis system displays, for example, the upper and lower limits of the constraint conditions and the optimal values of each axis in the objective function on each axis, colors each solution according to the fitness in the objective function, The sheaths may be displayed with different line types.

Hereinafter, the present embodiment will be described more specifically with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration example of the information analysis system according to the first embodiment. The information analysis system shown in FIG. 1 includes a user terminal 1 and a multipurpose optimization system 2.

The user terminal 1 includes an operation unit 11 and a display unit 12.

The operation unit 11 receives an input operation of a parameter reflected in the objective function for each purpose-based index from the user, and inputs input data by the operation to the GUI unit 211 of the multi-objective optimization system 2 described later. The operation unit 11 is realized by a mouse or a keyboard, for example. The user terminal 1 includes not only an input interface that accepts a direct user operation such as a mouse and a keyboard, but also a file system that inputs data via a file, a network card that inputs data via a network, and the like. Also good.

When the display unit 12 receives data for displaying a solution from the GUI unit 211 of the multi-purpose optimization system 2, the display unit 12 is various output interfaces that output an image for the user based on the data. The display unit 12 is realized by a display device, for example.

The multipurpose optimization system 2 includes a control unit 21 and a storage unit 22. The control unit 21 includes a GUI unit 211, a parameter setting unit 214, a solution calculation unit 215, and a display method determination unit 216. The storage unit 22 includes a program storage unit 221 and a data storage unit 222.

The GUI unit 211 receives the content of the parameter input by the user from the operation unit 11 of the user terminal 1 and outputs it to the parameter setting unit 211. In addition, when the GUI unit 211 receives setting information regarding various displays and a solution set from the display method determination unit 216, the GUI unit 211 generates display data based on the setting information and outputs the display data to the display unit 12.

When the parameter setting unit 214 receives the content of the parameter input by the user from the GUI unit 211, the parameter setting unit 214 reflects it in the objective function and outputs it to the solution calculation unit 215. The parameter setting unit 214 can directly use the parameters input for each objective-based index as parameters for the evaluation items in the objective function, but if necessary, the parameter values actually used in the objective function And convert it to the objective function.

When the parameter setting unit 214 reflects the content of the parameter input by the user in the objective function, the solution calculation unit 215 performs optimization calculation using an optimization program described later, and displays the result as a display method determination unit 216. Output to.

Upon receiving the optimization calculation result from the solution calculation unit 215, the display method determination unit 216 determines a display method for displaying the received optimization calculation result, and various settings for displaying the solution based on the determined display method The value is output to the GUI unit 211 to cause the GUI unit 211 to display a solution.

FIG. 2 is a block diagram illustrating a more detailed configuration example of the display method determination unit 216. As shown in FIG. 2, the display method determination unit 216 of this embodiment includes an axis display method determination unit 2161, an axis background display method determination unit 2162, and a solution display method determination unit 2163. The display method determination unit 216 determines various setting values that determine the display method of the solution in the graph for performing the solution display via these processing units.

The axis display method determination unit 2161 determines a display method related to the axis among the display methods of the solution in the graph, based on the values of parameters (including constraint conditions) used for the solution calculation. Here, for example, the arrangement of the axis, the length of the line of each axis, the thickness of the line of each axis, the interval between the axes, the direction of each axis, the presence or absence of the folding display of each axis, the folding point, the way of the scale, etc. Is determined.

The axis background display method determination unit 2162 determines a display method particularly regarding the axis background among the solution display methods in the graph based on the values of parameters (including constraints) used for the solution calculation. Here, for example, the presence / absence of background display, the type of background display (graphic, broken line, band, etc.), the shape of the figure displayed on the background of each axis, the installation position, the color, the color of the broken line displayed on the background of each axis, The line type, thickness, installation position of the band displayed on the background of each axis, color, etc. are determined. The axis background display includes display of information that is a point on each axis, and the axis background display method determination unit 2162 displays, for example, the upper limit and lower limit of the constraint condition, whether or not the optimum value is displayed, and the like. It also includes setting the mark and the installation position.

The solution display method determination unit 2163 determines, in particular, the display method of each solution among the solution display methods in the graph, based on the values of parameters (including constraint conditions) used for the solution calculation. Here, for example, the display color of each solution, its density, line type, line thickness, etc. are determined.

The GUI unit 211 draws a graph for displaying the solution on the screen in accordance with the solution display method determined by the display method determining unit 216 (more specifically, various setting values related to the graph for displaying the solution). The GUI unit 211 may generate image data including a graph depicting each solution on the display unit 12 of the user terminal 1 using, for example, a GUI function provided by the OS or a graph drawing application.

The program storage unit 221 includes an optimization program storage unit 2212. The optimization program storage unit 2212 is a storage device that stores an optimization program. Here, the optimization program is a program that actually performs optimization calculation by calling from the solution calculation unit 215. More specifically, the optimization program acquires data necessary for calculation of a solution from input data stored in a basic data storage unit 2221 described later in response to the call. In addition, the optimization program obtains a provisional solution set stored in the calculation result storage unit 2222, performs optimization calculation based on the temporary solution set, and outputs the result. Actually, the control unit 21 that has read the optimization program executes such processing according to the read optimization program. For example, the program storage unit 221 may store a plurality of optimization programs having different optimization calculation methods. The optimization program is converted into an executable format such as DLL and stored in the program storage unit 221.

The data storage unit 222 includes a basic data storage unit 2221 and a calculation result storage unit 2222.

The basic data storage unit 2221 is intended for the multi-objective optimization system 2 to solve, for example, a worker list, a task list, a worker's vacation schedule, and information on tasks that each worker can engage in a schedule problem. It is a storage device that stores basic data of the problem to be performed.

The calculation result storage unit 2222 is a storage device that stores all or part of the solutions calculated by the optimization program. When a part of the solution is stored, for example, a criterion such as storing only a solution called a Pareto optimal solution or an upper limit of the number of stored solutions may be provided.

In this embodiment, the GUI unit 211, the parameter setting unit 214, the solution calculation unit 215, and the display method determination unit 216 are realized by an information processing apparatus such as a CPU that operates according to a program. The program storage unit 221 and the data storage unit 222 are realized by a storage device.

Next, the overall operation of this embodiment will be described. FIG. 3 is a flowchart showing an example of the operation of the information analysis system of this embodiment. As a premise of the operation shown in FIG. 3, in the information analysis system of this embodiment, it is assumed that an objective function corresponding to an object to be optimized is set in advance. The setting of the objective function may be performed, for example, when the operator inputs the objective function using a format formulated for an optimization problem, such as a modeling language called AMPL. The objective function may be stored in a predetermined storage area in advance.

In the example shown in FIG. 3, the information analysis system first receives from the user input of parameters that are reflected in the objective function such as weights and constraint conditions for the objective-based index (step S11). For example, the GUI unit 211 may generate a parameter input screen that receives contents such as importance, upper limit value, and lower limit value as input parameters for each purpose-based index. Then, the GUI unit 211 may display the parameter input screen on the display unit 12 of the user terminal 1 and accept input of parameters via a user operation on the parameter input screen.

Hereafter, the problem of optimizing the schedule and arrangement of salesclerks at a certain store will be described as an example. In order to solve the optimization problem, it is assumed here that an objective function is set to find a solution that maximizes the effect when cost, long-term sales expectations, and customer satisfaction are used as evaluation items. To do. In the case of this example, the GUI unit 211, for example, according to a user operation, includes a constraint parameter and a weight parameter that represents the importance of three objective-based indicators such as cost, long-term sales expectation, and customer satisfaction. Input may be accepted. Here, in order to simplify the explanation, it is assumed that the objective function is linear and the optimization calculation is performed using the following equation (1).

maxmize (w ₁ C + w ₂ L + w ₃ S) (1)

In equation (1), C is the cost, L is the expected value of long-term sales, and S is the customer satisfaction. Further, w ₁ , w ₂ , and w ₃ are weights of the evaluation items (cost C, long-term sales expectation value L, customer satisfaction S) in the objective function. Since the lower cost is better, the weight w ₁ for the cost C is a negative value. Further, the weights w ₂ and w ₃ for the expected value L of long-term sales and customer satisfaction S are positive values. As an example, assume that w ₁ = −100, w ₂ = 70, and w ₃ = 50 are input by the operation unit 11. Here, it is assumed that C, L, and S do not care about units without losing generality, and that w ₁ , w ₂ , and w ₃ are the same as the importance of each index.

When the GUI unit 211 receives weight parameters w ₁ = −100, w ₂ = 70, and w ₃ = 50 input from the user via the operation unit 11, the GUI unit 211 outputs them to the parameter setting unit 214.

When the parameter setting unit 214 receives the contents of the parameters input by the user from the GUI unit 211, that is, the weight parameters w ₁ = −100, w ₂ = 70, and w ₃ = 50, the values are reflected in the objective function (step S12). In this example, the received weight parameter is set as the weight of each evaluation item in the objective function as it is.

When the contents of parameters (including constraint conditions) for the objective-based index set by the user are reflected in the objective function, the solution calculation unit 215 performs optimization calculation using the updated objective function, and displays the display target. Is obtained (step S13). In this example, the solution calculation unit 215 designates the updated objective function or various parameters set for the objective function and calls the optimization program to cause the optimization program to perform optimization calculation. A solution to be displayed is obtained as a return value. The optimization program includes, for example, weight parameters w ₁ = −100, w ₂ = 70, w ₃ = 50 in the objective function, a list of salesclerks stored in the basic data storage unit 2221, a list of roles, Using hourly wages, etc., salesclerk scheduling and the resulting costs, long-term sales expectations and customer satisfaction may be estimated. Here, for example, the optimization program may be a module that executes an algorithm that searches all possible combinations of schedules. The search algorithm used in the solution calculation method is not limited to the full search algorithm, and may be an algorithm that does not require an exact solution such as an approximate solution in order to shorten the calculation time. Furthermore, the solution calculation unit 215 may perform a filter operation such as using a Pareto optimal solution or setting an upper limit on the number of solutions in the optimization calculation.

Note that the solution calculation unit 215 can directly perform optimization calculation to calculate the solution without using the optimization program. When the solution is calculated, the solution calculation unit 215 outputs the calculation result of the solution and the parameters used for calculation of the solution to the display method determination unit 216, and requests setting of a graph for displaying the solution.

The display method determination unit 216 determines a solution display method based on a predetermined display rule or a display rule instructed by the user in response to a request from the solution calculation unit 215 (Steps S14-1 to S14-). 3). In the display method determining unit 216, as an example, first, the axis display method determining unit 2161 performs display setting regarding the axis among the setting items of the graph for performing the solution display (step S14-1). For example, the axis display method determination unit 2161 determines the display method related to the axis so that the contents of the parameters set for each evaluation item used for calculating the solution are visualized. More specifically, the axis display method determination unit 2161 determines the arrangement order of each axis, the arrangement interval, the line length, and the line length among the setting items of the graph according to the importance level and the constraint condition of each evaluation item. Values are set for setting items related to axes such as thickness, orientation, presence / absence of wrapping display, wrapping point / scale range, and interval.

Next, the axis background display method determination unit 2162 performs display settings related to the axis background (step S14-2). For example, the axis background display method determination unit 2162 determines a display method related to the axis background so that the contents of the parameters set for each evaluation item used for calculating the solution are visualized. More specifically, the axis background display method determination unit 2162 determines whether or not to display the background of each axis among the setting items of the graph, the type of background display, and the background display according to the importance level and the constraint conditions of each evaluation item. Axis such as figure shape, installation position, color determination or line color, line type, thickness, upper / lower limit of constraint condition, presence / absence of optimum value, display mark, installation position Set a value for the setting item related to the background.

Finally, the solution display method determination unit 2163 performs display settings related to the display of each solution (step S14-3). For example, the solution display method determination unit 2163 determines the display method of each solution so that the contents of the parameters set for each evaluation item used for calculating the solution are visualized. More specifically, the solution display method determination unit 2163 displays the color and density of lines and bars for displaying each solution among the setting items of the graph according to the fitness of the objective function of each solution, Set a value to the setting item related to the display of each solution such as the thickness of the line. Note that the axis display method determination unit 2161, the axis background display method determination unit 2162, and the solution display method determination unit 2163 may not be set so that the parameter contents are visualized. For example, when setting is made such that the parameter contents are visualized by any of the processing units, the other processing units may perform normal display.

When the determination process of the display method of the solution by the display method determination unit 216 is completed, the GUI unit 211 draws a graph for displaying the solution according to the contents of the setting items of the graph set in the determination process. Then, the GUI unit 211 outputs a graph display screen in which a graph is drawn on the display unit 12 of the user terminal 1 (step S15).

Here, when the user gives an instruction to change the parameter and redisplay (step S16), the GUI unit 211 switches the screen from the solution graph display screen to the parameter input screen, and again from the user the parameter setting screen. You may make it wait for the input of a value (it returns to step S11). At this time, the GUI unit 211 may retain information on the graphs displayed so far and display the information side by side with the solution display using new parameters.

Hereinafter, the display method determination process will be described using a specific example.

Display method example 1.
In this example, the improvement directions are aligned on each axis. The axis display method determination unit 2161 of the display method determination unit 216 determines whether or not the axis is inverted from the weight parameters (for example, w ₁ = −100, w ₂ = 70, w ₃ = 50) for each evaluation item in the objective function. decide. For example, the axis display method determination unit 2161 may determine whether to reverse the axis depending on whether the variable of the axis is a minimizing element or a maximizing element, assuming that the upper part of the axis is a more preferable position. In this example, each axis is associated with each evaluation item in the objective function. Therefore, the axis display method determination unit 2161 determines not to invert if the value of the weight parameter of the evaluation item corresponding to each axis is positive, and to invert if the value is negative. Specifically, since w ₁ is negative, the axis display method determination unit 2161 determines to invert the direction of the axis associated with the parameter w ₁ .

Information on the orientations of these axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and a graph display screen is displayed on the display unit 12. Thus, the user views the graph of FIG. Can do. FIG. 4 is an explanatory diagram showing an example of solution display according to this example. FIG. 4A shows an example of a solution display when the direction of each axis is set to an increasing direction regardless of the improvement direction of the evaluation item corresponding to the axis. FIG. 4B is an example of a solution display when the direction of each axis is reversed according to the improvement direction of the evaluation item corresponding to the axis.

In the example of FIG. 4A in which the direction of each axis is simply determined by the increasing direction of the value, it is difficult to determine which one of the solution a1, the solution a2, and the solution a3 is the optimal solution. On the other hand, in the example of FIG. 4B, the solution a1 and the solution a3 positioned higher on each axis are better than the solution a2 positioned lower than that. The solution is easy to understand. That is, according to this example, by making the improvement direction of each axis the same, it is possible to improve the ability to visually grasp the superiority or inferiority of the solution, and to improve the convincingness of the obtained solution.

Display method example 2.
In this example, the axes are displayed at different scales depending on the importance of the evaluation item. The axis display method determination unit 2161 of the display method determination unit 216 calculates the axis length (more specifically, from the weight parameters (for example, w ₁ = −100, w ₂ = 70, w ₃ = 50) for each evaluation item in the objective function. Specifically, the length of the axis line) is determined. For example, the axis display method determination unit 2161 may set the length of each axis i to a value proportional to the absolute value | w _i | of the weight parameter w _i of the evaluation item to which the axis corresponds.

Information on the lengths of these axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and a graph display screen is displayed on the display unit 12, so that the user sees the graph of FIG. be able to. FIG. 5 is an explanatory diagram illustrating an example of solution display according to this example. FIG. 5A is an example of a solution display when the length and scale of each axis are the same regardless of the importance of the evaluation item to which the axis corresponds. FIG. 5B shows an example of a solution display when the length of each axis is proportional to the importance of the evaluation item to which the axis corresponds.

In the example of FIG. 5A in which the axis length and the scale are the same for each axis, the importance of the axis is not reflected in the display, so which is the better solution between the solution a1 and the solution a3? May be difficult to understand. On the other hand, in the example of FIG. 5B, the length of the axis and the interval between the scales (the axis for each evaluation item so that the difference in the goal-based index in the important axis looks larger than the difference in the other axes. Is reflected in the scale. Therefore, it is easy to understand visually that the solution a1 is a better solution in terms of the parameters. That is, according to the present example, the GUI unit 211 displays the difference between the solutions in an emphasized manner according to the importance of the evaluation item corresponding to the goal-based index. The difference becomes clearer when the size is grasped by the integration of the solution display (area below the line segment). Therefore, the ability to visually grasp the superiority or inferiority of the solution can be improved, and the convincing property of the obtained solution can be improved.

Display method example 3.
This example is an example in which the axes are displayed with different intervals according to the importance of the evaluation item. The axis display method determination unit 2161 of the display method determination unit 216 uses parameters indicating the importance of each evaluation item in the objective function (in this example, weight parameters w ₁ = −100, w ₂ = 70, w ₃ = 50). Determine the axis spacing. For example, the axis display method determination unit 2161 may set the interval between the axes i and i + 1 to a value proportional to the value of | w _i | + | w _{i + 1} | using these two parameters. In addition, the axis display method determination unit 2161 adds parallel lines with widths proportional to | w _{i_min} | and | w _{i_max} | for the left and right end axes, respectively. Here, the axis display method determination unit 2161 adds parallel lines having a width proportional to | w ₁ | and | w ₃ |, respectively.

The information about the distance between the axes and the side lines is transmitted to the GUI unit 211, a graph based on the information is drawn, and the graph display screen is displayed on the display unit 12, so that the user can display the graph of FIG. Can see. FIG. 6 is an explanatory diagram illustrating an example of solution display according to this example. FIG. 6A shows an example of a solution display when the distance between the axes is the same regardless of the importance of the evaluation item corresponding to the axis. FIG. 6B shows an example of a solution display when the interval between the axes is proportional to the importance of the evaluation item corresponding to the axis and a side line is added.

In the example of FIG. 6A in which the axis spacing is the same for each axis, since the importance of the axis is not reflected in the display, it is difficult to determine which is the better solution between the solution a1 and the solution a3. There is a case. On the other hand, in the example of FIG. 6B, the difference in the goal-based index in the important axis is reflected in the axis interval so that it looks larger than the difference in the other axes. Furthermore, in the example of FIG. 6B, by adding a side line, it is easy to visually understand that the solution a1 is a better solution in terms of the parameters. In other words, according to the present example, the GUI unit 211 displays the difference in solution according to the importance of the axis in an emphasized manner, so that the magnitude of the degree that fits the objective function of each solution can be integrated ( The difference becomes more clear when grasped by the area below the line segment. Therefore, the ability to visually grasp the superiority or inferiority of the solution can be improved, and the convincingness of the obtained solution can be improved.

Display method example 4
In this example, when the optimum value of the evaluation item is an intermediate value, the axis is folded and displayed according to the optimum value. In this example, a case where the objective function is an optimization problem including cost, the degree of scheduled operation, and customer satisfaction will be described as an example of a problem for optimizing the schedule of a driver of a certain bus company. Here, it is assumed that the degree of scheduled operation and customer satisfaction are estimated according to the driver's performance at each time zone and route, and the manager has to consider the change in cost due to different hourly wages depending on the driver. The aim is to optimize the driver's schedule.

In this example, the operation unit 11 inputs weighting parameters representing the importance of each index with respect to the cost that is the purpose-based index, the degree of scheduled operation, and the customer satisfaction according to the user operation according to the user operation. Here, for simplicity, it is assumed that the calculation is linear, and the optimization calculation is performed using the following equation (2).

maxmize (w ₁ C + w ₂ T ′ + w ₃ S) (2)
However, T ′ = − max {T, −6T}

In Expression (2), C is a cost, T is a delay time with respect to a fixed time, and S is customer satisfaction. Further, w ₁ , w ₂ , and w ₃ are weights for each evaluation item (cost C, scheduled operation degree T ′, customer satisfaction S) in the objective function. In this example, the value after penalty of the delay time T, which is one of the evaluation items, is used for the scheduled operation degree T ′, which is a goal-based index. Here, the scheduled operation degree T ′, which is a goal-based index, is a constraint that T ′ = − max {T, −6T} (or T ′ = min {−T, 6T} can also be set) due to a penalty. It can be said that it is the delay time T on which the condition is imposed. Note that max {a, b} is a function that returns the larger value of a and b, and min {a, b} is a function that returns the smaller value of a and b. Thus, when the penalty is imposed on the evaluation item, the objective-based index and the evaluation item in the objective function may be different variables. In the following description, it is assumed that a constraint condition of T ′ = − max {T, −6T} is imposed as a delay time penalty.

The bus should have a small delay time, but I want to avoid leaving earlier than planned. For this reason, a penalty of 6 times plus is imposed on the degree of scheduled operation T ′ when the value is negative with respect to the delay time T. And since it is better that the cost and the degree of scheduled operation are small, the weights w ₁ and w ₂ for these are negative values. Further, the weight w ₃ for customer satisfaction is a positive value.

The GUI unit 211 receives a weight parameter indicating the importance of the goal-based index input by the operation unit 11 through the GUI. In this example, the GUI unit 211 obtains weight parameters for the cost C, the scheduled operation degree T ′, and the customer satisfaction degree S, which are goal-based indexes. Thereafter, as in the other examples, parameter setting processing, optimization calculation execution processing, and display method determination processing are performed. In this example, the received weight parameter is used after being converted into the weight parameter of the evaluation item in the objective function. For example, the weight parameter for the scheduled operation degree T ′ may be reflected in the objective function as a weight parameter for the evaluation item −max {T, −6T} after the penalty. Other indexes are reflected as they are as weight parameters of evaluation items.

In the execution process of the optimization calculation, for example, the weight parameter in the objective function, the constraint condition, the list of drivers stored in the basic data storage unit 2221, the list of buses to be ridden, the hourly rate of each driver are used. It is done. In this example, the constraint condition includes a penalty content for the delay time T, an upper limit value, a lower limit value, and the like. Using these pieces of information, the scheduling of the driver, the resulting cost, the degree of scheduled operation, and customer satisfaction are estimated. By using a branch and bound method, a result similar to an algorithm that traverses all possible schedule combinations is obtained. Here, as an example, the solution is calculated using the branch and bound method.

The axis display method determination unit 2161 of the display method determination unit 216 determines the presence / absence of the axis folding display and the folding point from the parameters used for calculating the solution and set for each evaluation item. The axis display method determination unit 2161 first determines where the optimum value of each evaluation item is based on the weight parameter w _i (i = 1, 2, 3) for each evaluation item in the objective function and the constraint condition. To do. Here, the axis display method determination unit 2161 calculates that the optimal value of cost C is infinitesimal, the optimal value of delay time T is zero, and the optimal value of customer satisfaction S is infinite. Then, based on the calculated optimum value, the axis display method determination unit 2161 determines that the axis corresponding to the delay time T is displayed to be folded back at zero. In addition, when the axis display method determination unit 2161 calculates that the penalty of the negative value is six times larger than the positive value at the delay time T, the axis corresponding to the delay time T is reduced in the scale of the axis at the negative value. It is determined to be 6 times the positive value.

Information related to the display of folding of the axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and a graph display screen is displayed on the display unit 12, so that the user sees the graph of FIG. 7B. be able to. FIG. 7 is an explanatory diagram illustrating an example of solution display according to this example. FIG. 7A shows an example of the solution display when each axis is not displayed. FIG. 7B is an example of a solution display when the display of each axis is displayed according to the optimum value of the evaluation item corresponding to the axis. In the example of FIG. 7B, on the axis corresponding to the scheduled operation evaluation item, the value positioned at the top is displayed as zero, and 10 points before the return are 60 points after the return. The scale is set and displayed so that the same scale interval is obtained.

Suppose that the evaluation items in the objective function, which is a real variable used to calculate the optimal solution, do not correspond one-to-one with the objective-based index, which is a weighting unit. In this case, in the example of FIG. 7A in which the axis is not folded back, the optimum value of the delay time, which is the evaluation item for the on-time operation index, is difficult to understand, so which one is the better solution? Confusing. On the other hand, in the example of FIG. 7B, the display is folded according to the optimum value of the axis, and the scale interval is changed before and after the folding in accordance with the penalty being different before and after the optimum value. Therefore, the difference between the indexes on the axes becomes easy to understand, and it becomes easy to visually understand that the solution a1 located on the upper side of each axis is a better solution in terms of the parameters. Yes. In other words, according to this example, the axis is folded and displayed in accordance with the optimal value of the axis, so that the degree of the degree of adaptability to the objective function of each solution is grasped by the integration of the solution display (area below the line segment). In some cases, the difference becomes clearer. Therefore, the ability to visually grasp the superiority or inferiority of the solution can be improved, and the convincing property of the obtained solution can be improved.

In addition, the folded display may be a loop-shaped shaft structure that is folded at a constant period as shown in FIG. 8, or may be a structure that is folded twice or more as shown in FIG. .

FIG. 8 shows an example in which a solution is displayed by applying a loop-like wrap around a certain axis. In the example shown in FIG. 8, the axis corresponding to the index T2 'indicating the scheduled operation is looped and displayed with a fixed period. The index T2 ′ indicating the scheduled service in this example is one of the evaluation items, for example, with respect to the departure portion T2 indicating how many minutes (mm) the departure time (hh: mm) from the predetermined bus stop is. It may be a value with a penalty. In this example, it is assumed that every hour at 0 minutes is the scheduled departure time. Then, the range from 0 to 59 is set as the value of the starting portion T2. At this time, the 0 minute departure may be set as the optimum value of the index T2 '. Otherwise, it imposes a greater penalty on departures that are considered to be late in the-direction (too earlier than the scheduled time) compared to departures that are considered to be in the + direction (exceeding the scheduled time). You may do it. In addition, the following formula | equation (3) is an example of the optimization calculation applied to this example.

maxmize (w ₁ C + w ₂ T2 ′ + w ₃ S) (3)
However, T2 ′ = max {− (T2% 60), 5 {(T2% 60) −60}}

In Equation (3), C is the cost, T2 is the starting portion, and S is the customer satisfaction. Further, w ₁ , w ₂ , and w ₃ are weights for each evaluation item (cost C, scheduled operation index T2 ′, customer satisfaction S) in the objective function. Moreover, a% b indicates the remainder (remainder) when a is divided by the natural number b. Further, the scheduled operation index T2 ′ has a worst value when the departure time is 50 minutes per hour. And if the departure amount is up to 50 minutes, it is regarded as a delay in the + direction, and the delayed amount x1 is regarded as a negative evaluation for T2 ', and the case where the departure amount is 51 to 59 minutes is regarded as a delay in the-direction. The premature amount × 5 times is regarded as a negative evaluation for T2 ′.

FIG. 9 is an explanatory diagram showing an example in which a solution is displayed by performing folding twice or more with respect to a certain axis. In the example shown in FIG. 9, the axis corresponding to the index T3 'indicating the scheduled operation is displayed by being folded back multiple times. The index T3 ′ indicating the scheduled operation in this example is one of the evaluation items, for example, a penalty with respect to the arrival time T3 in which the time of arrival at the predetermined bus stop (hh: mm) is converted in minutes. It may be a value imposed. In this example, the range from 0 to 1439 is the value of arrival time T3. In addition, when arriving at 0 minutes per hour, the arrival time at 0 minutes per hour is set as the optimum value because the connection with the train is particularly good. After that, the evaluation range is changed according to the time zone such as whether or not it is the commuting time zone, while the return evaluation is based on arrival at 30 minutes per hour. In addition, the following formula | equation (4) is an example of the optimization calculation applied to this example.

maxmize (w ₁ C + w ₂ T3 ′ + w ₃ S) (4)
However, T3 ′ = A [T3 / 60] × (| T3% 60−30 | −30)

In equation (4), C is the cost, T3 is the arrival time [minute conversion], and S is the customer satisfaction. W ₁ , w ₂ , and w ₃ are weights for each evaluation item (cost C, scheduled operation index T3 ′, customer satisfaction S) in the objective function. Moreover, a% b indicates the remainder (remainder) when a is divided by the natural number b. A / b indicates a quotient obtained by dividing a by a natural number b. A [0 to 23] is an array storing weights according to what time the arrival time is, and is assumed to be given in advance as one of the constraint conditions. For example, even if the delay times of two buses are the same, there are cases where it is desired to change the evaluation due to the time zone or other reasons. Even in such a case, it is possible to cope with the problem by providing two or more foldings having different folding intervals. The example shown in FIG. 9 shows an example in which the axis is expanded while shifting the folded point sideways, but the folded point in a loop shape as shown in FIG. 8 is overlapped with the original axis. It may be displayed.

Display method example 5
This example is an example in which the color, density, line thickness, and line type of the line displaying the solution are changed and displayed according to the degree of suitability of each solution for the objective function. The solution display method determination unit 2163 of the display method determination unit 216 calculates an objective function value indicating the degree of conformity to the objective function for each received solution. Then, the solution display method determination unit 2163 calculates the density of the color of the line displaying the solution from the calculated objective function value of each solution. For example, the solution display method determination unit 2163 may determine the density of a line for displaying each solution in accordance with the number of solutions to be displayed. The solution display method determination unit 2163 may determine the line density in proportion to the objective function value of each solution. The solution display method determination unit 2163 may determine the line density according to the order of the results sorted by the objective function value.

Information on the color density of the line displaying each of these solutions is transmitted to the GUI unit 211, a graph based on the information is drawn, and the graph display screen is displayed on the display unit 12, so that the user can display the graph shown in FIG. The graph of b) can be seen. Note that FIG. 10A is an example of solution display when lines displaying each solution are simply distinguished by line type. FIG. 10B shows an example of solution display when the density of the color of a line displaying each solution is distinguished depending on whether the objective function value of the solution is good or bad.

In the example of FIG. 10A in which the line to be displayed is simply distinguished by the line type, the solution can be distinguished, but since the importance of the axis is not reflected in the line type, which solution is the better solution It is difficult to see if there is. On the other hand, in the example of FIG. 10B, the difference in the objective-based index on the important axis is reflected in the line color density after being converted into the rank in the objective function value. For this reason, it is easy to visually understand that the solution a1 shown at the highest density is a better solution in terms of the parameters. In this example, the color density for displaying each solution is changed according to whether the objective function value of the solution is good or bad. However, the display method of each solution in which the order or difference is distinguished depending on whether the objective function value of the solution is good or bad is not limited to the method using the color density (from dark color to light color). According to the display rules according to the predetermined order and the difference, depending on the color (red → blue → yellow, etc.), line type (solid line → broken line → dotted line, etc.), line thickness (thick → thin), etc. Each solution may be displayed separately. That is, according to this example, the GUI unit 211 provides a difference in the display of each solution according to the degree of suitability of each solution with respect to the objective function, so that even when the axis display is difficult to understand the axis importance. The difference between each solution becomes clearer. Therefore, the ability to visually grasp the superiority or inferiority of the solution can be improved, and the degree close to the optimal solution can be grasped while grasping the value in the goal-based index in the parameter setting at that time.

Display method example 6
In this example, the background color of the axis and its density are changed according to the range of absolute constraints on each axis and the desired degree of objective function for the point value on each axis. is there. The axis background display method determination unit 2162 of the display method determination unit 216 extracts an absolute constraint condition range from the received constraint conditions of each axis. Furthermore, the axis background display method determination unit 2162 calculates a degree of evaluation item preferable as an objective function for each value (hereinafter referred to as a condition adaptability), and reflects it in the background color of each axis. For example, as an absolute condition, the cost must be avoided above a certain value, the expected value of long-term sales is not allowed to be below a certain value, the customer satisfaction is also above a certain level, etc. Suppose there are constraints. The axis background display method determination unit 2162 distinguishes areas that do not satisfy these absolute constraints by displaying the background of each axis in gray. In addition, the axis background display method determination unit 2162 distinguishes the background color of each axis by, for example, the density of red according to the condition conformance.

Information on the background colors of these axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and a graph display screen is displayed on the display unit 12. Thus, the user views the graph of FIG. be able to. FIG. 11 is an explanatory diagram illustrating an example of solution display according to this example. FIG. 11A shows an example of solution display in which the background of each axis is displayed in a single color. FIG. 11B shows a solution when the background of each axis is color-coded in a range that satisfies the absolute condition, and is further color-coded within the range that satisfies the condition according to the degree of satisfaction (that is, the condition adaptability). It is an example of a display.

In the example of FIG. 11A in which the background of each axis is displayed in a single color, it is difficult to understand that even if the solution is positioned higher on a certain axis, the other axes do not satisfy absolute constraints. On the other hand, in the example of FIG. 11B, the background color of each axis reflects the range that satisfies the constraint condition and the condition conformance within the range. Therefore, it is visually easy to simultaneously grasp a solution that satisfies the absolute constraint condition and a better solution. That is, according to the present example, the GUI unit 211 changes the background color according to the range satisfying the constraint condition for each axis and the condition conformance within the range. Therefore, even if the solution is positioned higher on a certain axis, the solution that the absolute constraint condition cannot be satisfied on the other axis becomes clearer. Therefore, it is possible to improve the ability to grasp a solution that satisfies visually absolute constraint conditions, to improve the ability to grasp a good solution, and to improve the convincingness of the obtained solution.

In the case where the axis orientation is displayed in the improved direction, for example, the boundary line indicating whether or not the absolute constraint condition is satisfied as shown in FIG. 12 or 13 is changed to a bar graph or a line graph. It may be added as a background display. As described above, the display method of this example is not limited to the method of color-coding the background color as long as the display can be distinguished from the display of each solution. As shown in FIG. 14, it is also possible to add a background display based on a color gradation according to the condition conformance within the range, which is a line graph.

In addition, since the display of each solution of the present embodiment is not limited to parallel coordinate display, for example, as shown in FIGS. 15 and 16, each solution is displayed as a bar graph for each axis, or displayed in a radar chart. Also in this case, the display method of this example is applicable. Note that the example shown in FIG. 16 shows an example in which a range that satisfies the absolute constraint condition is colored and displayed. In addition to this example, it is also possible to add a background display with a color gradation according to the condition conformance within the range.

Display method example 7.
This example is an example of displaying the contents of each solution after displaying the upper and lower limits of the range that satisfies the constraint condition in the evaluation item and the optimum value of each axis in the objective function. The axis background display method determination unit 2162 of the display method determination unit 216 extracts an absolute constraint condition range from the received constraint conditions for each axis, calculates an optimum value for each axis as an objective function, and calculates it. Display on each axis or in the background. For example, when the constraint condition is indicated by a section, the axis background display method determination unit 2162 may determine the type and position of the mark indicating the upper limit and the lower limit, and may determine the type and position of the mark of the optimum value. . In this example, the axis background display method determination unit 2162 writes the characters “upper limit” and “lower limit” on the left side of the axis as scales, and sets the optimum value as an asterisk.

Information on the types and positions of the marks to be displayed on the background of these axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and the graph display screen is displayed on the display unit 12, so that the user can display FIG. The graph of (b) can be seen. FIG. 17 is an explanatory diagram illustrating an example of solution display according to this example. Note that FIG. 17A is an example of a solution display in a case where information regarding constraint conditions is not displayed. FIG. 17B is an example of a solution display in which the upper limit and lower limit of the range satisfying the constraint condition and the optimum value of each axis in the objective function are displayed.

In the example of FIG. 17A that does not display information on constraint conditions, it is difficult to grasp a solution that satisfies the constraint conditions. On the other hand, in the example of FIG. 17B, a range that satisfies the constraint condition on the background color of each axis and an optimum value are displayed on the axis or on the axis background. Therefore, it is visually easy to simultaneously grasp a solution that satisfies the absolute constraint condition and a better solution. That is, according to the present example, the GUI unit 211 changes the background color according to the range satisfying the constraint condition for each axis and the condition conformance within the range. Therefore, even if the solution is positioned higher on a certain axis, the solution that the absolute constraint condition cannot be satisfied on the other axis becomes clearer. Therefore, it is possible to improve the ability to grasp a solution that satisfies visually absolute constraint conditions, to improve the ability to grasp a good solution, and to improve the convincingness of the obtained solution.

Display method example 8.
This example is an example in which a figure having a width corresponding to a degree of preference (condition suitability) as an objective function is displayed on the axis background for each point on the axis within the range of absolute constraints on each axis. . The axis background display method determination unit 2162 of the display method determination unit 216 extracts the range of absolute constraint conditions from the received constraint conditions for each axis, and the degree of condition conformance for each value of the evaluation item within the range. Is reflected in the figure displayed on the background of each axis. For example, the axis background display method determination unit 2162 sets the width of the graphic to zero in the background of each axis in an area that does not satisfy the absolute constraint condition. On the other hand, in the case of a region that satisfies the absolute constraint condition, the axis background display method determination unit 2162 increases the width of the figure in accordance with the condition suitability. The axis background display method determination unit 2162 may determine the maximum width according to the weight of each axis. For example, the cost has the largest weight and is preferably a small value. Therefore, the axis background display method determination unit 2162 may create a graphic having the maximum width at the lowest end of the axis corresponding to the cost, and designate it as a graphic to be displayed on the axis background. The axis background display method determination unit 2162 may determine the maximum width according to the weight of each axis.

Information on the background graphics of these axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and a graph display screen is displayed on the display unit 12, so that the user sees the graph of FIG. be able to. FIG. 18 is an explanatory diagram showing an example of a solution display according to this example. Note that FIG. 18A is an example of a solution display when information on the constraint condition is not displayed. FIG. 18B is an example of a solution display in which a graphic having a width corresponding to the range of absolute constraint conditions and the condition conformance within the range is displayed on the background of each axis. As shown in FIG. 18B, when the maximum width of the condition conformance of each axis is determined according to the weight, the axis background display method determining unit 2162 displays the ratio below the axis. May be.

In the example of FIG. 18A in which information on constraint conditions is not displayed, it is difficult to grasp a solution that satisfies the constraint conditions. On the other hand, in the example of FIG. 18B, a graphic having a width corresponding to the range of absolute constraint conditions and the degree of condition conformance is displayed in the background of each axis. Therefore, it is visually easy to simultaneously grasp a solution that satisfies the absolute constraint condition and a better solution. In other words, according to the present example, the GUI unit 211 changes the width of the graphic displayed on the background according to the condition conformance within the range satisfying the constraint condition for each axis. Therefore, a solution such as solutions a1 and a3 in this example that are positioned higher on a certain axis does not satisfy the absolute constraint condition on the other axis becomes clearer. Therefore, it is possible to improve the ability to grasp a solution that satisfies visually absolute constraint conditions, to improve the ability to grasp a good solution, and to improve the convincingness of the obtained solution.

In the example shown in FIG. 18, the example in which the change in the width of the figure is directly proportional or inversely proportional to the increase or decrease in the evaluation value on the axis is shown, but the change in the width is not a straight line but a free curve. Good. In such a case, the axis background graphic may be set by defining a function y = f (x) with vertical x and horizontal y.

As described above, according to this embodiment, even a judge who has little knowledge about the work of correcting each element can easily try to grasp the characteristics of each solution and analyze the optimal solution using various objective-based indicators. it can. The reason is that the user can operate the parameters with objective-based indexes such as cost and customer satisfaction instead of each element. In addition, the user can confirm the display of the solution that reflects the result using a graph that visualizes the difference in the weight of each index, the range of constraint conditions, and the difference in the degree of fitness for each index. Because it can.

In the above example, some basic examples are shown, but it is also possible to use a combination of these display methods.

Embodiment 2. FIG.
Next, a second embodiment of the present invention will be described. FIG. 19 is a block diagram illustrating a configuration example of the information analysis system according to the second embodiment. In the information analysis system shown in FIG. 19, compared with the first embodiment shown in FIG. 1, the control unit 21 of the multi-objective optimization system replaces the parameter setting unit 214 with a change operation receiving unit 212, a parameter adjustment unit, 213 is different. Also, the program storage unit 221 of the multi-objective optimization system is different in that it further includes a parameter calculation program storage unit 2211.

In the present embodiment, the GUI unit 211 receives change operation information regarding parameters and constraint conditions in the goal-based index via the operation unit 11 in response to a user operation on the solution display screen, and the change operation reception unit 212 receives the change operation information. A function for outputting is further included.

For example, the GUI unit 211 is a screen that can be changed to a parameter adjustment screen that can accept various changing operations in response to pressing of an instruction button for starting parameter adjustment by improving a graph display screen for displaying a solution. An interface may be prepared. As the change operation, for example, changing the length, changing the position, changing the shape of the component corresponding to the axis of the graph being displayed on the graph display screen, the mark on the axis, the figure displayed on the background of the axis, etc. And the like. For example, the GUI unit 211 may generate a parameter adjustment screen in which components such as a graph axis, a background graphic, and a scale mark on the currently displayed graph are converted into GUI components for graph drawing that can be operated by the user. The GUI unit 211 expands the converted GUI component in a drawing window. Furthermore, the GUI unit 211 may provide a screen interface that accepts operations such as addition, movement, and size change of GUI components in accordance with mouse operations and keyboard operations on the drawing window. It should be noted that which GUI component actually corresponds to which part may be linked using the component property information or the like when converting to the GUI component. In addition, when a new component is added, the GUI unit 211 determines which part actually corresponds based on the part type of the component itself and where it is added (positional relationship with other components). Just judge.

The change operation accepting unit 212 accepts the content of the operation performed by the user from the GUI unit 211, and outputs the changed information to the parameter adjusting unit 213. Here, the change operation accepting unit 212 may notify the component level change information (the type, size, position, etc. of the added / changed component). The change operation accepting unit 212 receives from the GUI unit 211 the parameters related to the optimization objective function specified by the content of the operation performed by the user and the input content to the GUI related to the constraint conditions, and transmits them to the parameter adjusting unit 213. For example, the change operation receiving unit 212 may receive information related to an operation using a notification event included in a GUI component for drawing a graph. Accepted information includes, for example, change of axis length, change of axis interval, change of solution color, change of axis direction, change of axis folding display method, change of axis background color and density, This is information regarding the change of the display method of the upper and lower limits and the optimum value on the axis, the change of the diamond display according to the axis, and the operation of changing the order of the axes.

The parameter adjustment unit 213 calculates values of the parameters and constraint conditions in the changed goal-based index based on the change information received from the change operation reception unit 212. The parameter adjustment unit 213 may use the parameter calculation program stored in the parameter calculation program storage unit 2211 to convert the change information into parameters and constraint conditions in the purpose-based index. The parameter adjustment unit 213 receives a parameter calculation result from the called parameter calculation program, and outputs the result to the solution calculation unit 215 to calculate a solution using a new parameter.

The parameter calculation program storage unit 2211 stores a parameter calculation program. The parameter calculation program uses the GUI information being displayed and the input information (in this case, the component level change information) to the GUI input by the user to determine the parameters in the objective-based index and the values in the constraint condition format. Is a program for calculating

The parameter calculation program uses, for example, a structure pointer for referring to predetermined component information as an argument, and necessary information (for example, the position of each component such as an axis, a background graphic, a mark on the scale, Get specified shape of color). In addition, the parameter calculation program performs a process of obtaining parameters and constraint conditions in the target-based index by performing a conversion process according to a predetermined rule based on the positional relationship, and outputting the result. Actually, the control unit 21 that has read the parameter calculation program executes such processing according to the read parameter calculation program. It is also possible to pass the parameter values and constraint conditions that are currently set as arguments to the parameter calculation program and use the arguments as a reference when converting GUI information into parameters or constraint conditions. For example, the program storage unit 221 may store a plurality of programs having different parameter calculation methods. The parameter calculation program is converted into an executable format such as DLL and stored.

The predetermined rule is, for example, a rule that when the direction of the axis is changed, the improvement direction of the evaluation item associated with the axis is reversed. The predetermined rule is that, for example, when the scale or length of the axis is changed, the ratio of the scale or length between the axes is changed to the ratio of the magnitude of importance between the evaluation items associated with each axis. The rule of reflecting may be used. Also, the predetermined rule is a rule that, for example, when the interval between the axes is changed, the ratio of the width between the axes is reflected in the ratio of the magnitude of the importance between the evaluation items associated with each axis. There may be. The predetermined rule is a rule that, for example, when the order of the axes is changed, the arrangement order of the axes is reflected in the ranking of the degree of importance between the evaluation items associated with each axis. May be.

Further, the predetermined rule may be a rule that, for example, when the axis is folded or the folding position is changed, the folding position is set to the optimum value of the evaluation item associated with the axis. In addition, for example, when the boundary line of an area displayed in a color-coded manner representing the range of the constraint condition is changed on the background of the axis, the predetermined rule corresponds to the scale on the axis indicated by the changed boundary line. The rule of changing the range of the constraint condition may be used. Further, it is assumed that the predetermined rule is such that, for example, the change in density is changed in the gradation area of the background of the axis representing the degree of condition conformance within the range of absolute constraint conditions. In this case, the predetermined rule may be a rule that the degree of density change after the change is regarded as a condition conformance within the range of the absolute constraint condition and reflected in the constraint condition and the importance level.

Suppose, for example, that the upper limit and lower limit of the constraint condition and the mark provided for displaying the optimum value in the objective function are added or the position is changed on the axis. In this case, the predetermined rule may be a rule that sets an upper limit or a lower limit of the constraint condition or an optimum value in the objective function according to the position after the change. In addition, for example, a component provided for the background of the axis as a figure with a width according to the degree of condition conformance within the range of absolute constraints on each axis, or the shape (mainly width and height) is Suppose it changes. In this case, the predetermined rule is a rule that calculates the upper limit, the lower limit, and the condition conformance level of the constraint condition according to the position and shape, and sets the constraint condition that satisfies the calculated upper limit, lower limit, and condition conformance level. Also good.

That is, in the first embodiment, the display method determination unit 216 performs a conversion process from the parameter or the constraint condition to the graph setting value in order to visualize the parameter or the constraint condition. Therefore, the conversion process to the parameter or the constraint condition can be realized by the reverse process of the conversion process.

In the present embodiment, the change operation accepting unit 212 and the parameter adjusting unit 213 are realized by an information processing apparatus such as a CPU that operates according to a program. Other points are the same as those in the first embodiment.

Next, the overall operation of this embodiment will be described. FIG. 20 is a flowchart showing an example of the operation of the information analysis system of this embodiment. As a premise of the operation shown in FIG. 20, in the information analysis system of the present embodiment, parameters and constraint conditions are set at least once, and a graph displaying the solution obtained as a result of optimization calculation for the parameters is a user terminal. It is assumed that the image is drawn on the first screen. Note that the setting values of parameters and constraint conditions may be default values.

In such a state, when the information analysis system receives an instruction to perform parameter adjustment from the user, the GUI unit 211 displays a parameter adjustment screen (step S21). Here, the parameter adjustment screen is a screen in which components such as the axis of the currently displayed graph, background graphics, and marks on the scale are converted into GUI components for graph drawing that can be operated by the user. The converted GUI component is expanded in a drawing window on this screen, and accepts operations such as addition, movement, and size change of the GUI component in accordance with mouse operation and keyboard operation on the drawing window.

Suppose that on the displayed parameter adjustment screen, the user performs a change operation such as adding a GUI component for drawing a graph or changing the position or size of a placed GUI component. In this case (Yes in step S <b> 22), the change operation accepting unit 212 accepts operation details performed by the user via the GUI unit 211. The change operation accepting unit 212 recognizes component level change information (the type, size, position, etc. of the added / changed component) from the received input information to the GUI, and outputs the information to the parameter adjustment unit 213. .

The parameter adjustment unit 213 that has received the component level change information calculates parameters and constraint conditions for the purpose-based index from the change information received from the change operation reception unit 212 (step S23). The parameter adjustment unit 213 may perform parameter conversion processing by selectively using a parameter adjustment program for performing conversion processing according to a rule corresponding to the display method adopted by the graph displayed before adjustment, for example. Good. Note that the parameter adjustment unit 213 can directly perform the parameter conversion calculation without using the parameter calculation program to calculate the changed parameter value and the constraint condition. The user can also specify conversion rules.

When the constraint condition and parameter value of the objective function after the change by the user operation are obtained by the conversion process, the parameter adjustment unit 213 reflects the obtained constraint condition and parameter value of the objective function after the change in the objective function (step S23). The subsequent processing is the same as in the first embodiment.

That is, when the constraint condition or parameter value of the objective function is updated, the solution calculation unit 215 obtains a solution by performing an optimization calculation using an optimization program based on the constraint condition and the parameter value (step) S13).

Next, the display method determination unit 216 sets a graph for displaying the solution (steps S14-1 to S14-3).

When the graph setting by the display method determining unit 216 is completed, the GUI unit 211 receives a display request from the display method determining unit 216, draws a graph for displaying the solution, and displays the graph on the display unit 12 of the user terminal 1. The graph display screen on which the graph is drawn is output (step S15).

Here, when the user gives an instruction to change the parameter again and display again (step S16), in this embodiment, the GUI unit 211 switches the screen from the solution graph display screen to the parameter adjustment screen, and again. You may make it receive the change operation with respect to the GUI component for graph drawing from a user (it returns to step S21). Note that the user may be able to select which one of the parameter input screen for directly inputting parameters and the parameter adjustment screen for changing parameters by operating the GUI component.

Hereinafter, a parameter adjustment method using a GUI component will be described using a specific example.

Parameter adjustment example 1.
In this example, when the user performs an operation to change the length or scale of the axis, the weight parameter indicating the importance of each evaluation item is changed accordingly. FIG. 21 is an explanatory diagram showing an example of parameter adjustment according to this example. In the following, a problem of optimizing the schedule and arrangement of a salesclerk at a certain store by performing the optimization calculation shown in the above-described equation (1) will be described as an example.

Before the operation, the length of the axis i set by the magnitude of the absolute value of w _i is equal so that the weight parameters w ₁ = −75, w ₂ = 75, and w ₃ = 75 for each evaluation item. (See FIG. 21A). Here, the user operates the operation unit 11 to change the length of the first axis corresponding to the cost, which is the first evaluation item, to 100/75 times, and the long-term, which is the second evaluation item. Operation to change the length of the second axis corresponding to the expected value of sales to 70/75 times, and change the length of the third axis corresponding to customer satisfaction, which is the third evaluation item, to 50/75 times (See FIG. 21B). The operation of changing the length of the axis is performed by, for example, a mouse drag operation on a GUI component for drawing a graph representing an axis (in this example, an arrow component representing an axis) or a property area of the component. It can be done by numerical input operation. At this time, the GUI unit 211 may automatically adjust so that the center points of the axes are aligned and displayed. The reference point for aligning the axes is not limited to the center point of the axes.

In addition, since the point position (evaluation item value) on the axis of each solution is not the target of the change operation, the GUI unit 211 may not display the solution during the change operation. Alternatively, the GUI unit 211 may perform interlocking processing such as changing the position on the solution axis in accordance with the change operation.

Here, it is assumed that C, L, and S do not care about the unit without losing generality, and that w ₁ , w ₂ , and w ₃ have the same importance as each index.

For example, when the GUI unit 211 detects that a change operation has been performed on a component representing an axis by receiving an event of the GUI component, the GUI unit 211 outputs that fact to the change operation reception unit 212. For example, when the change operation receiving unit 212 receives such information, the scale operation ratio of each axis (that is, the scale ratio 100/100 of the first axis) is obtained from the length information before the change and the length information after the change. 75 times, the second axis scale ratio 70/75 times, the third axis scale ratio 50/75 times), and outputs them to the parameter adjustment unit 213 as change information. Note that the change operation accepting unit 212 may output change information to the parameter adjustment unit 213 after waiting for a change reflection instruction from the user.

When the parameter adjustment unit 213 receives the change information from the change operation reception unit 212, the parameter change unit 213 uses the received change information and the information of the original parameters w ₁ = −75, w ₂ = 75, and w ₃ = 75 as arguments. Call the parameter calculation program. Then, the parameter adjustment unit 213 causes the parameter calculation program to calculate a parameter change value in the objective function. The parameter calculation program calculates the original parameter value and the scale ratio of each axis (that is, the scale ratio of the first axis is 100/75 times, the scale ratio of the second axis is 70/75 times, and the scale ratio of the third axis is 50/75 times. ) May be calculated as w ₁ = −100, w ₂ = 70, and w ₃ = 50 and output to the parameter adjustment unit 213. Here, parameters w ₁ = −100, w ₂ = 70, and w ₃ = 50 are obtained as changed parameter values.

The parameter adjustment unit 213 outputs new parameters w ₁ = −100, w ₂ = 70, and w ₃ = 50 obtained from the parameter calculation program to the solution calculation unit 215.

The solution calculation unit 215 executes optimization calculation using the optimization program based on the parameters w ₁ = −100, w ₂ = 70, w ₃ = 50 received from the parameter adjustment unit 213, and the changed Calculate the solution at the parameter. The optimization program includes parameters w ₁ = −100, w ₂ = 70, w ₃ = 50 related to the objective function, a list of store staff stored in the basic data storage unit 2221, a list of roles, hourly wages of each store staff, and the like. Used to evaluate store clerk scheduling, resulting costs, long-term sales expectations, and customer satisfaction. In the case of parameter change, the solution calculation unit 215 can perform optimization calculation without using a full search algorithm or the like by targeting only the difference due to the current change operation. At that time, the solution calculation unit 215 may use a previously calculated solution stored in the calculation result storage unit 2222.

The solution calculation unit 215 outputs the calculation result of the new solution obtained from the optimization program and the new parameters w ₁ = −100, w ₂ = 70, and w ₃ = 50 to the display method determination unit 216.

In the display method determination unit 216, the axis display method determination unit 2161 determines the length of each axis based on the received parameters w ₁ = −100, w ₂ = 70, and w ₃ = 50. For example, the axis display method determination unit 2161 may set the length of each axis i to a value proportional to the absolute value | w _i | of the parameter w _i .

Information on the lengths of these axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and a graph display screen is displayed on the display unit 12, so that the user sees the graph of FIG. be able to. FIG. 21 shows an example in which the element that derives the optimum value is not changed by the changing operation. In addition, when the weight with respect to an evaluation item is changed largely, the element which derives an optimal value may differ as a result of optimization calculation.

Parameter adjustment example 2.
In this example, when the user performs an operation of changing the axis interval, the importance of the evaluation item is changed accordingly. FIG. 22 is an explanatory diagram showing an example of parameter adjustment according to this example. Note that this example is not the length of the axis in the first parameter adjustment example described above, but the width of the axis (here, the width of the area allocated for display on the axis, not the width of the axis itself). The weighting parameter for each evaluation item is calculated based on the above-described basic operation, and the basic operation is the same as that of the first parameter adjustment example except that the parameter calculation method is different.

Before the operation, the width of the axis i set by the magnitude of the absolute value of w _i is equal, such as the weight parameters w ₁ = −75, w ₂ = 75, and w ₃ = 75 for each evaluation item. Suppose that there is (see FIG. 22A). In the example shown in FIG. 22, the background of the axis is colored according to the width of the axis so that the width of each axis can be visually understood.

Here, the user operates the operation unit 11 to change the width of the first axis corresponding to the cost, which is the first evaluation item, to 100/75 times, and the long-term sales, which is the second evaluation item. An operation of changing the width of the second axis corresponding to the expected value of 70/75 times, and an operation of changing the width of the third axis corresponding to the customer satisfaction as the third evaluation item to 50/75 times (See FIG. 22B). The axis width changing operation can be performed by, for example, a GUI component for drawing a graph representing the axis width (in this example, a block component provided by coloring the axis background or a width under the axis). Can be performed by a mouse drag operation or a numerical value input operation.

For example, when the GUI unit 211 detects that a change operation has been performed on the component representing the axis width by receiving an event of the GUI component, the GUI unit 211 outputs the change operation reception unit 212 to that effect. For example, when the change operation reception unit 212 receives such information, the scale ratio of the width of each axis (that is, the scale ratio of the first axis 100/75) from the information of the width before the change and the information of the width after the change. And the second axis scale ratio 70/75 times and the third axis scale ratio 50/75 times) are output to the parameter adjustment unit 213 as change information.

The subsequent processing is the same as in the first parameter adjustment example. FIG. 22C shows an example in which the solution obtained as a result of the optimization calculation using the changed parameter is displayed by a display method based on the changed parameter.

In the above example, an example in which the weight parameter is changed based on the width of the axis is shown, but the distance between the left and right axes can be used instead of the width of the axis.

Parameter adjustment example 3.
In this example, when the user performs an operation to change the axis folding position, the weight parameter indicating the importance of each evaluation item is changed accordingly. FIG. 23 is an explanatory diagram showing an example of parameter adjustment according to this example.

It is assumed that the objective function is the above equation (2) before the operation. That is, weight parameters w ₁ , w ₂ , w ₃ for each evaluation item (cost C, delay time T ′, customer satisfaction S) and penalty T ′ = − max {T, −6T} ( However, it is assumed that −10 <T <60) is given as a constraint condition, and the above equation (2) is used as the objective function. In the example shown in FIG. 23A, the result of the optimization calculation performed using such an objective function is displayed. In the example shown in FIG. 23 (a), the axis corresponding to the delay time T is folded back to zero, the scale of the axis with a negative value is displayed as 6 times the positive value, and the lower limit value of the delay time is set to -10. The upper limit value is displayed as +60.

Here, the user operates the operation unit 11 to change the folding position on the second axis corresponding to the delay time T, which is the second evaluation item, from 0 to +5, and to set the lower limit value +60 to +35. It is assumed that a change operation is performed (see FIG. 23B). The operation of changing the folding position can be realized, for example, by pressing the scale +5 of the GUI component for drawing a graph representing the axis and dragging it to the 0 position. Further, the operation for changing the return position can be performed by, for example, a numerical value input operation to the return scale column of the GUI component for drawing a graph. The upper limit value changing operation can be performed by, for example, a numerical value input operation in a turn scale column included in a graph drawing GUI component representing an axis.

For example, when the GUI unit 211 detects that a change operation has been performed on a component representing an axis by receiving an event of the GUI component, the GUI unit 211 outputs that fact to the change operation reception unit 212. For example, when the change operation reception unit 212 receives such information, the optimal value of the axis is changed from 0 to +5 based on the information on the return position before the change of the second axis and the information on the return position after the change. Recognize that. Further, the change operation accepting unit 212 recognizes that the upper limit value of the axis has been changed from +60 to +35 from the information on the upper limit value before the change of the second axis and the information of the upper limit value after the change. As a result of the change of the folding position and the change of the upper limit scale, the delay time T = −10 minus 15 from the optimum value +5 and the delay time T = + 35 minus +5 are the same degree. Has been. Therefore, the change operation accepting unit 212 recognizes that the scale at a value that is more negative than +5 is double that of a value that is more positive than +5. The change operation reception unit 212 outputs the above to the parameter adjustment unit 213 as change information. Note that the change operation accepting unit 212 may output change information to the parameter adjustment unit 213 after waiting for a change reflection instruction from the user.

In addition to this, as a user operation related to the wrapping display, a button etc. is prepared to enable / disable the axis wrapping display function, or a submenu selection operation by right-clicking etc. while pointing to a certain point on the axis The user can also specify that folding is performed at this point.

When the parameter adjustment unit 213 receives the change information from the change operation reception unit 212, the parameter adjustment unit 213 calls the parameter calculation program using the received change information, the original parameters, and the constraint conditions as arguments. Then, the parameter adjusting unit 213 calculates a parameter change value in the objective function. The parameter calculation program obtains T ′ = − max {2 (T−5), −4 (T−5)} (−10 ≦ T ≦ 35) as the penalty content for the delay time T, and based on that, The objective function is changed to the following equation (5).

maximize (w ₁ C + w ₂ × {−max {2 (T−5), −4 (T−5)}} + w ₃ S) (5)

The parameter adjustment unit 213 outputs a new objective function obtained from the parameter calculation program to the solution calculation unit 215. The solution calculation unit 215 performs optimization calculation using an optimization program based on the objective function received from the parameter adjustment unit 213, and calculates a solution for the changed parameter. The same applies to the case where the changed parameters and constraint conditions are output instead of the objective function.

The optimization program calculates a solution by using the weight parameter in the objective function, the constraint condition, the driver list stored in the basic data storage unit 2221, the list of buses to be ridden, the hourly wage of each driver, and the like. . In this example, the constraint condition includes a penalty content for the delay time T, an upper limit value, a lower limit value, and the like. The optimization program uses this information to estimate the driver's scheduling and the resulting cost, degree of scheduled operation, and customer satisfaction. Using a branch and bound method yields the same results as an algorithm that traverses all possible combinations of schedules, so the optimization program uses that branch and bound method to calculate the solution. To do. The solution calculation unit 215 can perform optimization calculation without using a full search algorithm or the like by targeting only the difference due to the current change operation. At that time, the solution calculation unit 215 may use a previously calculated solution stored in the calculation result storage unit 2222.

In the display method determination unit 216, the axis display method determination unit 2161 determines the presence / absence of the axis return display and the return point from the parameters used for calculating the solution and set for each evaluation item. . For example, the axis display method determination unit 2161 displays that the second axis corresponding to the delay time T is folded back at +5, and calculates that the penalty of the negative value is twice as large as the positive value. In this case, for the axis corresponding to the delay time T, the axis display method determination unit 2161 determines that the scale of the axis with a negative value from +5 is doubled with a positive value from +5.

Information on the lengths of these axes is transmitted to the GUI unit 211, a graph based on the information is drawn, and a graph display screen is displayed on the display unit 12, so that the user sees the graph of FIG. be able to. FIG. 23 shows an example in which the element that derives the optimum value is not changed by the changing operation. Note that, when the weights and constraint conditions for the evaluation items are greatly changed, the element for deriving the optimum value may differ as a result of the optimization calculation.

Parameter adjustment example 4.
In this example, when the user performs an operation to change the color range or color density in the axis background display, the constraint condition is changed or the weight for each evaluation item is changed accordingly. FIG. 24 is an explanatory diagram showing an example of parameter adjustment according to this example.

Suppose that before the operation, the range of absolute constraint conditions in each axis derived from the given constraint conditions and the condition conformance within the range are the contents shown in FIG. Here, it is assumed that the user performs an operation of pushing up the range of the gray area displayed on the background of the first axis corresponding to the cost, which is the first evaluation item, via the operation unit 11 (FIG. 24). (See (b)). Further, it is assumed that the user performs an operation for changing the gradient of the density of the red gradation area displayed on the background of the third axis corresponding to the customer satisfaction that is the third evaluation item via the operation unit 11. (See FIG. 24B.) These changing operations can be performed, for example, by a mouse drag operation on a GUI component for drawing a graph displaying an axis background, or a numerical value input operation on a property area of the component.

For example, when the GUI unit 211 detects that a change operation has been performed on the component displaying the axis background by receiving an event of the GUI component, the GUI unit 211 outputs that fact to the change operation receiving unit 212. For example, when receiving the information, the change operation receiving unit 212 recognizes the change contents from the length, position, and angle information before the change and the length, position, and angle information after the change, and changes them to the change information. To the parameter adjustment unit 213.

When the parameter adjustment unit 213 receives the change information from the change operation reception unit 212, the parameter adjustment unit 213 calls the parameter calculation program using the received change information, the original parameters, and the constraint conditions as arguments. Then, the parameter adjusting unit 213 calculates the parameter and the constraint condition change value in the objective function. The parameter calculation program calculates a new constraint condition and a weight for each evaluation item from the original parameter value and the constraint condition, the constraint condition after the change in each axis, and the fitness thereof, and the parameter adjustment unit 213 It may be output.

The parameter adjustment unit 213 outputs new parameters and constraint conditions obtained from the parameter calculation program to the solution calculation unit 215 to execute optimization calculation. The subsequent processing is the same as in the first parameter adjustment example. FIG. 24C shows an example in which the solution obtained as a result of the optimization calculation using the changed parameters is displayed by a display method based on the changed parameters.

Parameter adjustment example 5.
In this example, when the user performs an operation of changing the upper limit mark, lower limit mark, and optimum value mark displayed on the axis, this is reflected in the objective function. For example, when the solution display as shown in FIG. 17B is performed, in response to a user operation to change the upper limit mark, the lower limit mark, and the optimum value mark displayed on the axis, The parameter adjustment unit 213 changes the absolute constraint condition range or changes the objective function. Then, the parameter adjustment unit 213 obtains an optimal solution under new conditions and performs redisplay. Note that the change of the range of the constraint condition and the change of the objective function from the optimum value may be performed by the same method as in the third and fourth parameter adjustment examples, for example.

Parameter adjustment example 6.
In this example, when the user changes the graphic representing the condition conformance displayed on the background of each axis, the constraint condition and the weight for each evaluation item are changed accordingly. For example, when the solution display as shown in FIG. 18B is performed, the user changes the range in which the width of the diamond-shaped figure is zero, or changes the red width of the diamond-shaped figure. Accept the operation. At this time, the parameter adjustment unit 213 changes the range of absolute constraint conditions or changes the objective function in accordance with the user operation. FIG. 25 is an explanatory diagram showing an example of parameter adjustment according to this example.

Suppose that before the operation, the absolute constraint condition range in each axis derived from the given constraint conditions and the condition conformance within the range are as shown in FIG. Here, it is assumed that the user performs an operation of extending the height of the graphic displayed on the background of the first axis corresponding to the cost, which is the first evaluation item, via the operation unit 11 (FIG. 25). (See (b)). In addition, the user reduces the height of the graphic displayed on the background of the second axis corresponding to the degree of scheduled operation, which is the second evaluation item, via the operation unit 11 and increases the width in the lower part of the graphic. It is assumed that an operation of spreading is performed (see FIG. 25B). In addition, when the user performs an operation of expanding the width in the lower part of the graphic displayed on the background of the third axis corresponding to the third evaluation item, customer satisfaction, through the operation unit 11, It is assumed that an operation for changing the shape to a candle shape is performed (see FIG. 25B). These changing operations can be performed, for example, by a mouse drag operation on a GUI component for drawing a graph displaying an axis background, or a numerical value input operation on a property area of the component.

When the parameter adjustment unit 213 receives the change information from the change operation reception unit 212, the parameter adjustment unit 213 calls the parameter calculation program using the received change information, the original parameters, and the constraint conditions as arguments. Then, the parameter adjusting unit 213 calculates the parameter and the constraint condition change value in the objective function. The parameter calculation program uses the original parameter values and constraint conditions, the constraint conditions after the change in each axis, and the degree of conformance to the new constraint conditions (the range of absolute constraints and the conformance within the range. ) And a weight for each evaluation item may be calculated and output to the parameter adjustment unit 213.

The parameter adjustment unit 213 outputs new parameters and constraint conditions obtained from the parameter calculation program to the solution calculation unit 215 to execute optimization calculation. The subsequent processing is the same as in the first parameter adjustment example. FIG. 25C shows an example in which the solution obtained as a result of the optimization calculation using the changed parameters is displayed by a display method based on the changed parameters.

In the example shown in FIGS. 18B and 25, a linear diamond shape, a triangular shape, or a square shape is shown, but the shape of the shape is not limited to this. For example, it can be described by a free curve that is symmetrical with respect to each axis.

Parameter adjustment example 7.
In this example, when the user changes the order of the axes, this is reflected in the parameters in the objective function. For example, as in the example shown in FIG. 6B, it is assumed that the solutions are displayed in the order of axis arrangement and the axis interval according to the weight of each evaluation item. In this case, an operation for switching the second axis and the third axis is received from the user. At this time, the parameter adjustment unit 213 changes the values of the parameters w ₂ and w ₃ according to the user operation. Then, the parameter adjustment unit 213 obtains an optimal solution under new conditions and performs redisplay.

In the above example, the relationship between w ₂ and w ₃ before the operation is | w ₂ |> | w ₃ |, but the parameter adjustment unit 213 determines that | w ₃ |> | w ₂ | It may be changed so as to become. The operation is the same as that of the other parameter adjustment examples, except that the operation of the axis becomes the operation of the values of w ₂ and w ₃ .

The operations that can be performed in this example are not limited to the exchange of axes. The parameter adjustment unit 213 can also change the parameter value depending on the position of the axis. For example, the parameter adjustment unit 213 may slightly shift the axis 2 to the right side to set w ₂ = 75 to w ₂ = 70, and slightly shift to the left side to set w ₂ = 75 to w ₂ = 80.

As described above, according to the present embodiment, the user can perform optimization calculation by an easy-to-understand operation of changing the importance of an axis and the length of the axis. In addition, since it is possible to determine whether the order of the solutions is changed by such an operation, the user can easily request and simulate the optimum solution after changing the importance. In addition, according to the present embodiment, recalculation can be performed using only the difference in the movement of the position information of each axis point of the solution once displayed with the original setting. In this case, since it is not necessary to use again an optimization program that implements the algorithm for full search, the calculation time may be greatly improved.

Hereinafter, the outline of the present invention will be described. FIG. 26 is a block diagram showing an outline of the present invention. The multipurpose optimization system shown in FIG. 26 includes a display method determination unit 500.

The display method determination unit 500 (for example, the display method determination unit 216) displays the solution by displaying a solution by associating a plurality of evaluation items with one axis, respectively, in the graph for displaying the solution, the background of the axis, or each solution. The display reflects the importance or constraint conditions of each evaluation item, which are parameters used for solution calculation.

The display method determination means 500 is an axis display method determination means 501 that determines the axis display method in the graph for displaying the solution, based on the parameters used for the solution calculation and the importance of each evaluation item or the contents of the constraint conditions. (For example, the axis display method determination unit 2161) may be included.

Further, the display method determining means 500 is a parameter used for calculating the solution, and based on the importance of each evaluation item or the contents of the constraint condition, the axis background for determining the display method of the axis background in the graph for displaying the solution Display method determining means 502 (for example, axis background display method determining unit 2162) may be included.

Further, the display method determining means 500 is a solution display method for determining the display method of each solution in the graph for displaying the solution based on the importance of each evaluation item or the contents of the constraint conditions, which are parameters used for solution calculation. A determination unit 503 (for example, a solution display method determination unit 2163) may be included.

The axis display method determining means 501 includes, for example, the arrangement of the axes, the length of the lines of each axis, the thickness of the lines of each axis, the interval between the axes, the direction of each axis, whether or not each axis is folded, and the folding point Or, for at least one item of the scale method, the axis in the graph that displays the solution by setting the parameter that was used for solution calculation and that is set according to the importance of the evaluation item or the content of the constraint condition You may reflect the importance or constraint conditions of each evaluation item on the display.

Also, the axis background display method determining means 502 may include, for example, the shape of the graphic displayed on the background, the color of the graphic, the type of line displayed on the background, the thickness of the line, the position of the band displayed on the background, the color of the band, For each of the background display of the axis in the graph for displaying the solution, by setting the setting value according to the importance of the evaluation item or the content of the constraint condition, the parameter used for the solution calculation for at least one item The importance or constraint condition of the evaluation item can be reflected.

FIG. 27 is a block diagram showing another configuration example of the multi-objective optimization system of the present invention. As shown in FIG. 27, the multi-objective optimization system may further include a change operation accepting unit 601, a parameter calculating unit 602, a solution calculating unit 603, and a solution display unit 604.

In such a configuration, the change operation accepting unit 601 (for example, the change operation accepting unit 212) applies to the GUI component constituting the graph that performs the solution display by the display method determined by the display method determining unit. Changes by user operations may be accepted.

In addition, the parameter calculation unit 602 (for example, the parameter adjustment unit 213) calculates information about the change contents for the GUI component received by the change operation reception unit 601 and each solution displayed in the solution display before the change is received. The change value in the parameter used for the solution calculation may be calculated from the content of the parameter used in the above.

Further, the solution calculation unit 603 (for example, the solution calculation unit 215) may execute the optimization calculation again by using the changed parameter calculated by the parameter calculation unit 602.

In addition, the solution display unit 604 (for example, the GUI unit 211) redisplays the solution obtained by the optimization calculation by the solution calculation unit using the display method determined by the display method determination unit 500 based on the changed parameter. May be.

As mentioned above, although this invention was demonstrated with reference to embodiment and an Example, this invention is not limited to the said embodiment and Example. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

This application claims priority based on Japanese Patent Application No. 2012-124123 filed on May 31, 2012, the entire disclosure of which is incorporated herein.

The present invention is applicable not only for the purpose of optimization, but also for the purpose of simultaneously displaying the values of these indicators in order to grasp the contents of the trade-off information consisting of a plurality of options. For example, the present invention can be suitably applied to an application for simply visualizing information analysis results. Further, the present invention can be suitably applied to an application for inputting an objective function reflected in system control after grasping a solution space.

DESCRIPTION OF SYMBOLS 1 User terminal 11 Operation part 12 Display part 2 Multi-objective optimization system 21 Control part 211 GUI part 212 Change operation reception part 213 Parameter adjustment part 214 Parameter setting part 215 Solution calculation part 216 Display method determination part 2161 Axis display method determination part 2162 Axis Background display method determination unit 2163 Solution display method determination unit 22 Storage unit 221 Program storage unit 2211 Parameter calculation program storage unit 2212 Optimization program storage unit 222 Data storage unit 2221 Basic data storage unit 2222 Calculation result storage unit 500 Display method determination unit 501 Axis display method determination means 502 Axis background display method determination means 503 Solution display method determination means 601 Change operation acceptance means 602 Parameter calculation means 603 Solution calculation means 604 Solution display means

Claims

A multi-objective optimization system that displays the results of multiple solutions together with multiple evaluation items in multi-objective optimization,
When displaying a solution by associating each of the plurality of evaluation items with one axis, the parameters used for calculating the solution are displayed on the axis, background of the graph, or each solution in the graph for displaying the solution. A multi-objective optimization system characterized by comprising a display method determining means that reflects the importance or constraint condition of.
The display method determining means is
The axis display method determination means which determines the display method of the axis in the graph which is a parameter used for solution calculation, and displays the solution based on the importance of each evaluation item, or the contents of constraints. Multi-purpose optimization system.
The display method determining means is
2. An axis background display method determining means for determining a display method of an axis background in a graph for displaying a solution based on parameters used for calculating a solution and the importance of each evaluation item or the contents of constraint conditions. Or the multi-objective optimization system according to claim 2.
The display method determining means is
A solution display method determining means for determining a display method of each solution in a graph for displaying the solution based on the importance of each evaluation item or the content of the constraint condition, which is a parameter used for solution calculation. Item 4. The multi-objective optimization system according to any one of items 3 to 3.
Axis display method determining means determines the axis arrangement, the length of each axis line, the thickness of each axis line, the distance between the axes, the direction of each axis, whether or not each axis is folded, whether the folding point or scale The multi-objective optimization system according to claim 2, wherein a setting value is set for at least one item of the method, which is a parameter used for solution calculation and is set according to the importance of the evaluation item or the content of the constraint condition.
The axis background display method determining means includes at least one of the shape of the figure displayed on the background, the color of the figure, the type of line displayed on the background, the thickness of the line, the position of the band displayed on the background, and the color of the band. The multi-objective optimization system according to claim 3, wherein a setting value corresponding to the importance of the evaluation item or the content of the constraint condition is set for the item, which is a parameter used for solution calculation.
A change operation accepting means for accepting a change by a user operation with respect to a GUI component constituting a graph displaying a solution by the display method determined by the display method determining means;
In the parameters used for solution calculation, the information indicating the change contents for the GUI component received by the change operation receiving means and the parameter contents used for calculation of each solution displayed in the solution display before the change is received. The multi-objective optimization system according to any one of claims 1 to 6, further comprising parameter calculation means for calculating a change value.
Solution calculation means for performing optimization calculation again using the changed parameter calculated by the parameter calculation means;
The multi-objective optimum according to claim 7, further comprising solution display means for redisplaying a solution obtained by optimization calculation by the solution calculation means by a display method determined by a display method determination means based on a parameter after change. System.
An information analysis system that analyzes information by displaying results of multiple solutions together with multiple evaluation items,
When displaying a solution by associating each of the plurality of evaluation items with one axis, the parameters used for calculating the solution are displayed on the axis, background of the graph, or each solution in the graph for displaying the solution. An information analysis system characterized by comprising a display method determining means that reflects the importance or constraint condition of the information.
A multi-objective optimization method that performs multi-objective optimization by displaying results of a plurality of solutions together with a plurality of evaluation items,
When displaying a solution by associating each of the plurality of evaluation items with one axis, the parameters used for calculating the solution are displayed on the axis, background of the graph, or each solution in the graph for displaying the solution. A multi-objective optimization method characterized by reflecting the importance or constraint condition of.
The parameters used for solution calculation are the parameters used for solution calculation in the setting items that determine the display method of each axis in the graph that displays the solution based on the importance of each evaluation item or the contents of the constraint condition. The multipurpose according to claim 10, wherein the importance or the constraint condition of each evaluation item is reflected in the axis display in the graph for displaying the solution by setting a setting value according to the importance of the evaluation item or the content of the constraint condition. Optimization method.
The parameters used for calculating the solution are the setting items that determine the display method of the background of each axis in the graph that displays the solution based on the importance of each evaluation item or the contents of the constraint conditions. 11. By setting a setting value according to the importance of the evaluation item or the content of the constraint condition, the importance or the constraint condition of each evaluation item is reflected in the background display of the axis in the graph for displaying the solution. The multi-objective optimization method according to claim 11.
The parameters used for calculating the solution are the setting items that determine the display method of each solution in the graph for displaying the solution based on the importance of each evaluation item or the contents of the constraint condition. The importance level or the constraint condition of each evaluation item is reflected in the display of each solution in the graph for displaying the solution by setting a setting value according to the importance level of the evaluation item or the content of the constraint condition. Item 13. The multi-objective optimization method according to any one of Items 12.
If at least the axis, background of the graph or the display of each solution in the graph for which the solution is displayed reflect the importance or constraints of each evaluation item that are the parameters used in the solution calculation, the solution is displayed. Accepts changes made by user operations on the GUI components that make up the graph
A change value in a parameter used for solution calculation is calculated from information indicating a change content for the received GUI component and a parameter content used for calculation of each solution displayed in the solution display before the change is received.
Run the optimization calculation again using the calculated changed parameters,
Based on the parameters after the change, the parameters used for solution calculation in the axis, background of the graph or the display of each solution in the graph that displays the solution, reflecting the importance or constraint conditions of each evaluation item, The multi-objective optimization method according to claim 10, wherein a solution obtained by optimization calculation is displayed.
A multi-objective optimization program that performs multi-objective optimization by displaying the results of multiple solutions together with multiple evaluation items,
On the computer,
When displaying a solution by associating each of the plurality of evaluation items with one axis, the parameters used for calculating the solution are displayed on the axis, background of the graph, or each solution in the graph for displaying the solution. A multi-objective optimization program for executing processes that reflect the importance or constraints of the system.
On the computer,
If at least the axis, background of the graph or the display of each solution in the graph for which the solution is displayed reflect the importance or constraints of each evaluation item that are the parameters used in the solution calculation, the solution is displayed. A process of accepting a change by a user operation on a GUI component constituting a graph,
Processing for calculating a change value in a parameter used for solution calculation from information indicating the change content for the received GUI component and the content of the parameter used for calculation of each solution displayed in the solution display before the change is received ,
Processing to calculate the solution by the optimization calculation again using the calculated parameter after the change, and based on the parameter after the change, the solution is calculated for the axis, the background of the axis, or the display of each solution in the graph for displaying the solution. The multi-objective optimization program according to claim 15, wherein the process used to display the solution obtained by the optimization calculation is executed after reflecting the importance or constraint condition of each evaluation item.