JP2012174125A - Disaster prevention plan support system and program of the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a disaster prevention plan support system or the like capable of generating rules for performing the occurrence/non-occurrence evaluation of a disaster, an accident and repair corresponding to a risk degree rank, which is highly accurate and excellent in versatility with a natural disaster or an accident in a building structure or the like or repair as an object, in consideration of performing inspections and the repair in future.SOLUTION: The disaster prevention plan support system includes: a risk degree computing part 12 for computing a disaster occurrence risk degree of an observation/inspection object part using a boundary of occurrence/non-occurrence of a disaster analyzed by SVM; a representative data extraction part 13 for extracting representative data 20 using the risk degree; a rank setting part 14 for reconstructing a separation surface using the representative data 20 and setting the risk degree rank; a rough set generation part 26 for forming a rough set; and a rank-classified evaluation rule generation part 27 for extracting the rules of a disaster, an accident and repair for each risk degree rank from the rough set.

Description

  The present invention relates to earth and sand disasters that occur on slopes and mountain streams caused by rainfall, natural disasters such as disasters caused by earthquakes and volcanic activities, river disasters, and facilities and equipment such as roads, water and sewage systems, dams, ports, and sabos. Efficient and accurate with respect to the occurrence / non-occurrence of disasters / accidents / repairs for aging of buildings, bridges, steel towers, power plants, buildings, etc. and accidents and repairs associated with the above natural disasters In addition, it relates to a disaster prevention plan support system and program for supporting evaluation with excellent versatility.

In Japan, there is an urgent need for progress in disaster prevention projects to protect lives from natural disasters such as frequent sediment disasters. However, projects that target a huge number of sediment-related disaster risk points (hereinafter referred to as “dangerous places”) are not making progress as expected, and efforts are being made to improve the efficiency. On the other hand, if the risk of occurrence of a disaster at the target location (hereinafter referred to as “risk”) can be accurately grasped, it will be possible to examine the priority of the project, and it is considered to have a great effect on the above issues. It is done.
Natural disasters such as heavy rains, earthquakes, and volcanic activities have caused enormous damage to Japan every year. For example, about 70% of Japan's national land is mountainous, and there are many geologically vulnerable areas, and there are many steep landforms. It is one of the fatal natural disasters in Japan due to social conditions such as population growth in Japan and weather conditions such as typhoons and rainy seasons that are easily affected by heavy rains. However, precious lives have been lost and valuable assets have been destroyed. Currently, the maintenance rate due to hardware measures is as low as 25%.
The same applies to facilities and equipment such as roads, water and sewage systems, dams, and harbors, or aging of buildings such as bridges, steel towers, power plants, and buildings, and accidents and repairs associated with the natural disasters described above.
In response to such a situation, there are budgetary and time constraints to implement hardware countermeasures and repair / renewal work for all of these dangerous locations and facilities / equipment / buildings. The need for measures to cover delays in hard measures has been recognized, and there are points and places where natural disasters such as dangerous mountain streams that are widely scattered in the region are likely to occur, aging and natural disasters. The importance of strengthening prevention against occurrence of disasters, accidents and repairs by grasping the distribution of particularly dangerous parts and relatively safe parts among structures that may be damaged by accidents involved Yes.

Therefore, for example, research has been conducted to obtain highly accurate information by computer-processing data related to the occurrence or non-occurrence of actual disasters, accidents, and repairs, for example, to support planning of disaster prevention and repair business plans. Yes.
The present inventors have already disclosed Non-Patent Document 1 on a method for setting a breakage occurrence limit line, an evacuation reference line, and a warning reference line, which are discrimination boundaries between generated rain and non-occurrence rainfall, which are used for predicting the occurrence of landslide Announcing as shown.
In Non-Patent Document 1, there is a case where a radial basis function network excellent in non-linear discrimination (hereinafter abbreviated as RBFN) is used for the purpose of setting a high-precision generation limit line or the like without performing a linear approximation of complex natural phenomena. ), And proposes a method for setting a non-linear collapse critical rainfall line for each region. In the technology disclosed in Non-Patent Document 1, a non-linear collapse occurrence limit rainfall line is set by using RBFN and determining the optimum weight of the intermediate layer and the output layer using the learning function.
In Patent Document 1, “Disaster prevention comprehensive plan support system and its program” includes facilities such as natural disasters such as landslides on slopes and mountain streams caused by rainfall, roads, water and sewage systems, dams, ports, sabos, etc. Disaster prevention that can generate rules to efficiently and accurately evaluate occurrence / non-occurrence of disasters / accidents / repairs for accidents and repairs in buildings such as equipment, bridges, steel towers, power plants, buildings, etc. A comprehensive plan support system and its program are disclosed.
The disaster prevention comprehensive plan support system disclosed in Patent Document 1 uses separation plane data and observation / inspection data that form boundaries of occurrence / non-occurrence of disasters analyzed by a support vector machine (hereinafter referred to as SVM), Calculate the risk of occurrence of a disaster at the location subject to observation / inspection, extract representative data using this risk level, form a rough set using the representative data, and set rules for disasters, accidents, and repairs from this rough set. Extract.
In this way, combining SVMs and rough sets, natural disasters, roads and waterworks, dams, harbors, sabo and other facilities and equipment, bridges, steel towers, power plants, buildings and other aging accidents An invention is disclosed that can generate a rule for efficiently and accurately evaluating occurrence / non-occurrence of disaster / accident / repair for repair.
Furthermore, Non-Patent Document 2 created a rule for the occurrence / non-occurrence of debris flow by rough set using terrain / geological factors (hereinafter referred to as terrain factors) and rainfall factors. Research that pursues the factors is disclosed. This study explores objective disaster conditions by introducing mathematical methods.

JP 2010-129003 A

Kazumasa Kuramoto et al. 5: Research on the setting of the non-linear collapse occurrence limit rainfall line using the RBF network, No. 672 / VI-50, pp. 117-132, 2001.3 Masao Okamoto and 4 others: Research on important factors and rule extraction of sediment movement phenomenon by data mining using rough set, Journal of Sabo Society Vol. 54, no. 6, p. 4-15, 2002

However, the invention disclosed in Non-Patent Document 1 focuses on setting a disaster occurrence limit line, an evacuation reference line, and a warning reference line, and is grouped according to a specific region or a certain condition. In other regional groups, the objective was to objectively evaluate the extent to which the short-term rainfall index and long-term rainfall index reached the risk of disaster occurrence. Extremely speaking, the data of short-term and long-term rainfall index accumulated at the same point is input, and how much rain will cause a disaster based on the rain data accumulated at that point. Judgment was made.
Even though this is an objective and quantitative evaluation, it is possible to carry out individual specific evaluations for each region or group, but it is not a specific region but a general and universal common to all regions. There was a problem that it was difficult to evaluate.

  In addition, the invention disclosed in Patent Document 1 introduces the concept of rules (Rule) for evaluating the occurrence / non-occurrence of disasters / accidents / repairs by combining SVMs and rough sets, and priorities for disaster prevention and repair projects. Although it was possible to evaluate the occurrence / non-occurrence of disasters and accidents with high accuracy and high efficiency, the occurrence of accidents in facilities and buildings based on the occurrence / non-occurrence of disasters and aging・ The risk of non-occurrence is not simply separated into two categories, “danger” and “safety”, but also based on the degree of danger or safety, considering the development of future inspection and repair business. If you divide the rules and find the rules, the accuracy will be higher, and it will be easier to use when doing business. For this reason, the invention disclosed in Patent Document 1 that uniformly divides the document into two has raised the problem of lack of versatility, although accuracy and efficiency can be increased.

Further, in Non-Patent Document 2, there is a problem that a very diverse rule is created because it is intended to explain the occurrence / non-occurrence of as many disasters as possible only by the analysis result of the rough set. This is presumed that the influence of noisy data included in the population data (data in which a disaster has occurred / not occurred due to local and special conditions) is large.
If the rules are too diverse, it will be difficult to perform the evaluation efficiently even if the accuracy is high. In other words, the number of points and places that can be covered by the rules are reduced, and it becomes difficult to evaluate the occurrence / non-occurrence of disasters. That is, there has been a problem that it is impossible to carry out highly accurate and efficient evaluation on occurrence / non-occurrence of disasters.
In this application, “rule” means a combination of category sections for each factor in the rough set, and “rough set” means a category section for attribute values related to this factor for disaster / accident / repair occurrence factors. This is a set obtained by combining the category intervals for each factor to form a model and substituting observation and inspection data into this model, and is not completely separated by the subset formed by the combination of category intervals Means a set. In addition, in this application, “category section” refers to attribute values of factors causing disasters, accidents, repairs, such as observed values or inspection values such as rainfall, basin average slope, basin length, age, degree of damage, etc. And that section divided as a section. For example, when the factor is a basin average gradient, three ranges of 10 ° to 20 °, 20 ° to 30 °, and 30 ° to 40 ° are considered as examples of category sections. This also applies to east, west, south, north, etc. that do not have a physical order such as mountain stream direction.

  The present invention has been made in response to such a conventional situation, such as a landslide disaster that occurs on a slope or a mountain stream caused by rainfall, a natural disaster such as a disaster or a river disaster caused by an earthquake or a volcanic activity, and a road. Efficient and highly accurate for facilities and equipment such as water and sewage systems, dams, harbors, sabos, bridges, steel towers, power plants, buildings, etc. and accidents and repairs associated with the above natural disasters In addition, it is highly versatile, and in consideration of future inspections and repairs, rules (Rules) can be generated to evaluate the occurrence / non-occurrence of disasters / accidents / repairs according to the risk rank. The purpose is to provide a disaster prevention plan support system and its program.

In order to achieve the above object, the disaster prevention plan support system according to the first aspect of the present invention provides an occurrence factor of disaster / accident / repair (where the number of factors is n. N ≧ 2) at the observation / inspection target location, n N-dimensional observation / inspection data observed and inspected for each cause of disaster / accident / repair corresponding to the coordinate axes constituting the three-dimensional coordinate space is used as learning data, and disaster / accident / Separation plane that uses the occurrence of non-occurrence of repairs as a teacher value and analyzes the possibility of occurrence of non-occurrence of disasters / accidents / repairs using SVM to form the boundary of occurrence / non-occurrence of disasters / accidents / repairs And the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair observed / inspected at the location to be observed / inspected, and the n-dimensional observation / inspection data indicated by the n-dimensional observation / inspection data Locus in coordinate space A distance from a point to the separation plane, the distance that can be taken as either a positive value or a negative value when the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane A risk level calculation unit that calculates the risk of occurrence of disasters, accidents, and repairs at locations subject to observation and inspection;
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
And outputting the extracted rules for plan evaluation and support for disaster / accident prevention or repair selection.
In this application, in order to calculate the n-dimensional observation / inspection data observed and inspected for each cause of disaster, accident, and repair as learning data for analyzing the separation surface, and the distance to the separation surface The n-dimensional observation / inspection data for each factor may be the same or different. If they are different, they may be observed / inspected at the same point / location at different times, or may be observed / checked at different locations / locations. Further, in this application, “threshold” and “condition” are separately described, but this means that “threshold” is a concept that is limited only by numerical values, and “condition” is, for example, This is because the concept of including a logical expression including a threshold is also considered.

The disaster prevention plan support system according to claim 2 is a coordinate axis that constitutes an n-dimensional coordinate space for occurrence factors of disasters, accidents, and repairs (the number of factors is n. N ≧ 2) at the observation / inspection target locations. N-dimensional observation / inspection data observed and inspected for each cause of disaster / accident / repair is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection Use as a teacher value, analyze the possibility of occurrence / non-occurrence of disasters / accidents / repairs using SVM, and form the boundary of occurrence / non-occurrence of disasters / accidents / repairs, and the objects to be observed and inspected Coordinate points in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data using the n-dimensional observation / inspection data for each cause of the disaster, accident, or repair observed / inspected at a location To the separation surface The distance at which the coordinate point is located on the origin side of the n-dimensional coordinate space with respect to the separation plane can be taken as a positive value or a negative value as a disaster at the observation / inspection location.・ A risk calculator that calculates the accident / repair occurrence risk,
Based on a desired threshold for the disaster / accident / repair occurrence risk or a logical expression including the threshold, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster A representative data extraction unit for extracting, as representative data, n-dimensional observation / inspection data for each disaster / accident / repair occurrence factor corresponding to the accident / repair occurrence risk;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using SVM.・ A separation plane calculation unit that analyzes the representative data separation plane that forms the boundary between occurrence and non-occurrence of accidents and repairs,
Using the representative data separation surface analyzed by this separation surface calculation unit and the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair observed / inspected at the observation / inspection target location, The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, the coordinate point being on the origin side of the n-dimensional coordinate space with respect to the separation plane The risk level calculation unit that calculates the distance that can be taken as either a positive value or a negative value as a representative data risk level of the disaster / accident / repair occurrence of the observation / inspection point;
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence representative data risk level;
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
And outputting the extracted rules for plan evaluation and support for disaster / accident prevention or repair selection.

In addition, the disaster prevention plan support system according to the invention described in claim 3 is configured to detect occurrence factors of disasters / accidents / repairs (where the number of factors is n. N ≧ 2) in an observation / inspection location, and an n-dimensional coordinate space. The n-dimensional observation / inspection data observed and inspected for each cause of the disaster, accident, and repair corresponding to the coordinate axes constituting the learning data is used as learning data, and the occurrence of disaster, accident, and repair at the location subject to observation and inspection Separation plane calculation unit that uses non-occurrence as a teacher value and analyzes the separation plane that forms the boundary of occurrence of disaster / accident / repair occurrence / non-occurrence using SVM. When,
Using the separation surface analyzed by the separation surface calculation unit and the n-dimensional observation / inspection data for each factor of occurrence of the disaster, accident, and repair observed / inspected at the observation / inspection location, the n The distance from the coordinate point in the n-dimensional coordinate space indicated by the three-dimensional observation / inspection data to the separation plane, and the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane A risk calculation unit that calculates the distance that can be taken as either a positive value or a negative value as the risk of occurrence of a disaster, accident, or repair of the observation / inspection location,
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
And outputting the extracted rules for plan evaluation and support for disaster / accident prevention or repair selection.

In the disaster prevention plan support system according to claim 4, disaster / accident / repair occurrence factors (the number of factors is assumed to be n. N ≧ 2) in the observation / inspection locations correspond to coordinate axes constituting the n-dimensional coordinate space. N-dimensional observation / inspection data observed and inspected for each cause of occurrence of the disaster / accident / repair is used as learning data, and the occurrence / non-occurrence of disaster / accident / repair at the location subject to observation / inspection is a teacher value. A separation plane calculation unit that analyzes the separation plane that forms the boundary of occurrence / non-occurrence of disaster / accident / repair using SVM, and the possibility of occurrence / non-occurrence of disaster / accident / repair,
Using the separation surface analyzed by the separation surface calculation unit and the n-dimensional observation / inspection data for each factor of occurrence of the disaster, accident, and repair observed / inspected at the observation / inspection location, the n The distance from the coordinate point in the n-dimensional coordinate space indicated by the three-dimensional observation / inspection data to the separation plane, and the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane A risk calculation unit that calculates the distance that can be taken as either a positive value or a negative value as the risk of occurrence of a disaster, accident, or repair of the observation / inspection location,
Based on a desired threshold for the disaster / accident / repair occurrence risk or a logical expression including the threshold, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster A representative data extraction unit for extracting, as representative data, n-dimensional observation / inspection data for each disaster / accident / repair occurrence factor corresponding to the accident / repair occurrence risk;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using SVM. -The separation surface calculation unit for analyzing the representative data separation surface that forms the boundary of occurrence / non-occurrence of accidents / repairs;
Using the representative data separation surface analyzed by this separation surface calculation unit and the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair observed / inspected at the observation / inspection target location, The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, the coordinate point being on the origin side of the n-dimensional coordinate space with respect to the separation plane The risk level calculation unit that calculates the distance that can be taken as either a positive value or a negative value as a representative data risk level of the disaster / accident / repair occurrence of the observation / inspection point;
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence representative data risk level;
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
And outputting the extracted rules for plan evaluation and support for disaster / accident prevention or repair selection.

The disaster prevention plan support program according to claim 5 is a program executed by a computer to generate and output rules for occurrence / non-occurrence of disaster / accident / repair,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. The separation plane that analyzes the possibility of occurrence / non-occurrence using SVM to form the boundary of occurrence / non-occurrence of disaster / accident / repair, and each factor observed / inspected at the location to be observed / inspected n-dimensional observation / inspection data, and the distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane (provided that the coordinate point Against the separation surface The case where it is on the origin side of the n-dimensional coordinate space may be positive, or the case where it is on the origin side may be negative. Risk level calculation process,
A rank setting process for ranking n-dimensional observation / inspection data for each factor of occurrence of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
Is used for plan evaluation and support such as disaster / accident prevention or repair selection.

The disaster prevention plan support program according to claim 6 is a program executed by a computer to generate and output rules for occurrence / non-occurrence of disaster / accident / repair,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. The separation plane that analyzes the possibility of occurrence / non-occurrence using SVM to form the boundary of occurrence / non-occurrence of disaster / accident / repair, and each factor observed / inspected at the location to be observed / inspected n-dimensional observation / inspection data, and the distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane (provided that the coordinate point Against the separation surface The case where it is on the origin side of the n-dimensional coordinate space may be positive, or the case where it is on the origin side may be negative.) Is calculated as the disaster / accident / repair occurrence risk of the observation / inspection target location A first risk level calculating step,
Based on a desired threshold value or condition for the disaster / accident / repair occurrence risk, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster / accident / repair occurrence risk is extracted. A representative data extraction step of extracting n-dimensional observation / inspection data for each factor corresponding to the above as representative data;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using SVM. -Separation plane calculation process to analyze the representative data separation plane that forms the boundary of occurrence / non-occurrence of accidents / repairs,
Using the representative data separation surface analyzed in this separation surface calculation process, and the n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair observed / inspected in the observation / inspection location, The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, the coordinate point being on the origin side of the n-dimensional coordinate space with respect to the separation plane A second risk calculation step for calculating a distance that can be taken as either a positive value or a negative value as a representative data risk of occurrence of disaster / accident / repair at the location subject to observation / inspection;
Rank setting step for ranking the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the risk level of the disaster / accident / repair occurrence representative data;
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
Is used for plan evaluation and support such as disaster / accident prevention or repair selection.

The disaster prevention plan support program according to claim 7 is a program executed by a computer to generate and output rules for occurrence / non-occurrence of disaster / accident / repair,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. Separation plane calculation process to analyze the separation plane that forms the boundary of occurrence / non-occurrence of disaster / accident / repair using SVM for the possibility of occurrence / non-occurrence,
Using the separation surface analyzed in the separation surface calculation step and the n-dimensional observation / inspection data for each factor observed / inspected at the observation / inspection target location, the n-dimensional observation / inspection data is The distance from the coordinate point in the n-dimensional coordinate space to the separation plane (provided that the distance is positive when the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane) Or may be negative if it is on the origin side), a risk degree calculation step for calculating the disaster / accident / repair occurrence risk of the observation / inspection target part,
A rank setting process for ranking n-dimensional observation / inspection data for each factor of occurrence of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
Is used for plan evaluation and support such as disaster / accident prevention or repair selection.

The disaster prevention plan support program according to claim 8 is a program executed by a computer to generate and output rules for occurrence / non-occurrence of disaster / accident / repair,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. A first separation surface calculation step for analyzing a separation surface that forms a boundary of occurrence / non-occurrence of disaster / accident / repair using SVM for occurrence / non-occurrence;
Using the separation surface analyzed in the first separation surface calculation step and the n-dimensional observation / inspection data for each factor observed / inspected at the observation / inspection target location, the n-dimensional observation / inspection The distance from the coordinate point in the n-dimensional coordinate space indicated by the inspection data to the separation surface (however, this distance is the case where the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation surface). A first risk degree calculating step for calculating as a disaster / accident / repair occurrence risk of the observation / inspection target portion;
Based on a desired threshold value or condition for the disaster / accident / repair occurrence risk, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster / accident / repair occurrence risk is extracted. A representative data extraction step of extracting n-dimensional observation / inspection data for each factor corresponding to the above as representative data;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using SVM. A second separation plane calculation process for analyzing the representative data separation plane that forms the boundary of occurrence / non-occurrence of accidents / repairs;
Representative data separation surface analyzed in the second separation surface calculation step, n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair observed / inspected in the observation / inspection location; , The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, and the coordinate point is defined in the n-dimensional coordinate space with respect to the separation plane. A second risk calculation step of calculating a distance that can be taken as either a positive value or a negative value when it is on the origin side as a disaster / accident / repair occurrence representative data risk level of the observation / inspection target part;
Rank setting step for ranking the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the risk level of the disaster / accident / repair occurrence representative data;
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
Is used for plan evaluation and support such as disaster / accident prevention or repair selection.

The present invention does not use the observation / inspection data obtained at disaster risk points or facilities / equipment / buildings as a rough set as it is to evaluate the occurrence / non-occurrence of disasters / accidents / repairs as a preliminary stage.・ Risk level is set using the separation surface of inspection data and SVM analysis, then rough set is generated, and evaluation rule of occurrence / non-occurrence of disaster / accident / repair according to the rank of previous risk Therefore, it is possible to set a highly versatile evaluation rule according to the risk rank.
Furthermore, as a previous step, a part of the observation / inspection data is objectively extracted from the observation / inspection data and the SVM analysis as representative data, and the SVM analysis is performed again using the representative data. An invention for constructing a separation plane has also been created.
In the case of this invention, based on the separation plane by the representative data, the risk level is set for the observation / inspection data, and then a rough set is generated, and the disaster / accident / In order to set evaluation rules for the occurrence / non-occurrence of repairs, it is possible to set objective and highly accurate evaluation rules by deleting noise data contained in observation / inspection data, and at the same time, the risk rank A highly versatile evaluation rule can be set.
By setting a highly accurate and versatile evaluation rule according to the risk rank and outputting it, observation / inspection data of any hazardous location using this evaluation rule corresponds to any evaluation rule It is possible to set the risk level based on the corresponding evaluation rule, and it can be used for the evaluation of the necessity of repair and the occurrence / non-occurrence of disasters / accidents. It is possible to prioritize and support the planning of efficient disaster prevention plans.

Hereinafter, in the disaster prevention plan support system according to the embodiment of the present invention, SVM is used for extraction of representative data and risk setting for generation of disaster / accident / repair rules, and disaster / accident / Rough sets were used to generate repair rules. Therefore, before describing the disaster prevention plan support system according to the present embodiment, first, the SVM and the rough set will be described.
SVM is a technique proposed by Vapnik et al. In 1995 and is currently attracting attention as the most powerful pattern classification technique. In SVM, a data group that cannot be linearly separated is mapped to a high-dimensional feature space using a non-linear function, and a function of clearly separating the data group by making it linearly separable. At this time, a surface that separates data in the high-dimensional feature space is called a separation hyperplane. The concept of this separation hyperplane is shown in FIG. In the present application, this separation hyperplane is simply referred to as a separation surface.
Using this function, the degree of risk can be evaluated by calculating the distance between each data and the separation plane that determines the occurrence / non-occurrence of disasters / accidents / repairs. FIG. 2 shows the concept of calculating the distance between the separation plane and each data and evaluating the risk level. In this application, this risk assessment method is applied to represent the original observation / inspection data in determining the disaster / accident / repair rules (representing data with higher risk as disaster / accident / repair occurrence data) In the present embodiment, the case where the operation of extracting non-occurrence data (which has a higher degree of safety) is performed and the case where the operation is not performed are invented. In the embodiment, the case where representative data is extracted is mainly described.
Although it may not be necessary to explain the distance between the data point and the separation surface again, a perpendicular is drawn from the coordinates of the data point to the separation surface, and the length is defined as the distance. However, in FIG. 2, since the dangerous side is a positive direction and the distance (f (x)) is calculated from the separation plane toward the data point, that is, as a vector, the data point is the separation plane. The distance increases as a negative value as it moves away from the positive direction, and conversely as the data point moves away from the separation surface in the negative direction (closer to the coordinate origin). In this embodiment, although such a coordinate is adopted, on the contrary, when the safe side is a positive direction as a coordinate, the negative value is safer and the positive value is more positive. Needless to say, it can be dangerous.

Next, the rough set will be described. FIG. 3 shows a conceptual diagram of the rough set. Three categories are provided for the entire set using two factors (factor 1 and factor 2), and a model is formed in which each combination is considered to be divided into nine regions. As shown in FIG. 3, data relating to the occurrence / non-occurrence of disasters / accidents / repairs is plotted for this model. In this way, for example, category sections are set based on factors related to the occurrence / non-occurrence of disasters, accidents, and repairs, and the area divided into a matrix and the data plotted in the area are collectively called a “rough set”. .
In the rough set shown in FIG. 3, the subsets are not completely separated even if divided into nine regions as described above. Such a set that is not completely separated is called a rough set. When a set is divided by these factors, data in which all data in the same area is the same type is called consistent data (data in the hatched portion in FIG. 3), and mixed data is called contradictory data. When a set is divided by these factors, an area in which all data in the same area are the same type can be considered as a rule. In FIG. 3, rules A and B exist as occurrence rules, and rules C to F exist as non-occurrence rules.
Note that the degree of consistency defined by the following equation (1) is used as an index for evaluating a combination of factors used for setting a rule.

The degree of consistency is determined by replacing the category section name described above (for example, indicating the serial number assigned to each category section) as an attribute value, and determining the occurrence / non-occurrence of disaster / accident / repair attribute value (for example, (Indicates that the value is 1 when disaster / accident / repair occurs, and 0 when non-occurrence occurs)), individual data is classified, areas with inconsistent data with different decision attribute values are inconsistent areas, others are consistent areas As a result, the degree of matching is used as an index for evaluating the combination of factors in the region.
In addition, as an index for evaluating the accuracy and versatility of the set rule, the certainty factor and the support shown in equations (2) and (3) are used, respectively.

  For example, in FIG. 3, when the category divisions of factor 1 and factor 2 are 2 and 1, respectively, that is, the region of the lower middle segment is a non-occurrence rule, and the certainty in this region is 100% (3/3) Support is 14% (3/22). Therefore, all the locations that conform to this rule are composed only of data that does not occur, which means that accuracy is 100%, and the proportion of all data is 14%.

Next, explanation will be given on the expansion of the rule area.
When using rough sets for analysis of natural phenomena such as earth and sand disasters and deterioration of buildings, each of the created rules can only be explained in a small range and only a part of the database can be explained. It is often difficult to say. In such a case, it is effective to expand the rule area.
For example, when the hatching area shown in FIG. 4A is an occurrence rule, the concept of extending the rule area is that an area including a subset that is considered to be more dangerous as shown in FIG. By doing so (the concept of “above” is given to the category division), the number of occurrence points included in the rule increases, and the rule becomes versatile. The same applies to the non-occurrence rule, but in this case, the concept of “below” is given to the category division. Also in this application, it is possible to create a more versatile rule by using the extension of the rule area for the representative data extracted by the SVM.
In the above, SVM and rough set which are important concepts in describing the disaster prevention plan support system according to the embodiment of the present invention have been described.

Next, the disaster prevention plan support system according to the embodiment of the present invention will be described with reference to FIGS.
FIG. 5 is a configuration diagram of the disaster prevention plan support system, and FIG. 6 is a flowchart showing calculation steps executed in the disaster prevention plan support system.
In FIG. 5, the disaster prevention plan support system 1 includes an input unit 2, an SVM analysis unit 3, a rough set analysis unit 6, an analysis database 4, a database 5, and an output unit 7.
The input unit 2 stores the data shown in the analysis database 4 and the database 5 in advance, including the observation / inspection data 8 and the analysis condition data 9 stored in the analysis database 4 and the database 5, and stores them in a readable manner. Used to keep. Further, it is also used to directly input observation / inspection data 8 and analysis condition data 9 to the SVM analysis unit 3. Specific examples of the input unit 2 include a plurality of types of devices such as a keyboard, a mouse, a pen tablet, an optical reading device, or a receiving device that receives data from an analysis device such as a computer or a measurement device via a communication line. An apparatus that can be used properly according to the purpose is conceivable.
Further, as the output unit 7, specifically, a display device such as a CRT, liquid crystal, plasma or organic EL, a display device such as a printer device, and a transmitter such as a transmitter for transmitting to an external device, etc. Can be considered.
The SVM analysis unit 3 includes an analysis condition setting unit 10, a separation plane calculation unit 11, a risk level calculation unit 12, a representative data extraction unit 13, and a rank setting unit 14.
The analysis condition setting unit 10 sets analysis conditions for executing the analysis of the SVM. The analysis condition setting unit 10 reads the analysis condition data 9 directly input from the input unit 2 and sets the analysis condition data 9 as an analysis condition. The analysis condition data 9 stored in the analysis database 4 from the input unit 2 can be read from the analysis database 4 and set.
As an analysis condition in the SVM analysis, for example, in the case of the soft margin method described later, it is necessary to set two parameters of C, which indicates the degree to which misclassification is allowed, and r, which is the radius of Gaussian kernel (data influence degree). Yes, these are included as analysis conditions. Alternatively, in the calculation of the risk level described later, a threshold value for extracting representative data or a combination of threshold values, or a combination of these threshold values and a logical expression is included as an analysis condition. It is. In other words, the analysis condition is necessary in addition to the observation / inspection data 8, the separation surface function data 15 constituting the separation surface, or the risk function data 16 for executing the risk calculation when executing the SVM analysis. The analysis condition setting unit 10 is an element for setting these parameters in the SVM analysis.

As shown in step S1 of FIG. 6, the separation plane calculation unit 11 of the SVM analysis unit 3 generates / does not generate disasters / accidents / repairs in disaster / risk points and observation / inspection objects such as facilities / equipment / buildings. The data observed and inspected with respect to the observation / inspection as learning data 17 and the occurrence / non-occurrence of disaster / accident / repair at the observation / inspection location as teacher values are input to the separation plane function data 15 respectively. A separation plane that separates the possibility of occurrence of repair is calculated, and the separation plane data 18 obtained by the calculation is stored in the database 5 so as to be readable.
The learning data 17 is data observed and inspected in the observation / inspection object, but the contents of the data are numerical values observed and inspected for factors related to occurrence / non-occurrence of disasters / accidents / repairs. In the learning data 17 and the observation / inspection data 8, this coordinate axis is conceptualized as a dimension corresponding to the number of those factors (n, where n ≧ 2) and a multi-dimensional coordinate axis corresponding to that dimension. The learning data 17 and the observation / inspection data 8 are plotted in the coordinate space formed by, and the distance to the separation surface, which is also conceptualized in the n-dimensional space, is calculated by the risk calculating unit 12 described later. .
As described above, the learning data 17 and the observation / inspection data 8 may be the same data, or different data, that is, data obtained at different times with the same observed / inspected points / locations or different points / Any of the data obtained by observation / inspection at the location may be used.

In this embodiment, the separation surface function data 15 is stored in advance in the analysis database 4 and the learning data 17 is stored in the database 5 in advance, and the separation surface calculation unit 11 reads out and uses them. Even if it is not stored in the database 4 or the database 5, it may be input via the input unit 2 when the separation plane calculation unit 11 executes the analysis. Hereinafter, also in the case of using data stored in the analysis database 4 or the database 5 in the present embodiment, the data may be read out from these databases and used, or the input unit 2 may be externally provided as necessary. You may make it input via.
As shown in step S <b> 2 in FIG. 6, the risk level calculation unit 12 is based on a separation plane that is obtained as a result of calculation by the separation plane calculation unit 11 or that is configured by the separation plane data 18 stored in the database 5 in advance. The distance to the observation / inspection data 8 is calculated. More specifically, the separation plane data 18 and the observation / inspection data 8 are input to the risk function data 16 that is a function for calculating the distance, and the separation plane and the observation / inspection data 8 in the n-dimensional coordinate space. The distance from the plot point is calculated. This distance is stored in the database 5 so as to be readable as the risk data 19.
The concept of this distance is as described above with reference to FIG. Data regarding the function expressed as f (x) in FIG. 2 is shown as the risk function data 16 in FIG. Then, using this distance as a risk level, the risk level for the observation / inspection data 8 is determined. In the present application, for the sake of convenience, a risk index is used as a risk index, but it goes without saying that the safety index may be used as a safety index.

Next, as shown in step S3 in FIG. 6, the representative data extraction unit 13 sets the risk level set as desired with respect to the risk level (risk level data 19) calculated by the risk level calculation unit 12. A risk level is selected based on a threshold value or a condition, and observation / inspection data 8 corresponding to the selected risk level is extracted as representative data 20 from the database 5. The threshold value or condition for the degree of risk is stored in the analysis database 4 in advance as the analysis condition data 9 and read and used by the representative data extraction unit 13, or the representative data extraction unit 13 via the input unit 2 at the time of analysis It is good to read it.
The threshold value or condition for the risk level for extracting the representative data 20 will be described in more detail later.
The extracted representative data 20 is stored in the database 5 so as to be readable.
Next, the calculation is performed again by the separation surface calculation unit 11. As shown in step S <b> 4 in FIG. 6, the separation surface calculation unit 11 reads representative data 20 extracted using the risk data 19 from the database 5 and assigns it to the separation surface function data 15 as learning data. By doing so, the separation plane is calculated again. The representative data separation plane calculated again using the representative data is stored in the database 5 as readable data as representative data separation plane data 21.
Further, the risk level calculation unit 12 performs the calculation again. The risk level calculation unit 12 calculates the representative data risk level as shown in step S5 in FIG. In this calculation, the distance from the observation / inspection data 8 is calculated based on the separation plane obtained by the calculation result by the separation plane calculation unit 11 and constituted by the representative data separation plane data 21 stored in the database 5. . Specifically, the representative data separation plane data 21 and the observation / inspection data 8 are input to the risk function data 16 that is a function for calculating the distance, and the separation plane and the observation / inspection in the n-dimensional coordinate space are input. The distance from the plot point of data 8 is calculated. This distance is stored in the database 5 so as to be readable as representative data risk data 22.
At this time, as the observation / inspection data 8, the same data as the observation / inspection data 8 from which the representative data 20 is extracted may be used as described in the relationship between the learning data 17 and the observation / inspection data 8. Alternatively, different observation / inspection data 8 may be used. In other words, the data may be either data obtained at different times with the same observed / inspected points / locations or data obtained by observing / inspecting at different locations / locations. However, for the observation / inspection data as the final evaluation target, the risk rank is set using the separation plane set using the representative data 20, a rough set is generated, and an evaluation rule is generated. It is data.

Further, the rank setting unit 14 of the SVM analysis unit 3 reads the representative data risk data 22 from the database 5 as shown in Step S6 in FIG. The risk rank of observation / inspection data 8 to be set is set. The setting of the risk rank is determined based on a threshold for the representative data risk data 22 in the desired analysis condition data 9 read from the analysis database 4. The risk ranks for the set observation / inspection data 8 are stored in the database 5 so as to be read out as the risk rank data 23.
As a specific threshold value, the threshold value differs depending on how the risk rank is divided, that is, whether the rank is 4 rank or 6 rank. For example, in the case of 4 ranks, in addition to the two areas of safety and danger so far, “relatively safe” and “relatively dangerous” are added to give 4 ranks. In the SVM analysis, the distance from the observation / inspection data 8 to the separation surface is f (x), f (x) <-1 is “dangerous”, −1 ≦ f (x) <0 is “relatively dangerous”, 0 ≦ f (x) ≦ 1 is “relatively safe”, and −1 <f (x) is “safe”. Therefore, the threshold value is -1,0,1.
In the above description, the case where the representative data 20 is extracted has been described. For example, in the case where the observation / inspection data 8 is highly reliable and it is not necessary to delete the noise and extract the representative data 20, FIG. Instead of executing steps S3 to S5 in step 6, the risk rank 19 may be set in the rank setting unit 14 in step S6 after the risk level data 19 obtained in step S2.

The rough set generation unit 26 of the rough set analysis unit 6 reads the observation / inspection data 8 which is the original data of the representative data 20 to be evaluated from the database 5 as shown in step S7 in FIG. Generate. The observation / inspection data 8 may be the same as or different from the observation / inspection data 8 from which the representative data 20 is extracted, but here the observation / inspection data from which the representative data 20 is extracted. The description will be made assuming that the data is the same as the data 8.
Specifically, first, for an n-dimensional disaster / accident / repair occurrence factor corresponding to the observation / inspection data 8, a category section for an attribute value related to this factor is provided. This category section is as described when the rough set was explained earlier. This category section is set in advance for each disaster / accident / repair occurrence factor, and stored in the analysis database 4 as the analysis condition data 9 or output during analysis by the rough set generation unit 26. It is preferable to ask the user to ask about the setting of the category section via the unit 7 and to enable the input of the category section via the input unit 2. A model is formed by combining the category intervals for each factor. This is also as described for the rough set earlier. A combination of factors having a plurality of category sections is referred to as a model in the present application. Further, the contents of the observation / inspection data 8 which is the original data of the representative data 20 are substituted into this model to form a rough set and stored in the database 5 so as to be readable as the rough set data 24.

First, the operation and effect of extracting the representative data 20 to construct the separation plane, generating the rough set, and obtaining the rough set data 24 will be described with reference to FIGS.
FIG. 7A is a conceptual diagram of a separation plane constructed with data in a state before extracting representative data, and FIG. 7B is a conceptual diagram of a rule corresponding thereto. FIG. 8A is a conceptual diagram of a separation plane constructed with representative data, and FIG. 8B is a conceptual diagram of rules corresponding thereto. FIG. 9 is a conceptual diagram of representative data.
In the diagram shown in FIG. 7A, generated data and non-generated data plotted by two factors on the vertical axis and the horizontal axis indicated by a solid line are clearly separated into a dangerous side away from the origin and a safe side close to the origin. Therefore, the separation plane is calculated so as to sew between the data. This is because the separation plane is constructed with data that contains a lot of data (hereinafter referred to as noise data) in which disasters have occurred or are not occurring due to local or special conditions peculiar to the measurement point. It is.
When a rule is set using such a separation surface, the result is as shown in FIG. In FIG. 7B, the part surrounded by the arrows from the center to the right side and the upper side is shown as the risk rule set including the concept of expansion described with reference to FIG. The hatched part at the base of the arrow is the danger rule before expansion. Similarly, the part hatched in the lower right is the safety rule before expansion, and the part surrounded by two arrows extending from the safety rule to the left side and the lower side is the safety rule after expansion.
However, in the rule shown in FIG. 7B, the risk rule after expansion includes data evaluated as safe by the SVM, and conversely, the safety rule after expansion is also evaluated as dangerous by the SVM. Contradiction such as data is included.
Therefore, it was considered to construct the separation plane by extracting the representative data 20. When the data after extracting the representative data 20 is used, a separation plane as shown in FIG. 8A is set, and even when a rule is set, the rule set as the evaluation by SVM as shown in FIG. 8B. The contradiction has been resolved. By deleting noise data that causes such a contradiction, it is possible to provide a highly versatile rule that can be applied in as many dangerous locations as possible.
Therefore, how to eliminate noise data becomes important. Therefore, the threshold value for the degree of risk for extracting the representative data 20 left in the description of step S3 in FIG. 6 will be described with reference to FIG.
Since noise data has different phenomena (occurrence / non-occurrence) from surrounding data, a separation plane is constructed at the boundary between these data. Since these data are data close to the separation surface, the separation surface distance f (x) is | f (x) | ≦ 1. In the present application, using this characteristic, data satisfying | f (x) | ≦ 1 is deleted as noise data. That is, as the threshold value for the danger level, the separation surface distance is ± 1. FIG. 9 shows such a state. A portion (f (x) = 1, −1) having a separation surface distance of ± 1 is indicated by a dotted line with respect to the separation surface (f (x) = 0) indicated by the solid line. The area shown between the solid line and the dotted line is | f (x) | ≦ 1, and the data included in this area is deleted as noise data to extract representative data.
The extraction of representative data is effective not only to delete noise data but also to not disturb the ratio of generated / non-occurring data in the original database. Therefore, a threshold for these ratios can also be set as desired. A combination of these threshold values may be set as a condition by combining with a threshold value related to the distance of the separation surface.

The representative data 20 extracted in this way is generated data indicated by black squares and non-generated data indicated by white circles shown in FIGS. 8A and 8B, and rough set data 24 is In FIG. 8B, the representative data 20 is plotted as data in a region divided into a matrix by four category sections based on the factors on the vertical axis and the factors on the horizontal axis. ing.
In this description, since the rough set is described in a state where the representative data 20 is extracted, contradiction data does not exist in each model, but in the description of the rough set generation unit 26 in the previous step S7, Since the rough set is generated using the observation / inspection data 8 which is the original data, there is a possibility that inconsistent data is included.

Next, the rank-based evaluation rule generation unit 27 of the rough set analysis unit 6 uses the risk rank data 23 set by using the rank setting unit 14 of the SVM analysis unit 3 and the rough rank generated by the rough set generation unit 26. Using the aggregate data 24, rules regarding the occurrence / non-occurrence of disasters / accidents / repairs for each risk rank are generated, and the data is stored in the database 5 as readable evaluation rule data 25.
Specifically, the rank-specific evaluation rule generation unit 27 performs the category section of each factor included in the rough set model for each risk rank along the risk rank data 23 with respect to the rough set data 24. The certainty factor (the above-mentioned formula (2)) and the support (the above-mentioned formula (3)) in each rule which is a combination for each are calculated. And a factor satisfying either or both of the threshold value or condition set as desired for the certainty factor and / or the threshold value or condition set as desired for the support. Search and extract category segment combinations (rules). Accordingly, whether each rule is a single rule or a plurality of rules depends on the certainty factor and / or the support threshold value or condition. These certainty factors and / or support threshold values or conditions may be determined in advance and stored in the analysis database 4 as analysis condition data 9.
At that time, the above-mentioned rule expansion may be further executed. Also in this case, a certainty factor and / or support threshold value or condition for permitting the extension of the rule is determined in advance, stored as analysis condition data 9, and read during analysis. Good.
Alternatively, the certainty factor, the support, and the threshold value or conditions for this expansion may be input via the input unit 2 during the analysis. Of course, it is also possible to input via the input unit 2 when setting a rule.
When a combination (rule) of a factor that satisfies a certainty level and / or support threshold or condition and a category section is determined for each risk rank, the rank-specific evaluation rule generation unit 27 calculates the rank-specific evaluation rule data. 25 is stored in the database 5 so as to be readable.
The rules for each risk rank obtained in this way are output via the output unit 7. Therefore, it is possible to determine which evaluation rule the observation / inspection data of an arbitrary risk location corresponds to. Therefore, it is possible to set the risk level for the dangerous point based on the corresponding evaluation rule, and it can be used for evaluation of necessity of repair and occurrence / non-occurrence of disaster / accident.
For this reason, it is possible for users of the disaster prevention plan support system 1 according to the present embodiment to examine the risk of disasters / accidents and the priority of repairs, and support business plans related to them. It is.
FIG. 10 is a diagram schematically showing observation / inspection data 8 included in a rule set for each risk rank.
For the separation plane (f (x) = 0) indicated by the solid line, the upper right area from the plane indicated by the dotted line whose separation plane distance is -1 (f (x) =-1) is a dangerous area. The area from the plane indicated by the dotted line to the separation plane is a relatively dangerous area. Similarly, the lower left area from the plane indicated by the dotted line with the separation plane distance of +1 (f (x) = 1) is a safe area, and the area from the plane indicated by the dotted line to the separation plane is a relatively safe area. It is. An independent rule is applied to each of these four risk rank regions, and the observation / inspection data 8 is evaluated using the rule.
The significance of the technique according to the present embodiment shown in FIG. 10 is clarified by comparing with the state shown in FIG. 7B or FIG. 8B. That is, in the state shown in FIG. 7B, representative data is not extracted, and the rule is defined as a whole without setting the risk rank, so that contradictory data is included and the rule Are comprehensive and overall, the accuracy of evaluation of observation and inspection data is low. Further, in FIG. 8B, since the contradictory data is eliminated by extracting the representative data, the accuracy is improved, but the overall rule is set without setting the risk rank. This still affects the evaluation accuracy.
On the other hand, in the case shown in FIG. 10, a rule is set for each risk rank, and the evaluation area is subdivided. Therefore, the accuracy of evaluation for observation and inspection data is high. This contribution to the high evaluation accuracy is a contribution of factors that are technically different from the extraction of representative data. Therefore, adding the extraction of representative data can further improve the evaluation accuracy. is there.
In other words, the evaluation for each risk rank means that different category sections are set as rules depending on the rank difference, and moreover, by integrating the evaluation for each subdivided risk rank, It is possible to cover a highly accurate and broadly consistent area.

In addition to this, it goes without saying that each component included in the SVM analysis unit 3 and the rough set analysis unit 6 can output data that is calculated, extracted, generated, set, and the like to the output unit 7. Yes.
Further, in the present embodiment, the system invention has been described. However, considering the system shown in FIG. 5 as a general-purpose computer and executing the flowchart shown in FIG. In the explanation, the computer extracts the representative data 20 from the observation / inspection data 8 while executing each process, sets the risk rank based on the representative data 20, and performs disaster / accident / repair for each risk rank. It is established as a description of an embodiment of a program for generating a rule, and the operation according to the embodiment of this program, the effect according to the embodiment of the disaster prevention plan support system described earlier, It is the same as the effect.

Hereinafter, examples using specific data will be described.
In this embodiment, a debris flow disaster caused by the Nagasaki heavy rain on July 23, 1982 (hereinafter referred to as S57.7 Nagasaki heavy rain disaster) will be described (occurrence 150 mountain stream, non-occurrence 417 mountain stream). The topographic / geological factors and rainfall factors used in the explanation are the 18 factors shown in Table 1. Further, since the rough set cannot handle continuous value data, the continuous value data needs to be discretized into category values. Therefore, considering the maximum value and the minimum value of the data, the category division is determined based on five equally spaced divisions. Table 1 shows the category of each factor.

Using the above data, according to the procedure shown in the flowchart of FIG. 6, evaluation rules were set according to the disaster occurrence risk.
First, as analysis conditions in the SVM analysis, two parameters, C, which indicates the degree to which the soft margin method is allowed to be erroneously discriminated and r, which is the radius of Gaussian kernel (data influence degree), are set in advance. Since it is necessary to store them as analysis condition data 9 in FIG.
When performing an SVM analysis, a parameter study for performing an optimal analysis is required. In the SVM of the soft margin method, it is necessary to set two parameters of C which adjusts the degree of misclassification and r which is the radius of Gaussian kernel (data influence degree). In this embodiment, C = 5, 10, 50, 100, 200, 300, 400, 500, 8 cases, and r = 0.1, 0.5, 1, 2, 3, 4, 5, 7 cases, 56 in total. A parameter study was conducted in the case. In the parameter study, the accuracy of the separation plane constructed in each case (how exactly the data is classified) was verified by the hit rate defined by the following equation.

  Tables 2 and 3 show the results of the parameter study. Here, the parameter used for the risk setting by SVM is C in which the number of data serving as support vectors (data satisfying | f (x) | ≦ 1) is the smallest in the parameter set with a hit rate of 100.0%. = 500 and r = 4 (Tables 2 and 3). This was stored in the analysis database 4 as analysis condition data 9.

Next, regarding the extraction of representative data by the representative data extraction unit 13 described as step S3 in FIG. 6, the extraction of the representative data 20 is based on the f (x) value set using the previous parameters. To be implemented. Since the representative data 20 is data obtained by deleting the noise data close to the separation plane, the disaster occurrence data is more dangerous (f (x) value is smaller), and the non-occurrence data is safer Things (those with larger f (x) values) will be extracted. Here, the ratio of the representative data 20 to be extracted is set as described above because the method that does not disturb the ratio of generated / non-generated data in the original database is most effective as described above.
Focusing on the f (x) value calculated by the SVM analysis based on the above parameters, the number of generated data with the f (x) value being less than −1 was 32 (Table 3). Therefore, in order to maintain the relationship of the ratio of generated / non-occurrence data (150: 417) in all data, 87 non-occurrence data has a large f (x) among the data whose f (x) value is greater than 1. Was extracted as representative data.
That is, while selecting | f (x) | ≦ 1 as the selection threshold, the f (x) value is 1 from non-occurrence data so as to maintain the ratio of generated / non-occurrence data to all data. The representative data 20 is selected based on the condition that 87 pieces of data with larger f (x) are extracted as representative data. Usually, there are many non-occurrence data in occurrence data and non-occurrence data, so the condition to select when maintaining the ratio of occurrence / non-occurrence data is to select non-occurrence data after selecting the occurrence data first Of course, it may be set at the time, but the opposite is also possible.
Next, regarding the calculation of the representative data separation plane by the separation plane calculation unit 11 described as step S4 in FIG. 6, the accuracy of the separation plane constructed from the representative data 20 is shown in Table 4.
From the table, it was confirmed that the representative data 20 can be completely separated, and the hit rate for the observation / inspection data 8, which is the original data, is as high as 75.8%. Therefore, even when a separation plane constructed from the representative data 20 is used, it is considered that the risk level of the original data can be set with high accuracy.

Using the separation plane constructed from the representative data, the risk level was set for all the original data. Here, the relationship between the f (x) value calculated by SVM analysis and the degree of risk can be evaluated as dangerous if the f (x) value takes a negative value, and if the f (x) value takes a positive value. It can be evaluated as safe. In addition, since the data that becomes the support vector of | f (x) | ≦ 1 is a data group close to the separation plane, it is more dangerous than safety (data of −1 ≦ f (x) <0) or more dangerous. Can be considered as safe (0 ≦ f (x) ≦ + 1). Based on these, the teacher value in setting the evaluation rule according to the risk of disaster occurrence is “dangerous” when f (x) <− 1, “relatively dangerous” when −1 ≦ f (x) <0, and 0 ≦ f (x) ≤ + 1 is classified into four risk ranks, "relatively safe" and +1 <f (x) is classified as "safe".
FIG. 11 shows the relationship between the set risk rank and the occurrence rate of debris flow. From the figure, it was confirmed that the occurrence rate of debris flow increased as the risk rank increased, so the set risk rank is considered to be a reasonable rank that matches the actual disaster occurrence situation. In addition, it became clear that the risk set from the separation plane constructed using the representative data 20 is an effective system that can accurately represent the disaster occurrence situation in the original data.

Here, the meaning of setting a rule for each risk rank characteristic of the present invention will be described with reference to an embodiment.
In order to derive a good general-purpose and high-accuracy rule using a rough set as in the present invention, for example, in the case of a sediment disaster, a factor that includes a number of factors that are highly separable for the occurrence / non-occurrence of debris flows It is effective to set a rule using a combination of
In addition, considering that the rules to be set correspond to the risk rank divided into four rather than occurrence / non-occurrence, in addition to high separability against occurrence / non-occurrence of debris flow, it is divided into four It is necessary to select a factor with high separability for the given risk rank.
Therefore, first, selection of a factor having high separability will be described using another specific example. For example, 8 factors (high main stream length, basin length, basin width, stream azimuth, basin maximum slope, source head area, geological subclassification, effective rainfall) Since there is a half-life of 72 hr)), the separability with respect to the risk rank was verified. Here, the reason why the separability with respect to the risk rank is high is that both the ratio including danger and the ratio including relatively danger increase monotonously as the category value increases, as in the basin maximum slope shown in FIG. As a factor. As in the case of the maximum basin slope, FIG. 13 shows the factors that cause a monotonous increase in the proportion of danger, and FIG. 14 shows the factors that cause a monotonous increase in the proportion that contains danger. From the figure, it was confirmed that the four factors of basin length, basin width, basin maximum slope, and small geological classification are factors with high separability against the set risk rank.
In other words, by setting multiple risk ranks, it is possible to narrow down the factors with high separability from 8 factors to 4 factors and specify them (basin length, basin width, maximum basin slope, small geological classification). I understood it.

In the analysis in the invention using the rough set as in the present application, although the deterioration of the data quality can be prevented if the degree of matching is generally increased, the number of factors necessary for explaining the phenomenon increases, so the rule is determined as a result. It becomes complicated. On the other hand, if the degree of matching is lowered, the number of factors required to explain the phenomenon can be reduced, so that simplification of the rules can be expected.
In this embodiment, in order to prevent deterioration in data quality, the required level of consistency is set to 100.0%, and the combination of factors satisfying the required level is the minimum number of factors, and the risk rank Rule combination using a combination of many factors with high separability (steepest slope, basin width, basin shape ratio, stream direction, basin maximum slope, geological subclassification, effective rainfall (half-life 72 hours)) Set up.
In order to set a highly versatile rule, the rule area was expanded and the rules were combined based on the rule obtained from the analysis result of the rough set. Here, the required level of the extended rule used for the combination is only one having a certainty degree of 75.0% or more based on the accuracy of the separation plane used for setting the risk rank.

Table 5 shows the accuracy (hereinafter referred to as application accuracy) when the evaluation rule corresponding to the risk rank set using the analysis result of SVM is applied to the original database. Here, for the consistent data and contradictory data in the table, non-occurrence data is evaluated as consistent data, and the generated data is evaluated as contradictory data for the rule of “safe” or “relatively safe”.
On the other hand, for the “dangerous” or “relatively dangerous” rule, the generated data is evaluated as consistent data, and the non-occurrence data is evaluated as contradictory data. In the table, the category values of the factors that make up the rule indicate the maximum and minimum values included in each category. The notation "↓" and "↑" indicates the category value based on the concept of rule area expansion. The above means the above. In addition, “*” in the table means that any category value may be used. From Table 5, the set evaluation rules consist of seven factors: steepest bed slope, basin width, basin shape ratio, stream direction, basin maximum slope, subsurface geological classification, and effective rainfall (half-life 72 hours). . The accuracy of the rule with respect to the actual phenomenon is 78.7%, which is slightly higher than the accuracy (75.8%) of the separation surface used for setting the risk rank. In addition, since the support indicating the versatility of the rule is 100.0%, it is possible to explain all the cases of the S57.7 Nagasaki disaster with an accuracy of about 80% by using the rule. .
Focusing on the set rules, the characteristics of mountain streams that can be evaluated as “safe” and those that can be evaluated as “relatively safe” are relatively similar. A mountain stream that is characterized by the fact that it is a small-scale, north-facing mountain stream with andesitic tuff, pyroxene andesite, Tertiary formations, and lower plain diorite rocks, and has low long-term rainfall. Can be evaluated. Among them, the feature of “north-facing mountain stream” appears only in the rule of “safety”, and the feature of “mountain stream in which Tertiary layers and lower flat diorites are distributed” appears only in the rule of “relatively safe”. Therefore, it is considered to be an important index for classifying the risk level in low-risk mountain streams.
On the other hand, paying attention to the characteristics of mountain streams that are evaluated to be highly dangerous, these streams are "streams in areas where the bedrock gradient and basin maximum slope are steep, and where the amphibolite andesite is distributed." It has the characteristics. Among them, the feature of “area where amphibolite andesite is distributed” appears only in the “danger” rule, so it is considered to be an important indicator for classifying the danger level of high-risk mountain streams. . In addition, regarding the basin maximum slope and basin width, the “dangerous” rule is more dangerous than the “relatively dangerous” rule, and the difference between “relatively dangerous” and “dangerous” is clear. It appears in
Further, focusing on the relationship between the risk rank set by the evaluation rule and the occurrence rate of debris flow (FIG. 15), the occurrence rate of debris flow is as low as 12.5% at the location evaluated as “safe”, but the risk side The occurrence rate of debris flow has increased as the classification is made, and the occurrence rate of debris flow has risen to 87.5% at the dangerous place that is evaluated as the most dangerous.
Moreover, the graph which compared the occurrence rate of the debris flow based on the evaluation rule set by the method disclosed by patent document 1 with the occurrence rate of the debris flow based on the evaluation rule set in the disaster prevention plan support system which concerns on Example 1 is shown in FIG. Shown in If attention is paid to FIG. 16, in Example 1, the risk of a sediment disaster is divided into four categories of “safe”, “relatively safe”, “relatively dangerous”, and “dangerous”. It can also be confirmed that the consistency with respect to the occurrence / non-occurrence of actual disasters is improved compared to Patent Document 1 in which evaluation is performed in two categories of “danger” and “safety”. Therefore, the set evaluation rule is considered to be an effective rule that can grasp a risk rank with high consistency with respect to occurrence / non-occurrence of an actual disaster.
Furthermore, as can be understood from this embodiment, the rules are divided for each risk rank, and the factors constituting the rule may differ for each risk rank. It is clear that the excellent effect of clarifying the relationship with various factors is clearly demonstrated, and it can be said that this is one of the unique effects in the present invention.

Next, as Example 2, a case where the risk rank is changed from 4 ranks to 5 ranks in Example 1 will be described.
When considering 5 categories, “relatively safe” corresponding to 0 ≦ f (x) ≦ + 1 among the risk ranks in the evaluation rules of 4 categories is divided into two with f (x) = 0.5. F (x) <-1 is “dangerous”, −1 ≦ f (x) <0 is “relatively dangerous”, 0 ≦ f (x) ≦ + 0.5 is “relatively safe”, +0 .5 ≦ f (x) ≦ + 1 is classified into five safety ranks, “safety” and +1 <f (x) are classified as “especially safe”. FIG. 17 shows the relationship between the set risk rank and the occurrence rate of debris flow.
The evaluation rules were set according to the risk rank by the same procedure as that for setting the five evaluation rules. Table 6 shows the application accuracy of the set evaluation rules. Here, for the consistent data and contradictory data in the table, non-occurrence data is evaluated as consistent data and the generated data is evaluated as contradictory data for the rules on the safe side (particularly safe, safe, and relatively safe).
On the other hand, for the rules on the dangerous side (dangerous, relatively dangerous), the occurrence data is evaluated as consistent data, and the non-occurrence data is evaluated as contradiction data. Other notations in the table are the same as those in Table 5.

From Table 6, the five evaluation rules set are seven factors: steepest slope, basin width, valley depth ratio, stream direction, number of 0th valleys, maximum basin slope, and effective rainfall (half-life 72 hours). It is composed of The accuracy of the rule with respect to the actual phenomenon is 78.6%, which is almost the same as the evaluation rule of four categories. On the other hand, although the support indicating the versatility of the rule shows a high level of 98.1%, the result is slightly inferior to the support in the four-segment evaluation rule.
In this way, even if the rules set along the five risk ranks are used, as with the rules set along the four risk ranks, the risk of occurrence of disasters and accidents or the repair of buildings, etc. It is possible to quantitatively grasp whether it is necessary or not, and it is possible to provide support for determining the priorities for disaster prevention and repair projects with high accuracy and high versatility.

Next, as Example 3, an example of setting an evaluation rule relating to the necessity of repairing a structure using road inspection data of the Osaka Loop Line implemented in 2004 will be described.
There are nine factors related to the occurrence of repairs used in the study (bolt loosening, bolt deficiency, abnormal noise, drain pipe clogging, drain pipe damage, water leakage, water stop work, expansion / contraction body damage, rust / corrosion ) Is used. In addition, for the teacher value, the necessity of repair based on the judgment of a professional engineer is used.
In Examples 1 and 2 for landslide disasters, representative data excluding noise data was used because data containing a lot of noise that caused and did not occur due to local and special conditions was used. Although the risk level was set based on the separation plane constructed from the above, the target data in Example 3 is the data that the expert engineer determined whether or not repairs were necessary based on the inspection results. It is considered that there is very little data. Therefore, in the third embodiment, the risk level is set based on the separation plane constructed using all the target data.
In this case, with reference to FIG. 6, the risk level data 19 obtained in step S2 by omitting steps S3 to S5 is used to set the risk rank using the rank setting unit 14 in step S6. It corresponds to.
In setting the risk level using the SVM, as in the first embodiment, two parameters of C, which adjusts the degree of misclassification, and r, which is the radius of Gaussian kernel (data influence level), were examined. . Tables 7 and 8 show the results of the parameter study. From the table, it was confirmed that there were 12 combinations in which the number of data serving as support vectors (data satisfying | f (x) | ≦ 1) was the smallest in the parameter set with a hit rate of 100.0%. Here, since r is a parameter related to the influence degree of data, generally, a larger separation value can be constructed by using a larger value for r. On the other hand, C is a parameter that adjusts the degree of misclassification. Therefore, when a large value is adopted, a separation surface with less versatility that is closer to complete separation is constructed. Therefore, it is desirable to adopt a smaller value for C. Based on these, C = 50 and r = 5 were used in this example.
Using the separation plane constructed using the parameters, the risk level was set for all the original data. Here, the risk rank corresponding to the set risk is f (x) <-1 is “danger”, −1 ≦ f (x) <0 is “relatively dangerous”, and 0 ≦ The risk rank set by SVM and the necessity of repairs are determined by dividing f (x) ≤ +1 into four risk ranks of “relatively safe” and +1 <f (x) as “safe”. The relationship is shown in FIG. From the figure, it was confirmed that the repair location rate (the ratio of data that requires repair out of the corresponding data) increased as the risk rank increased, so the risk rank that was set was the actual repair It is considered that the rank is consistent with the necessity of

Evaluation rules were set according to such a risk rank. Here, in setting the rule, as shown in FIG. 18, since there is no data in which the actual phenomenon and the set risk rank contradict each other, all target data is used.
As a factor used for setting a rule, a combination of factors satisfying a matching degree of 100.0% is used. As a result of examining the consistency using the target data, there was only one combination of factors satisfying 100.0%.
Therefore, in this Example 3, the combination was not evaluated by examining the high separability factor, but the combination of the factor was adopted, and the rule was set by the analysis using the rough set. Here, the required level of the extended rule used for the combination is 100.0% confidence level for the “dangerous” and “safety” rules, and 80% confidence level for the “relatively dangerous” and “relatively safe” rules. Only 0% or more.
Table 9 shows the evaluation rules corresponding to the risk ranks set using the analysis results of SVM and the application accuracy thereof. Here, consistent data and contradictory data in the table are consistent data, and there is a necessity for repair for data that is considered not to be repaired for the time being against the rules of “safe” or “relatively safe”. Data that is supposed to be present is evaluated as contradictory data.
On the other hand, for “dangerous” or “relatively dangerous” rules, data that is considered to be in need of repair is evaluated as consistent data, and data that is not in need of repair for the time being is evaluated as inconsistent data. is doing.

From Table 9, the set evaluation rules are composed of six factors: bolt defects, abnormal noise, drainage dredge, that is, water leakage, water stoppage, and others (rust / corrosion).
The accuracy of the rule with respect to the actual phenomenon is 97.5%, which is a rule with sufficiently high accuracy although it is slightly inferior to the accuracy (100.0%) of the separation surface used for setting the risk rank. I can confirm that. Further, since the support indicating the versatility of the rule is 100.0%, it can be evaluated that the versatility of the rule is very high.
Focusing on the contents of the set rules, the features of the portion that can be evaluated as “safe” are the contents consistent with the actual phenomenon “no abnormality in various items”.
On the other hand, focusing on the content of rules other than “safety”, it can be confirmed that water leakage and water stoppage are very important evaluation indexes. If it is determined that the waterstop is B or C, it can be evaluated as relatively safe, so repair is not necessary for the time being. However, if the waterstop is A, the need for repair arises. It is considered that the need for repair is further increased when the “water leakage is determined as B or A” even in the location where “the water stop is determined as A” because it is in a very dangerous state.
Also, paying attention to the relationship between the risk rank set according to the evaluation rule and the necessity of repair (FIG. 19), the location evaluated as “safe” does not include the location that requires repair, but the risk side Since it can be confirmed that the proportion of places that require repair as they are classified, the set evaluation rule is considered to be a reasonable result.
Based on the above results, the evaluation rules set using six factors such as bolt failure, abnormal noise, drainage dredge, that is, water leakage, water stoppage, and others (rust / corrosion), and for each risk rank By evaluating the necessity of repair with different evaluation rules, without using these evaluation rules, the distance from the separation plane is simply used as the risk rank and the necessity of repair is the same. This means that high evaluation accuracy can be achieved.

As described above, as described in the present embodiment and examples, the risk rank by SVM is set, the evaluation rule according to the disaster occurrence risk rank using the rough set is set, and this is output. According to the disaster prevention plan support system 1 to be constructed, it can be expected that the construction and support of a risk assessment system and a disaster prevention plan for an arbitrary dangerous point can be supported by using this system.
Finally, the flow of the overall disaster prevention plan support will be described with reference to FIG. 20, including a schematic diagram obtained by simplifying the flowchart shown in FIG.
First, information on topography, geology, etc. is collected regarding the locations at risk of disaster. This is observation / inspection data 8 and learning data 17. This may be done by conducting a new survey and collecting necessary information. If there is existing debris flow dangerous mountain stream survey chart data, road disaster prevention general inspection data, etc., these can be used.
Next, the flow is surrounded by a dotted line. These flows correspond to the steps of the disaster prevention plan support system 1 or the disaster prevention plan support program according to the present embodiment.
The degree of risk is set by SVM using the collected data. In FIG. 6, when the representative data 20 is extracted, it corresponds to step S5, and when the representative data 20 is not extracted, it corresponds to step S2. In the present embodiment and Examples 1 and 2, since data containing a lot of noise was used, the risk level was set based on the separation plane constructed from the representative data 20 excluding the noise data. If it is data with little noise data or data that has been screened in advance, the degree of risk may be set based on a separation plane constructed from all target data.
Next, classification of the risk rank is performed based on the set risk (step S6 in FIG. 6), and evaluation rules are set according to the risk rank by analysis using the rough set (step S7 in FIG. 6). (Step S8 in FIG. 6). Here, it is considered that the convenience of the evaluation rule according to the obtained disaster occurrence risk is further improved by appropriately setting the number of classifications of the risk rank according to the user's request.
Next, by using the evaluation rules according to the risk of occurrence of disasters obtained by this series of flows for the survey data of any risk location, the survey data of any risk location can be Search for the evaluation rule corresponding to the dangerous part. Based on the evaluation rule, since the evaluation rule is set according to the disaster occurrence risk level, it is possible to grasp the risk rank related to an arbitrary risk location. If the degree of risk regarding an arbitrary dangerous spot can be grasped, the priority order between the dangerous spots can be established, and the planning of a disaster prevention plan such as repair work can be rationally and efficiently supported.

  It has a wide range of uses such as disaster prevention planning in public institutions such as local governments and disaster prevention centers, planning work related to the necessity of accident prevention or repair, and hazard map creation. It is also expected to be used as a teaching material for disaster prevention and accident prevention and evacuation drills at educational institutions, etc., and for private companies operating construction and civil engineering businesses, they will discover the needs of disaster prevention and maintenance management businesses and make proposals It can be used as a tool for business or as a shared tool for cooperation with public institutions, and can also be applied to research and development and design projects related to corporate disaster prevention and maintenance technologies.

It is a conceptual diagram which shows typically the basic concept of SVM. It is a two-dimensional image figure expressing the basic concept about the distance and risk with respect to the observation / inspection data by SVM. It is a conceptual diagram which shows typically the basic concept of a rough set. (A) And (b) is a conceptual diagram for demonstrating expansion of a rough set. It is a block diagram of the disaster prevention plan support system which concerns on this Embodiment. It is a flowchart which shows the process of the calculation performed in the disaster prevention plan assistance system which concerns on this Embodiment. (A) is a conceptual diagram of a separation plane constructed with data in a state before extracting representative data, and (b) is a conceptual diagram of a rule corresponding thereto. (A) is a conceptual diagram of a separation plane constructed with representative data, and (b) is a conceptual diagram of a rule corresponding thereto. It is a conceptual diagram of representative data. It is a figure which shows typically a mode that a rule is set for every risk rank and the observation and inspection data are included in the rule in the disaster prevention plan support system which concerns on this Embodiment. It is a graph which shows the relationship between the risk rank set in the disaster prevention plan assistance system which concerns on Example 1, and the incidence rate of a debris flow. It is a graph for demonstrating a factor with high separability with respect to a risk rank in the disaster prevention plan assistance system which concerns on Example 1. FIG. It is a graph for demonstrating the factor by which the ratio in which a danger is contained increases monotonously similarly to the basin maximum inclination in the disaster prevention plan assistance system which concerns on Example 1. It is a graph for demonstrating the factor by which the ratio which contains a danger comparatively monotonously increases similarly to the basin maximum inclination in the disaster prevention plan support system which concerns on Example 1. It is a graph which shows the relationship between the risk rank set by the evaluation rule in the disaster prevention plan support system which concerns on Example 1, and the incidence rate of a debris flow. It is the graph which compared the incidence rate of the debris flow based on the evaluation rule set by the disaster prevention plan support system which concerns on this Example 1, and the occurrence rate of the debris flow based on the evaluation rule set by the method currently disclosed by patent document 1. FIG. It is a graph which shows the relationship between the risk rank set by the evaluation rule in the disaster prevention plan support system which concerns on Example 2, and the incidence rate of a debris flow. It is a graph which shows the relationship between the value of the risk (f (x)) set by SVM in the disaster prevention plan assistance system which concerns on Example 3, and an applicable location. It is a graph which shows the relationship between the evaluation rule according to rank set by the rough set in the disaster prevention plan support system which concerns on Example 3, and an applicable location. It is a flowchart which shows the flow of the whole disaster prevention plan support including the schematic diagram which simplified the flowchart of FIG.

DESCRIPTION OF SYMBOLS 1 ... Disaster prevention plan support system 2 ... Input part 3 ... SVM analysis part 4 ... Analysis database 5 ... Database 6 ... Rough set analysis part 7 ... Output part 8 ... Observation / inspection data 9 ... Analysis condition data 10 ... Analysis condition setting part 11 ... Separation plane calculation unit 12 ... Risk level calculation unit 13 ... Representative data extraction unit 14 ... Rank setting unit 15 ... Separation plane function data 16 ... Risk level function data 17 ... Learning data 18 ... Separation plane data 19 ... Risk level data 20 ... Representative data 21 ... Representative data separation plane data 22 ... Representative data risk data 23 ... Risk rank data 24 ... Rough set data 25 ... Ran set evaluation rule data 26 ... Rough set generation unit 27 ... Ran evaluation rule generation unit

Claims (8)

No initial separation plane analysis + no representative data extraction Causes of occurrence of disasters / accidents / repairs at observation / inspection locations (the number of factors is n. N ≧ 2) corresponding to the coordinate axes that make up the n-dimensional coordinate space, N-dimensional observation / inspection data observed and inspected for each cause of occurrence of the disaster / accident / repair is used as learning data, and the occurrence / non-occurrence of disaster / accident / repair at the location subject to observation / inspection is used as a teacher value. , A separation plane that analyzes the possibility of occurrence / non-occurrence of disaster / accident / repair using a support vector machine (hereinafter referred to as SVM) to form the boundary of occurrence / non-occurrence of disaster / accident / repair, In the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data, using the n-dimensional observation / inspection data for each cause of the disaster, accident, or repair observed / inspected at the observation / inspection target location so The distance from the coordinate point to the separation plane, and the distance at which the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane can be taken as either a positive value or a negative value A risk calculation unit that calculates the risk of occurrence of disasters, accidents, and repairs at the locations subject to observation and inspection;
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
A disaster prevention plan support system characterized in that the extracted rules are output and used for plan evaluation and support such as disaster / accident prevention or repair selection. Initially no separation plane analysis + representative data extracted. Causes of occurrence of disasters, accidents, and repairs at the locations subject to observation and inspection (the number of factors is n. N ≧ 2) correspond to the coordinate axes that make up the n-dimensional coordinate space. N-dimensional observation / inspection data observed and inspected for each cause of occurrence of disasters / accidents / repairs is used as learning data, and occurrence / non-occurrence of disasters / accidents / repairs at the observation / inspection locations is used as a teacher value. , The possibility of occurrence / non-occurrence of disasters / accidents / repairs is analyzed using SVM, and the separation plane that forms the boundary of occurrence / non-occurrence of disasters / accidents / repairs and The n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair that has been inspected, and the separation plane from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data Distance to Thus, the distance that can be taken as either a positive value or a negative value when the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane is a disaster / accident / A risk calculation unit for calculating the risk of occurrence of repair;
Based on a desired threshold for the disaster / accident / repair occurrence risk or a logical expression including the threshold, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster A representative data extraction unit for extracting, as representative data, n-dimensional observation / inspection data for each disaster / accident / repair occurrence factor corresponding to the accident / repair occurrence risk;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using a support vector machine. A separation plane calculation unit that analyzes the representative data separation plane that forms the boundary of occurrence / non-occurrence of disasters / accidents / repairs,
Using the representative data separation surface analyzed by this separation surface calculation unit and the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair observed / inspected at the observation / inspection target location, The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, the coordinate point being on the origin side of the n-dimensional coordinate space with respect to the separation plane The risk level calculation unit that calculates the distance that can be taken as either a positive value or a negative value as a representative data risk level of the disaster / accident / repair occurrence of the observation / inspection point;
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence representative data risk level;
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
A disaster prevention plan support system characterized in that the extracted rules are output and used for plan evaluation and support such as disaster / accident prevention or repair selection. Initially with separation plane analysis + no representative data extraction Causes of occurrence of disasters / accidents / repairs at the locations subject to observation / inspection (the number of factors is n. N ≧ 2) corresponding to the coordinate axes constituting the n-dimensional coordinate space, N-dimensional observation / inspection data observed and inspected for each cause of occurrence of disasters / accidents / repairs is used as learning data, and occurrence / non-occurrence of disasters / accidents / repairs at the observation / inspection locations is used as a teacher value. , Separation surface calculation unit that analyzes the separation surface that forms the boundary of occurrence / non-occurrence of disaster / accident / repair using SVM, the possibility of occurrence / non-occurrence of disaster / accident / repair,
Using the separation surface analyzed by the separation surface calculation unit and the n-dimensional observation / inspection data for each factor of occurrence of the disaster, accident, and repair observed / inspected at the observation / inspection location, the n The distance from the coordinate point in the n-dimensional coordinate space indicated by the three-dimensional observation / inspection data to the separation plane, and the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane A risk calculation unit that calculates the distance that can be taken as either a positive value or a negative value as the risk of occurrence of a disaster, accident, or repair of the observation / inspection location,
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
A disaster prevention plan support system characterized in that the extracted rules are output and used for plan evaluation and support such as disaster / accident prevention or repair selection. Initial separation plane analysis + representative data extraction. Causes of occurrence of disasters / accidents / repairs at observation / inspection locations (the number of factors is n. N ≧ 2) correspond to the coordinate axes that make up the n-dimensional coordinate space. N-dimensional observation / inspection data observed and inspected for each cause of occurrence of the disaster / accident / repair is used as learning data, and the occurrence / non-occurrence of disaster / accident / repair at the location subject to observation / inspection is used as a teacher value. , Separation plane calculation unit that analyzes the separation plane that forms the boundary of occurrence / non-occurrence of disaster / accident / repair using SVM,
Using the separation surface analyzed by the separation surface calculation unit and the n-dimensional observation / inspection data for each factor of occurrence of the disaster, accident, and repair observed / inspected at the observation / inspection location, the n The distance from the coordinate point in the n-dimensional coordinate space indicated by the three-dimensional observation / inspection data to the separation plane, and the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane A risk calculation unit that calculates the distance that can be taken as either a positive value or a negative value as the risk of occurrence of a disaster, accident, or repair of the observation / inspection location,
Based on a desired threshold for the disaster / accident / repair occurrence risk or a logical expression including the threshold, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster A representative data extraction unit for extracting, as representative data, n-dimensional observation / inspection data for each disaster / accident / repair occurrence factor corresponding to the accident / repair occurrence risk;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using a support vector machine. The separation surface calculation unit for analyzing the representative data separation surface that forms the boundary of occurrence / non-occurrence of disaster / accident / repair,
Using the representative data separation surface analyzed by this separation surface calculation unit and the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair observed / inspected at the observation / inspection target location, The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, the coordinate point being on the origin side of the n-dimensional coordinate space with respect to the separation plane The risk level calculation unit that calculates the distance that can be taken as either a positive value or a negative value as a representative data risk level of the disaster / accident / repair occurrence of the observation / inspection point;
A rank setting unit that ranks n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the disaster / accident / repair occurrence representative data risk level;
A model is formed by combining preset category intervals for each of the n-dimensional disaster / accident / repair occurrence factors or combining category intervals input for each n-dimensional disaster / accident / repair occurrence factor. A rough set generation unit for forming a rough set by substituting n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair and the rank in the model;
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, a desired threshold value for the certainty level or a logical expression including the threshold value and / or a desired threshold value for the support or A rule generation unit for disaster / accident / repair that extracts the rule based on a logical expression including a threshold;
An output unit for outputting the extracted rule;
A disaster prevention plan support system characterized in that the extracted rules are output and used for plan evaluation and support such as disaster / accident prevention or repair selection. No initial separation plane analysis + no representative data extraction (program of claim 1)
A program executed by a computer to generate and output rules for occurrence / non-occurrence of disasters / accidents / repairs,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. The separation plane that analyzes the possibility of occurrence / non-occurrence using SVM to form the boundary of occurrence / non-occurrence of disaster / accident / repair, and each factor observed / inspected at the location to be observed / inspected n-dimensional observation / inspection data, and the distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane (provided that the coordinate point Against the separation surface The case where it is on the origin side of the n-dimensional coordinate space may be positive, or the case where it is on the origin side may be negative. Risk level calculation process,
A rank setting process for ranking n-dimensional observation / inspection data for each factor of occurrence of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
A disaster prevention plan support program characterized in that it is used for planning evaluation and support such as disaster / accident prevention or repair selection. Initially no separation plane analysis + representative data extraction (claim 2 program)
A program executed by a computer to generate and output rules for occurrence / non-occurrence of disasters / accidents / repairs,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. The separation plane that analyzes the possibility of occurrence / non-occurrence using SVM to form the boundary of occurrence / non-occurrence of disaster / accident / repair, and each factor observed / inspected at the location to be observed / inspected n-dimensional observation / inspection data, and the distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane (provided that the coordinate point Against the separation surface The case where it is on the origin side of the n-dimensional coordinate space may be positive, or the case where it is on the origin side may be negative.) Is calculated as the disaster / accident / repair occurrence risk of the observation / inspection target location A first risk level calculating step,
Based on a desired threshold value or condition for the disaster / accident / repair occurrence risk, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster / accident / repair occurrence risk is extracted. A representative data extraction step of extracting n-dimensional observation / inspection data for each factor corresponding to the above as representative data;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using a support vector machine. Separation surface calculation process to analyze the representative data separation surface that forms the boundary of occurrence / non-occurrence of disaster / accident / repair,
Using the representative data separation surface analyzed in this separation surface calculation process, and the n-dimensional observation / inspection data for each cause of occurrence of the disaster / accident / repair observed / inspected in the observation / inspection location, The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, the coordinate point being on the origin side of the n-dimensional coordinate space with respect to the separation plane A second risk calculation step for calculating a distance that can be taken as either a positive value or a negative value as a representative data risk of occurrence of disaster / accident / repair at the location subject to observation / inspection;
Rank setting step for ranking the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the risk level of the disaster / accident / repair occurrence representative data;
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
A disaster prevention plan support program characterized in that it is used for planning evaluation and support such as disaster / accident prevention or repair selection. Initial separation surface analysis + no representative data extraction (program of claim 3)
A program executed by a computer to generate and output rules for occurrence / non-occurrence of disasters / accidents / repairs,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. Separation plane calculation process to analyze the separation plane that forms the boundary of occurrence / non-occurrence of disaster / accident / repair using SVM for the possibility of occurrence / non-occurrence,
Using the separation surface analyzed in the separation surface calculation step and the n-dimensional observation / inspection data for each factor observed / inspected at the observation / inspection target location, the n-dimensional observation / inspection data is The distance from the coordinate point in the n-dimensional coordinate space to the separation plane (provided that the distance is positive when the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation plane) Or may be negative if it is on the origin side), a risk degree calculation step for calculating the disaster / accident / repair occurrence risk of the observation / inspection target part,
A rank setting process for ranking n-dimensional observation / inspection data for each factor of occurrence of the disaster / accident / repair according to the disaster / accident / repair occurrence risk,
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
A disaster prevention plan support program characterized in that it is used for planning evaluation and support such as disaster / accident prevention or repair selection. Initial separation plane analysis + representative data extraction (claim 4 program)
A program executed by a computer to generate and output rules for occurrence / non-occurrence of disasters / accidents / repairs,
The computer causes the disaster / accident / repair occurrence factor (the number of factors to be n. N ≧ 2) to correspond to the coordinate axes that make up the n-dimensional coordinate space, and the disaster / accident / repair The n-dimensional observation / inspection data observed / inspected for each occurrence factor is used as learning data, and the occurrence / non-occurrence of disasters / accidents / repairs at the locations subject to observation / inspection is used as a teacher value. A first separation surface calculation step for analyzing a separation surface that forms a boundary of occurrence / non-occurrence of disaster / accident / repair using SVM for occurrence / non-occurrence;
Using the separation surface analyzed in the first separation surface calculation step and the n-dimensional observation / inspection data for each factor observed / inspected at the observation / inspection target location, the n-dimensional observation / inspection The distance from the coordinate point in the n-dimensional coordinate space indicated by the inspection data to the separation surface (however, this distance is the case where the coordinate point is on the origin side of the n-dimensional coordinate space with respect to the separation surface). A first risk degree calculating step for calculating as a disaster / accident / repair occurrence risk of the observation / inspection target portion;
Based on a desired threshold value or condition for the disaster / accident / repair occurrence risk, the corresponding disaster / accident / repair occurrence risk is extracted, and the extracted disaster / accident / repair occurrence risk is extracted. A representative data extraction step of extracting n-dimensional observation / inspection data for each factor corresponding to the above as representative data;
Using this representative data as learning data, the occurrence / non-occurrence of disaster / accident / repair in this representative data is used as a teacher value, and the possibility of occurrence / non-occurrence of disaster / accident / repair is analyzed using a support vector machine. A second separation plane calculation process for analyzing the representative data separation plane that forms the boundary of occurrence / non-occurrence of disaster / accident / repair,
Representative data separation surface analyzed in the second separation surface calculation step, n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair observed / inspected in the observation / inspection location; , The distance from the coordinate point in the n-dimensional coordinate space indicated by the n-dimensional observation / inspection data to the separation plane, and the coordinate point is defined in the n-dimensional coordinate space with respect to the separation plane. A second risk calculation step of calculating a distance that can be taken as either a positive value or a negative value when it is on the origin side as a disaster / accident / repair occurrence representative data risk level of the observation / inspection target part;
Rank setting step for ranking the n-dimensional observation / inspection data for each occurrence factor of the disaster / accident / repair according to the risk level of the disaster / accident / repair occurrence representative data;
For the n-dimensional disaster / accident / repair occurrence factor, a category section for the attribute value related to this factor is provided, and a model is formed by combining the category sections for each factor, and for each occurrence factor of the disaster / accident / repair. Substituting n-dimensional observation / inspection data and the rank into this model to form a rough set,
As a rule corresponding to the rank, a combination of n-dimensional observation / inspection data for each cause of occurrence of the disaster, accident, and repair and a category section for each cause of the disaster, accident, and repair of the rough set including the rank Generating and calculating the certainty and support of disaster, accident and repair for each rule, and using the desired threshold or condition for the certainty and / or the desired threshold or condition for the support as a reference, The rule generation process for disasters, accidents, and repairs that extract rules,
An output step of outputting the extracted rule;
A disaster prevention plan support program characterized in that it is used for planning evaluation and support such as disaster / accident prevention or repair selection.