WO2017121053A1 - Method and system for evaluating collision degree between targets - Google Patents
Method and system for evaluating collision degree between targets Download PDFInfo
- Publication number
- WO2017121053A1 WO2017121053A1 PCT/CN2016/081495 CN2016081495W WO2017121053A1 WO 2017121053 A1 WO2017121053 A1 WO 2017121053A1 CN 2016081495 W CN2016081495 W CN 2016081495W WO 2017121053 A1 WO2017121053 A1 WO 2017121053A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- targets
- conflict
- target
- degree
- probability information
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Abstract
A method for evaluating a collision degree between targets, comprising: defining collision probability information between two targets (S11); making statistics about the collision probability information between the two targets using a sample data set (S12); and taking the collision probability information having been through the statistics as a collision degree between the two targets so as to quantize a collision between the targets (S13). According to the method and a system corresponding thereto, collision information about a target set in an approximate solution set is analysed with a statistical method, so whether there is a collision between targets and a degree of the collision can be identified in a high probability. With the increase of sample points, the identification rate can reach 100%, and the method is free from the influence by a Pareto front shape, i.e., the influence by the quality of an approximate solution set.
Description
本发明涉及信息技术领域，尤其涉及一种评估目标之间冲突程度的方法及其系统。The present invention relates to the field of information technology, and in particular, to a method and system for evaluating the degree of conflict between targets.
在工程、工业及科学等领域，常常会遇到目标优化的问题，现代进化多目标优化算法已成为处理此类问题最有效的方法。在多目标优化问题中，由于目标之间存在或强或弱的冲突性而使得算法不能像解决单目标优化问题一样获得唯一一个最优解，而是得到一组对于每个目标都折中的最优解集，通常这个最优解集称为Pareto最优解集，它在目标空间的像称为Pareto最优前沿。在这个最优解集里，解之间互不支配，从而又称彼此为非支配解；所谓支配解y是，一个解x Pareto支配y，当且仅当x相对于每一个目标的适应值都不差于y而且至少在一个目标上的适应值严格好于y。但是，在现实的多目标优化问题中，一般很难得到真实的Pareto最优解集，取而代之的是近似Pareto最优解集。In the fields of engineering, industry and science, the problem of target optimization is often encountered. Modern evolutionary multi-objective optimization algorithms have become the most effective way to deal with such problems. In the multi-objective optimization problem, because of the strong or weak conflict between the targets, the algorithm can not obtain the only one optimal solution as the single-objective optimization problem, but obtain a set of compromises for each target. The optimal solution set, usually called the Pareto optimal solution set, is called Pareto optimal frontier in the target space. In this optimal solution set, the solutions do not dominate each other, which is also called each other as a non-dominated solution; the so-called dominating solution y is, a solution x Pareto dominates y, if and only if x is adapted to each target Not worse than y and the fitness at least on one target is strictly better than y. However, in the realistic multi-objective optimization problem, it is generally difficult to obtain the true Pareto optimal solution set, and instead the approximate Pareto optimal solution set is replaced.
在求解多目标优化问题中，目前为止，多数这方面的研究只针对2到3个目标的多目标优化问题(Multi-objective Optimization Problems，MOPs)，而对于多于3个目标的优化问题，Farina等人于2002年提出高维目标优化问题(Many-Objective Optimization Problems，MOOPs)的概念用于区别前者。在目标优化问题中，目标之间的关系或冲突或非冲突，或两种关系并存。大量理论与实验分析得出，当目标个数增多时，并且存在冲突性的目标大量存在时，现代绝大多数基于Pareto占优的进化多目标优化算法处理此类问题就会碰到一系列的困难：随着目标个数增加，进化种群中非支配个体的比例快速增长，使得Pareto占优关系区分个体优劣能力下降，令算法的选择压力变小而无法收敛到Pareto最优前沿附近；而且，随着目标个数增多算法搜索空间也跟着增大，从
而需要较大规模进化种群才可获得覆盖整个Pareto最优前沿的近似解集，这无疑增加算法的复杂度及决策困难；由于彼此之间存在冲突的目标的增加造成Pareto前沿维数增加，这就带来决策过程的可视化问题。带来这些问题的主要原因是高维目标优化问题需要同时进行优化的目标大多数属于相互冲突的。冲突即当一个目标被改善时另一目标的性能却下降，非冲突即当一个目标被改善时另一目标也同时得到改善。可见，现有进化多目标优化算法处理高维目标优化问题所遇到的困难的根本原因是在众多目标中存在相互冲突的目标占高比例，行业里常称此为“维数灾难”问题。In solving multi-objective optimization problems, most of the research in this area has only focused on multi-objective optimization problems (MOPs) of 2 to 3 targets, while for more than 3 target optimization problems, Farina In 2002, the concept of Many-Objective Optimization Problems (MOOPs) was proposed to distinguish the former. In the goal optimization problem, the relationship between the goals is either conflicting or non-conflicting, or both. A large number of theoretical and experimental analysis shows that when the number of targets increases, and there are a large number of conflicting targets, most modern modern evolutionary multi-objective optimization algorithms based on Pareto will encounter a series of problems. Difficulties: As the number of targets increases, the proportion of non-dominated individuals in the evolutionary population increases rapidly, which makes the Pareto dominant relationship distinguish the superiority and inferior ability of the individual, making the selection pressure of the algorithm smaller and unable to converge near the Pareto optimal frontier; As the number of targets increases, the algorithm search space also increases, from
However, a larger-scale evolutionary population is needed to obtain an approximate solution set covering the entire Pareto optimal frontier, which undoubtedly increases the complexity and decision-making difficulty of the algorithm; the Pareto frontier dimension increases due to the increase of conflicting targets. It brings the visualization of the decision process. The main reason for these problems is that the goals of high-dimensional target optimization problems that need to be optimized at the same time are mostly conflicting. Conflict is the performance of another target when one goal is improved, and the other goal is also improved when one goal is improved. It can be seen that the root cause of the difficulties encountered by the existing evolutionary multi-objective optimization algorithm in dealing with high-dimensional target optimization problems is that there are conflicting targets in many targets, which is often referred to as “dimensional disaster” in the industry.
因此，亟需设计一种评估目标之间冲突程度的方法，以提高评估概率。Therefore, it is urgent to design a method to assess the degree of conflict between targets to increase the probability of evaluation.
发明内容Summary of the invention
有鉴于此，本发明的目的在于提供一种评估目标之间冲突程度的方法及其系统，旨在解决现有技术中进化高维多目标优化时识别概率低的问题。In view of this, an object of the present invention is to provide a method and system for evaluating the degree of conflict between targets, aiming at solving the problem of low recognition probability in the evolution of high-dimensional multi-objective optimization in the prior art.
本发明提出一种评估目标之间冲突程度的方法，包括：The present invention proposes a method for assessing the degree of conflict between targets, including:
定义两个目标之间的冲突概率信息；Define conflict probability information between two targets;
利用样本数据集统计出所述两个目标之间的冲突概率信息；Using the sample data set to calculate conflict probability information between the two targets;
将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。The statistical conflict probability information is used as the degree of conflict between the two targets to quantify the conflict between the targets.
优选的，所述定义两个目标之间的冲突概率信息的步骤包括：Preferably, the step of defining conflict probability information between two targets includes:
在给定的一个高维目标优化问题中，设P是从目标集合Φ与其自身的笛卡尔积(即Φ×Φ)映射到[0，1]的映射，即P:Φ×Φ→[0，1]，那么称P为目标之间的冲突概率映射，因此，
P(fi,fj)称为目标fi与目标fj之间的冲突概率信息。In a given high-dimensional target optimization problem, let P be a mapping from the target set Φ and its own Cartesian product (ie Φ × Φ) to [0, 1], ie P: Φ × Φ → [0 , 1], then P is the collision probability map between the targets, therefore, P(f i , f j ) is called collision probability information between the target f i and the target f j .
优选的，所述样本数据集包括采用进化算法生成的近似解。Preferably, the sample data set includes an approximate solution generated using an evolutionary algorithm.
优选的，所述利用样本数据集统计出所述两个目标之间的冲突概率信息的
步骤包括：Preferably, the using the sample data set to calculate the conflict probability information between the two targets
The steps include:
假设X为一个高维目标优化问题的可行解集合，给定一个种群POP＝{x1,…,xN},xi∈X，那么若从种群POP中任选一对个体(xk,xl)，其中k≠l，按照排列组合的定义，则有
种取法，如果其中有
对个体满足以下条件：[fi(xk)≥fi(xl)]∧[fj(xk)≥fj(xl)]或者[fi(xk)≤fi(xl)]∧[fj(xk)≤fj(xl)]；那么，目标fi与目标fj在X上的冲突概率信息可由以下公式计算：Suppose X is a feasible solution set for a high-dimensional target optimization problem. Given a population POP={x 1 ,..., x N }, x i ∈X, then if you choose a pair of individuals from the population POP (x k , x l ), where k≠l, according to the definition of the permutation combination, there is Kind of method, if there is The following conditions are satisfied for the individual: [f i (x k ) ≥ f i (x l )] ∧ [f j (x k ) ≥ f j (x l )] or [f i (x k ) ≤ f i (x l )]∧[f j (x k )≤f j (x l )]; Then, the collision probability information of the target f i and the target f j on X can be calculated by the following formula:
另一方面，本发明还提供一种评估目标之间冲突程度的系统，包括：In another aspect, the present invention also provides a system for assessing the degree of conflict between targets, comprising:
定义模块，用于定义两个目标之间的冲突概率信息；a definition module for defining conflict probability information between two targets;
统计模块，用于利用样本数据集统计出所述两个目标之间的冲突概率信息；a statistics module, configured to calculate, by using a sample data set, conflict probability information between the two targets;
量化模块，用于将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。And a quantization module, configured to use the calculated conflict probability information as a conflict degree between the two targets to quantify the conflict between the targets.
优选的，所述定义模块具体用于在给定的一个高维目标优化问题中，设P是从目标集合Φ与其自身的笛卡尔积(即Φ×Φ)映射到[0，1]的映射，即P:Φ×Φ→[0，1]，那么称P为目标之间的冲突概率映射，因此，
P(fi,fj)称为目标fi与目标fj之间的冲突概率信息。Preferably, the definition module is specifically configured to, in a given high-dimensional target optimization problem, set P to map from the target set Φ and its own Cartesian product (ie, Φ×Φ) to [0, 1]. , that is, P: Φ × Φ → [0, 1], then P is called the collision probability map between the targets, therefore, P(f i , f j ) is called collision probability information between the target f i and the target f j .
优选的，所述样本数据集包括采用进化算法生成的近似解。Preferably, the sample data set includes an approximate solution generated using an evolutionary algorithm.
优选的，所述统计模块，具体用于假设X为一个高维目标优化问题的可行解集合，给定一个种群POP＝{x1,…,xN},xi∈X，那么若从种群POP中任选一对个体(xk,xl)，其中k≠l，按照排列组合的定义，则有
种取法，如果其中有
)对个体满足以下条件：[fi(xk)≥fi(xl)]∧[fj(xk)≥fj(xl)]或者[fi(xk)≤fi(xl)]∧[fj(xk)≤fj(xl)]；那么，目标fi与目标fj在X上的冲突概
率信息可由以下公式计算：Preferably, the statistical module is specifically used to assume that X is a feasible solution set of a high-dimensional target optimization problem, given a population POP={x 1 ,..., x N }, x i ∈X, then if the population is from the population POP chooses a pair of individuals (x k , x l ), where k≠l, according to the definition of the permutation combination, there is Kind of method, if there is The following conditions are satisfied for the individual: [f i (x k ) ≥ f i (x l )] ∧ [f j (x k ) ≥ f j (x l )] or [f i (x k ) ≤ f i ( x l )]∧[f j (x k )≤f j (x l )]; Then, the collision probability information of the target f i and the target f j on X can be calculated by the following formula:
本发明提供的技术方案能在处理高维目标优化问题时，采用统计方法分析目标集合在近似解集上的冲突信息，这种方式能够高概率地识别目标之间是否冲突以及冲突的程度，随着样本点的增加，识别率可以达到100％，并且得出该方法不受Pareto前沿形状影响，即不受近似解集质量的影响。The technical solution provided by the invention can analyze the conflict information of the target set on the approximate solution set by using the statistical method when dealing with the high-dimensional target optimization problem, and the method can recognize the conflict between the targets and the degree of the conflict with high probability. With the increase of sample points, the recognition rate can reach 100%, and the method is not affected by the Pareto frontier shape, that is, it is not affected by the approximate solution set quality.
图1为本发明一实施方式中评估目标之间冲突程度的方法流程图；1 is a flow chart of a method for evaluating a degree of conflict between targets in an embodiment of the present invention;
图2为本发明一实施方式中评估目标之间冲突程度的系统结构示意图。FIG. 2 is a schematic structural diagram of a system for evaluating the degree of conflict between targets in an embodiment of the present invention.
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
本发明具体实施方式提供了一种评估目标之间冲突程度的方法，主要包括如下步骤：A specific embodiment of the present invention provides a method for evaluating the degree of conflict between targets, which mainly includes the following steps:
S11、定义两个目标之间的冲突概率信息；S11. Define conflict probability information between the two targets;
S12、利用样本数据集统计出所述两个目标之间的冲突概率信息；S12. Calculate conflict probability information between the two targets by using a sample data set;
S13、将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。S13. Calculate the conflict probability information as the conflict degree between the two targets to quantify the conflict between the targets.
本发明在处理高维目标优化问题时，采用统计方法分析目标集合在近似解集上的冲突信息，这种方式能够高概率地识别目标之间是否冲突以及冲突的程度，随着样本点的增加，识别率可以达到100％，并且得出该方法不受Pareto前沿形状影响，即不受近似解集质量的影响。The invention adopts a statistical method to analyze the conflict information of the target set on the approximate solution set when dealing with the high-dimensional target optimization problem, which can recognize the conflict between the targets and the degree of the conflict with high probability, with the increase of the sample points. The recognition rate can reach 100%, and the method is not affected by the Pareto frontier shape, that is, it is not affected by the approximate solution set quality.
以下将对本发明所提供的一种评估目标之间冲突程度的方法进行详细说
明。In the following, a method for assessing the degree of conflict between targets provided by the present invention will be described in detail.
Bright.
请参阅图1，为本发明一实施方式中评估目标之间冲突程度的方法流程图。Please refer to FIG. 1 , which is a flowchart of a method for evaluating the degree of conflict between targets in an embodiment of the present invention.
在步骤S11中，定义两个目标之间的冲突概率信息。In step S11, conflict probability information between the two targets is defined.
在本实施方式中，所述定义两个目标之间的冲突概率信息的步骤S11包括：在给定的一个高维目标优化问题(Many-Objective Optimization Problem，MOOP)中，设P是从目标集合Φ与其自身的笛卡尔积(即Φ×Φ)映射到[0，1]的映射，即P:Φ×Φ→[0，1]，那么称P为目标之间的冲突概率映射，因此，
P(fi,fj)称为目标fi与目标fj之间的冲突概率信息。In this embodiment, the step S11 of defining the conflict probability information between the two targets includes: in a given one-dimensional optimization problem (MOOP), setting P as the target set Φ and its own Cartesian product (ie Φ × Φ) map to the map of [0, 1], that is, P: Φ × Φ → [0, 1], then P is called the collision probability map between the targets, therefore, P(f i , f j ) is called collision probability information between the target f i and the target f j .
在步骤S12中，利用样本数据集统计出所述两个目标之间的冲突概率信息。In step S12, the conflict probability information between the two targets is counted using the sample data set.
在本实施方式中，所述样本数据集包括采用进化算法生成的近似解，其中，该进化算法包括第二代遗传算法(NSGA2)，除此之外还可以包括其他的进化算法，在此不做限定。In this embodiment, the sample data set includes an approximate solution generated by an evolutionary algorithm, where the evolutionary algorithm includes a second generation genetic algorithm (NSGA2), and may include other evolutionary algorithms, and Make a limit.
在本实施方式中，所述利用样本数据集统计出所述两个目标之间的冲突概率信息的步骤S12包括：In this embodiment, the step S12 of using the sample data set to calculate the conflict probability information between the two targets includes:
假设X为一个高维目标优化问题的可行解集合，给定一个种群POP＝{x1,…,xN},xi∈X，那么若从种群POP中任选一对个体(xk,xl)，其中k≠l，按照排列组合的定义，则有
种取法，如果其中有
)对个体满足以下条件：[fi(xk)≥fi(xl)]∧[fj(xk)≥fj(xl)]或者[fi(xk)≤fi(xl)]∧[fj(xk)≤fj(xl)]；那么，目标fi与目标fj在X上的冲突概率信息可由以下公式计算：Suppose X is a feasible solution set for a high-dimensional target optimization problem. Given a population POP={x 1 ,..., x N }, x i ∈X, then if you choose a pair of individuals from the population POP (x k , x l ), where k≠l, according to the definition of the permutation combination, there is Kind of method, if there is The following conditions are satisfied for the individual: [f i (x k ) ≥ f i (x l )] ∧ [f j (x k ) ≥ f j (x l )] or [f i (x k ) ≤ f i ( x l )]∧[f j (x k )≤f j (x l )]; Then, the collision probability information of the target f i and the target f j on X can be calculated by the following formula:
在本实施方式中，由上述公式可以得出高维目标优化问题中两两目标之间的冲突概率信息。
In the present embodiment, the conflict probability information between the two targets in the high-dimensional target optimization problem can be obtained from the above formula.
在步骤S13中，将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。In step S13, the statistical conflict probability information is used as the degree of conflict between the two targets to quantify the conflict between the targets.
在本实施方式中，该度量方法具有客观性，根据度量结果用于识别冲突目标的准确性不受近似解集质量的影响。In the present embodiment, the metric method has objectivity, and the accuracy for identifying the conflicting target according to the metric result is not affected by the quality of the approximate solution set.
本发明在处理高维目标优化问题时，采用统计方法分析目标集合在近似解集上的冲突信息，这种方式能够从客观的角度分析目标之间的冲突度，还能够高概率地识别目标之间是否冲突以及冲突的程度，随着样本点的增加，识别率可以达到100％，并且得出该方法不受Pareto前沿形状影响，即不受近似解集质量的影响。The invention uses the statistical method to analyze the conflict information of the target set on the approximate solution set when dealing with the high-dimensional target optimization problem, which can analyze the conflict degree between the targets from an objective perspective, and can also identify the target with high probability. Whether the conflict and the degree of conflict, with the increase of sample points, the recognition rate can reach 100%, and the method is not affected by the Pareto frontier shape, that is, it is not affected by the approximate solution set quality.
本发明具体实施方式还提供一种评估目标之间冲突程度的系统10，主要包括：The specific embodiment of the present invention further provides a system 10 for evaluating the degree of conflict between targets, which mainly includes:
定义模块11，用于定义两个目标之间的冲突概率信息；a definition module 11 for defining conflict probability information between two targets;
统计模块12，用于利用样本数据集统计出所述两个目标之间的冲突概率信息；The statistics module 12 is configured to use the sample data set to calculate conflict probability information between the two targets;
量化模块13，用于将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。The quantization module 13 is configured to use the statistical conflict probability information as the conflict degree between the two targets to quantify the conflict between the targets.
本发明在处理高维目标优化问题时，采用统计方法分析目标集合在近似解集上的冲突信息，这种方式能够高概率地识别目标之间是否冲突以及冲突的程度，随着样本点的增加，识别率可以达到100％，并且得出该方法不受Pareto前沿形状影响，即不受近似解集质量的影响。The invention adopts a statistical method to analyze the conflict information of the target set on the approximate solution set when dealing with the high-dimensional target optimization problem, which can recognize the conflict between the targets and the degree of the conflict with high probability, with the increase of the sample points. The recognition rate can reach 100%, and the method is not affected by the Pareto frontier shape, that is, it is not affected by the approximate solution set quality.
请参阅图2，所示为本发明一实施方式中评估目标之间冲突程度的系统10的结构示意图。在本实施方式中，评估目标之间冲突程度的系统10包括定义模块11、统计模块12以及量化模块13。Referring to FIG. 2, a schematic structural diagram of a system 10 for evaluating the degree of conflict between targets in an embodiment of the present invention is shown. In the present embodiment, the system 10 for assessing the degree of conflict between targets includes a definition module 11, a statistics module 12, and a quantization module 13.
定义模块11，用于定义两个目标之间的冲突概率信息。A definition module 11 is defined for defining conflict probability information between two targets.
在本实施方式中，定义模块11具体用于在给定的一个高维目标优化问题(Many-Objective Optimization Problem，MOOP)中，设P是从目标集合Φ与
其自身的笛卡尔积(即Φ×Φ)映射到[0，1]的映射，即P:Φ×Φ→[0，1]，那么称P为目标之间的冲突概率映射，因此，
P(fi,fj)称为目标fi与目标fj之间的冲突概率信息。In the present embodiment, the definition module 11 is specifically used to give a Cartesian product from the target set Φ and its own in a given Many-Objective Optimization Problem (MOOP) (ie, Φ× Φ) maps to [0,1], ie P:Φ×Φ→[0,1], then P is called the collision probability map between targets, therefore, P(f i , f j ) is called collision probability information between the target f i and the target f j .
统计模块12，用于利用样本数据集统计出所述两个目标之间的冲突概率信息。The statistics module 12 is configured to use the sample data set to calculate conflict probability information between the two targets.
在本实施方式中，所述样本数据集包括采用进化算法生成的近似解，其中，该进化算法包括第二代遗传算法(NSGA2)，除此之外还可以包括其他的进化算法，在此不做限定。In this embodiment, the sample data set includes an approximate solution generated by an evolutionary algorithm, where the evolutionary algorithm includes a second generation genetic algorithm (NSGA2), and may include other evolutionary algorithms, and Make a limit.
在本实施方式中，统计模块12，具体用于假设X为一个高维目标优化问题的可行解集合，给定一个种群POP＝{x1,…,xN},xi∈X，那么若从种群POP中任选一对个体(xk,xl)，其中k≠l，按照排列组合的定义，则有
种取法，如果其中有
)对个体满足以下条件：[fi(xk)≥fi(xl)]∧[fj(xk)≥fj(xl)]或者[fi(xk)≤fi(xl)]∧[fj(xk)≤fj(xl)]；那么，目标fi与目标fj在X上的冲突概率信息可由以下公式计算：In this embodiment, the statistic module 12 is specifically configured to assume that X is a feasible solution set of a high-dimensional target optimization problem, given a population POP={x 1 , . . . , x N }, x i ∈X, then Select a pair of individuals (x k , x l ) from the population POP, where k≠l, according to the definition of the permutation combination, there is Kind of method, if there is The following conditions are satisfied for the individual: [f i (x k ) ≥ f i (x l )] ∧ [f j (x k ) ≥ f j (x l )] or [f i (x k ) ≤ f i ( x l )]∧[f j (x k )≤f j (x l )]; Then, the collision probability information of the target f i and the target f j on X can be calculated by the following formula:
在本实施方式中，由上述公式可以得出高维目标优化问题中两两目标之间的冲突概率信息。In the present embodiment, the conflict probability information between the two targets in the high-dimensional target optimization problem can be obtained from the above formula.
量化模块13，用于将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。The quantization module 13 is configured to use the statistical conflict probability information as the conflict degree between the two targets to quantify the conflict between the targets.
在本实施方式中，该度量方法具有客观性，根据度量结果用于识别冲突目标的准确性不受近似解集质量的影响。In the present embodiment, the metric method has objectivity, and the accuracy for identifying the conflicting target according to the metric result is not affected by the quality of the approximate solution set.
本发明在处理高维目标优化问题时，采用统计方法分析目标集合在近似解集上的冲突信息，这种方式能够从客观的角度分析目标之间的冲突度，还能够高概率地识别目标之间是否冲突以及冲突的程度，随着样本点的增加，识别率
可以达到100％，并且得出该方法不受Pareto前沿形状影响，即不受近似解集质量的影响。The invention uses the statistical method to analyze the conflict information of the target set on the approximate solution set when dealing with the high-dimensional target optimization problem, which can analyze the conflict degree between the targets from an objective perspective, and can also identify the target with high probability. Whether there is conflict and the degree of conflict, with the increase of sample points, the recognition rate
It can reach 100% and it is concluded that the method is not affected by the Pareto frontier shape, ie it is not affected by the approximate solution set quality.
值得注意的是，上述实施例中，所包括的各个单元只是按照功能逻辑进行划分的，但并不局限于上述的划分，只要能够实现相应的功能即可；另外，各功能单元的具体名称也只是为了便于相互区分，并不用于限制本发明的保护范围。It should be noted that, in the foregoing embodiment, each unit included is only divided according to functional logic, but is not limited to the above division, as long as the corresponding function can be implemented; in addition, the specific name of each functional unit is also They are only used to facilitate mutual differentiation and are not intended to limit the scope of the present invention.
另外，本领域普通技术人员可以理解实现上述各实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成，相应的程序可以存储于一计算机可读取存储介质中，所述的存储介质，如ROM/RAM、磁盘或光盘等。In addition, those skilled in the art can understand that all or part of the steps of implementing the above embodiments may be completed by a program to instruct related hardware, and the corresponding program may be stored in a computer readable storage medium. Storage medium, such as ROM/RAM, disk or CD.
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。
The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.
Claims (8)
- 一种评估目标之间冲突程度的方法，其特征在于，所述方法包括：A method for assessing a degree of conflict between targets, characterized in that the method comprises:定义两个目标之间的冲突概率信息；Define conflict probability information between two targets;利用样本数据集统计出所述两个目标之间的冲突概率信息；Using the sample data set to calculate conflict probability information between the two targets;将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。The statistical conflict probability information is used as the degree of conflict between the two targets to quantify the conflict between the targets.
- 如权利要求1所述的评估目标之间冲突程度的方法，其特征在于，所述定义两个目标之间的冲突概率信息的步骤包括：The method of evaluating the degree of conflict between targets according to claim 1, wherein the step of defining conflict probability information between the two targets comprises:在给定的一个高维目标优化问题中，设P是从目标集合Φ与其自身的笛卡尔积(即Φ×Φ)映射到[0，1]的映射，即P:Φ×Φ→[0，1]，那么称P为目标之间的冲突概率映射，因此，P(fi,fj)称为目标fi与目标fj之间的冲突概率信息。In a given high-dimensional target optimization problem, let P be a mapping from the target set Φ and its own Cartesian product (ie Φ × Φ) to [0, 1], ie P: Φ × Φ → [0 , 1], then P is the collision probability map between the targets, therefore,
P(f i , f j ) is called collision probability information between the target f i and the target f j . - 如权利要求2所述的评估目标之间冲突程度的方法，其特征在于，所述样本数据集包括采用进化算法生成的近似解。The method of evaluating the degree of conflict between targets as claimed in claim 2, wherein the sample data set comprises an approximate solution generated using an evolutionary algorithm.
- 如权利要求3所述的评估目标之间冲突程度的方法，其特征在于，所述利用样本数据集统计出所述两个目标之间的冲突概率信息的步骤包括：The method for estimating the degree of conflict between targets according to claim 3, wherein the step of using the sample data set to calculate the conflict probability information between the two targets comprises:假设X为一个高维目标优化问题的可行解集合，给定一个种群POP＝{x1,…,xN},xi∈X，那么若从种群POP中任选一对个体(xk,xl)，其中k≠l，按照排列组合的定义，则有种取法，如果其中有
对个体满足以下条件：[fi(xk)≥fi(xl)]∧[fj(xk)≥fj(xl)]或者[fi(xk)≤fi(xl)]∧[fj(xk)≤fj(xl)]；那么，目标fi与目标fj在X上的冲突概率信息可由以下公式计算：Suppose X is a feasible solution set for a high-dimensional target optimization problem. Given a population POP={x 1 ,..., x N }, x i ∈X, then if you choose a pair of individuals from the population POP (x k , x l ), where k≠l, according to the definition of the permutation combination, there is
Kind of method, if there isThe following conditions are satisfied for the individual: [f i (x k ) ≥ f i (x l )] ∧ [f j (x k ) ≥ f j (x l )] or [f i (x k ) ≤ f i (x l )]∧[f j (x k )≤f j (x l )]; Then, the collision probability information of the target f i and the target f j on X can be calculated by the following formula:
- 一种评估目标之间冲突程度的系统，其特征在于，所述评估目标之间冲 突程度的系统包括：A system for assessing the degree of conflict between targets, characterized in that The degree of system includes:定义模块，用于定义两个目标之间的冲突概率信息；a definition module for defining conflict probability information between two targets;统计模块，用于利用样本数据集统计出所述两个目标之间的冲突概率信息；a statistics module, configured to calculate, by using a sample data set, conflict probability information between the two targets;量化模块，用于将统计出的冲突概率信息作为所述两个目标之间的冲突度，以将目标之间的冲突进行量化。And a quantization module, configured to use the calculated conflict probability information as a conflict degree between the two targets to quantify the conflict between the targets.
- 如权利要求5所述的评估目标之间冲突程度的系统，其特征在于，所述定义模块具体用于在给定的一个高维目标优化问题中，设P是从目标集合Φ与其自身的笛卡尔积(即Φ×Φ)映射到[0，1]的映射，即P:Φ×Φ→[0，1]，那么称P为目标之间的冲突概率映射，因此，P(fi,fj)称为目标fi与目标fj之间的冲突概率信息。The system for evaluating the degree of conflict between targets according to claim 5, wherein the definition module is specifically configured to set a P from the target set Φ and its own flute in a given high-dimensional target optimization problem. The card product (ie Φ × Φ) maps to the mapping of [0, 1], that is, P: Φ × Φ → [0, 1], then P is called the collision probability map between the targets, therefore,
P(f i , f j ) is called collision probability information between the target f i and the target f j . - 如权利要求6所述的评估目标之间冲突程度的系统，其特征在于，所述样本数据集包括采用进化算法生成的近似解。A system for assessing the degree of conflict between targets as claimed in claim 6 wherein said sample data set comprises an approximate solution generated using an evolutionary algorithm.
- 如权利要求7所述的评估目标之间冲突程度的系统，其特征在于，所述统计模块，具体用于假设X为一个高维目标优化问题的可行解集合，给定一个种群POP＝{x1,…,xN},xi∈X，那么若从种群POP中任选一对个体(xk,xl)，其中k≠l，按照排列组合的定义，则有种取法，如果其中有
对个体满足以下条件：[fi(xk)≥fi(xl)]∧[fj(xk)≥fj(xl)]或者[fi(xk)≤fi(xl)]∧[fj(xk)≤fj(xl)]；那么，目标fi与目标fj在X上的冲突概率信息可由以下公式计算：The system for evaluating the degree of conflict between targets according to claim 7, wherein the statistical module is specifically configured to assume that X is a feasible solution set of a high-dimensional target optimization problem, given a population POP={x 1 ,...,x N },x i ∈X, then if a pair of individuals (x k ,x l ) are selected from the population POP, where k≠l, according to the definition of the permutation combination, there is
Kind of method, if there isThe following conditions are satisfied for the individual: [f i (x k ) ≥ f i (x l )] ∧ [f j (x k ) ≥ f j (x l )] or [f i (x k ) ≤ f i (x l )]∧[f j (x k )≤f j (x l )]; Then, the collision probability information of the target f i and the target f j on X can be calculated by the following formula:
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610029033.8A CN105719064A (en) | 2016-01-15 | 2016-01-15 | Method and system for evaluating confliction degree between targets |
| CN201610029033.8 | 2016-01-15 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2017121053A1 true WO2017121053A1 (en) | 2017-07-20 |
Family
ID=56147618
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2016/081495 WO2017121053A1 (en) | 2016-01-15 | 2016-05-10 | Method and system for evaluating collision degree between targets |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN105719064A (en) |
| WO (1) | WO2017121053A1 (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107977701A (en) * | 2017-09-01 | 2018-05-01 | 深圳大学 | A kind of online object space division methods, device and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060004683A1 (en) * | 2004-06-30 | 2006-01-05 | Talbot Patrick J | Systems and methods for generating a decision network from text |
| CN104021392A (en) * | 2014-01-27 | 2014-09-03 | 河南大学 | Conflict evidence fusion method based on vector measurement |
| CN104462826A (en) * | 2014-12-11 | 2015-03-25 | 云南师范大学 | Method for detecting and measuring multi-sensor evidence conflicts on basis of matrix singular value decomposition |
-
2016
- 2016-01-15 CN CN201610029033.8A patent/CN105719064A/en active Pending
- 2016-05-10 WO PCT/CN2016/081495 patent/WO2017121053A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060004683A1 (en) * | 2004-06-30 | 2006-01-05 | Talbot Patrick J | Systems and methods for generating a decision network from text |
| CN104021392A (en) * | 2014-01-27 | 2014-09-03 | 河南大学 | Conflict evidence fusion method based on vector measurement |
| CN104462826A (en) * | 2014-12-11 | 2015-03-25 | 云南师范大学 | Method for detecting and measuring multi-sensor evidence conflicts on basis of matrix singular value decomposition |
Non-Patent Citations (1)
| Title |
|---|
| YANG, MING ET AL.: "Measuring Dynamic Degree and Conflict Degree between Objectives for Dynamic Multi-Objective TSP", GEOMATICS AND INFORMATION SCIENCE OF WUHAN UNIVERSITY, vol. 33, no. 2, 29 February 2008 (2008-02-29) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105719064A (en) | 2016-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sun et al. | Cervical cancer diagnosis based on random forest | |
| Kočišová et al. | Discriminant analysis as a tool for forecasting company's financial health | |
| CN109543763B (en) | Raman spectrum analysis method based on convolutional neural network | |
| JP4391506B2 (en) | Evaluation apparatus, evaluation method, and evaluation program | |
| Algamal et al. | High dimensional logistic regression model using adjusted elastic net penalty | |
| Bommert et al. | A multicriteria approach to find predictive and sparse models with stable feature selection for high-dimensional data | |
| CN112036476A (en) | Data feature selection method and device based on two-classification service and computer equipment | |
| CN112785420A (en) | Credit scoring model training method and device, electronic equipment and storage medium | |
| CN107463797B (en) | Biological information analysis method and device for high-throughput sequencing, equipment and storage medium | |
| JP6300588B2 (en) | Risk analysis apparatus, method and program for anonymized data | |
| JP5715445B2 (en) | Quality estimation apparatus, quality estimation method, and program for causing computer to execute quality estimation method | |
| WO2017121053A1 (en) | Method and system for evaluating collision degree between targets | |
| EP4227855A1 (en) | Graph explainable artificial intelligence correlation | |
| CN110837853A (en) | Rapid classification model construction method | |
| CN111783088B (en) | Malicious code family clustering method and device and computer equipment | |
| CN115146890A (en) | Enterprise operation risk warning method and device, computer equipment and storage medium | |
| Lyu et al. | A new semiparametric transformation approach to disease diagnosis with multiple biomarkers | |
| CN113971984A (en) | Classification model construction method and device, electronic equipment and storage medium | |
| CN116451771B (en) | Image classification convolutional neural network compression method and core particle device data distribution method | |
| JP2018151913A (en) | Information processing system, information processing method, and program | |
| Chen et al. | A comparative study of discrimination methods for credit scoring | |
| CN108735295B (en) | Blood analysis method and terminal equipment based on regression tree model | |
| Pollard et al. | Supervised distance matrices | |
| CN115728396A (en) | Acoustic emission signal feature extraction method based on pressure-bearing equipment activity defect monitoring | |
| Verschueren | Regularized Buckley–James Method: A Comprehensive Review and Applications |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 16884613 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 13/11/2018) |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 16884613 Country of ref document: EP Kind code of ref document: A1 |