An unmanned system completion degree parameter solving method based on an entropy weight method

Abstract

The present application belongs to the field of artificial intelligence, especially the field of autonomy evaluation of intelligent unmanned system such as intelligent unmanned aerial vehicle, and specifically relates to a method for solving unmanned system completion degree parameters based on entropy weight method, which can be applied to intelligent unmanned aerial vehicle manufacturing, comprising: constructing a multi-dimensional completion degree evaluation model containing task execution effect and task reliability, wherein the task execution effect comprises single task indexes such as target recognition rate, false alarm rate and task completion time, and the task reliability comprises task success rate, and meanwhile, the dimension difference of original data is eliminated through normalization processing; and solving the completion degree parameters based on the entropy weight method. The present application objectively weights the evaluation indexes by using the entropy weight method, automatically determines the weight distribution by analyzing the discrete degree of index value, and strictly adapts the weight distribution to the actual performance of unmanned system. The present application significantly improves the objectivity and accuracy of evaluation results, and provides a standardized technical means for autonomy quantitative evaluation of unmanned system.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence, especially the field of autonomous assessment of unmanned systems such as intelligent drones. Specifically, it is a method for solving the completion parameters of unmanned systems based on the entropy weight method, which can be applied to the manufacturing of intelligent unmanned aerial vehicles. Background Technology

[0002] Unmanned systems are typically designed as autonomous systems, capable of independently executing tasks, adapting to environmental changes, and achieving predetermined objectives. Advances in autonomous system technology have enabled unmanned systems to achieve a higher degree of autonomous operation, reducing human intervention and improving overall system efficiency. With the widespread use of unmanned systems, methods for evaluating their performance have emerged.

[0003] However, current assessments of the autonomy of unmanned systems generally rely on subjective weighting methods (such as expert scoring and the analytic hierarchy process), which lack sufficient quantification and objectivity. The assessment of the autonomy of unmanned systems is a crucial step in measuring their intelligence level. Currently, the industry mainly relies on subjective weighting methods such as expert scoring. The core problems with these methods are: strong subjectivity (weights depend on expert experience, and different assessors may give significantly different results), poor adaptability (when the task scenario changes, the original weights may no longer be applicable and need to be readjusted), and insufficient quantification (it is difficult to accurately reflect the true performance of the unmanned system, resulting in low credibility of the assessment results). Summary of the Invention

[0004] This invention addresses the technical shortcomings of subjective weighting methods in autonomous system assessment, such as insufficient quantification and lack of objectivity. It innovatively proposes a method for solving unmanned system completion parameters based on the entropy weight method, comprising the following steps:

[0005] Step 1: Construct a multi-dimensional completion evaluation model that includes task execution performance and task reliability. The task execution performance includes single task indicators such as target recognition rate, false alarm rate, and task completion time. The task reliability includes task success rate. At the same time, normalization processing is used to eliminate the dimensional differences of the original data.

[0006] Step 2: Solve for the completion parameter based on the entropy weight method.

[0007] Furthermore, in step 1,

[0008] Target recognition rate and false alarm rate They are respectively:

[0009]

[0010]

[0011] In the formula, Indicates the number of correctly identified targets; Indicates the number of incorrectly identified targets; This indicates the total number of targets actually contained in the image;

[0012] single-flight completion rate of drones for:

[0013]

[0014] In the formula, The standard time to complete the task; The actual time it took for the drone to complete the mission; These are the weight coefficients corresponding to task completion time, target recognition rate, and false alarm rate, respectively, and satisfy the following conditions: ;

[0015] Task success rate for:

[0016]

[0017] In the formula, Indicates the number of times the task was successful. Indicates the total number of times the operation was performed;

[0018] The overall completion rate is:

[0019]

[0020] In the formula, This indicates the proportion of the overall completion rate to the effect achieved in a single instance.

[0021] Furthermore, step 2 specifically includes:

[0022] Step 2.1: Construct a single completion matrix and task success rate matrix ; respectively and Standardization is performed to obtain the single-time completion matrix. Indicator values ​​and task success rate matrix Indicator values:

[0023]

[0024] ;

[0025] Step 2.2: Calculate the single-time completion matrix respectively. index value and task success rate matrix index value Specific gravity:

[0026]

[0027] ;

[0028] Step 2.3: Calculate the weight of the effect of a single task, specifically including:

[0029] First, calculate the entropy values ​​of the single-task effect weight and the success rate of multiple tasks:

[0030]

[0031] ;

[0032] Then calculate the difference coefficient from the entropy value:

[0033]

[0034]

[0035] Then, calculate the weight of the effect of a single task. :

[0036] ;

[0037] Step 2.4, for The second evaluation constructs a task completion time matrix. Recognition rate matrix and false alarm rate matrix ;right Standardize the data to obtain the task completion time matrix index values. Recognition rate matrix index value And false alarm rate matrix index values :

[0038]

[0039]

[0040] ;

[0041] Step 2.5: Calculate the task completion time matrix index values. Recognition rate matrix index value And false alarm rate matrix index values Specific gravity:

[0042]

[0043]

[0044] ;

[0045] Calculate the entropy values ​​for task completion time, target recognition rate, and false alarm rate:

[0046]

[0047]

[0048] ;

[0049] Calculate the difference coefficient from the entropy value:

[0050]

[0051]

[0052] ;

[0053] Step 2.6: Calculate the completion time weight. Recognition rate weight and false alarm rate weight :

[0054]

[0055]

[0056] .

[0057] Furthermore, it also includes:

[0058] Set completion threshold ;

[0059] when If the task is completed successfully, it is considered a success; otherwise, it is considered a failure.

[0060] The significant advantages of this invention are:

[0061] 1) Breaking through the limitations of traditional subjective weighting, it automatically generates weight allocation schemes based on the inherent characteristics of the data;

[0062] 2) It innovatively solved the problem of fusing and calculating real-time performance data of a single task with long-term reliability data;

[0063] 3) Rigorous mathematical derivation and standardized processing ensure the comparability of indicators with different dimensions;

[0064] 4) The evaluation results can be directly used for the classification of autonomous levels and performance optimization of unmanned systems.

[0065] In summary, this invention significantly improves the objectivity and accuracy of the autonomous assessment results of unmanned systems, and provides a standardized technical means for the quantitative assessment of the autonomousness of unmanned systems. Attached Figure Description

[0066] Figure 1 This is a flowchart of the process for solving the completion parameter based on the entropy weight method in this invention. Detailed Implementation

[0067] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0068] This invention objectively quantifies the weights of various indicators based on actual data generated during UAV missions, effectively avoiding human error introduced by traditional subjective weighting methods. Faced with the complex characteristics of multiple indicators in UAV evaluation, the entropy weighting method determines weights based on the dispersion of indicator values. The greater the dispersion, the greater the impact of the indicator on the overall evaluation result. This weighting mechanism ensures that the weight allocation is strictly adapted to the actual performance of the UAV. Simultaneously, this method eliminates dimensional differences in indicators through standardization, constructing a unified dimensionless evaluation system at the mathematical level, suitable for integrating evaluation data from two different dimensions: single-mission effectiveness and multi-mission success rate.

[0069] This invention aims to quantify the completion level of unmanned systems through data-driven analysis, thereby comprehensively improving the assessment of their autonomy. By continuously optimizing the system through data accumulation, this invention holds promise for paving new paths for the future development of unmanned systems and promoting their widespread application and advancement.

[0070] The core technical solution of this invention includes:

[0071] 1. Completion modeling stage: Construct a multi-dimensional evaluation system that includes task execution performance (such as single task indicators such as target recognition rate, false alarm rate, and task completion time) and task reliability (such as long-term indicators such as task success rate), and eliminate the dimensional differences of the original data through normalization processing.

[0072] 2. Parameter solving stage of the entropy weight method:

[0073] 1) Calculate the information entropy value of each indicator;

[0074] 2) Determine the index weights based on the entropy value;

[0075] 3) Establish a weighted comprehensive evaluation function.

[0076] This invention provides a method for solving the completion degree parameter of unmanned systems based on the entropy weight method, and constructs a comprehensive and objective completion degree calculation model. It is implemented through the following two steps: completion degree modeling and the method for solving the completion degree parameter based on the entropy weight method. The implementation process is as follows:

[0077] Step 1: Completion modeling.

[0078] Completion rate reflects the overall performance of a drone in completing a mission, involving multiple aspects such as mission completion time, target recognition accuracy, and mission success rate. The level of completion rate directly affects the practical value and reliability of the drone; therefore, accurately calculating the completion rate is of great significance for assessing the autonomy of a drone.

[0079] Completion rate involves two aspects: the effectiveness of a single mission and the success rate of multiple missions. For a single mission, the primary prerequisite for judging mission success is whether the drone can safely return after completing the mission. If the drone loses control or crashes during the return journey, even if the previous reconnaissance mission was successfully completed, the entire mission should be considered a failure. Secondly, whether the drone can avoid collisions during the mission is also an important criterion for evaluating mission completion. Frequent collisions not only affect the drone's flight attitude and target tracking capabilities but may also cause damage to the aircraft, reducing the drone's safety and reliability. Therefore, obstacle avoidance capability is another prerequisite for drone mission completion. Based on the above two prerequisites, the drone's mission completion time, target recognition accuracy, and false alarm rate become the key factors determining the level of completion. Generally speaking, the shorter the time taken for the drone to complete the mission, the higher the target recognition accuracy, and the lower the false alarm rate, the higher its completion rate.

[0080] For multiple missions, the success rate of the drone in performing the same task also needs to be considered; that is, the proportion of times the drone successfully completes the task when it is repeatedly executed. A higher success rate indicates better reliability and stability of the drone, and thus a higher level of completion. By comprehensively considering both the effectiveness of a single mission and the success rate across multiple missions, a more comprehensive and objective assessment of the drone's mission execution capabilities can be achieved.

[0081] In the above process, the target recognition rate and false alarm rate The calculation formulas are as follows:

[0082]

[0083]

[0084] In the formula, Indicates the number of correctly identified targets; Indicates the number of incorrectly identified targets; This represents the total number of targets actually contained in the image. Based on the target recognition rate and false alarm rate, the single-shot completion rate of the UAV is... It can be calculated using the following formula:

[0085]

[0086] In the formula, The standard time to complete the task; The actual time it took for the drone to complete the mission; These are the weight coefficients corresponding to task completion time, target recognition rate, and false alarm rate, respectively, and satisfy the following conditions: . The value range is [0,1], with a larger value indicating a higher completion rate. The formula squares the contributions of the target recognition rate and false alarm rate, ensuring that a high completion score can be achieved even with a slight increase in task completion time when the recognition rate is high and the false alarm rate is low. Conversely, sacrificing recognition accuracy for shorter completion times will significantly reduce the completion score. A completion threshold needs to be set to determine whether the task was successfully completed. .when If the task is completed successfully, it is considered a success; otherwise, it is considered a failure. The value needs to be determined based on actual application requirements and evaluation criteria. The task success rate can be calculated by counting the number of times the task is completed. The calculation formula is as follows:

[0087]

[0088] In the formula, Indicates the number of times the task was successful. This represents the total number of executions. Combining the effectiveness of a single task and the success rate of multiple tasks, the overall completion rate (based on multiple evaluations) is calculated using the following formula:

[0089]

[0090] In the formula, This indicates the proportion of the overall completion rate to the effect of a single successful task, which is usually less than the proportion of the task success rate.

[0091] Step 2: Solve for the completion parameter based on the entropy weight method.

[0092] like Figure 1 As shown, this is for calculating the weight of a single task's effect. And the weight of success rate of multiple tasks For the single completion rate in the formula and success rate of multiple tasks Perform entropy weight analysis.

[0093] 201. First, construct the single-time completion matrix. (Number of trials) and task success rate matrix (number of evaluations) Then, respectively for and Standardization processing

[0094]

[0095]

[0096] Obtained after standardization and All are positive indicators.

[0097] 202. Calculate the weight of the indicator values:

[0098]

[0099]

[0100] 203. Calculate the entropy value of the effect of a single task and the success rate of multiple tasks:

[0101]

[0102]

[0103] The difference coefficient is obtained from the entropy value:

[0104]

[0105]

[0106] Calculate the weight of the effect of a single task :

[0107]

[0108] 204. Use the same approach to solve for the weights of the three indicators in the single completion rate. .

[0109] First of all, for The second evaluation constructs a task completion time matrix. Recognition rate matrix and false alarm rate matrix Then, respectively for The following standardization processes were performed respectively:

[0110]

[0111]

[0112]

[0113] 205. Obtained after standardization , and All are positive indicators. Based on these indicator values, the weight of each indicator value is calculated using the following formulas:

[0114]

[0115]

[0116]

[0117] The entropy values ​​for completion time, recognition rate, and false alarm rate are calculated using the following formula:

[0118]

[0119]

[0120]

[0121] The difference coefficient is obtained from the entropy value:

[0122]

[0123]

[0124]

[0125] 206. Finally, the completion time weight can be calculated. Recognition rate weight and false alarm rate weight :

[0126]

[0127]

[0128]

[0129] Thus, this invention has calculated the completion time weight using the entropy weight method. Recognition rate weight False alarm rate weight and the weight of single completion rate A comprehensive and objective completion calculation model was constructed. This model fully considers various key indicators during the task execution process, enabling a comprehensive evaluation of task completion.

[0130] This invention constructs a multi-dimensional completion evaluation model that includes the effect of a single task and the success rate of multiple tasks. It employs the entropy weighting method to objectively assign weights to evaluation indicators, automatically determining the weight allocation by analyzing the dispersion of indicator values, ensuring that the weight allocation strictly matches the actual performance of the unmanned system. This method eliminates dimensional differences through data standardization, establishing a dimensionless evaluation system and achieving the fusion calculation of evaluation data from different dimensions. Compared to traditional subjective weighting methods, this invention significantly improves the objectivity and accuracy of the evaluation results, providing a standardized technical means for the quantitative evaluation of the autonomy of unmanned systems.

[0131] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for solving the completion parameters of an unmanned system based on the entropy weight method, characterized in that, Includes the following steps: Step 1: Construct a multi-dimensional completion evaluation model that includes task execution performance and task reliability. The task execution performance includes single-task indicators such as target recognition rate, false alarm rate, and task completion time. The task reliability includes the task success rate. Simultaneously, normalization processing is used to eliminate the dimensional differences in the original data. Specifically, the target recognition rate... and false alarm rate They are respectively: In the formula, Indicates the number of correctly identified targets; Indicates the number of incorrectly identified targets; This indicates the total number of targets actually contained in the image; single-flight completion rate of drones for: In the formula, The standard time to complete the task; The actual time it took for the drone to complete the mission; These are the weight coefficients corresponding to task completion time, target recognition rate, and false alarm rate, respectively, and satisfy the following conditions: ; Task success rate for: In the formula, Indicates the number of times the task was successful. Indicates the total number of times the operation was performed; The overall completion rate is: In the formula, This indicates the proportion of the overall completion rate relative to the effect achieved in a single instance. Step 2: Solve for the completion parameters based on the entropy weight method, specifically including: Step 2.1: Construct a single completion matrix and task success rate matrix ; respectively and Standardization is performed to obtain the single-time completion matrix. Indicator values ​​and task success rate matrix Indicator values: ； Step 2.2: Calculate the single-time completion matrix respectively. index value and task success rate matrix index value Specific gravity: ； Step 2.3: Calculate the weight of the effect of a single task, specifically including: First, calculate the entropy values ​​of the single-task effect weight and the success rate of multiple tasks: ； Then calculate the difference coefficient from the entropy value: Then, calculate the weight of the effect of a single task. : ； Step 2.4, for The second evaluation constructs a task completion time matrix. Recognition rate matrix and false alarm rate matrix ;right Standardize the data to obtain the task completion time matrix index values. Recognition rate matrix index value And false alarm rate matrix index values : ； Step 2.5: Calculate the task completion time matrix index values. Recognition rate matrix index value And false alarm rate matrix index values Specific gravity: ； Calculate the entropy values ​​for task completion time, target recognition rate, and false alarm rate: ； Calculate the difference coefficient from the entropy value: ； Step 2.6: Calculate the completion time weight. Recognition rate weight and false alarm rate weight : 。 2. The method according to claim 1, characterized in that, Also includes: Set completion threshold ; when If the task is completed successfully, it is considered a success; otherwise, it is considered a failure.

Images