Team state evaluation method based on index dimension system

Abstract

The invention discloses a team state evaluation method based on an index dimension system. The method comprises the following steps: firstly, constructing a comprehensive and customized evaluation system through a preset evaluation dimension and an evaluation index corresponding to the evaluation dimension; then obtaining target evaluation index scores of the team based on the evaluation indexes, converting an evaluation system into computable data, accumulating the target evaluation index scores based on the evaluation dimensions to obtain evaluation dimension scores corresponding to the evaluation dimensions of the current team, and comprehensively reflecting the overall performance of the team in the evaluation dimensions; obtaining an evaluation dimension mapping score through equal ratio mapping processing based on the evaluation dimension score in different years, and eliminating annual difference; the evaluation dimension mapping score is visually processed and displayed through a radar map, and improved strategies such as team states, auxiliary decision making and resource allocation are visually presented.

Description

Technical Field

[0001] The present invention relates to the field of team status assessment, and in particular to a team status assessment method based on an indicator dimension system. Background Art

[0002] In today's rapidly evolving business environment, the overall state of a team is crucial to project success and corporate growth. However, there is currently a lack of a scientific, comprehensive, and visual team status assessment method and system that can accurately measure team performance across various dimensions, promptly identify team issues and areas of progress or regression, and provide robust data support for team management and decision-making.

[0003] Traditional team status assessment involves manually recording scores across various dimensions and visualizing them. However, this manual recording method is time-consuming and labor-intensive, and it fails to provide a detailed breakdown of dimensions, making it difficult to reflect the structural characteristics of team capabilities.

[0004] At the same time, when calculating the scores of various evaluation indicators, due to human recording omissions (such as missing the number of training participations in social service capabilities) or system collection errors (such as not capturing horizontal project completion data), key evaluation indicator scores are missing. For example, a teacher team missed the social training score, which directly led to a 25% drop in the total score of the social service capability dimension. The traditional method uses a simple mean filling, which fails to consider the inherent correlation between evaluation indicators (such as the synergy between school-enterprise cooperation and professional construction capabilities), resulting in distorted evaluation dimension scores. Existing technologies lack a dynamic filling mechanism based on the correlation of historical data and cannot effectively repair the impact of missing data on evaluation results.

[0005] Furthermore, the traditional approach to filling missing values ​​involves directly analyzing the correlation coefficients between evaluation indicators using the Pearson correlation coefficient, thus filling in missing values ​​without considering key factors such as internal team management adjustments and fluctuations in resource input. Research has shown that when using the correlation coefficient calculated using the Pearson correlation coefficient to fill in missing values ​​and then calculate the evaluation dimension scores, the standard deviation of the scientific research capability dimension scores can increase by 40% when the team changes or adjusts its research direction. This makes it impossible to guarantee the robustness of the correlation coefficient and, therefore, the accuracy of the evaluation dimension scores. Summary of the Invention

[0006] The purpose of the present invention is to provide a team status assessment method based on an indicator dimension system, which solves the above-mentioned technical problems pointed out in the prior art.

[0007] The present invention provides a team status assessment method based on an indicator dimension system, comprising the following steps:

[0008] Preset evaluation dimensions and corresponding evaluation indicators;

[0009] Obtaining an initial evaluation index score for the team based on the evaluation index; analyzing and processing the initial evaluation index score by identifying missing values ​​and combining it with historical evaluation index scores through an adaptive particle swarm optimization algorithm to obtain a target evaluation index score;

[0010] Based on the accumulation of the scores of each target evaluation indicator in each evaluation dimension, the evaluation dimension scores corresponding to each evaluation dimension of the current team are obtained;

[0011] Based on the geometric mapping of the evaluation dimension scores in different years, the evaluation dimension mapping scores are obtained;

[0012] The evaluation dimension mapping scores are visualized and displayed through radar charts.

[0013] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages:

[0014] According to the analysis of the team status assessment method based on the indicator dimension system provided by the present invention, in specific application, first, the assessment dimensions and the assessment indicators corresponding to the assessment dimensions are preset; the target assessment indicator scores of the team are obtained based on the assessment indicators; a comprehensive and customized assessment system is constructed by presetting the assessment dimensions and the assessment indicators corresponding to the assessment dimensions, covering core areas such as the team's professional capabilities and social services, and avoiding assessment blind spots; then, the target assessment indicator scores of the team are obtained based on the assessment indicators, the assessment system is converted into computable data, the actual performance data is collected based on the standard of step S1, and points are assigned according to the rules. The assessment dimension scores corresponding to the assessment dimensions of the current team are obtained by accumulating the scores of the target assessment indicators based on each assessment dimension, and the overall performance of the team in each assessment dimension is comprehensively reflected (the assessment dimension scores corresponding to the assessment dimensions of the current team are obtained based on the scores of the target assessment indicators of each assessment dimension; and the assessment dimension mapping scores are obtained by geometric mapping based on the assessment dimension scores in different years to eliminate annual differences;

[0015] Finally, the evaluation dimension mapping scores are visualized and displayed through radar charts, which can directly present the team status and assist in decision-making. Radar charts can also provide visual insights and help improve strategies such as resource allocation.

[0016] Moreover, in the subsequent implementation process, the Pearson correlation coefficient was combined with the inertia fluctuation factor of the evaluation indicators obtained by adaptive particle swarm optimization to calculate and analyze the correlation coefficients between the evaluation indicators, making the calculation of the scores of the team's evaluation dimensions more robust. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1This is a schematic diagram of the main process of a team status assessment method based on an indicator dimension system;

[0018] Figure 2 This is a visualization simulation diagram of a radar chart of evaluation dimension mapping scores in a team status evaluation method based on an indicator dimension system;

[0019] Figure 3 A schematic diagram simulating the impact of internal team management adjustments and personnel changes on overall evaluation indicators in a team status evaluation method based on an indicator dimension system;

[0020] Figure 4 Obtain an operational flow diagram for analyzing the inertia fluctuation factor in a team status assessment method based on an indicator dimension system;

[0021] Figure 5 A schematic diagram of the simulation effect of the fixed step size of the traditional PSO and the dynamically changing particle motion step size in a team status assessment method based on an indicator dimension system;

[0022] Figure 6 This is a schematic diagram of the simulation of iterative convergence acceleration trend in a team status assessment method based on an indicator dimension system;

[0023] Figure 7 A schematic diagram of the simulation of iterative oscillation suppression trend in a team status assessment method based on an indicator dimension system;

[0024] Figure 8 This is a simulation diagram of the first curve and the second curve under sampling frequency in a team status assessment method based on an indicator dimension system. DETAILED DESCRIPTION

[0025] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] The present invention will be further described in detail below through specific embodiments in conjunction with the accompanying drawings.

[0027] Example 1

[0028] like Figure 1 As shown, the first embodiment of the present invention provides a team status assessment method based on an indicator dimension system, including the following steps:

[0029] Step S1: presetting evaluation dimensions and corresponding evaluation indicators;

[0030] It should be noted that the above-mentioned embodiment of the present application determines a plurality of evaluation dimensions for evaluating the status of the team according to the nature, goals and characteristics of the work of the team. For example, for a teaching and scientific research team, evaluation dimensions such as political quality, professional construction ability, curriculum construction ability, teaching ability, internship and training guidance ability, scientific research ability and social service ability can be included. These evaluation dimensions should be able to fully reflect the performance of the team in all aspects. For each determined evaluation dimension, the specific scoring evaluation index items are further refined, and clear scoring standards and scores are set for each evaluation index item. For example, under the scientific research ability evaluation dimension, specific evaluation index items such as published papers, horizontal topics, and vertical topics can be set, and different scores can be assigned to them, such as 6 points for ordinary papers and 40 points for core journals.

[0031] Step S2: obtaining the team's initial evaluation index score based on the evaluation index; analyzing and processing the initial evaluation index score by identifying missing values ​​and combining it with historical evaluation index scores through an adaptive particle swarm optimization algorithm to obtain a target evaluation index score;

[0032] Step S3: Accumulate the scores of each target evaluation indicator based on each evaluation dimension to obtain the evaluation dimension scores corresponding to each evaluation dimension of the current team;

[0033] It should be noted that the above embodiment of the present application collects the actual performance data of the team on each evaluation indicator item and scores it according to the set scoring criteria. Then, the scores of each evaluation indicator item under the same evaluation dimension are accumulated to obtain the total score of the evaluation dimension as the score of the evaluation dimension. For example, a teacher team scored 180 points in the scientific research ability evaluation dimension.

[0034] Step S4: performing geometric mapping processing based on the evaluation dimension scores in different years to obtain the evaluation dimension mapping scores;

[0035] It should be noted that, in order to facilitate the comparison of the performance of teams in different years on the same evaluation dimension, the above-mentioned embodiment of the present application needs to geometrically map the scores of different years to a unified interval [0, 100]. The specific method is to find the highest score and the lowest score of the evaluation dimension over the years, and then convert the scores of other years according to a certain mathematical formula so that they are mapped to the interval [0, 100]. For example, assuming that the highest score of a certain evaluation dimension in the past three years is 95 points and the lowest score is 60 points, and the original score of the evaluation dimension in a certain year is 80 points, then the score calculation formula after mapping is: Based on this formula, we can calculate:

[0036] Step S5: Visualize the evaluation dimension mapping scores through a radar chart.

[0037] It should be noted that the above-mentioned embodiment of the present application visualizes the final scores of the team in each evaluation dimension through a radar chart. Each axis of the radar chart represents an evaluation dimension, and the length of the axis represents the score size of the evaluation dimension. Through the radar chart, you can intuitively see the performance of the team in different evaluation dimensions, as well as the relative relationship between the evaluation dimensions. For example, if a team scores high in the innovation ability evaluation dimension and low in the project progress evaluation dimension, then a shape that is longer in one direction and shorter in another direction will appear in the radar chart, clearly showing the strengths and weaknesses of the team.

[0038] For example, as shown in Table 1 below, when evaluating the status of a teaching team, we first determine the evaluation dimensions as political quality, professional development ability, curriculum development ability, teaching ability, internship and training guidance ability, scientific research ability, and social service ability. There are a total of seven evaluation dimensions. Then, for each evaluation dimension, we further refine the scoring evaluation indicators as follows:

[0039]

[0040]

[0041] Furthermore, the scores are quantified and the evaluation dimension scores are accumulated. The relevant data of the team at a certain stage are collected, and each scoring evaluation indicator item is scored according to the above scoring criteria.

[0042] Further, geometric mapping, using the above formula to perform geometric mapping to [0, 100]

[0043] Finally, the visualization is done using a radar chart as follows Figure 2 shown.

[0044] The above-mentioned embodiment of the present application first constructs a comprehensive and customized evaluation system by presetting evaluation dimensions and evaluation indicators corresponding to the evaluation dimensions, covering core areas such as the team's political literacy, professional ability, and social services, to avoid evaluation blind spots; then, by obtaining the team's target evaluation indicator scores based on the evaluation indicators, the evaluation system is converted into computable data, and actual performance data (such as the number of published papers and the number of trainings participated in) is collected based on the standards of step S1, and points are assigned according to the rules. By accumulating the scores of each target evaluation indicator based on each evaluation dimension, the evaluation dimension scores corresponding to each evaluation dimension of the current team are obtained, which comprehensively reflects the overall performance of the team in each evaluation dimension;

[0045] By performing geometric mapping based on the evaluation dimension scores in different years, we can obtain the evaluation dimension mapping scores, eliminate annual differences, and achieve comparability across time. Through annual mapping comparison, we can support long-term trend analysis (such as the annual improvement or fluctuation of scientific research capabilities). Finally, we can visualize the evaluation dimension mapping scores through radar charts to intuitively present the team status and assist in decision-making. Radar charts can be used to provide visual insights and help formulate improvement strategies such as training and resource allocation.

[0046] During the execution of S2 above, when recording the scores of each evaluation indicator of the team, there may be data missing due to human or non-human factors. For example, in the social service capacity evaluation dimension, there are two evaluation indicators: participation in social training and horizontal topics. However, during the statistical process, due to human factors, the score of the evaluation indicator of participation in social training was omitted, resulting in missing evaluation indicator scores, which in turn led to deviations in the evaluation dimension scores during the execution of S3 of the current team. Therefore, obtaining the scores of the evaluation indicators of S2 is crucial for subsequent team analysis.

[0047] Specifically, in step S2, the target evaluation index score is obtained by analyzing and processing the initial evaluation index score by identifying missing values ​​and combining the historical evaluation index scores with the adaptive particle swarm optimization algorithm, including the following steps:

[0048] Step S21: obtaining a sampling frequency interval based on the communication overhead corresponding to each sampling frequency and the historical evaluation index score fluctuation data corresponding to the sampling frequency; sampling based on the sampling frequency interval to obtain the historical evaluation index score of each evaluation index corresponding to each evaluation dimension;

[0049] Step S22: Based on the current evaluation index score, the target evaluation index score is obtained by combining the correlation analysis and missing value filling processing processed by the adaptive particle swarm optimization algorithm according to the historical evaluation index score.

[0050] It should be noted that the above-mentioned embodiment of the present application uses historical evaluation index scores to establish a historical data benchmark to provide a data basis for subsequent dynamic analysis and data correction; then, based on the current score combined with historical data, correlation analysis and missing value filling are performed to improve the accuracy and rationality of the current score and solve the problem of incomplete data.

[0051] Specifically, in step S22, based on the current evaluation index score, the target evaluation index score is obtained according to the historical evaluation index score combined with the correlation analysis and missing value filling processing processed by the adaptive particle swarm optimization algorithm, including the following operation steps:

[0052] Step S221: identifying correlation coefficients between evaluation indicators based on the historical evaluation indicator scores by performing correlation analysis combined with adaptive particle swarm optimization processing;

[0053] Step S222: performing missing value identification based on the currently acquired evaluation index scores to obtain missing evaluation indexes;

[0054] It should be noted that the above-mentioned missing value identification methods, such as null value detection, directly detect NaN, NULL, or empty strings in the data. For example, if the value of an evaluation indicator field is empty or a placeholder (such as "-999"), it is marked as a missing evaluation indicator.

[0055] Step S223: performing missing value filling processing on the missing evaluation indicator based on the correlation coefficient to obtain the missing evaluation indicator score of the missing indicator;

[0056] Step S224: The missing evaluation indicator score based on the missing indicator is integrated with the current evaluation indicator score to obtain a complete target evaluation indicator score.

[0057] It should be noted that the above embodiment of the present application is aimed at a missing evaluation indicator in the current evaluation, and then the missing information is marked, for example, the evaluation indicator score of the enterprise survey is missing; then, according to the correlation coefficient between each evaluation indicator, n evaluation indicators that are strongly correlated with the evaluation indicator of the current missing information (strong correlation, that is, the first n evaluation indicators after the correlation coefficient is sorted from high to low) are retrieved, and the missing value filling process is performed by weighted filling (with the correlation coefficient as the weight) through the correlation coefficient of the evaluation indicator of the current missing information and the n evaluation indicators with strong correlation. For example, the current missing value is x1 (the evaluation indicator score of the enterprise survey is missing), and the evaluation indicators with strong correlation are x2 (80 points for participating in the school-enterprise cooperation seminar) and x3 (90 points for new professional development), and the correlation coefficient between the missing value and x2 is 0.3, and the correlation coefficient between the missing value and x3 is 0.6; through the weighted filling method, the score of the missing value x1 is:

[0058]

[0059] The above embodiment of the present application first uses the historical evaluation index scores to calculate the correlation coefficients between the indicators and quantify the degree of their mutual influence; further, it directly detects explicit missingness through rule matching (null values, placeholders such as "-999") and marks missing evaluation indicators. For example, if the survey score field of a certain enterprise is "NULL", the system automatically marks it as missing; for missing indicators, select the indicators with the first n correlation coefficients (such as x2, x3), use the correlation coefficients as weights, and calculate the estimated values ​​in combination with the existing scores; finally, merge the filled missing values ​​(such as x1=86.67) with the current non-missing values ​​(such as x2=80, x3=90) to form a target evaluation indicator set without missing values, which provides complete data support for the calculation of subsequent evaluation dimension scores and the evaluation of team status.

[0060] Specifically, in step S221, based on the historical evaluation index scores, a correlation analysis is performed in combination with an adaptive particle swarm optimization process to identify correlation coefficients between the evaluation indicators, including the following steps:

[0061] Step S2211: constructing an evaluation indicator matrix based on all the historical evaluation indicator scores;

[0062] Step S2212: Calculate the Pearson correlation coefficient for any two elements in the evaluation index matrix;

[0063] Step S2213: Based on the Pearson correlation coefficient and the inertia fluctuation factor of the historical evaluation index score obtained by adaptive particle swarm optimization processing (the inertia factor is used to measure the continuity and stability of a team's historical performance on a certain evaluation index. If the historical data performance of the "i-th team" on the "j-th evaluation index" has small fluctuations and a continuous trend, it means that the evaluation index has strong inertia and a high inertia factor value. On the contrary, if the historical performance fluctuates greatly, the inertia factor value is low.) calculate to obtain the correlation coefficient.

[0064] The correlation coefficient is calculated as follows:

[0065]

[0066] Among them, γ m,n Represents the correlation coefficient between evaluation index m and evaluation index n; a m,n It represents the Pearson correlation coefficient between the evaluation index m and the evaluation index n; N represents the number of evaluation indicators in the current evaluation dimension; f represents the inertial fluctuation factor of the evaluation index n.

[0067] It should be noted that, in the above embodiment of the present application, the Pearson correlation coefficient a m,nIt represents the Pearson correlation coefficient between the evaluation index m and the evaluation index n, quantifies the linear correlation between the two evaluation indicators, and reflects whether the evaluation indicator n changes synchronously (positive correlation or negative correlation) when the evaluation index m increases; the average Pearson correlation coefficient is used to For the "mth evaluation indicator", first calculate the average Pearson correlation coefficient between it and all other evaluation indicators in the same evaluation dimension, which can be understood as the overall correlation level of the evaluation indicator m in the current evaluation dimension, reflecting its overall synergy with other evaluation indicators; the inertia fluctuation factor f reflects the degree of stability of the evaluation indicator; the inertia factor is calculated based on historical data and is determined by cyclical volatility (standard deviation) and autocorrelation coefficient (trend continuity); an evaluation indicator with a higher inertia factor indicates that its historical value fluctuates less and the trend is continuous; while an evaluation indicator with larger fluctuations has a lower inertia factor; after introducing f into the product, the correlation between the evaluation indicators can be weighted according to their respective stability, and more stable evaluation indicators have a greater impact on the overall correlation; using the summation operation By accumulating all the evaluation indicators n in the entire evaluation dimension, the contribution of the relevant evaluation indicators after inertia weighting of each evaluation indicator is superimposed to form the final correlation coefficient Y m,n , ensuring that all relevant factors are comprehensively reflected;

[0068] In the application process of the above embodiment of the present application, the Pearson correlation coefficient a is directly used. m,n When using the inertia fluctuation factor f to measure the relationship between evaluation indicators, the stability of the data of each evaluation indicator itself will be ignored. When the historical data of an evaluation indicator fluctuates violently, its correlation may be affected by large noise. By introducing the inertia fluctuation factor f, stable evaluation indicators (higher inertia) will be given higher weights, so that the correlation coefficient not only reflects the linear correlation, but also corrects the uncertainty caused by data fluctuations.

[0069] The Pearson correlation coefficients of all evaluation indicators in the same evaluation dimension are averaged to ensure the integrity of the evaluation. The corresponding inertia factor f is then multiplied to take into account the internal stability of each evaluation indicator data. Finally, all evaluation indicators are accumulated to integrate the contribution of each evaluation indicator after stability weighting, thus obtaining a more objective correlation measure between evaluation indicators.

[0070] In a multidimensional evaluation indicator system, each evaluation indicator may come from different business fields and data characteristics. The above embodiment of the present application adapts to the actual situation of different volatility differences of different evaluation indicators by using the inertia factor as an adjustment parameter. Furthermore, the corrected correlation coefficient can more accurately reflect the actual dependency relationship between the evaluation indicators, providing a more powerful statistical basis for subsequent missing value filling, trend analysis, evaluation weight allocation, etc.

[0071] In summary, the above-mentioned implementation of this application captures the linear relationship between the evaluation indicators through the average Pearson correlation coefficient; introduces the inertia fluctuation factor f to perform weighted adjustment on the stability of the evaluation indicators; and finally comprehensively considers the contributions of different evaluation indicators through weighted accumulation to obtain an evaluation indicator correlation coefficient that better reflects the actual situation.

[0072] Research shows that if a team readjusts its internal management system, workflow, or evaluation direction for a particular evaluation indicator, the score for that indicator may be significantly affected, manifesting as large fluctuations in historical data. For example, if a teaching or research team frequently changes evaluation criteria or makes major adjustments to a key project in a short period of time, the score data may fluctuate significantly. In addition, when there are major adjustments to core team members, or when key positions are frequently replaced, or when there are large fluctuations or interruptions in the investment of resources such as funds, equipment, and time for a particular evaluation indicator, the score data will also fluctuate dramatically.

[0073] like Figure 3 As shown, Figure 3 The scoring of a certain evaluation indicator is simulated for 9 evaluation cycles. The vertical bold line in 2021-Q3 represents internal adjustments, and the vertical bold line in 2022-Q3 represents team personnel changes, reflecting the impact of internal management adjustments and personnel changes on the overall evaluation indicators.

[0074] Therefore, the stability of evaluation indicator scores is usually determined by factors such as a sound internal management system, stable team members, continuous and sufficient resource investment, and a relatively stable external environment;

[0075] Data volatility often reflects the impact of factors such as frequent internal management adjustments, drastic personnel turnover, fluctuating resource inputs, and sudden changes in the external environment, policies, or market conditions.

[0076] However, when evaluating team status, if one only relies on the current evaluation indicator scores and simple correlation analysis, while ignoring the impact of internal team management adjustments, personnel replacements, changes in resource inputs, etc. on the evaluation indicator scores, then directly using historical data for predictions or filling in missing values ​​will cause the evaluation results to be out of line with actual performance.

[0077] Specifically, if Figure 4 As shown, in step S2213, the acquisition of the inertia fluctuation factor includes the following steps:

[0078] Step S22131: Calculate the periodic volatility value (the periodic volatility value is the standard deviation) of the j-th evaluation indicator of the i-th team over a preset time period; and calculate the autocorrelation coefficient using the historical evaluation indicator scores;

[0079] It should be noted that, in the above embodiment of the present application, the standard deviation of the j-th evaluation indicator of the i-th team is calculated based on a preset time period (e.g., quarter or year) as the periodic volatility value; at the same time, the autocorrelation coefficient is calculated using the historical evaluation indicator scores. The autocorrelation coefficient measures the linear correlation between data at different time points in the time series, reflecting whether the change trend of the evaluation indicator over a period of time is continuous;

[0080] By analyzing cyclical volatility and autocorrelation coefficients, we identify evaluation indicators that can maintain relatively stable performance even in the face of internal management adjustments, personnel turnover, or fluctuations in resource input, providing basic data support for the subsequent determination of initial candidate values.

[0081] Step S22132: determining an initial candidate value based on the periodic volatility value and the autocorrelation coefficient (the initial candidate value is an initial value that may be an inertia volatility factor);

[0082] The above-mentioned embodiment of the present application determines the initial candidate value based on the periodic volatility value and autocorrelation coefficient obtained in S22131; the selection of the initial candidate value is based on the understanding of the volatility of historical data and trend analysis to ensure that the selected value can better reflect the historical stability and continuity characteristics of the evaluation indicator.

[0083] Step S22133: Initializing the particle swarm, taking the initial candidate value as the initial particle; calculating the fitness corresponding to the initial particle using a preset fitness function based on the initial particle;

[0084] It should be noted that initializing the particle swarm in the above-mentioned embodiment of the present application means creating multiple potential inertia fluctuation factor values ​​(each called a "particle"), and using a preset fitness function to calculate the fitness corresponding to each particle to measure the contribution of a certain particle (i.e., the inertia fluctuation factor value) to the overall model prediction accuracy; by initializing the particle swarm and calculating the fitness, the optimal inertia fluctuation factor value is explored from multiple angles, thereby increasing the possibility of finding the optimal parameters.

[0085] Step S22134: When it is determined that the fitness does not satisfy the fitness less than or equal to the fitness threshold (if it is determined that the fitness satisfies the fitness, the initial particle corresponding to the fitness when satisfied is directly output as the inertia fluctuation factor), the fitness is sorted from high to low, and the initial particles corresponding to the top N fitnesses are selected and projected onto a two-dimensional plane based on a preset projection rule to obtain the two-dimensional coordinates corresponding to each initial particle;

[0086] It should be noted that in the above embodiment of the present application, if the current fitness does not meet the set threshold, all particles are sorted from high to low according to fitness, and the top N particles with the highest fitness are selected for projection operation on a two-dimensional plane and converted into a two-dimensional coordinate form for further processing, focusing on the most promising particles, reducing the search space while ensuring the quality of the solution, and facilitating the subsequent optimization process.

[0087] Step S22135: After updating the two-dimensional coordinates with the preset particle motion speed and particle motion change step, back-projection is performed based on the preset projection rule to obtain the first particle; based on the first particle, return to S22133 and repeat the execution until the fitness satisfies the fitness less than or equal to the fitness threshold, and the inertia fluctuation factor is output.

[0088] It should be noted that the above-mentioned embodiment of the present application updates the position of the selected particles according to the preset particle movement speed and step size, and then converts it back to the original problem space through reverse projection to form new particles, and returns the newly formed particles to step S22133 for re-execution until an inertia fluctuation factor whose fitness meets the requirements is found; by iteratively updating the position of the particles, the optimal solution is gradually approached to ensure that the final inertia fluctuation factor can not only accurately reflect the historical stability of the evaluation indicators, but also effectively improve the prediction performance of the entire model.

[0089] The above-mentioned embodiment of the present application obtains the inertia fluctuation factor through analysis. By introducing the inertia fluctuation factor, the impact of abnormal data fluctuations caused by factors such as sudden changes in the external environment, policies or market conditions on the final evaluation results is avoided, thereby making the target evaluation indicator score obtained based on historical data more in line with the actual situation.

[0090] In the above-mentioned embodiments of the present application, the scores of the evaluation indicators reflect the team's performance in each evaluation dimension. When the historical data volatility of different evaluation indicators varies, for example, scientific research ability may fluctuate greatly, while political quality may be relatively stable. If a fixed step size is used, the search may not be precise enough for the evaluation indicators with large fluctuations, or the search may be too slow for the stable evaluation indicators. Therefore, adjusting the step size according to the characteristics of the evaluation indicators can more efficiently find the optimal solution.

[0091] Traditional PSO (Particle Swarm Optimization) algorithms use fixed step sizes or simple adaptive rules, failing to account for the differences in data characteristics across different evaluation metrics. In a multidimensional, complex evaluation metric system, this "one-size-fits-all" approach can lead to: unstable searches for high-volatility metrics (due to excessively large step sizes, resulting in oscillations); or inefficient searches for low-volatility metrics (due to excessively small step sizes, resulting in slow convergence); Team status assessment involves multiple evaluation dimensions (such as scientific research and teaching ability), and the scoring criteria, data distribution, and business logic for these dimensions vary significantly. For example, "scientific research ability" may follow a long-tail distribution (with a few teams scoring high and the majority scoring low), while "teaching ability" may approximate a normal distribution. A fixed step size cannot adapt to this heterogeneity, requiring a customized step size strategy tailored to the characteristics of each evaluation metric.

[0092] Therefore, the evaluation index score is closely related to the step size of particle movement. The evaluation index includes the actual performance and historical changes of the team in various evaluation dimensions (such as teaching, scientific research, social services, etc.). These data can reflect the overall level and volatility of the current team as feedback. By integrating this feedback information into the particle swarm optimization process, the step size can be dynamically adjusted: when the evaluation index shows that the team is in good condition and the data is relatively stable, a larger step size of particle movement change is allowed to speed up the search and jump out of the local optimum; if the team performance is weak or the data fluctuates greatly, a smaller step size is used for fine search, which helps to avoid search errors caused by excessive step size ... Figure 5 As shown, the technical solution adopted in the embodiment of the present application is compared with the fixed step size of the traditional PSO by dynamically changing the particle motion step size, and the convergence speed of the present application (lower curve) and the convergence speed of the traditional PSO (upper curve) on the objective function value.

[0093] See the subsequent steps for details.

[0094] Specifically, in step S22135, the particle motion change step length is obtained by analyzing the evaluation index score;

[0095] Specifically, the acquisition of the particle motion change step size includes the following steps:

[0096] Step S221351: collecting window evaluation index scores for each evaluation index under continuous time including the current time period; performing calculation and analysis based on the window evaluation index scores to obtain statistical data; the statistical data includes the standard deviation of the window evaluation index scores (measurement of the dispersion of the scores), the autocorrelation coefficient of the window evaluation index scores (calculation of the correlation of scores at adjacent time points, reflecting trend continuity), and the real-time coefficient of variation of the window evaluation index scores (eliminating the dimension effect, reflecting relative volatility);

[0097] Step S221352: Initialize the particle motion change step s0;

[0098] Step S221353: Calculating based on the initialized particle motion change step s0 and the statistical data to obtain the particle motion change step s1 to be determined in the next iteration;

[0099] The calculation method of the particle motion change step size s1 to be determined in the next iteration is:

[0100] s1=s0·(1+μ β )·e -ασγδλ ;

[0101] Where s0 is the initialization particle motion change step size; σ is the standard deviation of the window evaluation index score; μ is the autocorrelation coefficient of the window evaluation index score; δ is the real-time variation coefficient of the window evaluation index score; α, β, γ and λ are adjustment parameters used to control the nonlinear weight of each factor;

[0102] It should be noted that, in the above embodiment of the present application, the basic step length s0 is the initial particle motion change step length, representing the basic moving distance of the particle before any adjustment factors are applied; the autocorrelation term (1+μ β ), μ is the autocorrelation coefficient of the evaluation index score within the window, reflecting the continuity and trend consistency of historical data;

[0103] When μ is large, it means that the evaluation index shows strong continuity. β This effect is amplified to increase the particle's step length, which is conducive to rapid and in-depth search in the stable region; the parameter β is used to increase or decrease the influence of the autocorrelation effect; the exponential decay term e -ασγδλ In the formula, σ represents the standard deviation of the evaluation index scores within the window, reflecting the data volatility; δ is the real-time coefficient of variation, reflecting the intensity of the real-time changes in the data; when σ or δ is large, it means that the data fluctuations or real-time changes are strong, and the value of the exponential term will be significantly reduced, thereby reducing the motion step size to avoid the search out of control due to excessive step size when the state fluctuates violently; the parameters α, γ, and λ respectively control the nonlinear influence of the standard deviation and the coefficient of variation when adjusting the step size. These parameters can be used to carefully characterize the impact of data fluctuations on the step size.

[0104] Step S221354: continuously record and analyze the fitness f obtained by M iterations in the process of obtaining the inertia fluctuation factor; based on the fitness f, modify the step size s1 of the particle motion change to be determined to obtain the step size S of the target particle motion change.

[0105] It should be noted that the above embodiment of the present application first collects window evaluation index scores within a continuous time window, and then calculates statistical data based on the window evaluation index scores, converting subjective "volatility" into objective statistical indicators, providing a basis for dynamic step size adjustment, and comprehensively capturing the stability characteristics of the evaluation indicators through the triple perspectives of standard deviation, autocorrelation coefficient, and coefficient of variation;

[0106] Furthermore, during the iteration process, the step size of the particle motion to be determined is controlled by statistical data, and the step size is adjusted accordingly according to the fluctuation of the data. When the evaluation indicators show that the team status is good and the data is relatively stable, a larger step size of particle motion change is allowed to speed up the search and escape the local optimum; if the team status is weak or the data fluctuates greatly, a smaller step size is used for fine search, which helps to avoid search errors caused by excessive step size; and, further, the step size is corrected through the fitness during continuous iteration to improve the robustness of the algorithm.

[0107] Specifically, in step S221355, the step length s1 of the particle motion change to be determined is corrected based on the fitness f to obtain the step length S of the target particle motion change, including the following steps:

[0108] Step S2213551: determining an iteration trend based on the fitness f; the iteration trend includes a convergence acceleration trend and an oscillation suppression trend;

[0109] The method of judging the iteration trend is that if the fitness f gradually decreases with the number of consecutive iterations, the iteration trend is a convergence acceleration trend (such as Figure 6 If the fitness f increases and decreases alternately under the number of consecutive iterations, the iteration trend is an oscillation suppression trend (as shown in Figure 7 shown);

[0110] Step S2213552: Calculating the target particle motion change step length S based on the iterative trend;

[0111] The calculation method of the target particle motion change step size S is:

[0112]

[0113] Where, f t-1 is the fitness at the previous iteration; f t is the fitness at the last iteration; M is the number of iterations of the fitness f obtained in the process of continuous recording and analysis to obtain the inertia fluctuation factor;

[0114] It should be noted that, in the above-mentioned embodiment of the present application, for the calculation of the step size of the target particle motion change with the convergence acceleration trend, as the number of fitness f continuously recorded during the iteration increases, the logarithmic function becomes slightly larger, reflecting that the step size can be appropriately increased when the system converges stably for a long time, thereby speeding up the search speed; when M increases, the logarithmic term increases, but the growth is slow (the logarithmic function is progressive), thereby ensuring that the step size is not too fast or too large, and avoiding crossing the optimal solution; thus, when it is determined that the overall trend has developed in the direction of convergence, a larger step size is used to quickly jump to a more optimal position;

[0115] Under the oscillation suppression trend, |f in the numerator t -f t-1 | represents the absolute difference between the fitness of the last two iterations. When this value is large, it means that the fitness fluctuates violently, that is, the system is oscillating. The f in the denominator t-1 It is used to normalize this fluctuation, which is equivalent to measuring the relative degree of change; the normalized difference is smoothed through the square root form (that is, large fluctuations are suppressed), and then the whole is in a fractional structure of 1 / (1+·). When the relative fluctuation is large, the entire fractional value becomes smaller, thereby making the step size S smaller, avoiding further instability caused by excessive step size jumps in the oscillation area; when the relative fluctuation is small, the fraction tends to 1, and the step size basically remains at s1, retaining a more refined search capability.

[0116] Research shows that during the execution of step S21, the communication overhead of the system (communication overhead refers to the amount of data transmitted by the system in each sampling period, bandwidth occupancy or delay, etc.) will fluctuate with the change of sampling frequency, and the historical evaluation index score will also fluctuate with the change of sampling frequency. When the communication overhead is high, the system will occupy more bandwidth, thereby bringing a larger data load, affecting the efficiency of subsequent data analysis; and when the historical evaluation index score fluctuates greatly, it proves that there are large differences in the historical evaluation index score, and there are more factors such as team environment or internal adjustments within the team that cause scoring uncertainty. Therefore, it is necessary to analyze each evaluation index score. The premise of the correlation coefficient between indicators is to obtain historical evaluation indicator scores with greater reference value, and to obtain historical evaluation indicator scores with greater reference value, it is necessary to obtain a more accurate sampling frequency; and when sampling data, traditional sampling methods are often through random sampling and other processing methods, which cannot provide an accurate data basis for the analysis of correlation relationships. Based on this, the embodiment of the present application proposes to obtain more robust historical evaluation indicator scores based on the analysis of sampling frequency, communication overhead and fluctuations in historical evaluation indicator scores, so as to provide an accurate data basis for the analysis of the correlation coefficient of subsequent evaluation indicators. For details, please see the subsequent steps S211-step S21x.

[0117] Specifically, in step S21, the sampling frequency interval is obtained based on the communication overhead corresponding to each sampling frequency and the historical evaluation index score fluctuation data corresponding to the sampling frequency, including the following steps:

[0118] Step S211: acquiring communication overhead at different sampling frequencies by sampling at different sampling frequencies; and acquiring volatility at different sampling frequencies by calculating the ratio of the standard deviation of the historical evaluation index scores to the mean of the historical evaluation index scores based on the historical evaluation index scores obtained by sampling at different sampling frequencies;

[0119] Step S212: establishing a first curve and a second curve in the same coordinate system based on the communication overheads at different sampling frequencies and the fluctuation rates at different sampling frequencies;

[0120] The first curve is a communication overhead curve at different sampling frequencies; the second curve is a volatility curve at different sampling frequencies; the X-axes of the first curve and the second curve are both sampling frequencies (f1, f2, f3, ..., fn);

[0121] Step S213: calculating a communication overhead variation trend between adjacent candidate sampling frequency points based on the first curve; calculating a fluctuation rate variation trend between adjacent candidate sampling frequency points based on the second curve;

[0122] It should be noted that the communication overhead change trend and the volatility change trend in the above-mentioned embodiment of the present application can be obtained by calculating the difference between adjacent points on the curve, that is, using the values ​​(communication overhead value or volatility value) at two adjacent candidate sampling frequency points to calculate the difference, that is, v2-v1 (v is the communication overhead value) or l2-l1 (l is the volatility value); when the difference is greater than 0, it is judged that the trend is an upward trend; when the difference is less than 0, it is judged that the trend is a downward trend; when the difference is equal to 0, it is judged that the trend is stable.

[0123] Step S214: Determine a simultaneous downward trend interval based on the communication overhead change trend and the volatility change trend corresponding to the continuous candidate sampling frequencies; determine the candidate sampling frequency interval corresponding to the simultaneous downward trend interval as the target sampling frequency interval (i.e., the above-mentioned sampling frequency interval).

[0124] It should be noted that in the above-mentioned embodiment of the present application, by selecting the communication overhead change rate and volatility change rate that are simultaneously in a downward trend under continuous candidate sampling frequencies as a simultaneous downward trend interval, at this time, the subsequent sampling frequency interval corresponding to the simultaneous downward trend interval is the target sampling frequency interval.

[0125] According to the above embodiment of the present application, the communication overhead and volatility obtained by sampling through the execution of the above step S211 are shown in Table 1.

[0126]

[0127]

[0128] By establishing the curve in step S212, as shown in FIG. Figure 8 As shown;

[0129] Further, based on Figure 8 The first curve and the second curve are used to obtain the communication overhead change rate and the fluctuation change rate of each candidate sampling frequency point through the operation of step S213, and trend analysis is performed to obtain:

[0130] From f1 to f3:

[0131] Communication overhead decreased from 119 to 101, and volatility decreased from 24.9% to 21.2%;

[0132] From f6 to f8:

[0133] Communication overhead decreased from 154 to 132, and volatility decreased from 31% to 27.1%;

[0134] From f12 to f13:

[0135] Communication overhead decreased from 72 to 59, and volatility decreased from 21.4% to 20.4%;

[0136] From f16 to f17:

[0137] Communication overhead dropped from 50 to 35, and volatility dropped from 26.9% to 26%;

[0138] In these four stages, both curves showed a significant downward trend;

[0139] For example, from f3 to f4:

[0140] Although volatility continues to decrease (27.1→24.9), communication overhead begins to increase (132→134);

[0141] Based on this, the target sampling frequency range is determined:

[0142] The commonly decreasing intervals are from f1 to f3, f6 to f8, f12 to f13, and f16 to f17, i.e., the target sampling frequency intervals are f1-f3, f6-f8, f12-f13, and f16-f17. Therefore, the above step of "obtaining the sampling frequency interval based on the analysis of the communication overhead corresponding to each sampling frequency and the historical evaluation index score fluctuation data corresponding to the sampling frequency" can obtain an optimal target sampling frequency interval (determining the candidate sampling frequency interval corresponding to the simultaneously decreasing trend interval as the target sampling frequency interval). The communication overhead and volatility corresponding to this special sampling frequency interval are both on a downward trend, so the historical evaluation index scores corresponding to this sampling frequency interval are the most reliable and have practical value.

[0143] Finally, sampling is performed based on the target sampling frequency interval to obtain the historical evaluation index scores of each evaluation index corresponding to each evaluation dimension (used for analyzing correlation relationships, so it does not need to consider missingness); however, since the current evaluation index scores are missing due to human factors, the embodiment of the present application believes that it is necessary to combine the correlation analysis and missing value filling processing processed by the adaptive particle swarm optimization algorithm to obtain the target evaluation index score.

[0144] In summary, the present invention proposes a team status assessment method based on an indicator dimension system, which first constructs a comprehensive and customized assessment system by presetting assessment dimensions and assessment indicators corresponding to the assessment dimensions; then, the target assessment indicator scores of the team are obtained based on the assessment indicators, and the assessment system is converted into computable data; the assessment dimension scores corresponding to the current team's assessment dimensions are obtained by accumulating the scores of the target assessment indicators based on each assessment dimension, which comprehensively reflects the overall performance of the team in each assessment dimension; the assessment dimension mapping scores are obtained by geometric mapping based on the assessment dimension scores in different years, and the annual differences are eliminated; the assessment dimension mapping scores are visualized and displayed through radar charts, which intuitively present the team status, assist in decision-making, and use radar charts to provide visual insights to help formulate improvement strategies such as training and resource allocation;

[0145] During the specific implementation process, historical evaluation index scores are used to establish a historical data benchmark to provide a data basis for subsequent dynamic analysis and data correction. Then, based on the current score combined with historical data, correlation analysis and missing value filling are carried out to improve the accuracy and rationality of the current score and solve the problem of incomplete data.

[0146] Furthermore, using historical evaluation indicator scores, we calculate the correlation coefficients between indicators and quantify their mutual influence. We also directly detect explicit missing values ​​through rule matching and mark missing evaluation indicators. For example, if a company's survey score field is "NULL," the system will automatically mark it as missing. For missing indicators, we select the first n indicators with the correlation coefficient, use the correlation coefficient as the weight, and calculate an estimated value based on the existing score. Finally, we merge the filled missing values ​​with the current non-missing values ​​to form a target evaluation indicator set with no missing values, providing complete data support for the subsequent calculation of evaluation dimension scores and the assessment of team status.

[0147] Furthermore, during the subsequent implementation process, the Pearson correlation coefficient was combined with the inertia fluctuation factor of the evaluation indicators obtained through adaptive particle swarm optimization to calculate and analyze the correlation coefficients between the evaluation indicators, making the calculation of the team's evaluation dimension scores more robust.

[0148] Among them, the fluctuation of evaluation index scores caused by internal team management is also taken into account, and a more robust inertia fluctuation factor is obtained by adaptively changing the step size of particle motion.

[0149] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. A person skilled in the art may modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A team status assessment method based on an indicator dimension system, characterized in that: The steps are as follows: Preset evaluation dimensions and corresponding evaluation indicators; Obtaining an initial evaluation index score for the team based on the evaluation index; analyzing and processing the initial evaluation index score by identifying missing values ​​and combining it with historical evaluation index scores through an adaptive particle swarm optimization algorithm to obtain a target evaluation index score; Based on the accumulation of the scores of each target evaluation indicator in each evaluation dimension, the evaluation dimension scores corresponding to each evaluation dimension of the current team are obtained; Based on the geometric mapping of the evaluation dimension scores in different years, the evaluation dimension mapping scores are obtained; The evaluation dimension mapping scores are visualized and displayed through radar charts.

2. A team status assessment method based on an indicator dimension system according to claim 1, characterized in that: The step of identifying missing values ​​of the initial evaluation index score and analyzing and processing the historical evaluation index score using an adaptive particle swarm optimization algorithm to obtain a target evaluation index score comprises the following steps: Obtain a sampling frequency interval based on the communication overhead corresponding to each sampling frequency and the historical evaluation index score fluctuation data corresponding to the sampling frequency; obtain the historical evaluation index score of each evaluation index corresponding to each evaluation dimension by sampling based on the sampling frequency interval; Based on the current evaluation index score, the target evaluation index score is obtained according to the historical evaluation index score combined with the correlation analysis and missing value filling processing processed by the adaptive particle swarm optimization algorithm.

3. A team status assessment method based on an indicator dimension system according to claim 2, characterized in that: The acquisition of the sampling frequency interval based on the communication overhead corresponding to each sampling frequency and the historical evaluation index score fluctuation data corresponding to the sampling frequency includes the following steps: By sampling at different sampling frequencies, the communication overhead at different sampling frequencies is obtained; and based on the historical evaluation index scores obtained by sampling at different sampling frequencies, the volatility at different sampling frequencies is obtained by calculating the ratio of the standard deviation of the historical evaluation index scores to the mean of the historical evaluation index scores; Establishing a first curve and a second curve in the same coordinate system based on the communication overhead at the different sampling frequencies and the fluctuation rate at the different sampling frequencies; Calculate the communication overhead change trend between adjacent candidate sampling frequency points based on the first curve; calculate the fluctuation rate change trend between adjacent candidate sampling frequency points based on the second curve; Determining a simultaneous decreasing trend interval based on the communication overhead change trend and the volatility change trend corresponding to the continuous candidate sampling frequencies; The candidate sampling frequency interval corresponding to the simultaneous downward trend interval is determined as the target sampling frequency interval.

4. A team status assessment method based on an indicator dimension system according to claim 3, characterized in that: The method of obtaining a target evaluation index score based on the current evaluation index score and the historical evaluation index score in combination with the correlation analysis and missing value filling processing processed by the adaptive particle swarm optimization algorithm includes the following steps: Identifying correlation coefficients between evaluation indicators based on the historical evaluation indicator scores by performing correlation analysis combined with adaptive particle swarm optimization processing; Identify missing values ​​based on the currently obtained evaluation index scores to obtain missing evaluation indicators; Performing missing value filling processing on the missing evaluation indicator based on the correlation coefficient to obtain a missing evaluation indicator score of the missing indicator; The missing evaluation indicator score based on the missing indicator is fused with the current evaluation indicator score to obtain a complete target evaluation indicator score.

5. A team status assessment method based on an indicator dimension system according to claim 4, characterized in that: The process of identifying the correlation coefficients between the evaluation indicators based on the historical evaluation indicator scores by combining the correlation analysis with the adaptive particle swarm optimization process comprises the following steps: Constructing an evaluation indicator matrix based on all the historical evaluation indicator scores; Calculate the Pearson correlation coefficient for any two elements in the evaluation index matrix; The correlation coefficient is calculated based on the Pearson correlation coefficient combined with the inertia fluctuation factor of the historical evaluation index score obtained by adaptive particle swarm optimization.

6. A team status assessment method based on an indicator dimension system according to claim 5, characterized in that: The correlation coefficient is calculated as follows: Among them, γ m,n Represents the correlation coefficient between evaluation index m and evaluation index n; a m,n It represents the Pearson correlation coefficient between the evaluation index m and the evaluation index n; N represents the number of evaluation indicators in the current evaluation dimension; f represents the inertial fluctuation factor of the evaluation index n.

7. A team status assessment method based on an indicator dimension system according to claim 6, characterized in that: The acquisition of the inertia fluctuation factor includes the following steps: Calculate the periodic volatility value of the jth evaluation indicator of the i-th team over a preset time period; and calculate the autocorrelation coefficient using the historical evaluation indicator scores; determining an initial candidate value based on the periodic volatility value and the autocorrelation coefficient; Initializing the particle swarm, using the initial candidate value as the initial particle; Calculating the fitness corresponding to the initial particles using a preset fitness function based on the initial particles; When it is determined that the fitness does not satisfy the fitness less than or equal to the fitness threshold, the fitness is sorted from high to low, and the initial particles corresponding to the top N fitness are selected and projected onto a two-dimensional plane based on a preset projection rule to obtain the two-dimensional coordinates corresponding to each initial particle; After updating the two-dimensional coordinates with a preset particle motion speed and a particle motion change step size, back-projection is performed based on a preset projection rule to obtain a first particle; Based on the first particle, the process returns to the initial particle for re-iteration until the fitness satisfies that the fitness is less than or equal to the fitness threshold, and the inertia fluctuation factor is output.

8. A team status assessment method based on an indicator dimension system according to claim 7, characterized in that: The acquisition of the particle motion change step length includes the following steps: Collecting window evaluation index scores of each evaluation index under continuous time including the current period; performing calculation and analysis based on the window evaluation index scores to obtain statistical data; Initialize the particle motion change step s0; Calculating the particle motion change step length s1 to be determined in the next iteration based on the initialized particle motion change step length s0 and the statistical data; The fitness f obtained by M iterations in the process of continuously recording and analyzing the inertia fluctuation factor is obtained; based on the fitness f, the step length s1 of the particle motion change to be determined is corrected to obtain the step length S of the target particle motion change.

9. A team status assessment method based on an indicator dimension system according to claim 8, characterized in that: The step of correcting the to-be-determined particle motion change step size s1 based on the fitness f to obtain the target particle motion change step size S includes the following steps: An iteration trend is determined based on the fitness f; the iteration trend includes a convergence acceleration trend and an oscillation suppression trend; The target particle motion change step length S is calculated based on the iterative trend.

10. A team status assessment method based on an indicator dimension system according to claim 9, characterized in that: The method for judging the iteration trend is that if the fitness f gradually decreases with the number of consecutive iterations, the iteration trend is a convergence acceleration trend; if the fitness f alternately increases and decreases with the number of consecutive iterations, the iteration trend is an oscillation suppression trend.

Images