A team performance evaluation method and system for regional layout optimization

Abstract

The application discloses a team performance evaluation method and system for regional layout optimization, and the method comprises the following steps: collecting heart rate variability (HRV) and electroencephalogram (EEG) data of team members in different preset regional layouts in real time; extracting a pure cognitive load feature vector; constructing a high-order dynamic interaction model based on a graph neural network (GNN) to generate a dynamic coordination state atlas; establishing a nonlinear dynamic system equation adjusted by an online reinforcement learning strategy network to calculate the Lyapunov exponent spectrum of the coordination state atlas; based on the cognitive load features and the coordination stability indexes, constructing a multi-objective optimization model containing double constraints of task performance and coordination stability, solving a Pareto optimal solution set, and outputting an optimal regional layout scheme considering efficiency and robustness. The application realizes accurate mapping from micro physiological state to macro team efficiency, and provides a scientific decision basis for team management in a complex physical space.

Description

Technical Field

[0001] This invention belongs to the fields of human factors engineering, computational social science and complex system dynamics, and in particular relates to a team performance evaluation method and system for regional layout optimization. Background Technology

[0002] In modern organizational management and human-computer interaction design, the impact of physical space layout on team performance is increasingly recognized. However, existing team performance evaluation systems primarily rely on post-event statistical indicators such as task completion time and error rate, lacking the ability to perceive and predict the team's internal micro-psychological and physiological states in real time. The fundamental reason for this is that traditional evaluation systems lack technical solutions capable of high-precision, low-latency synchronous acquisition of non-stationary physiological signals such as HRV and EEG from multiple members, and also lack algorithmic models for fusing and analyzing these signals with interactive event flows within a unified timeframe.

[0003] Limitations of existing assessment methods Traditional team performance evaluations typically employ questionnaires or simple task performance scores (such as accuracy and speed). While these methods reflect the final results, they exhibit significant lag and fail to reveal the implicit cumulative cognitive load effects caused by inappropriate spatial layout among individual team members during complex collaborative tasks. For example, in physical education or competitive sports, although heart rate monitoring devices are used to assess exercise intensity, they are mostly used for individual load monitoring, and few studies have applied this high-precision physiological signal acquisition to spatial layout optimization decisions for team collaboration.

[0004] The shortcomings of team interaction modeling Existing technologies for processing team interaction data often employ static social network analysis (SNA) or simple frequency statistics. This approach ignores the time-series dynamics and semantic complexity of team interactions. Especially in heterogeneous teams or high-pressure environments, asymmetric influence and deep semantic interactions among members are often simplified to linear relationships, leading to assessments that deviate from the true collaborative state. Because these methods are built on static graphs or temporal aggregation graphs, their models cannot characterize the instantaneous changes in interaction intensity and causal direction, thus exhibiting inherent limitations in predicting the temporal evolution of team states, particularly in identifying the critical points of collaborative breakdown.

[0005] Lack of quantitative characterization of cooperative stability Current evaluation models mostly focus on "maximizing task performance" while neglecting the system's "robustness." In complex systems theory, team collaboration can be viewed as a nonlinear dynamic system. When external disturbances (such as sudden changes in task difficulty) or internal load fluctuations exceed a certain threshold, the system may slide from orderly collaboration to chaotic instability. Existing technology lacks a mathematical tool to quantify this "colour stability decay" by analyzing the system's evolutionary trajectory, making it difficult to determine whether a certain layout merely improves efficiency or at the expense of the system's robustness.

[0006] Bottlenecks in data fusion and computing Existing multi-source data fusion schemes often remain at the level of simple data splicing, lacking adaptive filtering and decoupling algorithms for the non-stationary characteristics of physiological signals. Heart rate variability (HRV) and electroencephalogram (EEG) signals contain rich information about cognitive load, but are greatly affected by environmental noise and individual differences. If traditional Fourier transform or wavelet transform is used directly for analysis, spectral leakage or aliasing artifacts are easily introduced, resulting in insufficient accuracy in the calculation of the cognitive load index. Summary of the Invention

[0007] To address the shortcomings of the existing technology, this invention provides a team performance evaluation method for regional layout optimization, comprising the following steps: S1. Construct at least two virtual region layout models with different spatial topologies, and synchronously configure physiological data acquisition and interactive recording modules for each member; S2. Drive the team to perform the same collaborative task, and synchronously collect multimodal physiological time series data, task performance time series data and team interaction event stream data; S3. Perform Kalman filtering and missing data completion on the physiological time series data to obtain a smooth and complete physiological signal sequence; S4. Based on the physiological signal sequence, the individual cognitive load dynamic index of each member is calculated online using multi-scale variational mode decomposition and Bayesian adaptive weighting. S5. Input the team interaction event stream into the pre-trained "graph neural network + temporal attention" high-order interaction encoder to dynamically generate an asymmetric directed graph and calculate the team coupling strength of each member. S6. Using the cognitive load index and coupling strength as state variables, the nonlinear dynamic system with parameters adjusted online through a reinforcement learning policy network is solved to obtain the maximum Lyapunov exponent and the layout fitness decay factor. S7. Integrate the task performance time series data within the task cycle to obtain the static team performance total score; S8. With the objectives of minimizing the layout fitness decay factor and maximizing the total static performance score, establish a multi-objective optimization model and solve for the Pareto optimal solution set. S9. Based on the decision-maker preferences determined by the analytic hierarchy process, select the optimal layout from the Pareto solution set and output a regional layout optimization analysis report.

[0008] The multimodal physiological time-series data is acquired using a wireless wearable sensor array at a sampling frequency of no less than 256Hz. This sampling frequency configuration is used to ensure that the gamma band (typically >30Hz) in the electroencephalogram (EEG) signal and the high-frequency components in heart rate variability can be effectively resolved in order to capture millisecond-level cognitive load fluctuations.

[0009] In S3, the Kalman filter establishes a linear state-space model for each physiological signal and uses a prediction-update mechanism to fill in missing values, ensuring the continuity and reliability of subsequent model inputs.

[0010] The physiological signals in S4 include heart rate variability (HRV) and electroencephalogram (EEG) data.

[0011] In S4, the multi-scale variational mode decomposition adaptively adjusts the penalty factor α at different scales to match the center frequency and bandwidth of different rhythmic components in physiological signals (such as δ, θ, α, β, and γ waves in EEG), thereby suppressing mode aliasing and extracting rhythmic components that are strongly correlated with cognitive load.

[0012] In S5, the high-order interactive encoder first vectorizes the interactive events and constructs a dynamic temporal graph, then aggregates neighborhood information using a graph attention network, and captures temporal evolution using LSTM, finally outputting node embeddings and attention coefficients as edge weights.

[0013] In S8, the NSGA-II algorithm is used to solve the Pareto front, and in S9, AHP is used to assign weights to "cooperative stability" and "task performance" to determine the final layout scheme by weighted summation or ideal point method.

[0014] The collaborative task is divided into four stages: "information collection and sharing, independent analysis and decision-making, conflict identification and negotiation, and collaborative operation and execution". Quantifiable micro-performance indicators are collected accordingly and dynamically weighted and integrated into time-series data of single task performance.

[0015] The layout fitness decay factor and the static team performance score together constitute a repeatable quantitative indicator, which is used for horizontal comparison and optimization iteration between different spatial topology layouts.

[0016] This invention also proposes a team performance evaluation system for regional layout optimization using the above-mentioned method, comprising: Virtual layout and task-driven module; Multi-source data synchronous acquisition module; Data robustness preprocessing module with built-in Kalman filter; The individual cognitive load calculation module incorporates multi-scale variational mode decomposition and Bayesian weighted units. Team dynamic coupling analysis module, with built-in graph neural network + temporal attention encoder; An adaptive collaborative stability solving module, with a built-in reinforcement learning policy network and a GPU parallel Runge-Kutta engine, is used to solve nonlinear dynamic systems online and output the layout fitness decay factor; The multi-objective performance evaluation and report generation module is used to perform NSGA-II optimization, AHP weight decision-making, and output regional layout optimization analysis reports.

[0017] This invention effectively removes environmental noise and irrelevant components from physiological signals by introducing multi-scale VMD decomposition and Bayesian adaptive weighting, significantly improving the sensitivity and accuracy to instantaneous high-load states and load accumulation effects. By utilizing Lyapunov exponential spectroscopy to quantify cooperative stability, it can provide early warning of the critical point at which the team system transitions from order to disorder (instability), offering a forward-looking decision-making basis for layout adjustments. Specifically: This invention enhances the objectivity and real-time nature of assessments. It eliminates the lag inherent in subjective questionnaires by acquiring multimodal physiological signals (heart rate variability, EEG power ratio, etc.) in real time, combined with the preprocessing capabilities of edge computing nodes, enabling millisecond-level updates to the cognitive state of team members. This makes the assessment process independent of hindsight, relying instead on objective physiological data streams, significantly improving the reliability and validity of the assessment results.

[0018] This invention enhances the representation of complex interactions. By constructing a high-order interaction encoder based on graph neural networks (GNNs), it can capture complex temporal dependencies and deep semantic information among members. Compared to traditional static network analysis, this method can dynamically identify key opinion leader (KOL) nodes and accurately quantify their influence weights at different stages, thus more realistically reflecting the power structure and collaboration patterns within the team.

[0019] This invention achieves a dynamic trade-off between "efficiency" and "stability." It introduces a multi-objective optimization framework that simultaneously considers both static task performance scores and dynamic collaborative stability indicators (layout fitness decay factors). This addresses the one-sidedness of traditional performance-only evaluations. By generating a Pareto-optimal solution set, managers can select the regional layout scheme that best maintains the team's long-term stability while ensuring a certain level of task completion, thus avoiding the risk of team collapse due to excessive pursuit of short-term efficiency.

[0020] Breakthrough at the algorithm level. In terms of algorithm implementation, this invention adopts state estimation and missing data completion technology based on Kalman filtering, which ensures that the system can still maintain a continuous and smooth data stream input when the physiological signal acquisition equipment experiences a brief failure or data packet loss, thus guaranteeing the stability and continuity of the subsequent dynamic model solution. Attached Figure Description

[0021] The above and other objects, features, and advantages of exemplary embodiments of the present disclosure will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the present disclosure are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 This is a flowchart illustrating a team performance evaluation method for regional layout optimization according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating a team performance evaluation system for regional layout optimization according to an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0023] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0024] It should be understood that although the terms first, second, third, etc., may be used to describe... in the embodiments of the present invention, these... should not be limited to these terms. These terms are only used to distinguish... For example, first... may also be referred to as second... without departing from the scope of the embodiments of the present invention, and similarly, second... may also be referred to as first...

[0025] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0026] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0027] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device that includes said element.

[0028] The optional embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0029] Example 1 like Figure 1 As shown, this invention discloses a team performance evaluation method for regional layout optimization, comprising the following steps: S1. Construct at least two virtual region layout models with different spatial topologies, and synchronously configure physiological acquisition and interactive recording modules for each member to provide a controllable experimental environment for subsequent data acquisition. S2 drives the team to perform the same collaborative task, and based on the configuration in S1, synchronously collects multimodal physiological time-series data, task performance time-series data and team interaction event stream data; S3. For the physiological time series data collected in S2, Kalman filtering and missing data completion are performed to obtain a smooth and complete physiological signal sequence to ensure the continuity and reliability of subsequent signal analysis. S4. Based on the physiological signal sequence processed in S3, multi-scale variational mode decomposition and Bayesian adaptive weighting are used to calculate the individual cognitive load dynamic index of each member online, so as to decouple the pure features representing the cognitive state from the original signal. S5. Input the team interaction event stream data collected in S2 into the pre-trained "graph neural network + temporal attention" high-order interaction encoder to dynamically generate an asymmetric directed graph and calculate the team coupling strength of each member in order to quantify the dynamic collaborative relationship between members. S6. Using the cognitive load index obtained in S4 and the coupling strength obtained in S5 as state variables, construct a nonlinear dynamic system whose parameters are adjusted online by a reinforcement learning policy network. By simulating the evolution of the team's collaborative state, solve for its maximum Lyapunov exponent to obtain the layout fitness decay factor that characterizes the system's robustness. S7. Integrate the task performance time series data over the task cycle to obtain the static team performance total score. S8. With the objectives of minimizing the layout fitness decay factor and maximizing the total static performance score, establish a multi-objective optimization model and solve for the Pareto optimal solution set. S9. Based on the decision-maker preferences determined by the analytic hierarchy process, select the optimal layout from the Pareto solution set and output a regional layout optimization analysis report.

[0030] The multimodal physiological time-series data is acquired using a wireless wearable sensor array at a sampling frequency of no less than 256Hz. This sampling frequency configuration is used to ensure that the gamma band (typically >30Hz) in the electroencephalogram (EEG) signal and the high-frequency components in heart rate variability can be effectively resolved in order to capture millisecond-level cognitive load fluctuations.

[0031] In S3, the Kalman filter establishes a linear state-space model for each physiological signal and uses a prediction-update mechanism to fill in missing values, ensuring the continuity and reliability of subsequent model inputs.

[0032] The physiological signals in S4 include heart rate variability (HRV) and electroencephalogram (EEG) data.

[0033] In S4, the multi-scale variational mode decomposition adaptively adjusts the penalty factor α at different scales to match the center frequency and bandwidth of different rhythmic components in physiological signals (such as δ, θ, α, β, and γ waves in EEG), thereby suppressing mode aliasing and extracting rhythmic components that are strongly correlated with cognitive load.

[0034] In S5, the high-order interactive encoder first vectorizes the interactive events and constructs a dynamic temporal graph, then aggregates neighborhood information using a graph attention network, and captures temporal evolution using LSTM, finally outputting node embeddings and attention coefficients as edge weights.

[0035] In S8, the NSGA-II algorithm is used to solve the Pareto front, and in S9, AHP is used to assign weights to "cooperative stability" and "task performance" to determine the final layout scheme by weighted summation or ideal point method.

[0036] The collaborative task is divided into four stages: "information collection and sharing, independent analysis and decision-making, conflict identification and negotiation, and collaborative operation and execution". Quantifiable micro-performance indicators are collected accordingly and dynamically weighted and integrated into time-series data of single task performance.

[0037] The layout fitness decay factor and the static team performance score together constitute a repeatable quantitative indicator, which is used for horizontal comparison and optimization iteration between different spatial topology layouts.

[0038] Example 2 This invention proposes a team performance evaluation method for regional layout optimization, comprising the following steps: S1: Construct at least two virtual region layout models with different spatial topologies. The virtual region layout models include a strongly separated layout and a weakly separated layout. In each layout model, configure at least one physiological signal acquisition device and a task interaction behavior recording module for each member N in the team. S2: In the virtual area layout model, team members are driven to perform preset typical collaborative tasks. Within the task execution cycle T, multimodal physiological time-series data B of all members are synchronously collected and timestamped. N (t), Task performance time series data P N (t) and team interaction event stream data E N (t); S3: For the multimodal physiological time-series data B N (t) Perform state estimation and missing data completion preprocessing based on Kalman filtering to obtain a smooth and complete physiological signal sequence, including the ECG RR interval sequence H. N (t) and the power ratio sequence R of α waves to β waves in EEG signals N (t); S4: Based on the preprocessed physiological signal sequence, calculate the individual cognitive load dynamic index for each member N. The calculation formula is as follows: ; in, Let N be the individual cognitive load dynamic index at time t. δ represents the length of the dynamic integration time window, and δ is the width parameter of the Gaussian kernel function. For time integration variables, These are the k-th and l-th intrinsic mode functions extracted after multi-scale variational mode decomposition of the input signal, respectively, where K and L are the total number of modes decomposed from the ECG and EEG signals, respectively. These are the Bayesian adaptive weights of the k-th ECG modal component and the l-th EEG modal component at time t, respectively. These weights are determined based on the posterior probabilities obtained through Bayesian inference using historical data and current signal features, and are used to dynamically adjust the contribution of different physiological features in cognitive load assessment. S5: Transfer the team interaction event stream data E N(t) Input a pre-trained high-order interactive encoder based on graph neural network and temporal attention mechanism to construct a team dynamic interactive network, where nodes are members and asymmetric directed edge weights are between nodes. The encoder is based on the time window The semantic information, temporal dependencies, and higher-order interaction patterns of the interaction events between member i and member j are dynamically generated, and based on this network, the time window for each member N is calculated. Team coupling strength within ; S6: Establish a set of nonlinear dynamic system equations with parameters adjusted online by a reinforcement learning policy network to describe the dynamic index of the individual's cognitive load. Coupling strength with the team The time-varying coupling relationship between them is determined by solving the maximum Lyapunov exponent of this system of equations within the period T. The core indicator characterizing the overall collaborative stability of the team was obtained—the layout fitness decay factor. ; S7: Transfer the task performance time series data P N (t) The static team performance score is obtained by integrating over period T. ; S8: Decrease the layout fitness factor With the aforementioned static team performance score As two optimization objectives, a multi-objective optimization problem is constructed, and Pareto optimal solution set describing the performance trade-off between the two layouts is obtained by solving this problem. S9: Based on the Pareto optimal solution set and the decision-maker preferences determined by the Analytic Hierarchy Process (AHP), generate a comprehensive team layout performance evaluation result for each virtual region layout model, and output a regional layout optimization analysis report containing quantitative comparison data, Pareto front distribution map and optimization suggestions.

[0039] Among them, the multi-scale variational mode decomposition function in step S4 The specific implementation involves solving a multi-scale constrained variational problem to transform the original physiological signal f(t) (i.e., H) into a single signal. N (t) or R N (t) can be decomposed into a series of eigenmode functions with different center frequencies and bandwidths. The variational problem is: The constraints are To ensure complete reconstruction, among which The center frequencies of each mode are... The imaginary unit, This is a scale-dependent penalty factor used to balance reconstruction fidelity and modal bandwidth. For the Dirac function, This represents the convolution operation; the method adaptively adjusts the penalty factor at different scales. This technology enables refined separation of different frequency components in physiological signals. Compared with traditional single-scale decomposition, it can more effectively suppress modal aliasing and noise interference, thereby extracting purer rhythmic components that are strongly correlated with cognitive negative content.

[0040] Among them, the Bayesian adaptive weights in step S4 The update process is as follows: the posterior distribution of the previous time step is used as the prior of the current time step, and the intrinsic mode function (IMF) of the current time step is used as the prior. k Based on the likelihood relationship between energy and cognitive load labeled data (t), Bayes' theorem is used to... The posterior probability is calculated and normalized before being used as the weight. The cognitive load annotation data can be obtained offline through auxiliary signals such as heart rate and pupil diameter during preset high-difficulty cognitive task stages.

[0041] The specific implementation of the high-order interactive encoder based on graph neural network and temporal attention mechanism in step S5 is as follows: First, each interactive event (including initiator, receiver, timestamp, and text / voice content) is vectorized to form an event sequence; second, a dynamic temporal graph is constructed, where nodes represent team members, and the edges of the graph are updated at each time step based on the interactive events that occur; then, a graph attention network (GAT) is used as the basic unit to aggregate the neighborhood information of each node, where the attention coefficients are... The importance of neighbor node j to node i is characterized through a shared attention mechanism. Finally, the graph attention network is combined with a Long Short-Term Memory (LSTM) network to capture the evolution of the graph structure over time. The node embeddings output by the encoder are used to calculate the team coupling strength. The dynamically calculated attention weights are directly used as asymmetric edge weights. .

[0042] The core differential equation of the nonlinear dynamic system equations that are online adjusted by the reinforcement learning policy network in step S6 is: ; where Γ N (t) represents the collaborative state variable of member N at time t; Let be the team interaction state of member M at time τ; α(t) and β(t) are the dynamic parameters of the online adjustment of the reinforcement learning policy network; σ(t) is the time delay variable; Environmental noise; Define the coupling function; define the reward function R at time t. tThe negative value of the placement fitness decay factor is a weighted sum of task performance, used to guide the policy network in adjusting parameters α(t) and β(t); the reward function R... t Designed to match the maximum Lyapunov index The negative rate of change is related, that is By maximizing cumulative rewards, the policy network learns an optimal set of dynamically adjusted parameters, thereby reducing the placement fitness decay factor. It dynamically tends to a minimum during task execution. In the formula, k = 1, 2, ..., K. H K represents the eigenmode function index of the electrocardiogram signal. H The total number of ECG modes; l=1,2,...,K E K represents the eigenmode function index of the EEG signal. E This represents the total number of EEG modalities. and These are the amplitudes of the k-th ECG modal component and the l-th EEG modal component at time t, respectively. and These are the corresponding Bayesian adaptive weights.

[0043] Let λ max (t) represents the maximum Lyapunov exponent of the nonlinear dynamic system at time t, characterizing the local stability of the system at that time. The layout fitness decay factor is defined as: , which is the average value of the maximum Lyapunov exponent over the task period T. The smaller the value, the more stable the team coordination under this layout. In the formula, λ is the layout fitness decay factor mentioned in S6, which is the time-domain average of the maximum Lyapunov exponent over the task period T. The smaller the value, the more stable the team coordination under this layout and the stronger the system robustness.

[0044] Specifically, the state estimation and missing data completion preprocessing based on Kalman filtering in step S3 is performed as follows: for each physiological time-series signal (such as H) of each member... N (t) Establish a linear state-space model , where x t The hidden state vector z contains information such as the true value of the signal and its first derivative. t These are the noisy signal values ​​actually observed, where A is the state transition matrix and H is the observation matrix. These are process noise and observation noise, respectively; when a new data point z is received... t At that time, the Kalman filter update step is executed to obtain the optimal estimate of the current state. If no data is received at a certain point in time, then the prediction step is used. A predicted value is generated and used as supplementary data, thereby ensuring the continuity and reliability of the data sequence input into subsequent models.

[0045] Specifically, the multi-objective optimization and decision-making method described in steps S8 and S9 involves using the Non-Dominated Sorting Genetic Algorithm II (NSGA-II) as the solver, with different virtual region layout models as input variables, to minimize the layout fitness decay factor. And maximize the static team performance score Using the objective function as the objective function, iterative optimization is performed to obtain a set of solutions that cannot dominate each other, namely the Pareto optimal solution set. Subsequently, domain experts or managers are invited to use the Analytic Hierarchy Process (AHP) to construct a hierarchical structure of evaluation criteria, and to compare "cooperative stability" and "task performance" pairwise to obtain a quantified weight preference vector. Finally, this weight vector is applied to the Pareto optimal solution set, and the unique best layout scheme that best meets the decision-maker's preferences is selected from it by weighted summation or the ideal point method.

[0046] Example 3 like Figure 2 As shown, this invention also proposes a team performance evaluation system for regional layout optimization, comprising: The virtual layout and task-driven module is used to construct the at least two virtual area layout models and drive the team to perform the typical collaborative tasks. The multi-source data synchronous acquisition module is used to synchronously acquire and timestamp-align the multimodal physiological time-series data, task performance time-series data, and team interaction event stream data of all members; The data robustness preprocessing module integrates a Kalman filter for performing state estimation and missing data completion. The individual cognitive load calculation module is used to perform multi-scale variational mode decomposition and Bayesian adaptive weighted calculation to generate the individual cognitive load dynamic index for each member. The team dynamic coupling analysis module integrates a high-order interactive encoder based on graph neural networks and temporal attention mechanisms to perform calculations to construct a team dynamic interaction network and calculate the team coupling strength. The adaptive cooperative stability solving module integrates a reinforcement learning policy network and a high-performance numerical computing engine to perform the establishment and solution of the nonlinear dynamic system equations with online parameter adjustment by reinforcement learning, and to calculate the layout fitness decay factor. The multi-objective performance evaluation and report generation module is used to construct and solve multi-objective optimization problems, and generate the final regional layout optimization analysis report by combining the analytic hierarchy process (AHP).

[0047] The multi-source data synchronization acquisition module includes a software time synchronization service based on Network Time Protocol (NTP) or Precise Time Protocol (PTP). This service periodically calibrates the local clocks of all physiological signal acquisition devices and task interaction behavior recording modules to ensure that all acquired data streams have sub-millisecond timestamp consistency before being uploaded to the central server. Furthermore, the module also includes an edge computing gateway for performing preliminary buffering, formatting, and timestamp alignment of the raw data at the data acquisition end, reducing the data processing load on the central server.

[0048] The adaptive cooperative stability solving module is equipped with a graphics processing unit (GPU) as a dedicated coprocessor. The high-performance numerical computing engine in the module adopts a fourth-order Runge-Kutta algorithm based on a GPU parallel computing architecture to perform large-scale parallel numerical integration of the nonlinear dynamic system equations. Furthermore, the engine uses an orthogonalization method based on continuous QR decomposition to track the evolution of the system's Jacobian matrix, ensuring that the complete Lyapunov exponent spectrum can be calculated stably and efficiently in long-period integration, thereby meeting the near real-time requirements of the evaluation process.

[0049] The typical collaborative task is designed as a composite task flow comprising four stages: "information gathering and sharing," "independent analysis and decision-making," "conflict identification and negotiation," and "collaborative operation and execution"; the task performance time-series data P N The generation method for (t) is as follows: at each stage of the task, micro-performance indicators related to the core capabilities of that stage are recorded, such as "information transmission accuracy" in the information sharing stage, "decision quality score" in the independent decision-making stage, "consensus reaching time" in the conflict negotiation stage, and "goal completion rate" in the collaborative execution stage; finally, through a dynamic weighting function that automatically switches with the task stages, these multi-dimensional micro-performance time-series indicators are merged into a single task performance time-series data P. N (t).

[0050] Example 4 This disclosure provides a non-volatile computer storage medium storing computer-executable instructions that can perform the steps described in the above embodiments.

[0051] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0052] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0053] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (AN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0054] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0055] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0056] The preferred embodiments of the present invention have been described above to make the spirit of the present invention clearer and easier to understand, and are not intended to limit the present invention. All modifications, substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope summarized by the appended claims.

Claims

1. A team performance evaluation method for regional layout optimization, characterized in that, Includes the following steps: S1. Construct at least two virtual region layout models with different spatial topologies, and synchronously configure physiological acquisition and interactive recording modules for each member to provide a controllable experimental environment for subsequent data acquisition. S2 drives the team to perform the same collaborative task, and based on the configuration in S1, synchronously collects multimodal physiological time-series data, task performance time-series data and team interaction event stream data; S3. For the physiological time series data collected in S2, Kalman filtering and missing data completion are performed to obtain a smooth and complete physiological signal sequence to ensure the continuity and reliability of subsequent signal analysis. S4. Based on the physiological signal sequence processed in S3, multi-scale variational mode decomposition and Bayesian adaptive weighting are used to calculate the individual cognitive load dynamic index of each member online, so as to decouple the pure features representing the cognitive state from the original signal. S5. Input the team interaction event stream data collected in S2 into the pre-trained "graph neural network + temporal attention" high-order interaction encoder to dynamically generate an asymmetric directed graph and calculate the team coupling strength of each member in order to quantify the dynamic collaborative relationship between members. S6. Using the cognitive load index obtained in S4 and the coupling strength obtained in S5 as state variables, a nonlinear dynamic system with parameters adjusted online by a reinforcement learning policy network is constructed. By simulating the evolution of the team's collaborative state, its maximum Lyapunov exponent is solved to obtain the layout fitness decay factor that characterizes the robustness of the system. S7. Integrate the task performance time series data within the task cycle to obtain the static team performance total score; S8. With the objectives of minimizing the layout fitness decay factor and maximizing the total static performance score, establish a multi-objective optimization model and solve for the Pareto optimal solution set. S9. Based on the decision-maker preferences determined by the analytic hierarchy process, select the optimal layout from the Pareto solution set and output a regional layout optimization analysis report.

2. The method as described in claim 1, characterized in that, The multimodal physiological time-series data were acquired using a wireless wearable sensor array at a sampling frequency of no less than 256 Hz. This sampling frequency configuration was designed to ensure that the gamma band in the EEG signal and the high-frequency components in heart rate variability could be effectively resolved in order to capture millisecond-level cognitive load fluctuations.

3. The method as described in claim 1, characterized in that, The Kalman filter described in S3 establishes a linear state-space model for each physiological signal and uses a prediction-update mechanism to fill in missing values, ensuring the continuity and reliability of subsequent model inputs.

4. The method as described in claim 1, characterized in that, The physiological signals mentioned in S4 include heart rate variability (HRV) and electroencephalogram (EEG) data.

5. The method as described in claim 1, characterized in that, The multi-scale variational mode decomposition in S4 adaptively adjusts the penalty factor α at different scales to match the center frequency and bandwidth of different rhythmic components in the physiological signal, thereby suppressing mode aliasing and extracting rhythmic components that are strongly correlated with cognitive load.

6. The method as described in claim 1, characterized in that, The temporal attention mechanism in the high-order interactive encoder adopts a multi-head self-attention structure, which dynamically captures the long-range dependencies of interactive events in the time dimension by calculating the correlation between query vector, key vector and value vector.

7. The method as described in claim 1, characterized in that, In S8, the NSGA-II algorithm is used to solve the Pareto front. In S9, AHP is used to assign weights to "cooperative stability" and "task performance", and the final layout scheme is determined by weighted summation or ideal point method.

8. The method as described in claim 1, characterized in that, The collaborative task is divided into four stages: "information collection and sharing, independent analysis and decision-making, conflict identification and negotiation, and collaborative operation and execution". Quantifiable micro-performance indicators are collected accordingly and dynamically weighted and integrated into time-series data of single task performance.

9. The method as described in claim 1, characterized in that, The layout fitness decay factor and the static team performance score together constitute a repeatable quantitative indicator, which is used for horizontal comparison and optimization iteration between different spatial topology layouts.

10. A team performance evaluation system for implementing the method of any one of claims 1–9 for region layout optimization, comprising: Virtual layout and task-driven module; Multi-source data synchronous acquisition module; Data robustness preprocessing module with built-in Kalman filter; The individual cognitive load calculation module incorporates multi-scale variational mode decomposition and Bayesian weighted units. Team dynamic coupling analysis module, with built-in graph neural network + temporal attention encoder; An adaptive collaborative stability solving module, with a built-in reinforcement learning policy network and a GPU parallel Runge-Kutta engine, is used to solve nonlinear dynamic systems online and output the layout fitness decay factor; The multi-objective performance evaluation and report generation module is used to perform NSGA-II optimization, AHP weight decision-making, and output regional layout optimization analysis reports.

Images