CN-122018676-A - Aviation man-machine cooperative task allocation system and method based on cognitive load prediction

CN122018676ACN 122018676 ACN122018676 ACN 122018676ACN-122018676-A

Abstract

The invention provides an aviation man-machine cooperative task distribution system and method based on cognitive load prediction. By dynamically optimizing task allocation, pilot cognitive load is reduced, and task execution efficiency and flight safety are improved. The system realizes accurate monitoring and dynamic response of the cognitive state of the pilot through multidimensional data fusion, intelligent algorithm optimization and man-machine interaction innovation, and remarkably improves the intelligent level of aviation task execution.

Inventors

Liang Deda
ZHANG HUATIAN
ZHANG XU
QIU XUYI
DONG CHEN
Chang Lifan
YANG XIZHONG
Yuan Feiran

Assignees

中国航空无线电电子研究所

Dates

Publication Date: 20260512
Application Date: 20251227

Claims (9)

1. The aviation man-machine cooperative task distribution system based on cognitive load prediction is characterized by comprising a load prediction module, a task arbitration module and an adaptive training module, wherein the load prediction module is used for realizing dynamic monitoring and prediction of a pilot cognitive state by establishing a pilot cognitive model and fusing multi-source heterogeneous data including physiological data, flight history data and personalized characteristic data of the pilot, the task arbitration module is used for carrying out priority assessment on tasks from three dimensions of safety, timeliness and operation complexity based on a dynamic task priority matrix and realizing smooth transition of task distribution through a multi-mode interaction system, and the adaptive training module is used for generating a personalized training scheme based on cognitive load characteristics and optimizing training effects through a pressure situation simulation system so as to improve response capability of the pilot in a complex environment.
2. The cognitive load prediction-based aviation man-machine cooperative task allocation system according to claim 1, wherein the load prediction module adopts an LSTM-RNN-based load trend prediction algorithm, combines an attention mechanism and a graph rolling network, and predicts the cognitive load trend in a rolling manner by taking 5 seconds as a time window.
3. The aviation man-machine cooperative task allocation system based on cognitive load prediction according to claim 2 is characterized in that the LSTM-RNN algorithm is used for constructing a cognitive load prediction model by collecting physiological data, flight task data and personality characteristic data of pilots in real time and outputting a load change trend, and the model is used for realizing accurate prediction of cognitive load through multi-source data fusion and space-time characteristic extraction and supporting dynamic adjustment of prediction thresholds to adapt to different task environments.
4. The system for allocating the cooperative tasks of aviation and man-machine based on cognitive load prediction according to claim 1, wherein the task arbitration module calculates the priority weights of the tasks through a fuzzy analytic hierarchy process and an entropy method, generates an optimal dynamic task sequence, and updates the task queue in real time.
5. The aviation man-machine cooperative task distribution system based on cognitive load prediction is characterized in that the multi-mode interaction system specifically refers to a mixed instruction distribution system of voice interaction and tactile feedback, the voice interaction system adopts a natural language generation engine based on a transducer architecture to generate instructions conforming to cognitive habits of pilots, the tactile feedback device is integrated on a steering column, and encoded tactile signals are output through a high-frequency vibration module to ensure that the pilots receive task switching information within 0.5 seconds.
6. The aviation man-machine cooperative task allocation system based on cognitive load prediction according to claim 1, wherein the task arbitration module comprises an autopilot system takeover degree dynamic adjustment algorithm supporting gradual switching from full manual to full automatic, the algorithm adjusts the takeover degree of the autopilot system at a rate of 5% per second according to real-time cognitive load and task requirements of a pilot based on deep reinforcement learning and Markov Decision Process (MDP), when the cognitive load exceeds a threshold value, the system gradually increases the autopilot takeover proportion, and when the load falls back, the manual operation authority is restored at a rate of 3%/second, so that dynamic balance of man-machine cooperation is achieved.
7. The aviation man-machine cooperative task allocation system based on cognitive load prediction is characterized in that the self-adaptive training module generates a personalized training scheme based on cognitive load feature analysis to optimize task execution capacity of pilots, and through a transfer learning algorithm, the system can identify weak links of the pilots on cognitive load and design training contents in a targeted manner to improve performance of the pilots in high-load tasks.
8. The aviation man-machine cooperative task allocation system based on cognitive load prediction is characterized in that the self-adaptive training module develops a meta-cognitive monitoring technology, acquires self-evaluation data of pilots through an intelligent interactive interface and compares and analyzes the self-evaluation data with a system evaluation result, the system can dynamically optimize a training scheme by using a Bayesian optimization algorithm, help the pilots to better know self states, correct self-cognition deviation and improve the capability of coping with complex flight tasks, and in addition, the training data adopts a storage mechanism based on a blockchain technology to ensure the safety and traceability of the data.
9. An aviation man-machine cooperative task distribution method based on cognitive load prediction is characterized in that an aviation man-machine cooperative task distribution system based on cognitive load prediction is adopted, and the aviation man-machine cooperative task distribution method based on cognitive load prediction comprises the following steps: predicting cognitive load of the pilot through a load prediction module; When the cognitive load is predicted to exceed a preset threshold value within 5-10 seconds in the future, immediately starting a task arbitration module; The task arbitration module generates an optimal dynamic task sequence and updates the task queue in real time; and meanwhile, according to the predicted cognitive load of the pilot, an automatic driving system is adopted to take over a degree dynamic adjustment algorithm, so that the gradual switching from full manual to full automatic is realized.

Description

Aviation man-machine cooperative task allocation system and method based on cognitive load prediction Technical Field The invention belongs to the technical field of aviation man-machine cooperation, and particularly relates to an aviation man-machine cooperation task distribution system and method based on cognitive load prediction. Background Along with the rapid development of aviation technology, the complexity and diversity of modern aviation tasks are remarkably increased, pilots need to process heterogeneous mass data from multiple sources such as an onboard system, air traffic control communication, meteorological information, navigation data, control decisions and the like when performing tasks, and meanwhile, fast and accurate decisions need to be made under complex and changeable external environments and emergency situations. This presents an unprecedented challenge to the pilot's cognitive abilities. The conventional aeronautical man-machine cooperative system mostly adopts a fixed task allocation mode, namely, tasks are allocated according to preset task priorities and operation flows. However, the rigid man-machine interaction mode has obvious defects that firstly, the rigid man-machine interaction mode cannot adapt to the cognitive state of a pilot in the process of executing a task, cognitive load overload or resource waste is easy to cause, secondly, a fixed task allocation strategy is difficult to cope with sudden situations and rapid changes of task environments, and finally, the mode ignores the influence of individual differences of the pilot on the task execution effect. These problems severely limit the safety and execution efficiency of the aeronautical tasks. In an actual flight mission, the pilot's cognitive load is complexly affected by multidimensional factors. Task complexity, information density, time pressure, multi-task parallel requirements and the like directly determine the cognitive requirement level from the task dimension, meteorological conditions (such as turbulence and visibility), airspace situations (such as air traffic density and threat target distribution), electromagnetic environments (such as communication interference and radar signal strength) and other external factors can significantly influence the cognitive load from the environment dimension, and physiological states (such as fatigue degree, stress level and biological rhythm) of pilots, psychological characteristics (such as working memory capacity, attention distribution capacity and emotion stability) and professional skill levels (such as task experience and special emotion treatment capacity) of pilots can have important influence on the cognitive load from the individual dimension. When cognitive load exceeds the pilot's capacity, operational errors, reaction delays, and even task failure may result. Studies have shown that in complex aviation tasks, more than 60% of human error is not relevant to cognitive load management. Therefore, how to monitor the cognitive state of the pilot in real time and optimize the task allocation according to the dynamic change of the cognitive state becomes a key scientific problem to be solved in the field of aviation man-machine cooperation. The existing cognitive load monitoring technology mainly has the following limitations that firstly, the traditional task allocation system lacks of intelligence and self-adaption and cannot be dynamically adjusted according to the real-time cognitive state of a pilot. Although some researches try to introduce a machine learning algorithm to perform task allocation optimization, the methods often have the following problems that (1) model training data and actual task environments have differences, so that generalization capability is insufficient, (2) influence of individual differences of pilots on task execution effects is ignored, (3) adaptability to dynamic changes of task environments is lacking, (4) multi-source heterogeneous data (such as physiological data, task data and environmental data) cannot be integrated effectively for comprehensive analysis, and (5) instantaneity and predictability are lacking, and early warning and intervention are difficult to perform before cognitive load overload. These problems severely restrict the application effect of the existing methods in practical aviation tasks. Disclosure of Invention The invention aims to solve the problem that a fixed task allocation mode in the existing aviation man-machine cooperative system cannot adapt to the dynamic cognitive state of a pilot, and provides an aviation man-machine cooperative task allocation system and method based on cognitive load prediction. The system realizes accurate monitoring and dynamic response of the cognitive state of the pilot through multidimensional data fusion, intelligent algorithm optimization and man-machine interaction innovation, and remarkably improves the intelligent level of aviation task execution. According to th