A neural computing-based pilot safety risk analysis method and system

Abstract

This invention discloses a method and system for pilot safety risk analysis based on neural computation. The method includes: identifying safety risk factors; establishing a mapping relationship between safety risk factors and unsafe pilot behaviors; establishing a structural model of the cortico-basal ganglia-thalamus neural circuit; establishing AMPA and GABA neural synapse computational models based on the theories of double exponential synapses and α-β synapses, respectively; establishing a computational model of the cortico-basal ganglia-thalamus neural circuit; calculating the thalamic discharge rate under the influence of different safety risk factors, and simulating the regulatory mechanisms of dopamine and acetylcholine during the formation of safety risks. This invention conducts safety analysis from the perspective of pilot neurodynamics, providing key physiological basis and theoretical support for improving risk analysis models and optimizing system design by analyzing the neural regulatory mechanisms of acetylcholine and dopamine during the formation of safety risks.

Description

Technical Field

[0001] This invention belongs to the aviation field and relates to pilot safety risk analysis, specifically to a pilot safety risk analysis method and system based on neural computing. Background Technology

[0002] The widespread application of intelligent interaction technology in cockpits has improved the efficiency of human-machine systems, but it has also brought new challenges to safety risk analysis. Existing research on safety risks in intelligent human-machine interaction mainly focuses on analyzing the intelligent human-machine interaction process in the cockpit, modeling pilot situational awareness and cognitive processes, and analyzing the oscillation rhythms of pilot neural signals. These methods can describe the formation process of safety risks in intelligent human-machine interaction from a macroscopic perspective. However, the neurodynamic mechanisms of pilots during the formation of safety risks remain unclear. The microscopic neural activities of pilots during operation, such as the firing patterns of neurons in the brain, changes in neurotransmitter levels, and synaptic plasticity, have a direct impact on their cognitive abilities, decision-making, and behavioral responses. Revealing the microscopic neurodynamic mechanisms influencing the formation of safety risks in intelligent human-machine interaction can analyze pilots' operational behaviors that drive or inhibit potential risks in different situations, providing crucial physiological evidence and theoretical support for improving risk analysis models and optimizing system design. Summary of the Invention

[0003] Purpose of the invention: In order to overcome the shortcomings of the existing technology, this invention provides a pilot safety risk analysis method and system based on neural computing, which solves the problem that traditional safety risk analysis models only focus on macro factors such as behavior and environment and lack analysis of the individual neurophysiological state of pilots.

[0004] Technical Solution: To achieve the above objectives, this invention provides a pilot safety risk analysis method based on neural computing, comprising the following steps:

[0005] S1: Identify safety risk factors;

[0006] S2: Establish a mapping relationship between safety risk factors and unsafe behaviors of pilots, and abstract visual, auditory, tactile and vestibular senses into electrical stimulation of the cortex;

[0007] S3: Establish a structural model of the cortical-basal ganglia-thalamic neural circuit;

[0008] S4: Based on the theories of bi-exponential synapses and α-β synapses, computational models of AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) and GABA (γ-aminobutyric acid) neural synapses were established respectively.

[0009] S5: By combining the structural model of the cortico-basal ganglia-thalamus neural circuit, the AMPA neural synapse computational model, and the GABA neural synapse computational model, a computational model of the cortico-basal ganglia-thalamus neural circuit is established.

[0010] S6: Using a computational model of the cortical-basal ganglia-thalamic neural circuit, the thalamic discharge rate under the influence of different safety risk factors is calculated, simulating the regulatory mechanisms of dopamine and acetylcholine in the formation of safety risks.

[0011] Furthermore, the safety risk factors in step S1 include air traffic control, airborne systems, and the environment;

[0012] Air traffic control safety risk factors include information transmission errors, read-back-hear-back errors, failure to acknowledge or respond to instructions, and non-standard terminology;

[0013] Transmission error: A transmission error refers to a situation where information is interfered with or lost during communication between pilots and air traffic controllers. Such errors may stem from technical problems such as signal instability, noise interference, or frequency mismatch.

[0014] Read-back-hear-back error: A read-back-hear-back error occurs when a pilot misreads an air traffic controller's instructions, or when a controller fails to detect an error while listening to a pilot's confirmation. This error may stem from pilot distraction, fatigue, or the effects of ambient noise.

[0015] Failure to acknowledge or respond to instructions: This refers to a pilot's failure to acknowledge an air traffic controller's instructions as required, or the air traffic controller's failure to receive acknowledgment within the stipulated time. This situation may prevent controllers from determining whether the pilot has received and understood the instructions, thus hindering effective subsequent air traffic management.

[0016] Non-standard terminology: Non-standard terminology refers to the use of terms or expressions that do not conform to aviation communication standards during communication. This may stem from the language habits of pilots or air traffic controllers, cultural differences, or impromptu expressions in emergency situations.

[0017] Airborne systems refer to the various subsystems on an aircraft. Pilots send commands to these subsystems via the cockpit human-machine interface to further control flight parameters and heading, ensuring safe flight. Safety risk factors for airborne systems include design flaws, technical system malfunctions, and human-machine trust issues.

[0018] Design flaws: Design flaws mainly refer to unreasonable aspects of the cockpit human-machine system interface design, information presentation methods, and operation logic design. These include unintuitive and inconsistent interfaces, and complex operation of automated systems, which can easily lead to confusion or misoperation by pilots during operation.

[0019] Technical system failure: This category of risk covers the malfunction or erroneous triggering of technical systems such as autopilot systems, avionics systems, and communication systems.

[0020] Human-machine trust: Human-machine trust is mainly reflected in the following two aspects. Automation dependence: Pilots over-rely on automated / intelligent systems; Distrust of automation: Some pilots are skeptical of automated / intelligent systems, leading to excessive monitoring or frequent intervention.

[0021] Starting from the pilot's primary sensory systems, environmental safety risk factors include visual load, auditory load, tactile perception, and vestibular perception.

[0022] Visual load: Visual load refers to the situation in which pilots need to process a large amount of information in a complex visual environment.

[0023] Auditory load: Auditory load refers to a pilot’s ability to process auditory information in the presence of noise or multiple audio signal interference.

[0024] Tactile perception: Tactile perception is the pilot's ability to sense the response of control equipment through tactile feedback.

[0025] Vestibular perception: Vestibular perception is involved in a pilot's sense of balance and spatial awareness.

[0026] Furthermore, the structural model of the cortico-basal ganglia-thalamus neural circuit in step S3 includes direct pathway, indirect pathway, and superdirect pathway structures;

[0027] The direct pathway's structure is cortex → striatum → medial globus pallidus → thalamus → cortex. Dopamine binds to dopamine D1 receptors in the striatum, activating Gs-type G protein of G protein-coupled receptors, thereby enhancing adenylate cyclase activity, increasing cyclic adenosine monophosphate synthesis, and activating protein kinase A. Protein kinase A, by regulating ion channels and other downstream effectors, increases the activity of GABAergic neurons in the striatum, inhibits the activity of the medial globus pallidus, reduces its inhibitory effect on the thalamus, and promotes the transmission of motor signals from the thalamus to the cortex.

[0028] The indirect pathway's structure is: cortex → striatum → lateral globus pallidus → hypothalamic nucleus → medial globus pallidus → thalamus → cortex. Dopamine binds to dopamine D2 receptors in the striatum, activating Gi-type G protein of G protein-coupled receptors, inhibiting adenylate cyclase, reducing cyclic adenosine monophosphate levels, and decreasing protein kinase A activity. This effect weakens the inhibitory effect of GABAergic neurons in the striatum, thereby enhancing the activity of the lateral globus pallidus. The lateral globus pallidus sends glutamatergic signals through the hypothalamic nucleus (subthalamic nucleus), exciting the medial globus pallidus, which further inhibits the transmission of motor signals from the thalamus to the cortex through its GABAergic neurons.

[0029] The superdirect pathway's structure is: cortex → hypothalamic nucleus → medial globus pallidus → thalamus → cortex. GABAergic neurons in the cerebral cortex project directly to the hypothalamic nucleus, forming the superdirect pathway through extensive connections between the hypothalamic nucleus and the medial globus pallidus. GABAergic neurons in the medial globus pallidus directly regulate thalamic nuclei, while the thalamus connects to the cerebral cortex via GABAergic nerve fibers. This connection triggers a dynamic cycle of "inhibition-enhancement-inhibition" in the electrical activity of cortical neurons. Therefore, changes in the activity of GABAergic neurons in the medial globus pallidus, the basal ganglia output nucleus, directly alter the final efferent information from the basal ganglia, thus precisely initiating and stopping motor plans while inhibiting unnecessary movements.

[0030] Furthermore, the construction of the AMPA and GABA neural synapse computational models in step S4 includes:

[0031] Neurons connect via synapses, forming signal transmission pathways. The action potential of the presynaptic neuron is transmitted to the postsynaptic neuron after a certain time delay. Therefore, the synaptic current is calculated as follows:

[0032] I syn =g max s(ν-E syn )

[0033] Among them, I syn It is the postsynaptic current, g max and E syn denoted as maximum synaptic conductance and synaptic back electromotive force, respectively, and s represents synaptic variable;

[0034] AMPA synaptic channels are simulated by bi-exponential synapses:

[0035]

[0036] Where, τ r and τ d These are the rise time and the decay time, respectively. Normalization factor

[0037] GABA synaptic channels are simulated by α-β type synapses:

[0038]

[0039] Where s is the synaptic variable, α is the synaptic activation rate constant, β is the deactivation rate constant, and [T] is the concentration of neurotransmitters released by the presynaptic neuron. When the synapse is triggered, [T] is usually a fixed constant.

[0040] The dopamine coefficient is the multiplier of the maximum synaptic conductance of all neural synaptic models in the direct pathway, and the acetylcholine coefficient is the multiplier of the maximum synaptic conductance of all neural synaptic models in the indirect pathway. Both are coefficients used for testing.

[0041] Furthermore, in step S5, the Hodgkin-Huxley equation describes the membrane potential changes of each neuron, and the AMPA and GABA models describe how neurons interact through excitatory and inhibitory synapses to achieve complex communication between neurons.

[0042] The neuronal model of the computational model of the cortical-basal-thalamic neural circuit is expressed as follows:

[0043] The Hodgkin-Huxley model is described by the following differential equation:

[0044]

[0045] Where V represents the membrane potential and C is the membrane capacitance. and g leak These are the maximum conductivity values ​​of the sodium ion channel, potassium ion channel, and leakage channel, respectively, E. Na E K and E leak These are the inversion potentials of the corresponding channels, and I(t) is the external input current;

[0046] The dynamic behavior of ion channels is represented by the following set of equations:

[0047]

[0048] Where m, h, and n represent the gating variables of the sodium ion channel and the potassium ion channel, respectively, and α x and β x It is a rate constant dependent on the membrane potential, and its form is as follows:

[0049]

[0050]

[0051] Furthermore, the method for calculating the average thalamic discharge rate in step S6 includes the following steps:

[0052] A1: Calculate the instantaneous discharge rate

[0053]

[0054] Among them, t i δ represents the i-th firing time of the neuron, and δ is the Dirac function used to capture the firing event at a specific time.

[0055] A2: Calculate the average discharge rate

[0056]

[0057] Where t0 is the start time of the time window, and T is the duration of the window.

[0058] This invention provides a pilot safety risk analysis system based on neural computing, comprising:

[0059] The safety risk analysis module is used to analyze the correlation between unsafe behaviors of pilots and safety risks during flight;

[0060] The model building module is used to build neural computational models based on the direct, indirect, and superdirect pathways of the cortico-basal ganglia-thalamus neural circuit; the neural computational models include the Hodgkin-Huxley model, AMPA, and GABA synapse models;

[0061] The analysis module is used to calculate the average thalamic discharge rate under different cortical current stimulation and different acetylcholine and dopamine coefficient stimulation, and to analyze the regulatory mechanism of acetylcholine and dopamine in the formation of safety risks.

[0062] Beneficial effects: Compared with existing technologies, this invention conducts safety analysis from the perspective of pilot neurodynamics, solving the problem that traditional safety risk analysis models only focus on macro-factors such as behavior and environment and lack analysis of individual pilot neurophysiological states. By analyzing the neural regulation mechanism of acetylcholine and dopamine in the formation of safety risks, this invention provides key physiological basis and theoretical support for improving risk analysis models and optimizing system design. Attached Figure Description

[0063] Figure 1 This is a schematic flowchart of the method of the present invention;

[0064] Figure 2 A structural model diagram of the cortical-basal ganglia-thalamic neural circuit;

[0065] Figure 3 This is a picture of the experimental site;

[0066] Figure 4 The graph shows the average thalamic discharge rate as a function of the experimental groups. Detailed Implementation

[0067] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0068] like Figure 1 As shown, this invention provides a pilot safety risk analysis method based on neural computing, comprising the following steps:

[0069] S1: Identify security risk factors;

[0070] Safety risk factors include air traffic control, airborne systems, and the environment;

[0071] Air traffic control safety risk factors include information transmission errors, read-back-hear-back errors, failure to acknowledge or respond to instructions, and non-standard terminology;

[0072] Transmission error: A transmission error refers to a situation where information is interfered with or lost during communication between pilots and air traffic controllers. Such errors may stem from technical problems such as signal instability, noise interference, or frequency mismatch.

[0073] Read-back-hear-back error: A read-back-hear-back error occurs when a pilot misreads an air traffic controller's instructions, or when a controller fails to detect an error while listening to a pilot's confirmation. This error may stem from pilot distraction, fatigue, or the effects of ambient noise.

[0074] Failure to acknowledge or respond to instructions: This refers to a pilot's failure to acknowledge an air traffic controller's instructions as required, or the air traffic controller's failure to receive acknowledgment within the stipulated time. This situation may prevent controllers from determining whether the pilot has received and understood the instructions, thus hindering effective subsequent air traffic management.

[0075] Non-standard terminology: Non-standard terminology refers to the use of terms or expressions that do not conform to aviation communication standards during communication. This may stem from the language habits of pilots or air traffic controllers, cultural differences, or impromptu expressions in emergency situations.

[0076] Airborne systems refer to the various subsystems on an aircraft. Pilots send commands to these subsystems via the cockpit human-machine interface to further control flight parameters and heading, ensuring safe flight. Safety risk factors for airborne systems include design flaws, technical system malfunctions, and human-machine trust issues.

[0077] Design flaws: Design flaws mainly refer to unreasonable aspects of the cockpit human-machine system interface design, information presentation methods, and operation logic design. These include unintuitive and inconsistent interfaces, and complex operation of automated systems, which can easily lead to confusion or misoperation by pilots during operation.

[0078] Technical system failure: This category of risk covers the malfunction or erroneous triggering of technical systems such as autopilot systems, avionics systems, and communication systems.

[0079] Human-machine trust: Human-machine trust is mainly reflected in the following two aspects. Automation dependence: Pilots over-rely on automated / intelligent systems; Distrust of automation: Some pilots are skeptical of automated / intelligent systems, leading to excessive monitoring or frequent intervention.

[0080] Starting from the pilot's primary sensory systems, environmental safety risk factors include visual load, auditory load, tactile perception, and vestibular perception.

[0081] Visual load: Visual load refers to the situation in which pilots need to process a large amount of information in a complex visual environment.

[0082] Auditory load: Auditory load refers to a pilot’s ability to process auditory information in the presence of noise or multiple audio signal interference.

[0083] Tactile perception: Tactile perception is the pilot's ability to sense the response of control equipment through tactile feedback.

[0084] Vestibular perception: Vestibular perception is involved in a pilot's sense of balance and spatial awareness.

[0085] S2: Establish a mapping relationship between safety risk factors and unsafe behaviors of pilots, and abstract visual, auditory, tactile and vestibular senses into electrical stimulation of the cortex;

[0086] The mapping relationship between safety risk factors and unsafe pilot behaviors is shown in Table 1;

[0087] Table 1. Mapping relationship between safety risk factors and unsafe pilot behaviors.

[0088]

[0089] S3: Establish a structural model of the cortical-basal ganglia-thalamic neural circuit;

[0090] like Figure 2 As shown, the structural model of the cortico-basal ganglia-thalamus neural circuit includes direct pathways, indirect pathways, and superdirect pathway structures;

[0091] The direct pathway's structure is cortex → striatum → medial globus pallidus → thalamus → cortex. Dopamine binds to dopamine D1 receptors in the striatum, activating Gs-type G protein of G protein-coupled receptors, thereby enhancing adenylate cyclase activity, increasing cyclic adenosine monophosphate synthesis, and activating protein kinase A. Protein kinase A, by regulating ion channels and other downstream effectors, increases the activity of GABAergic neurons in the striatum, inhibits the activity of the medial globus pallidus, reduces its inhibitory effect on the thalamus, and promotes the transmission of motor signals from the thalamus to the cortex.

[0092] The indirect pathway's structure is: cortex → striatum → lateral globus pallidus → hypothalamic nucleus → medial globus pallidus → thalamus → cortex. Dopamine binds to dopamine D2 receptors in the striatum, activating Gi-type G protein of G protein-coupled receptors, inhibiting adenylate cyclase, reducing cyclic adenosine monophosphate levels, and decreasing protein kinase A activity. This effect weakens the inhibitory effect of GABAergic neurons in the striatum, thereby enhancing the activity of the lateral globus pallidus. The lateral globus pallidus sends glutamatergic signals through the hypothalamic nucleus (subthalamic nucleus), exciting the medial globus pallidus, which further inhibits the transmission of motor signals from the thalamus to the cortex through its GABAergic neurons.

[0093] The superdirect pathway's structure is: cortex → hypothalamic nucleus → medial globus pallidus → thalamus → cortex. GABAergic neurons in the cerebral cortex project directly to the hypothalamic nucleus, forming the superdirect pathway through extensive connections between the hypothalamic nucleus and the medial globus pallidus. GABAergic neurons in the medial globus pallidus directly regulate thalamic nuclei, while the thalamus connects to the cerebral cortex via GABAergic nerve fibers. This connection triggers a dynamic cycle of "inhibition-enhancement-inhibition" in the electrical activity of cortical neurons. Therefore, changes in the activity of GABAergic neurons in the medial globus pallidus, the basal ganglia output nucleus, directly alter the final efferent information from the basal ganglia, thus precisely initiating and stopping motor plans while inhibiting unnecessary movements.

[0094] S4: Based on the theories of bi-exponential synapses and α-β synapses, computational models of AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) and GABA (γ-aminobutyric acid) neural synapses were established respectively.

[0095] Neurons connect via synapses, forming signal transmission pathways. The action potential of the presynaptic neuron is transmitted to the postsynaptic neuron after a certain time delay. Therefore, the synaptic current is calculated as follows:

[0096] I syn =g max s(ν-E syn )

[0097] Among them, I syn It is the postsynaptic current, g max and E syn denoted as maximum synaptic conductance and synaptic back electromotive force, respectively, and s represents synaptic variable;

[0098] AMPA synaptic channels are simulated by bi-exponential synapses:

[0099]

[0100] Where, τ r and τ d These are the rise time and the decay time, respectively. Normalization factor

[0101] GABA synaptic channels are simulated by α-β type synapses:

[0102]

[0103] Where s is the synaptic variable, α is the synaptic activation rate constant, β is the deactivation rate constant, and [T] is the concentration of neurotransmitters released by the presynaptic neuron. When the synapse is triggered, [T] is usually a fixed constant.

[0104] The dopamine coefficient is the multiplier of the maximum synaptic conductance of all neural synaptic models in the direct pathway, and the acetylcholine coefficient is the multiplier of the maximum synaptic conductance of all neural synaptic models in the indirect pathway. Both are coefficients used for testing.

[0105] S5: The Hodgkin-Huxley equation describes the membrane potential changes of each neuron, while the AMPA and GABA models describe how neurons communicate through excitatory and inhibitory synapses. Combining the structural model of the cortico-basal ganglia-thalamus neural circuit, the AMPA synaptic computational model, and the GABA synaptic computational model, a computational model of the cortico-basal ganglia-thalamus neural circuit is established, expressed as follows:

[0106] The Hodgkin-Huxley model is described by the following differential equation:

[0107]

[0108] Where V represents the membrane potential and C is the membrane capacitance. and g leak These are the maximum conductivity values ​​of the sodium ion channel, potassium ion channel, and leakage channel, respectively, E. Na E K and E leak These are the inversion potentials of the corresponding channels, and I(t) is the external input current;

[0109] The dynamic behavior of ion channels is represented by the following set of equations:

[0110]

[0111] Where m, h, and n represent the gating variables of the sodium ion channel and the potassium ion channel, respectively, and α x and β x It is a rate constant dependent on the membrane potential, and its form is as follows:

[0112]

[0113]

[0114] The number of neuron groups and synaptic connection parameters of the cortical-basal ganglia-thalamus neural circuit computational model are set as shown in Tables 2 and 3.

[0115] Table 2. Number of neurons in each population

[0116]

[0117] Table 3. Parameter settings for AMPA and GABA synaptic connections.

[0118]

[0119] S6: Using a computational model of the cortical-basal ganglia-thalamic neural circuit, the thalamic discharge rate under the influence of different safety risk factors is calculated, simulating the regulatory mechanisms of dopamine and acetylcholine in the formation of safety risks.

[0120] The method for calculating the average discharge rate of the thalamus includes the following steps:

[0121] A1: Calculate the instantaneous discharge rate

[0122]

[0123] Among them, t i δ represents the i-th firing time of the neuron, and δ is the Dirac function used to capture the firing event at a specific time.

[0124] A2: Calculate the average discharge rate

[0125]

[0126] Where t0 is the start time of the time window, and T is the duration of the window.

[0127] To achieve the above method, the present invention also provides a pilot safety risk analysis system based on neural computing, comprising:

[0128] The safety risk analysis module is used to analyze the correlation between unsafe behaviors of pilots and safety risks during flight;

[0129] The model building module is used to build neural computational models based on the direct, indirect, and superdirect pathways of the cortico-basal ganglia-thalamus neural circuit; the neural computational models include the Hodgkin-Huxley model, AMPA, and GABA synapse models;

[0130] The analysis module is used to calculate the average thalamic discharge rate under different cortical current stimulation and different acetylcholine and dopamine coefficient stimulation, and to analyze the regulatory mechanism of acetylcholine and dopamine in the formation of safety risks.

[0131] The dopamine coefficient is the multiplier of the maximum synaptic conductance of all neural synaptic models in the direct pathway, and the acetylcholine coefficient is the multiplier of the maximum synaptic conductance of all neural synaptic models in the indirect pathway. Both are coefficients used for testing.

[0132] To verify the effectiveness and actual effect of the present invention, this embodiment uses experiments for verification, as detailed below:

[0133] In this embodiment, a civil aircraft cockpit intelligent interaction scenario is selected as the test object. Two types of safety risk factors are designed: technical system failure and human-machine trust. These are presented through an automatic flight control system. The experiment adopts a two-factor repeated measures within-group design paradigm, with the factors being system reliability and operator proficiency. The experimental setting is as follows: Figure 3 As shown, a total of 21 men participated in the experiment. Electrocardiogram (ECG) and electromyography (EMG) sensors worn by the subjects recorded their bioelectrical characteristics throughout the experiment. The effectiveness of the invention was verified by analyzing the correlation between the average thalamic discharge rate output by the cortico-basal ganglia-thalamus neural circuit calculation model and the physiological data observed in the experiment.

[0134] Changing the magnitude of cortical stimulation current simulates the improvement of subject proficiency; decreasing the dopamine coefficient and increasing the acetylcholine coefficient simulates the impact of technical system failure and human-machine trust on pilots under different system reliability conditions. Parameters used for testing the model are shown in Table 4, and the average thalamic discharge rate for different experimental groups is as follows: Figure 4 As shown.

[0135] Table 4 Parameters of the Test Model

[0136]

[0137] The results of the within-group effects F-test for the model simulation results are shown in Table 5. The analysis of variance results show that the average thalamic discharge rate has significant differences in the main effects of system reliability and operational proficiency (p = 0.0009, p = 0.037), indicating that the established model can simulate the impact of different levels of safety risk factors on pilots.

[0138] Table 5 shows the results of the analysis of variance for the model.

[0139] Experimental factors F value p-value <![CDATA[Effect size η 2 > Test strength 1-β System reliability 62.16 0.000972 0.857330 0.99 Operational proficiency 8.344771 0.037378 0.115087 0.11

[0140] The correlation results between the average thalamic discharge rate and experimental data are shown in Table 6. The correlation analysis results show that the average thalamic discharge rate is significantly correlated with heart rate, muscle activation (biceps brachii), and the proportion of systemic control time. The correlation between the average thalamic discharge rate and the proportion of systemic control time is relatively high (Rho = -0.9667), indicating that as the proportion of systemic control time decreases, the subjects need to continuously process more complex operational procedures and decision-making tasks, leading to an increase in thalamic excitation.

[0141] Table 6. Correlation calculation results between the average thalamic discharge rate and experimental data.

[0142]

Claims

1. A pilot safety risk analysis method based on neural computation, characterized in that, Includes the following steps: S1: Identify security risk factors; S2: Establish a mapping relationship between safety risk factors and unsafe behaviors of pilots, and abstract visual, auditory, tactile and vestibular senses into electrical stimulation of the cortex; S3: Establish a structural model of the cortical-basal ganglia-thalamic neural circuit; S4: Based on the theories of bi-exponential synapses and α-β synapses, computational models of AMPA and GABA neural synapses are established respectively; S5: By combining the structural model of the cortico-basal ganglia-thalamus neural circuit, the AMPA neural synapse computational model, and the GABA neural synapse computational model, a computational model of the cortico-basal ganglia-thalamus neural circuit is established. S6: Using a computational model of the cortical-basal ganglia-thalamic neural circuit, the thalamic discharge rate under the influence of different safety risk factors is calculated, and the regulatory mechanisms of dopamine and acetylcholine in the formation of safety risks are simulated. The safety risk factors in step S1 include air traffic control, airborne systems, and the environment; Air traffic control safety risk factors include information transmission errors, read-back-hear-back errors, failure to acknowledge or respond to instructions, and non-standard terminology; Safety risk factors for airborne systems include design flaws, technical system failures, and human-machine trust issues. Environmental safety risk factors include visual load, auditory load, tactile perception, and vestibular perception; The structural model of the cortico-basal ganglia-thalamus neural circuit in step S3 includes direct pathway, indirect pathway, and superdirect pathway structures; The direct pathway is structured as follows: cortex → striatum → medial globus pallidus → thalamus → cortex. Dopamine binds to dopamine D1 receptors on the striatum, activating Gs G protein of G protein-coupled receptors, thereby enhancing the activity of adenylate cyclase, increasing the synthesis of cyclic adenosine monophosphate, and activating protein kinase A. Protein kinase A increases the activity of GABAergic neurons in the striatum by regulating ion channels and other downstream effectors, inhibits the activity of the medial globus pallidus, reduces its inhibitory effect on the thalamus, and promotes the transmission of motor signals from the thalamus to the cortex. The indirect pathway is structured as follows: cortex → striatum → lateral globus pallidus → hypothalamic nucleus → medial globus pallidus → thalamus → cortex. Dopamine binds to dopamine D2 receptors in the striatum, activating Gi-type G protein of G protein-coupled receptors, inhibiting adenylate cyclase, reducing cyclic adenosine monophosphate levels, and decreasing protein kinase A activity. The lateral globus pallidus sends glutamatergic signals through the hypothalamic nucleus, exciting the medial globus pallidus, which further inhibits the transmission of motor signals from the thalamus to the cortex through its GABAergic neurons. The route structure of the superdirect pathway is cortex → hypothalamic nucleus → medial pallidus → thalamus → cortex. Glu-genergic neurons in the cerebral cortex project directly to the hypothalamic nucleus, and the superdirect pathway is formed through extensive connections between the hypothalamic nucleus and the medial pallidus. GABAergic neurons in the medial pallidus project to directly regulate the thalamic nuclei, while the thalamus is connected to the cerebral cortex through Glu-genergic nerve fibers. Step S4 involves constructing computational models of AMPA and GABA neural synapses, including: Neurons connect via synapses, forming signal transmission pathways. The action potential of the presynaptic neuron is transmitted to the postsynaptic neuron after a time delay. Therefore, the synaptic current is calculated as follows: I syn =g max s(vE syn ) Among them, I syn It is the postsynaptic current, g max and E syn denoted as maximum synaptic conductance and synaptic back electromotive force, respectively, and s represents synaptic variable; AMPA synaptic channels are simulated by bi-exponential synapses: Where, τ r and τ d These are the rise time and the decay time, respectively. Normalization factor GABA synaptic channels are simulated by α-β type synapses: Where s is the synaptic variable, α is the synaptic activation rate constant, β is the deactivation rate constant, and [T] is the concentration of neurotransmitters released by the presynaptic neuron. When the synapse is triggered, [T] is a fixed constant. The neuronal model of the cortical-basal ganglia-thalamic neural circuit computational model in step S5 is expressed as follows: The Hodgkin-Huxley model is described by the following differential equation: Where V represents the membrane potential and C is the membrane capacitance. and g leak These are the maximum conductivity values ​​of the sodium ion channel, potassium ion channel, and leakage channel, respectively, E. Na E K and E leak These are the inversion potentials of the corresponding channels, and I(t) is the external input current; The dynamic behavior of ion channels is represented by the following set of equations: Where m, h, and n represent the gating variables of the sodium ion channel and the potassium ion channel, respectively, and α x and β x It is a rate constant dependent on the membrane potential, and its form is as follows:

2. The pilot safety risk analysis method based on neural computing according to claim 1, characterized in that, The method for calculating the average thalamic discharge rate in step S6 includes the following steps: A1: Calculate the instantaneous discharge rate Among them, t i δ represents the i-th firing time of the neuron, and δ is the Dirac function used to capture the firing event at a specific time. A2: Calculate the average discharge rate Where t0 is the start time of the time window, and T is the duration of the window.

3. A pilot safety risk analysis system based on neural computing according to claim 1 or 2, characterized in that, include: The safety risk analysis module is used to analyze the correlation between unsafe behaviors of pilots and safety risks during flight; The model building module is used to build neural computational models based on the direct, indirect, and superdirect pathways of the cortico-basal ganglia-thalamus neural circuit. The analysis module is used to calculate the average thalamic discharge rate under different cortical current stimulation and different acetylcholine and dopamine coefficient stimulation, and to analyze the regulatory mechanism of acetylcholine and dopamine in the formation of safety risks.

4. The pilot safety risk analysis system based on neural computing according to claim 3, characterized in that, The neural computational models include the Hodgkin-Huxley model, AMPA, and GABA synaptic models.

Images