A safety helmet intelligent control method and system based on electroencephalogram features

Abstract

The application discloses a safety helmet intelligent control method and system based on electroencephalogram features, generates cognitive load frequency domain features by real-time electroencephalogram signal spectrum decomposition, and establishes an electroencephalogram control space; scans the electroencephalogram control space to identify frequency band coupling imbalance points and reversely deduce collaborative intervention time, and forms a multi-channel adjustment mode; maps a stimulation frequency range to generate a frequency matching domain and establishes a resonance intervention window through resonance analysis; monitors the frequency band in the resonance window to obtain an energy transfer trajectory, constructs a risk propagation chain to identify a cascading failure node and forms a key blocking anchor point; analyzes the risk acceleration of the key blocking anchor point to form a state deterioration rate, fuses the frequency domain features to generate a dynamic delay tolerance, and generates variable speed control parameters through fast and slow grading; compares the variable speed control parameters with a threshold value to generate a control deviation signal, generates a hedging stimulation waveform through anti-phase modulation, and forms a phase compensation execution instruction, and precise monitoring and safe intervention of a cognitive state are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of brain-computer interface control, in particular to a safety helmet intelligent control method and system based on electroencephalogram features. BACKGROUND

[0002] In high-risk industrial scenes such as coal mines, construction, and chemical industry, the cognitive state of workers directly affects the safety of operation. Fatigue, distraction, or cognitive overload can all lead to operational errors and trigger safety accidents. Traditional safety management relies on operation time limits and regular rest systems, but these methods cannot reflect the real cognitive state of workers in real time and are difficult to cope with individual differences and dynamic changes in work load.

[0003] Electroencephalogram signals, as a direct physiological indicator of brain cognitive activity, provide a new technical approach for real-time monitoring of the cognitive state of workers. However, existing methods still face many challenges: how to extract the cognitive load feature representation from multi-band brain wave data and establish the corresponding relationship between different fatigue levels and electroencephalogram patterns; when signs of fatigue or decreased attention are detected in workers, how to determine the best warning intervention time and choose the appropriate stimulation method and intensity; when the stimulation output parameters deviate from the preset range due to various factors, how to detect and correct these deviations in a timely manner to avoid ineffective warnings or excessive stimulation. SUMMARY

[0004] The present application provides a safety helmet intelligent control method and system based on electroencephalogram features, which establishes an electroencephalogram control space through spectral decomposition to realize hierarchical judgment of cognitive state, identifies frequency band imbalance and reversely deduces intervention timing to realize forward-looking early warning, constructs key intervention nodes through resonance analysis and risk propagation chain, generates dynamic response strategies through state deterioration rate fusion, and realizes adaptive compensation of parameter deviation through anti-phase modulation, providing intelligent safety protection for high-risk industrial operation.

[0005] The present application provides a safety helmet intelligent control method and system based on electroencephalogram features, which establishes an electroencephalogram control space through spectral decomposition to realize hierarchical judgment of cognitive state, identifies frequency band imbalance and reversely deduces intervention timing to realize forward-looking early warning, constructs key intervention nodes through resonance analysis and risk propagation chain, generates dynamic response strategies through state deterioration rate fusion, and realizes adaptive compensation of parameter deviation through anti-phase modulation, providing intelligent safety protection for high-risk industrial operation.

[0006] Collecting real-time electroencephalogram signals of workers, performing spectral decomposition on the real-time electroencephalogram signals to generate cognitive load frequency domain features, and establishing an electroencephalogram control space based on the cognitive load frequency domain features;

[0007] Scanning the electroencephalogram control space to form a multi-band energy distribution matrix, identifying frequency band coupling imbalance points along the multi-band energy distribution matrix, reversely deducing cooperative intervention timing via the frequency band coupling imbalance points, and forming a multi-channel adjustment mode based on the cooperative intervention timing;

[0008] Obtaining a safety helmet stimulation frequency range and a target electroencephalogram frequency band, mapping the stimulation frequency range to the electroencephalogram control space to generate a frequency matching domain, and establishing a resonance intervention window based on the target electroencephalogram frequency band and the resonance analysis of the frequency matching domain;

[0009] Using the multi-channel adjustment mode to perform frequency band monitoring in the resonance intervention window to obtain an energy migration trajectory, constructing a risk propagation chain according to the energy migration trajectory to identify a cascading failure node, performing a blocking priority evaluation on the cascading failure node to form a key blocking anchor point;

[0010] Performing risk acceleration analysis on the key blocking anchor point to form a state deterioration rate, using the state deterioration rate to fuse the cognitive load frequency domain feature to determine a response time and generate a dynamic delay tolerance, and performing fast and slow grading according to the dynamic delay tolerance to generate a variable speed control parameter;

[0011] Comparing the variable speed control parameter with a preset safety threshold to generate a control deviation signal, performing anti-phase modulation on the control deviation signal to generate a hedging stimulation waveform, and forming a phase compensation execution instruction based on the hedging stimulation waveform.

[0012] The second aspect of the present application provides a safety helmet intelligent control system based on electroencephalogram features, comprising:

[0013] A signal acquisition module is configured to acquire real-time electroencephalogram signals of workers, perform frequency spectrum decomposition on the real-time electroencephalogram signals to generate cognitive load frequency domain features, and establish an electroencephalogram control space based on the cognitive load frequency domain features.

[0014] A frequency band analysis module is configured to scan the electroencephalogram control space to form a multi-frequency band energy distribution matrix, identify a frequency band coupling imbalance point along the multi-frequency band energy distribution matrix, reversely deduce a cooperative intervention opportunity via the frequency band coupling imbalance point, and form a multi-channel adjustment mode based on the cooperative intervention opportunity.

[0015] A resonance matching module is configured to obtain a safety helmet stimulation frequency range and a target electroencephalogram frequency band, map the stimulation frequency range to the electroencephalogram control space to generate a frequency matching domain, and establish a resonance intervention window based on the target electroencephalogram frequency band and the resonance analysis of the frequency matching domain.

[0016] A propagation tracking module is configured to use the multi-channel adjustment mode to perform frequency band monitoring in the resonance intervention window to obtain an energy migration trajectory, construct a risk propagation chain according to the energy migration trajectory to identify a cascading failure node, perform a blocking priority evaluation on the cascading failure node to form a key blocking anchor point.

[0017] a response scheduling module, configured to perform risk acceleration analysis on the key blocking anchor point to form a state deterioration rate, fuse the cognitive load frequency domain features with the state deterioration rate to determine a response time and generate a dynamic delay tolerance, perform fast and slow grading according to the dynamic delay tolerance to generate a variable speed control parameter, and perform fast and slow grading according to the dynamic delay tolerance to generate a variable speed control parameter;

[0018] a phase control module, configured to compare the variable speed control parameter with a preset safety threshold to generate a control deviation signal, perform inverse phase modulation on the control deviation signal to generate a hedging stimulation waveform, and form a phase compensation execution instruction based on the hedging stimulation waveform.

[0019] The beneficial effects of the present application are embodied in the following points: first, the cognitive load frequency domain features are extracted through spectral decomposition, and an electroencephalogram control space containing a normal working layer, a pre-warning transition layer and an out-of-control intervention layer is established, and then the frequency band coupling imbalance point is identified by scanning the electroencephalogram control space, and the collaborative intervention opportunity is deduced reversely, realizing a complete closed loop from static cognitive state grading to dynamic imbalance early warning, which not only can accurately determine the current cognitive load level of the operator, but also can predict the deterioration trend of the cognitive state in advance through the imbalance characteristics of the coupling relationship between frequency bands, and identify the best intervention window before the operator has completely entered the dangerous state. Secondly, the frequency matching domain is generated by mapping the stimulation frequency range, the synchronous frequency component is extracted by performing phase locking, the coupling degree is screened, the peak resonance point is obtained, and the resonance intervention window is determined, then the energy migration trajectory is obtained in the window, and the risk propagation chain is identified to identify the cascade failure node, realizing the precise matching of the intervention frequency and the inherent frequency of the electroencephalogram, and the precise positioning of the risk node, and through the resonance effect, a larger awakening effect is obtained with smaller stimulation intensity, and targeted intervention is implemented at the key link most likely to trigger a chain collapse. Finally, the state deterioration rate is formed by performing risk acceleration analysis on the key blocking anchor point, the dynamic delay tolerance is generated by fusing the cognitive load frequency domain features, and the variable speed control parameter is generated by performing fast and slow grading, then the control deviation signal is generated by comparing the variable speed control parameter with the safety threshold, and the hedging stimulation waveform is generated by performing inverse phase modulation on the control deviation signal, and the phase compensation execution instruction is formed, realizing the dynamic grading of the intervention response speed and the adaptive compensation of the stimulation parameter deviation, which can flexibly adjust the intervention response time according to the emergency degree of the cognitive state deterioration, and can detect and correct the deviation of the stimulation parameter in real time through the inverse phase modulation technology, ensuring that the stimulation always remains in a safe and effective range.

[0020] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0021] The drawings herein show specific examples of the technical solutions described in the present application, and constitute part of the specification together with the specific embodiments, for explaining the technical solutions, principles and effects of the present application.

[0022] Unless specifically stated or otherwise as clear from the context of the following detailed description, the same reference numerals and characters can be used throughout the description and the figures to indicate like features, where applicable.

[0023] Figure 1 is a flow diagram of a safety helmet intelligent control method based on electroencephalogram features.

[0024] Figure 2 is a structure block diagram of a safety helmet intelligent control system based on electroencephalogram features. DETAILED DESCRIPTION

[0025] In the following description, for purposes of explanation and not limitation, specific details are set forth, such as a particular sequence of acts or the like, in order to provide a thorough understanding of the embodiments of the application. However, it will be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known methods, devices, and circuits are omitted so as not to obscure the description of the present application with unnecessary detail.

[0026] It is to be understood that the terminology "includes", "has", "holds", "contains" or "comprising", when used in this specification and in the following claims, indicates the presence of the described features, integers, steps, operations, elements, and / or components but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0027] It is also to be understood that the terminology "and / or" when used in this specification and in the following claims, refers to at least one of the items, or any combination of one or more of the items, and includes all possible combinations of one or more of the items.

[0028] As used in this specification and in the claims, the terms "if" and "when" can be interpreted to mean "upon" or "in response to a determination" or "in response to a detection" depending on the context. Similarly, the phrase "if it is determined" or "if [a described condition or event] is detected" can be interpreted to mean "upon determining" or "in response to determining" or "upon detecting [the described condition or event]" or "in response to detecting [the described condition or event]" depending on the context.

[0029] In addition, in the description of the application and in the following claims, the terms "first", "second", "third", etc. are used only to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

[0030] Reference throughout this application to "one embodiment" or "some embodiments" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrase "in one embodiment" or "in some embodiments" in various places throughout this specification are not necessarily all referring to the same embodiment, however, but can refer to one or more but not all embodiments. The terms "including," "comprising," "carrying," "having," "containing," and variations thereof are meant to encompass the item listed thereafter, but do not exclude additional, unrecited items. The terms "a" or "an," as used herein in the disclosure, mean "one or more" unless otherwise indicated.

[0031] The technical solutions of the embodiments of the present application are introduced as follows.

[0032] As shown in the figure, the embodiment of the present application provides a safety helmet intelligent control method based on electroencephalogram features, including the following steps S110-S160: Figure 1

[0033] Step S110, collect the real-time electroencephalogram signal of the worker, perform frequency spectrum decomposition on the real-time electroencephalogram signal to generate cognitive load frequency domain features, and establish an electroencephalogram control space based on the cognitive load frequency domain features.

[0034] Specifically, the real-time electroencephalogram signal of the worker is collected. Electroencephalogram acquisition electrodes are deployed on the inside of the forehead of the safety helmet, and the electrodes are in contact with the scalp surface using flexible conductive materials. The electroencephalogram acquisition system records the brain electrical activity of the worker in real time during work, including brain wave signals in five frequency bands of δ, θ, α, β, and γ, and the sampling frequency is set to 256 Hz. The original signal is preprocessed, the industrial power equipment and lighting system generated power frequency interference is eliminated by using a notch filter, the low frequency drift caused by head movement and muscle activity is removed by using a high pass filter, and high frequency noise is removed by using a low pass filter. When the worker is working normally, the β wave activity in the frontal lobe region remains at a moderate level, indicating that the attention is focused and the cognitive load is appropriate. When the worker starts to be tired after a long time of continuous work, the low frequency θ wave and α wave energy gradually rises, and the β wave activity weakens. These changes in brain wave patterns reflect the transition from a state of alertness to a state of fatigue and drowsiness. When the miner is operating large mining equipment or handling sudden danger, the β wave and γ wave are active at the same time, indicating that the brain is undergoing high-intensity information processing and decision-making activities. The β wave and γ wave of the high-altitude worker also appear to be synergistically enhanced when lifting heavy components, reflecting the increase in cognitive load under high-risk operations.

[0035] ​The real-time electroencephalogram signal is subjected to spectral decomposition to generate cognitive load frequency domain features. The five frequency bands in the collected signal are subjected to fast Fourier transform respectively, converting the time domain signal into frequency domain representation. The power spectral density of each frequency band in each time window is calculated, which reflects the energy distribution of the signal at different frequency components. The time sequence variation characteristics of the power spectral density are analyzed to identify the frequency bands with rising, falling or stable energy. Under the relatively relaxed working condition of inspection or monitoring, the signal shows high and stable α wave power spectral density, indicating low cognitive load. When receiving instructions to troubleshoot equipment failure or handle abnormal situations, the signal shows rapid rise of frontal lobe β wave power spectral density, indicating that the cognitive system is activated into working state. Based on the power spectral density of the signal, the energy ratio between different frequency bands is calculated, including θ / α ratio and β / α ratio, which reflect the cognitive load level and alertness state. When the operator is monitoring complex production processes or operating precision instruments, the signal analysis shows a significant increase in β / α ratio, indicating increased cognitive load and high concentration of attention. When the operator starts to feel sleepy at the end of the night shift, the signal shows a gradual rise in θ / α ratio, indicating a transition of the cognitive system from wakefulness to drowsiness. The power spectral density, energy ratio and time sequence variation characteristics of the signal are integrated into cognitive load frequency domain features.

[0036] In some embodiments, the cognitive load frequency domain features are used to establish an electroencephalogram control space, including: performing energy spectrum analysis on the cognitive load frequency domain features to extract energy distribution features; calibrating a boundary transition threshold based on the energy distribution features; monitoring the fluctuation amplitude of the cognitive load frequency domain features according to the boundary transition threshold to establish a hierarchical mapping rule, the hierarchical mapping rule including a normal working layer, a pre-warning transition layer and a loss-of-control intervention layer; and constructing an electroencephalogram control space using the hierarchical mapping rule.

[0037] The energy spectrum analysis is performed on the cognitive load frequency domain features to extract the energy distribution characteristics. Based on the cognitive load frequency domain features, the proportion of energy in each frequency band in the total energy is calculated, which reflects the contribution of different frequency bands to the overall brain activity. During normal work, the proportion of alpha and beta wave energy is observed to be relatively balanced in the frequency domain features, indicating that the cognitive system is in a moderate activation state. When the workers start to fatigue, the frequency domain features show that the proportion of theta wave energy gradually increases, while the proportion of beta wave energy decreases, which marks the transition of cognitive state to drowsiness. The time series trend of the proportion of energy in the frequency domain features is analyzed to identify the frequency bands with increasing, decreasing or stable proportions. When the workers receive emergency tasks, the frequency domain features show that the proportion of beta wave energy rapidly increases, indicating that the cognitive load increases rapidly. The energy ratio data in the frequency domain features, including the θ / α ratio and the β / α ratio, are extracted, combined with the proportion of energy in each frequency band and the time series change characteristics, and integrated into the energy distribution characteristics. The energy distribution characteristics include the proportion of energy in each frequency band, the energy ratio between frequency bands, and the time series trend of these indicators, which describe the energy distribution of cognitive load reflected by the frequency domain features.

[0038] The energy distribution characteristics are used to determine the boundary transition threshold. The transition characteristics of the energy distribution characteristics between different working states are analyzed to identify the jump characteristics during state transition. When the workers transition from rest state to working state, the distribution characteristics show that the proportion of beta wave energy rapidly increases, the proportion of alpha wave energy decreases accordingly, and the β / α ratio significantly increases, indicating the occurrence of state transition. When transitioning from normal working state to fatigue state, the distribution characteristics show that the proportion of theta wave energy continuously increases and crosses the critical level, which is the boundary transition threshold. By analyzing the distribution characteristics of a large number of workers during state transition, the statistically significant transition threshold is identified. For the boundary between normal working state and fatigue state, the proportion of theta wave energy is extracted from the distribution characteristics. When the proportion of theta wave energy crosses 25%, the workers start to show signs of fatigue, so 25% is set as the boundary transition threshold for fatigue state. For the boundary between normal working state and high load state, the proportion of beta wave energy is extracted from the distribution characteristics. When the proportion of beta wave energy crosses 45%, it corresponds to the start of complex task processing, so 45% is set as the boundary transition threshold for high load state. When the miners process the gas concentration abnormal alarm, the distribution characteristics show that the proportion of beta wave energy rapidly crosses the threshold of 45%, entering the high load state. By analyzing the transition interface of the distribution characteristics, the key boundary transition thresholds are determined, each threshold corresponding to a state transition interface.

[0039] The fluctuation amplitude of the cognitive load frequency domain feature according to the boundary transition threshold is used to establish a hierarchical mapping rule. The real-time collected frequency domain features are compared with the boundary transition threshold. The energy proportion of each frequency band and the energy ratio are extracted from the frequency domain features, and the degree of deviation of these indicators from the threshold is calculated. This deviation is the fluctuation amplitude. When the frequency domain feature shows that the theta wave energy proportion is 20%, the fluctuation amplitude deviating from the theta wave threshold (25%) is -20%. When the frequency domain feature shows that the beta wave energy proportion is 50%, the fluctuation amplitude deviating from the beta wave threshold (45%) is +11%. Based on the size of the fluctuation amplitude, a hierarchical mapping rule is established. The mapping rule divides the electroencephalogram control space into three levels: the normal working layer corresponds to the fluctuation amplitude of the frequency domain feature within ±10%. At this time, the energy distribution of each frequency band is balanced, the beta / alpha ratio is within a reasonable range, and the theta wave energy proportion is low, indicating that the working personnel have moderate cognitive load, concentrated attention, and no obvious signs of fatigue. The pre-warning transition layer corresponds to the fluctuation amplitude of the frequency domain feature within ±10%-30%. At this time, the energy distribution starts to approach the threshold but has not completely crossed it. After several hours of continuous work, the frequency domain feature shows that the theta wave energy proportion is 22%, and the fluctuation amplitude is -12%. According to the mapping rule, it is determined that it is in the pre-warning transition layer, indicating that the worker starts to feel tired but can still work. The out-of-control intervention layer corresponds to the fluctuation amplitude of the frequency domain feature exceeding ±30%. At this time, the energy distribution has crossed the threshold. In extreme fatigue, the frequency domain feature shows that the theta wave energy proportion exceeds 32%, and the fluctuation amplitude is +28%. According to the mapping rule, it is determined to enter the out-of-control intervention layer, and immediate warning or protective measures need to be taken. The above three-level division and determination criteria constitute a complete hierarchical mapping rule.

[0040] The EEG control space is constructed by using a hierarchical mapping rule. The multi-dimensional space of cognitive load frequency domain features is divided into three hierarchical regions according to the hierarchical mapping rule. In the EEG control space, the normal working layer occupies the central region, which corresponds to the fluctuation amplitude ±10% range defined by the mapping rule, and represents the frequency domain feature range of the worker in the ideal working state. The pre-warning transition layer is located between the central region and the boundary region, which corresponds to the fluctuation amplitude ±10%-30% range defined by the mapping rule, and represents the frequency domain feature range of the state starting to deviate but not yet reaching the dangerous level. The out-of-control intervention layer is located in the boundary region, which corresponds to the fluctuation amplitude greater than ±30% range defined by the mapping rule, and represents the frequency domain feature range of the state deviating from the normal range and needing timely intervention. Clear boundary conditions are set for each level, and the boundary conditions are based on the comprehensive judgment of the energy ratio of each frequency band and the energy distribution. A buffer zone is set between the levels, which is used to smooth the state transition process and avoid frequent level jumps. In a complete work shift, the position of the frequency domain features in the EEG control space will move with the change of the cognitive state. At the beginning of the shift, the worker is full of energy, and the frequency domain features are located in the center of the normal working layer. As the work time extends, the frequency domain features gradually drift to the pre-warning transition layer. If no timely rest is given, the frequency domain features will eventually enter the out-of-control intervention layer, at which time a forced rest instruction or adjustment of work arrangement should be given according to the judgment of the mapping rule.

[0041] In step S120, the EEG control space is scanned to form a multi-frequency band energy distribution matrix, a frequency band coupling imbalance point is identified along the multi-frequency band energy distribution matrix, a collaborative intervention opportunity is deduced in reverse through the frequency band coupling imbalance point, and a multi-channel adjustment mode is formed based on the collaborative intervention opportunity.

[0042] Specifically, the EEG control space is scanned to form a multi-frequency band energy distribution matrix. Each region in the EEG control space is scanned in real time, and the energy values of each frequency band at different time points are extracted. The energy of each frequency band is arranged in time order and frequency band dimension to construct a two-dimensional energy distribution matrix. The rows of the matrix correspond to different time sampling points, and the columns correspond to five EEG frequency bands. The matrix element value represents the energy intensity of the corresponding time point and frequency band. The energy distribution matrix is normalized to scale the energy values of each frequency band to the range of 0-1, eliminating the difference in energy magnitude of different frequency bands and making the energy of each frequency band comparable. The characteristics of the energy of each frequency band in the matrix with time are analyzed. At the beginning of the shift, the energy distribution matrix shows that the β wave and α wave energy is relatively balanced, and the energy distribution among the frequency bands is stable. As the continuous work time extends, the values in the θ wave corresponding column in the matrix gradually increase, and the values in the β wave corresponding column gradually decrease, reflecting the evolution process of the cognitive state from wakefulness to fatigue. When the worker suddenly encounters an emergency and needs to make a quick decision, the values in the β wave and γ wave corresponding columns in the matrix rapidly rise synchronously, showing a sudden increase in cognitive load.

[0043] Identify the frequency band coupling imbalance point along the multi-band energy distribution matrix. Analyze the synergy relationship between different frequency bands in the energy distribution matrix, and calculate the energy correlation between frequency bands. Under normal working conditions, the energy changes of different frequency bands usually present a synergistic mode, that is, when the energy of certain frequency bands rises, the energy of other specific frequency bands decreases accordingly, maintaining the overall energy balance. Identify the time points in the matrix that deviate from the normal synergistic mode. These time points correspond to the time when the energy distribution between frequency bands loses balance. Calculate the deviation of the energy of each frequency band from its baseline level. The baseline level is the average energy value of the frequency band under normal working conditions. When the energy of a certain frequency band deviates from the baseline by more than 30%, it is marked as an imbalance frequency band. When multiple frequency bands simultaneously deviate significantly and the deviation direction does not conform to the normal synergistic mode, the time point is marked as a frequency band coupling imbalance point. After a long time of work in the mine, the energy distribution matrix shows that the theta wave energy continues to rise while the beta wave energy continues to decline, but the alpha wave energy does not adjust accordingly according to the normal mode, and the synergy relationship between the three frequency bands is broken, identifying the frequency band coupling imbalance point. When operating complex equipment, the worker is suddenly disturbed by external strong interference, and the beta and gamma wave energies fluctuate sharply and the fluctuation mode loses synchronicity, and the coupling relationship between frequency bands is momentarily imbalanced.

[0044] In some embodiments, the reverse derivation of the synergistic intervention opportunity via the frequency band coupling imbalance point comprises: extracting the duration data from the frequency band coupling imbalance point; identifying the deterioration acceleration gradient based on the duration data; determining the high-risk time period according to the time when the deterioration acceleration gradient exceeds the preset threshold; and moving the starting point of the high-risk time period by a set buffer duration to obtain the synergistic intervention opportunity.

[0045] Extract the duration data from the frequency band coupling imbalance point. Calculate the duration of each frequency band coupling imbalance point from its appearance to its disappearance, which reflects the persistence of the imbalance state. Analyze the time sequence variation of the duration, and identify whether the duration presents an increasing trend. Shorter and less fluctuating duration indicates stronger cognitive regulation ability of the worker, who can quickly recover from the imbalance state. Longer and continuously increasing duration indicates weakened cognitive regulation ability, and the imbalance state gradually solidifies. The duration of the imbalance point appearing at the beginning of the shift is shorter, and the worker usually recovers to normal after a short rest or task switching. As the continuous work time prolongs, the duration of the imbalance point appearing later significantly prolongs, indicating that the cognitive recovery ability of the worker decreases with the accumulation of fatigue. Record the duration values of each imbalance point, arrange them in chronological order, and construct the time sequence of the duration data.

[0046] An acceleration gradient of cognitive state deterioration is identified based on the duration data. A regression analysis is performed on the time series of the duration data to fit a trend line T(t)=k×t+b of the duration over time, where T(t) is a predicted value of the duration at time t, k is a slope of the trend line, t is a time variable, and b is an intercept. The slope k of the trend line is the acceleration gradient of the cognitive state deterioration, and the greater the slope is, the faster the cognitive state deteriorates. At the beginning of the work, the duration data is relatively stable, and the acceleration gradient of the cognitive state deterioration is close to zero. After entering the fatigue accumulation stage, the duration data starts to accelerate, and the acceleration gradient of the cognitive state deterioration calculated by the regression analysis significantly increases. When the acceleration gradient of the cognitive state deterioration suddenly increases, it indicates that the cognitive state of the worker is rapidly deteriorating, and timely intervention is needed. When the miner is dealing with an emergency, multiple unbalanced points of rapid growth of the duration appear in a short time, the acceleration gradient of the cognitive state rapidly rises, indicating that the cognitive state is about to enter an out-of-control state.

[0047] A high-risk time period is determined according to a time point at which the acceleration gradient exceeds a preset threshold. A safety threshold of the acceleration gradient is set, and the threshold corresponds to a critical point of the transition of the cognitive state from controllable to out-of-control. The safety threshold is dynamically adjusted according to the danger level of the work type and the individual characteristics of the worker, and a stricter threshold is used for a high-risk work scene, and a looser threshold is used for a low-risk work scene. The value change of the acceleration gradient is monitored in real time, and when the gradient value exceeds the safety threshold, the time point is marked as the starting point of the high-risk time period. From the starting point, the acceleration gradient is tracked backward until the acceleration gradient falls below the safety threshold or the worker is effectively intervened, and the time interval is the high-risk time period. The high-risk time period corresponds to a period of rapid deterioration of the cognitive state of the worker and rapid rise of the safety risk. When the acceleration gradient first exceeds the safety threshold during continuous work of the worker, the time point and the subsequent time period are determined as the high-risk time period, indicating that the worker is about to enter or has entered a dangerous state. In the high-risk time period, the probability of operation errors, judgment errors, or accidents of the worker significantly increases, and protective measures need to be taken immediately.

[0048] The starting point of the high-risk period is moved forward by a buffer time length as the collaborative intervention opportunity. The buffer time length is set before the starting point of the high-risk period, and the buffer time length is dynamically determined according to the danger of the work type and the speed of cognitive state deterioration. For work tasks with higher danger, a longer buffer time length is set to ensure that the intervention measures have enough time to take effect. For the case where the cognitive state deteriorates faster, a longer buffer time length is set to avoid the intervention measures being implemented too late. The collaborative intervention opportunity is obtained by subtracting the buffer time length from the starting point of the high-risk period. Implementing preventive intervention measures at the collaborative intervention opportunity can prevent further deterioration of the cognitive state before the worker enters a high-risk state. When operating dangerous equipment, the worker predicts that the deterioration acceleration gradient will exceed the safety threshold, and issues a rest reminder and suggests pausing the dangerous operation at the collaborative intervention opportunity, successfully avoiding the worker continuing the high-risk operation in a fatigued state. When the miner is working underground, the multi-channel adjustment is started at the collaborative intervention opportunity before the high-risk period, and through the combination of sound reminders, lighting enhancement and task adjustment, the deterioration of the cognitive state is effectively delayed.

[0049] A multi-channel adjustment mode is formed based on the collaborative intervention opportunity. According to the brain electrical feature type corresponding to the collaborative intervention opportunity, a corresponding intervention strategy combination is selected. For imbalance caused by fatigue, a combination of intervention measures such as rest reminders, task switching and environmental adjustment is adopted. For imbalance caused by cognitive overload, a combination of intervention measures such as task simplification, auxiliary support and rhythm regulation is adopted. A multi-channel adjustment mode is established, and each channel corresponds to different types of intervention means. The sound reminder channel issues voice or warning sound through the built-in speaker in the safety helmet to remind the worker to rest or adjust the state. The visual warning channel flashes the LED indicator light on the face mask of the safety helmet to send a signal to the worker and the surrounding colleagues that the state is abnormal. The environmental control channel sends instructions to the environmental control equipment in the work area to adjust the lighting brightness, ventilation intensity or temperature setting to improve the work environment to relieve fatigue. The task management channel sends suggestions to the work scheduling platform to prompt the scheduling personnel to arrange the workers to rotate and rest or adjust the work tasks. When the worker is working at a high altitude and shows signs of attention dispersion, a safety prompt is played through the sound reminder channel, a warning symbol is displayed on the face mask through the visual warning channel, and a state report is sent to the ground monitoring personnel, forming a multi-level collaborative intervention. When the miner shows signs of fatigue underground, not only is he reminded to pay attention to safety through sound, but also instructions are sent to the underground lighting equipment to increase the local lighting brightness, and suggestions are made to the scheduling center to arrange the miner to ascend the well for rest.

[0050] In step S130, the stimulation frequency range of the safety helmet and the target brain electrical frequency band are obtained, the stimulation frequency range is mapped to the brain electrical control space to generate a frequency matching domain, and the resonance intervention window is established based on the target brain electrical frequency band and the frequency matching domain resonance analysis.

[0051] Specifically, the safety helmet stimulation frequency range and the target brain electrical frequency band are obtained. The stimulation output device types equipped by the safety helmet are identified, including sound generators, vibration motors, and LED flicker lights, etc. The working frequency range of each stimulation device is obtained, the sound generator usually covers the low to medium frequency range of 200-4000Hz, the vibration motor covers the very low frequency range of 5-50Hz, and the LED flicker light covers the low frequency range of 1-30Hz. The frequency ranges of various types of stimulation devices together constitute the overall stimulation frequency range of the safety helmet. In the fatigue state of the worker, the safety helmet plays a prompt sound of a specific frequency through the sound generator, and the frequency of the prompt sound is designed to effectively activate the beta wave activity in the frontal lobe region. According to the current cognitive state of the worker and the intervention target, the target brain electrical frequency band that needs to be adjusted is determined. When the fatigue and drowsiness symptoms of the worker are detected, the target brain electrical frequency band is set to beta wave (13-30Hz), and the intervention target is to enhance the beta wave activity to improve alertness. When the cognitive load of the worker is too high, the target brain electrical frequency band is set to alpha wave (8-13Hz), and the intervention target is to enhance the alpha wave activity to relieve cognitive tension. The theta wave energy of the miner significantly increases in the later period of night shift work, and the beta wave is set as the target brain electrical frequency band, and the beta wave activity is improved through appropriate stimulation frequency to help the miner recover to a sober state. When the worker is in a highly tense state, the alpha wave is set as the target brain electrical frequency band, and the alpha wave is enhanced through rhythmic stimulation to relieve cognitive overload.

[0052] The mapping of the stimulation frequency range to the electroencephalogram control space generates a frequency matching domain. Based on the stimulation frequency range, the correspondence between the stimulation frequency and the electroencephalogram frequency band is analyzed, and different frequencies of external stimulation have different effects on different electroencephalogram frequency bands. The frequency range of each stimulation device of the safety helmet is projected into the electroencephalogram control space, and the projection process considers the stimulation transmission path and the brain response characteristics. The action area of the stimulation frequency is calibrated in the electroencephalogram control space, and the action area represents the range of electroencephalogram states that can be effectively affected by a specific stimulation frequency. The mid-frequency component of the sound stimulation mainly affects the area corresponding to the beta wave in the electroencephalogram control space, and can effectively activate the alert state. The very low frequency component of the vibration stimulation forms a wide area of influence covering multiple frequency bands in the electroencephalogram control space, and has a regulating effect on the delta wave, theta wave and alpha wave. The low-frequency component of the LED flicker light mainly acts on the theta wave and alpha wave area in the electroencephalogram control space, and is suitable for relaxation and soothing intervention. Under different cognitive states, the electroencephalogram characteristics of the workers are located at different positions in the control space, and the most suitable stimulation frequency range is selected according to the current position. When the worker is in the pre-warning transition layer and drifts towards the drowsiness area, the mid-frequency range of the sound stimulation is activated, a frequency matching domain covering the beta wave area is formed in the electroencephalogram control space, and the worker's electroencephalogram characteristics are guided to return to the normal working layer. The frequency matching domain represents the area in the electroencephalogram control space where the stimulation frequency range has a good matching relationship with the target electroencephalogram frequency band, and the implementation of stimulation in this area can obtain better intervention effect. By mapping the frequency range of different types of stimulation devices to the electroencephalogram control space, a collaborative intervention mechanism of multi-modal stimulation is established, and precise regulation of the cognitive state of the worker is realized.

[0053] In some embodiments, the resonance intervention window is established based on the target electroencephalogram frequency band resonance analysis of the frequency matching domain, including: performing phase-locked extraction of the frequency component synchronized with the target electroencephalogram frequency band on the frequency matching domain; generating a high-coupling frequency set through coupling degree screening processing of the frequency component; obtaining a peak resonance point by matching degree analysis of the high-coupling frequency set and the target electroencephalogram frequency band; and determining a resonance intervention window by setting a time span to both sides of the peak resonance point.

[0054] The phase lock extraction is performed on the frequency matching domain to extract the frequency components synchronized with the target brain electrical frequency band. The phase relationship between each stimulation frequency in the frequency matching domain and the target brain electrical frequency band is analyzed, and the phase relationship reflects the timing synchronization degree of the stimulation frequency and the target brain electrical frequency band. The phase lock degree of the stimulation frequency and the target brain electrical frequency band is calculated, and the high phase lock degree indicates that the stimulation frequency can effectively drive the oscillation of the target brain electrical frequency band. The stimulation frequencies with the phase lock degree exceeding the synchronization threshold are extracted, and these frequencies maintain the phase lock relationship with the target brain electrical frequency band. The synchronization threshold is set according to the intervention intensity requirement, and the threshold is set to 0.7 for a strong intervention scenario and 0.5 for a mild intervention scenario. The beta wave of the worker presents a stable oscillation mode at a certain moment, and the oscillation frequency is 18 Hz. All the stimulation frequencies in the scanning frequency matching domain are scanned, and the frequency components phase-locked with the beta wave oscillation mode are identified, including the 1800 Hz and 3600 Hz frequency components of the sound stimulation. These frequency components can produce a synchronous resonance effect with the beta wave and are candidate frequencies for implementing intervention. The higher the phase lock degree, the more significant the driving effect of the stimulation frequency on the target brain electrical frequency band, and the ideal intervention effect can be obtained with smaller stimulation intensity.

[0055] The high coupling frequency set is generated by performing coupling degree screening processing on the frequency components. The energy coupling degree of the extracted synchronization frequency components is analyzed, and the coupling degree reflects the energy transmission efficiency of the stimulation frequency to the target brain electrical frequency band. The coupling degree C of each frequency component is calculated, where ΔE is the energy change amount of the target brain electrical frequency band, and P_s is the stimulation power. The frequency component with high coupling degree can cause a large energy response of the target brain electrical frequency band with smaller stimulation intensity, and the frequency component with low coupling degree is difficult to effectively change the brain electrical state even if stronger stimulation is applied. The coupling degree screening threshold C_threshold is set to 2.0, and the frequency component with coupling degree exceeding the threshold is included in the high coupling frequency set. When the worker receives stimulation of different frequencies, some frequencies can quickly activate the beta wave and continuously increase the energy of the beta wave. The coupling degree C of a certain frequency is 4.5, which makes the energy of the beta wave rise by 25%, and is identified as a high coupling frequency. Although other frequencies are also phase-locked with the beta wave, the coupling degree is only 0.8, which only makes the energy of the beta wave rise by 5%, and is excluded by screening. The high coupling frequency set contains stimulation frequencies with strong effect on the target brain electrical frequency band, and is a frequency resource library for implementing precise intervention.

[0056] The high-coupling frequency set is matched with the target brain electrical frequency band to obtain a peak resonance point. The closeness of each frequency in the high-coupling frequency set to the dominant frequency of the target brain electrical frequency band is analyzed, and the closeness is quantified by the reciprocal of the frequency difference. A small frequency difference indicates that the stimulation frequency is highly matched with the inherent frequency of the target brain electrical frequency band, and can induce resonance effect. The comprehensive matching degree of each frequency in the high-coupling frequency set is calculated, which integrates the information of the three dimensions of phase locking degree, energy coupling degree and frequency closeness. The stimulation frequency with the highest comprehensive matching degree is identified as the peak resonance point, and the optimal intervention effect can be obtained at this frequency point. When the dominant frequency of the beta wave of the worker is 18 Hz, the stimulation frequency in the high-coupling frequency set that is most matched with the dominant frequency is searched. The 1800 Hz frequency component of the sound stimulation corresponds to the 18 Hz frequency of the brain electrical oscillation, and the two show a frequency ratio relationship of 100 times. The comprehensive matching degree of this frequency is the highest, and it is determined as the peak resonance point. By accurately positioning the peak resonance point, the intervention stimulation is ensured to be best matched with the brain electrical characteristics of the worker, and the intervention effect is maximized.

[0057] A resonance intervention window is determined according to the time span extending to both sides of the peak resonance point. The peak resonance point is taken as the center frequency, and a resonance frequency band is formed by extending to both sides of the frequency axis. The extension range considers the frequency fluctuation characteristics and individual differences of the target brain electrical frequency band. The time sequence stability of the target brain electrical frequency band of the worker is analyzed. When the frequency stability is high, a narrow resonance frequency band is set, and the extension range is ±5% of the frequency range. When the frequency fluctuation is large, a wide resonance frequency band is set, and the extension range is ±15% of the frequency range. The time span of continuously applying stimulation within the resonance frequency band is determined, and the time span is determined according to the intervention target and the brain electrical response speed. For fast wake-up type intervention, a short time span of 10-30 seconds is set to avoid overstimulation. For continuous adjustment type intervention, a long time span of 60-180 seconds is set to maintain stable intervention effect. The determined time span is defined as the resonance intervention window, which represents the time range within which the stimulation frequency corresponding to the peak resonance point can obtain good resonance effect. During the process of receiving intervention stimulation, the brain electrical state of the worker will change dynamically, and the evolution of the characteristics of the target brain electrical frequency band is monitored in real time. When the energy of the target brain electrical frequency band rises to the target level or the brain electrical characteristics of the worker move out of the effective range of the resonance intervention window, the stimulation parameters are adjusted or the stimulation is terminated in time to avoid weakening of the intervention effect or negative impact. By reasonably setting the frequency range and time span of the resonance intervention window, safe operation is ensured while avoiding discomfort caused by overstimulation.

[0058] In step S140, the energy transfer trajectory is obtained by performing frequency band monitoring in the resonance intervention window using the multi-channel adjustment mode, the cascade failure node is identified by constructing a risk propagation chain according to the energy transfer trajectory, and the key blocking anchor point is formed by performing blocking priority evaluation on the cascade failure node.

[0059] Specifically, the energy transfer trajectory is obtained by performing frequency band monitoring in the resonance intervention window using the multi-channel adjustment mode. After the resonance intervention window is started, the worker is stimulated by sound, vibration or light through the multi-channel adjustment mode. The energy changes of the five brain electrical frequency bands of δ, θ, α, β and γ are monitored synchronously, and the energy time sequence data of each frequency band before, during and after the stimulation is recorded. The evolution mode of the energy of each frequency band on the time axis is analyzed, and the characteristics of the energy rising, falling or transferring between different frequency bands are identified. After receiving the sound stimulation, the worker's θ wave energy initially decreases, and then the β wave energy gradually rises, and this energy transfer process from low frequency band to high frequency band reflects the change of the cognitive state from drowsiness to wakefulness. The energy transfer trajectory diagram is drawn, the horizontal axis of the trajectory diagram is time, the vertical axis is the energy of each frequency band, and the curve shows the dynamic flow process of the energy between different frequency bands. In the initial stage of receiving the intervention stimulation, the energy transfer trajectory of the worker shows that the θ wave energy rapidly decreases, the α wave energy temporarily rises and then stabilizes, the β wave energy continuously rises and finally stabilizes at a higher level, and the entire energy transfer process undergoes complete evolution from fatigue recovery to wakefulness stability. The energy transfer trajectory obtained by monitoring contains the complete change information of the energy of each frequency band over time.

[0060] In some embodiments, the risk propagation chain is constructed according to the energy transfer trajectory to identify the cascade failure node, including: performing propagation direction window division in the energy transfer trajectory to generate a set of direction windows; based on the set of direction windows, a main propagation channel is obtained by performing propagation path planning; the main propagation channel is processed by intensity modulation to generate a propagation load; and the cascade relationship is mapped using the propagation load to identify the cascade failure node.

[0061] The propagation direction window set is generated by dividing the energy migration trajectory into propagation direction windows. The direction characteristics of energy flow in the energy migration trajectory are analyzed to identify the time period in which energy is transferred from the source frequency band to the target frequency band. The trajectory is divided into multiple propagation direction windows according to the direction of energy flow, and each window corresponds to a time interval with a clear energy propagation direction. Within the propagation direction window, energy mainly flows out from a specific frequency band and flows into another specific frequency band, and the flow direction remains relatively stable. At the initial stage of receiving intervention stimulation, energy is mainly transferred from θ wave to α wave, and this time period constitutes a propagation direction window. Subsequently, energy is mainly transferred from α wave to β wave, forming another propagation direction window. The start time, end time, source frequency band, and target frequency band information of each propagation direction window are extracted to construct the data structure of the propagation direction window. All propagation direction windows are arranged in chronological order to form the direction window set. The direction window set describes the complete time sequence of energy propagation between different frequency bands, for example, the direction window set of a certain miner contains three main windows: θ wave to α wave transfer, α wave to β wave transfer, and β wave stable.

[0062] The main propagation channel is obtained by planning the propagation path based on the direction window set. The continuity of energy propagation in the direction window set is analyzed to identify the energy transfer link between frequency bands. Multiple windows in which energy flows out from the same source frequency band and multiple windows in which energy flows into the same target frequency band are associated to construct the propagation path between frequency bands. The frequency band links with large energy flow and long propagation duration in the propagation path are identified, and these links constitute the main propagation path. During the recovery process from fatigue state, the main propagation path of the worker is usually θ wave→α wave→β wave, and energy flows from low frequency band to high frequency band in sequence, which corresponds to the normal cognitive recovery mechanism. The time sequence relationship in the direction window set is analyzed, and the energy transmission amount between each frequency band is calculated, which is equal to the energy drop of the source frequency band. The path with the largest energy transmission amount is extracted as the main propagation channel, which carries the main flow in the energy migration process and is the core channel for the intervention mechanism to work. When a miner receives intervention, the main propagation channel is θ→α→β. When a high-altitude worker experiences cognitive tension, the main propagation channel is β→α, and energy flows back from high frequency band to low frequency band to achieve cognitive relaxation.

[0063] The main propagation channel is intensity modulated to generate propagation load. The energy flow of each propagation link in the main propagation channel is obtained, and the energy flow is equal to the energy change amount flowing from the source frequency band to the target frequency band per unit time. The time sequence variation characteristics of the energy flow are analyzed to identify the time periods of stable flow, enhanced flow, and weakened flow. Stable flow indicates smooth energy propagation, enhanced flow indicates accelerated energy propagation, and weakened flow indicates blocked energy propagation. The propagation resistance of each propagation link is evaluated, and the propagation resistance reflects the difficulty of energy transfer between frequency bands. High resistance indicates that energy transfer is difficult, and low resistance indicates that energy transfer is smooth. The propagation resistance is estimated by the ratio of the energy flow to the source frequency band energy change rate. A small ratio indicates high resistance. The propagation load of each propagation link is obtained, and the propagation load L = F x T / R, where F is the energy flow, T is the propagation duration, and R is the propagation resistance coefficient. High propagation load indicates that the link has a large amount of energy transfer task, and high propagation load may cause the propagation channel to be blocked or collapsed. When workers receive strong stimulation, the propagation load from theta wave to alpha wave increases rapidly. If the load exceeds the carrying capacity threshold of the link, energy propagation will be blocked, causing theta wave energy to accumulate, and the intervention effect will be weakened. By monitoring the propagation load of each link in real time, the propagation bottleneck can be found in time, and the intervention strategy can be adjusted.

[0064] The propagation load is used to identify the cascading failure nodes. The dependency relationship between each propagation link in the main propagation channel is analyzed to identify the influence of upstream link failure on downstream link. When the upstream link propagation load is too high to cause propagation blockage, the downstream link will lack energy supply, causing a chain reaction of downstream link failure. Mark the link whose propagation load exceeds the carrying threshold as a potential failure link, and the carrying threshold is dynamically set according to the type of work. Analyze the influence of potential failure links on downstream links. The load change of upstream links causes corresponding changes in downstream link load, and the influence degree is quantified by the ratio of downstream load change to upstream load change. The link with large influence degree has a strong dependency relationship. The link located at the key position of the main propagation channel and with high propagation load is the cascading failure node. Once these nodes fail, it will trigger a chain reaction of multiple downstream links, causing the overall collapse of cognitive regulation mechanism. The propagation link from theta wave to alpha wave becomes a cascading failure node when the worker receives intervention stimulation in an extremely fatigued state, and the load exceeds the carrying threshold. After identifying the node, the stimulation strategy is adjusted by reducing the stimulation intensity and prolonging the stimulation time, which relieves the propagation pressure of the link and avoids the occurrence of cascading failure.

[0065] The blocking priority assessment of the cascading failure nodes forms the critical blocking anchor points. The position and influence range of each cascading failure node in the risk propagation chain are analyzed. The nodes located upstream of the propagation chain affect multiple subsequent links, and the nodes located downstream of the propagation chain have limited influence range. The consequences caused by the failure of each node are evaluated. The failure of some nodes will lead to a local decline in cognitive function, and the failure of other nodes will lead to a complete collapse of cognition. The blocking priority P_i of each node is calculated as P_i = w1 × U_i + w2 × S_i, where P_i is the blocking priority of node i, U_i is the upstream degree of the node in the propagation chain, S_i is the severity of the node failure, and w1 and w2 are weight coefficients. The upstream degree U_i is quantified by the number of links from the node to the end of the propagation chain, and the severity S_i is quantified by the percentage of cognitive function decline caused by the node failure. The nodes with high blocking priority correspond to the critical blocking anchor points. Implementing targeted interventions at these nodes can achieve the maximum risk control effect with the least cost. After a continuous night shift, the theta wave energy continues to rise and becomes the starting node of the risk propagation chain. This node affects multiple downstream frequency bands and has high failure severity, and is identified as a critical blocking anchor point. By increasing the combined intensity of sound and vibration stimulation, the further rise of theta wave energy is successfully blocked, and the development of cognitive state to deep fatigue is avoided.

[0066] In step S150, the state deterioration rate is formed by performing risk acceleration analysis on the critical blocking anchor points. The state deterioration rate is used to fuse the cognitive load frequency domain features to determine the response time and generate a dynamic delay tolerance. According to the dynamic delay tolerance, the fast and slow grading is performed to generate the variable speed control parameters.

[0067] Specifically, the state deterioration rate is formed by performing risk acceleration analysis on the critical blocking anchor points. The electroencephalogram feature time series data corresponding to the critical blocking anchor points are extracted, and the change speed of each frequency band energy at the node is analyzed. The time derivative of the energy change speed is obtained, and the derivative reflects the acceleration of the energy change. The positive acceleration indicates that the energy change speed is increasing, and the state deterioration trend is intensifying. The negative acceleration indicates that the energy change speed is slowing down, and the state deterioration trend is slowing down. In the early stage of fatigue, the theta wave energy slowly rises, and the deterioration acceleration is small. After entering the deep fatigue stage, the theta wave energy rises significantly faster, and the deterioration acceleration increases significantly. The risk acceleration values at the critical blocking anchor points are counted. The larger the risk acceleration, the faster the cognitive state deterioration speed, and the shorter the time window left for intervention measures. During the continuous operation of the miner in the underground mine, when the risk acceleration of the critical blocking anchor point suddenly increases, it indicates that the cognitive state of the miner is deteriorating rapidly, and emergency intervention measures need to be taken immediately. The risk acceleration is converted into the state deterioration rate, which quantifies the time required for the cognitive state to deteriorate from the current level to the dangerous level.

[0068] In some embodiments, the response time determination generates a dynamic delay tolerance by fusing the cognitive load frequency domain features with the state deterioration rate, including: performing a change trend analysis on the state deterioration rate to identify an accelerated mutation point; cross-referencing the accelerated mutation point with the cognitive load frequency domain features to generate an urgency index; dividing a tolerance time level according to the urgency index; and forming a dynamic delay tolerance based on the tolerance time level.

[0069] The change trend analysis on the state deterioration rate identifies an accelerated mutation point. The time series curve of the state deterioration rate is analyzed to identify the time point at which the deterioration rate suddenly jumps. Before the accelerated mutation point, the state deterioration rate is relatively stable or changes slowly, and the deterioration process exhibits linear or sub-linear characteristics. At the accelerated mutation point, the state deterioration rate rapidly increases in a short time, and the deterioration process changes to super-linear or exponential growth. By calculating the change amplitude of the deterioration rate, the time point at which the change amplitude suddenly increases is identified as the accelerated mutation point. The accelerated mutation point marks the turning point at which the cognitive state changes from slow deterioration to rapid deterioration, and is a key early warning signal for the implementation of intervention measures. During continuous work, the initial fatigue symptoms of workers are not obvious and the state deterioration rate is low. After entering a certain critical time, fatigue deepens rapidly and the state deterioration rate rises sharply. This critical time is the accelerated mutation point. When miners work underground, an accelerated mutation point often appears when the working time approaches the personal tolerance limit. After that, cognitive function rapidly declines, and reaction speed and judgment accuracy decrease significantly. By identifying the accelerated mutation point, the key transition time of the cognitive state can be captured in time to provide early warning for rapid response.

[0070] For example, the cross-referencing of the accelerated mutation point with the cognitive load frequency domain features to generate an urgency index includes: extracting time features of the accelerated mutation point to obtain mutation time features; synchronously sampling the cognitive load frequency domain features based on the mutation time features to extract synchronization features; evaluating the correlation degree between the mutation time features and the synchronization features to generate a coupling response coefficient; and forming an urgency index according to the coupling response coefficient.

[0071] The time feature of the acceleration mutation point is extracted to obtain the mutation time feature. The time position, mutation amplitude and mutation speed of the acceleration mutation point are extracted. The time position represents the time when the mutation occurs, the mutation amplitude represents the change amount of the deterioration rate before and after the mutation point, and the mutation speed represents the speed of the jump of the deterioration rate. By analyzing the size of the mutation amplitude and the mutation speed, it is found that the large mutation amplitude and the fast mutation speed represent the severe deterioration jump of the cognitive state, and the small mutation amplitude and the slow mutation speed represent the relatively smooth state deterioration. When the worker encounters a sudden stress event, the state deterioration rate may increase sharply in a short time, the mutation amplitude and the mutation speed are large, and a significant mutation time feature is formed. In contrast, during the normal fatigue accumulation process of the worker, the change of the state deterioration rate is relatively smooth, and the mutation feature is not obvious. The vector representation of the mutation time feature is constructed, which contains information of multiple dimensions such as time position, mutation amplitude and mutation speed, and provides a quantitative basis for subsequent cross-correlation analysis.

[0072] Based on the mutation time feature, the cognitive load frequency domain feature is synchronously sampled to extract the synchronization feature. The time position corresponding to the acceleration mutation point is located, and the cognitive load frequency domain feature data is extracted at the time position. Synchronous sampling ensures that the mutation time feature and the cognitive load frequency domain feature are accurately aligned on the time axis, avoiding the correlation analysis error caused by time deviation. The frequency domain features obtained by synchronous sampling are analyzed, including the energy values of each frequency band, the energy ratio values and the cooperative relationship between frequency bands. Key frequency domain indicators reflecting the cognitive load state are extracted, such as θ / α ratio, β / α ratio and energy entropy. The cognitive load frequency domain feature of the worker at the acceleration mutation point reflects the cognitive background condition triggering rapid deterioration. If the proportion of θ wave energy is already high, it indicates that the worker is in a deep fatigue state before the mutation occurs. If the energy distribution of each frequency band is relatively balanced, it indicates that the mutation may be caused by external sudden events rather than internal fatigue accumulation. The vector representation of the synchronization feature is constructed, which contains the numerical values of multiple key frequency domain indicators, and provides the state information of the cognitive load for cross-correlation analysis.

[0073] The coupling response coefficient is generated by correlating the mutation moment feature and the synchronization feature. The correlation between the mutation moment feature vector and the synchronization feature vector is analyzed, and the Pearson correlation coefficient or mutual information method is used to quantify the correlation strength of the two. High correlation indicates that the acceleration of state deterioration is closely related to the imbalance of cognitive load, and the two form a vicious cycle of mutual promotion. Low correlation indicates that the acceleration of state deterioration is mainly driven by external factors, and the relationship with the current cognitive load state is weak. An evaluation model of the coupling response coefficient is established, which comprehensively considers the strength of the mutation feature, the severity of the synchronization feature, and the correlation between the two. The coupling response coefficient C = r x (T_mag + F_load) / 2 is calculated, where r is the correlation coefficient of the mutation feature and the synchronization feature, T_mag is the normalized value of the mutation amplitude, and F_load is the normalized value of the cognitive load imbalance. The higher the coupling response coefficient indicates that the cognition is in a dangerous positive feedback loop, and the state deterioration and load imbalance intensify each other. If the worker's cognitive load frequency domain feature shows severe imbalance while the state is deteriorating, and the two are highly correlated, the coupling response coefficient is high, indicating that the cognition is collapsing rapidly. When a miner encounters an abnormal gas concentration alarm in the underground mine, if his cognitive state is already on the edge of fatigue, a sudden stress event will trigger rapid state deterioration. At this time, the mutation feature and the load imbalance are highly coupled, generating a high coupling response coefficient.

[0074] An urgency index is formed according to the coupling response coefficient. The coupling response coefficient comprehensively reflects the coordination degree of state deterioration rate and cognitive load imbalance, which is the core indicator for evaluating the urgency. Combined with other auxiliary factors such as the danger of the task, the severity of the environment, and the historical state record of the worker, the urgency index is modified. For high-risk operation scenarios, even if the coupling response coefficient is moderate, the urgency index needs to be increased to ensure safety margin. For low-risk operation scenarios, the sensitivity of the urgency index can be appropriately reduced to avoid excessive reaction affecting normal operation. A comprehensive evaluation formula of the urgency index U = w1 x C + w2 x R_task + w3 x E_env is established, where U is the urgency index, C is the coupling response coefficient, R_task is the task danger coefficient, E_env is the environmental severity coefficient, and w1, w2, and w3 are weight coefficients and satisfy w1 + w2 + w3 = 1. The weight coefficients are dynamically adjusted according to the type of operation and actual situation. When a miner is working in the underground mine, if the coupling response coefficient shows rapid state deterioration and severe cognitive load imbalance, combined with the high-risk characteristics of the underground operation and the closed nature of the underground environment, a high urgency index is generated, triggering a rapid response mechanism. When a worker is performing routine equipment inspection in an open area on the ground, even if there is a slight coupling response, due to the low task danger and good environmental conditions, the generated urgency index is relatively low, allowing for a gentle intervention method.

[0075] The tolerance time level is classified according to the urgency index. The urgency index is set to a threshold value, and the urgency index space is divided into three levels of high urgency, medium urgency and low urgency. The high urgency level corresponds to the case where the urgency index exceeds the high threshold value, indicating that the cognitive state is rapidly deteriorating and the current load is at a dangerous level, and immediate intervention is required. The medium urgency level corresponds to the case where the urgency index is in the medium range, indicating that the cognitive state is deteriorating but there is a short intervention preparation time. The low urgency level corresponds to the case where the urgency index is low, indicating that the cognitive state is deteriorating slowly and the current load is acceptable, and there is sufficient time to implement intervention. The corresponding tolerance time level is set for each urgency level, the high urgency level corresponds to extremely short tolerance time, the medium urgency level corresponds to short tolerance time, and the low urgency level corresponds to standard tolerance time. When the worker is working at high altitude and suddenly shows signs of severe distraction of attention, the urgency index enters the high urgency level, and the tolerance time level is set to extremely short, requiring that the intervention measures must take effect within a very short time. Through the classification mechanism, differentiated response to different urgency levels is achieved, ensuring rational allocation of resources.

[0076] The dynamic delay tolerance is formed based on the tolerance time level. The tolerance time level is converted into a specific time value to form the dynamic delay tolerance. The dynamic delay tolerance corresponding to the extremely short tolerance time requires that the intervention response must be started within a very short time, the dynamic delay tolerance corresponding to the short tolerance time allows a short preparation time, and the dynamic delay tolerance corresponding to the standard tolerance time allows a relatively ample planning and execution time. The dynamic delay tolerance is adjusted in real time as the cognitive state of the worker and the environmental conditions change, and when the state deterioration rate increases or the cognitive load increases, the dynamic delay tolerance is shortened to speed up the intervention response. When the state tends to be stable or the cognitive load decreases, the dynamic delay tolerance is extended to avoid excessive frequent intervention. When the miner is working underground, the safety helmet continuously monitors the change of his cognitive state, and dynamically adjusts the delay tolerance according to the fluctuation of the urgency index, ensuring that the intervention measures are timely and effective and avoid excessive frequent intervention, while ensuring safety and minimizing the interference to normal operation. Through the dynamic adjustment mechanism, adaptive management of the cognitive state of the worker is achieved.

[0077] The variable-speed control parameters are generated according to the dynamic delay tolerance. The intervention response is divided into a fast response range, a standard response range and a slow response range according to the length of the dynamic delay tolerance. The fast response range corresponds to the case of extremely short dynamic delay tolerance, and the preset emergency intervention scheme needs to be started immediately, and a high-intensity multi-channel stimulation combination is adopted. The standard response range corresponds to the case of moderate dynamic delay tolerance, and there is time to evaluate the specific state of the worker and select the most appropriate intervention strategy. The slow response range corresponds to the case of longer dynamic delay tolerance, and a mild and gradual intervention method can be adopted to avoid strong stimulation interference on the worker. The control parameters are set for each response range, including stimulation intensity, stimulation duration, stimulation frequency and multi-channel combination method. When a miner encounters an emergency danger in the underground mine and shows signs of cognitive overload, the dynamic delay tolerance is detected to be extremely short, and the fast response range is immediately switched to, the maximum intensity sound alarm and vibration reminder combination is started, and the worker's alertness is quickly improved. The worker starts to be sleepy at the end of the night shift, but the state deterioration rate is slow, and the slow response range is adopted, and through gradually increasing environmental lighting adjustment and intermittent mild sound prompts, the worker's clear state is gradually restored. Through the control of the range, precise intervention response is realized for different emergency degrees.

[0078] In step S160, the variable-speed control parameters are compared with the preset safety threshold to generate a control deviation signal, the control deviation signal is inversely phase-modulated to generate a counter-stimulation waveform, and a phase compensation execution instruction is formed based on the counter-stimulation waveform.

[0079] Specifically, the variable-speed control parameters are compared with the preset safety threshold to generate a control deviation signal. The index values in the variable-speed control parameters are extracted, including stimulation intensity, stimulation duration and stimulation frequency. The upper and lower limits of the preset safety threshold are obtained, and the safety threshold is set according to the danger level of the work type and the individual characteristics of the worker. Each control parameter is compared with the corresponding safety threshold to identify the parameters that exceed the safety range. When a control parameter exceeds the safety upper limit or is lower than the safety lower limit, the parameter is in a deviation state. When a worker is working at a high altitude due to extreme fatigue, the stimulation intensity generated according to the fast response range may exceed the individual upper limit of the worker, and at this time the stimulation intensity parameter deviates from the safety threshold. When a miner is working in the underground mine, if it is detected that the cognitive state deteriorates rapidly, the generated stimulation frequency may deviate from the optimal resonance frequency range and cannot effectively activate the target brain electrical frequency band. The deviation direction and deviation amplitude of each deviation parameter are recorded, the deviation direction is divided into positive deviation and negative deviation, and the deviation amplitude reflects the severity of the deviation. A control deviation signal is constructed, and the control deviation signal contains information such as the type of the deviation parameter, the deviation direction, the deviation amplitude and the deviation duration.

[0080] In some embodiments, the anti-phase modulation of the control deviation signal to generate the counter-stimulus waveform comprises: performing phase-domain unfolding of the control deviation signal to obtain a dominant phase component; performing reverse detection of the dominant phase component to identify a counter-stimulus trigger point; constructing a reverse compensation sequence based on the counter-stimulus trigger point; and performing phase compensation modulation of the control deviation signal using the reverse compensation sequence to generate the counter-stimulus waveform.

[0081] The control deviation signal is phase-domain unfolded to obtain a dominant phase component. The control deviation signal is phase analyzed to extract the main phase component of the deviation signal. By analyzing the phase energy distribution, the phase component with the largest energy contribution is identified as the dominant phase component. The dominant phase component represents the core phase characteristics of the control deviation and reflects the main oscillation mode of the deviation signal on the time axis. When the worker receives the stimulus, if the stimulus parameters periodically deviate from the safe range, the deviation signal presents a specific periodic phase pattern, and the dominant phase component corresponds to the starting phase of the deviation period. The starting phase is a key reference point for implementing counter-stimulus intervention. Analyzing the timing characteristics of the dominant phase component provides a benchmark for subsequent reverse detection and compensation design. When the miner is working underground, if the stimulus frequency continuously deviates from the target frequency band, the dominant phase component shows obvious phase drift characteristics, and the deviation trend can be accurately grasped by tracking the characteristics.

[0082] For example, the reverse detection of the dominant phase component to identify a counter-stimulus trigger point comprises: performing phase deviation detection on adjacent phase intervals using the dominant phase component as a detection reference to obtain deviation distribution data; performing reverse amplitude analysis on the deviation distribution data to generate an amplitude distribution; and continuously positioning the reverse position using the amplitude distribution to obtain positioning accuracy, and determining the position as the counter-stimulus trigger point when the positioning accuracy meets the preset phase requirement.

[0083] The dominant phase component is used as a detection reference to perform phase deviation detection on adjacent phase intervals to obtain deviation distribution data. The phase value of the dominant phase component at each time is extracted, and the phase value is used as a detection reference phase. The adjacent phase interval of the dominant phase component is defined, and the deviation of each phase component from the detection reference phase in the adjacent interval is analyzed. The phase deviation is equal to the phase value of each phase component minus the detection reference phase value, and the deviation is positive, indicating that the component phase leads the dominant phase, and the deviation is negative, indicating that it lags behind the dominant phase. The phase deviation values at each time are counted to form the deviation distribution data. The deviation distribution data describes the phase dispersion degree and distribution pattern of the control deviation signal around the dominant phase. During the process of receiving the stimulus, if the deviation of the stimulus parameters leads to the loss of coordination of multiple frequency bands, the deviation distribution data shows that the phase components are highly dispersed, and the deviation value distribution range is large. When the stimulus parameters remain within the safe range, the phase components are concentrated around the dominant phase, and the deviation value distribution range is small.

[0084] The reverse amplitude analysis is performed on the deviation distribution data to obtain an amplitude distribution. The signal amplitudes corresponding to each deviation value in the deviation distribution data are analyzed. The phase component with a large deviation value and a large signal amplitude has a more significant impact on the deviation behavior. The positions where the deviation sign changes in the deviation distribution data are extracted. The change in the deviation sign indicates that the phase component changes from a leading dominant phase to a lagging dominant phase, or vice versa. The change characteristics of the signal amplitudes at the positions where the deviation sign changes are analyzed. The positions where the amplitudes change significantly are marked as reverse amplitude candidate points. The reverse amplitude candidate points correspond to the time points at which the phase and amplitude of the deviation signal change significantly. These time points are the turning points of the behavior pattern of the deviation signal. The amplitudes of each reverse amplitude candidate point are counted to form an amplitude distribution. The amplitude distribution describes the distribution law of the reverse phenomenon at different amplitude levels. When the phase and amplitude of the deviation signal both show reverse characteristics, the amplitude distribution of the fatigue-induced abnormal stimulation response of the miner in the underground mine shows a clear double-peak or multi-peak structure, and the peak positions correspond to the key reverse time points.

[0085] The reverse amplitude distribution is used to continuously locate the reverse position to obtain positioning accuracy. The amplitude peak positions are identified in the amplitude distribution. The peak positions correspond to the time points at which the reverse phenomenon is most significant. The time points corresponding to the peak positions are located to obtain accurate time points. The peak positions are continuously tracked and located. The peak positions are repeatedly detected in consecutive sampling periods to evaluate the time stability of the peak positions. The peak positions remain stable in the multiple sampling periods, indicating high positioning accuracy. The peak positions fluctuate greatly between different sampling periods, indicating low positioning accuracy. When the time fluctuation range of the peak positions is less than a preset threshold and the peak amplitude meets the minimum requirement, the position is confirmed as a reliable counter-stimulus point. When the amplitude distribution shows clear peaks and the continuous positioning accuracy meets the requirements, the position of the counter-stimulus point is accurately identified when the stimulation parameter deviation is continuously monitored by the worker during high-risk operations. The compensation mechanism is ensured to be started at the best time. The counter-stimulus point marks the key time point for implementing reverse compensation. Injecting the reverse-phase signal at these time points can obtain the best counter-stimulating effect.

[0086] The reverse compensation sequence is constructed based on the counter-stimulus points. The time position and corresponding dominant phase value of each counter-stimulus point are extracted, and the dominant phase value represents the phase state deviating from the signal at the stimulus point. The dominant phase value is reversed, and the reversed phase is different from the original phase by π. The reversed phase corresponds to the phase of the counter-stimulus waveform. The reverse compensation pulse is generated at each counter-stimulus point, the phase of the compensation pulse is the reversed phase, the amplitude is determined according to the deviation amplitude and the compensation requirement, and the duration is set according to the deviation duration. The time interval between each counter-stimulus point is analyzed, and the short interval indicates that the deviation occurs frequently, and the long interval indicates that the deviation occurs occasionally. For the frequent deviation, the amplitude of the compensation pulse is enhanced to strengthen the counteracting effect. For the occasional deviation, the compensation pulse with the standard amplitude is used to avoid excessive intervention. The reverse compensation pulses at each counter-stimulus point are arranged in time sequence to form a reverse compensation sequence. The reverse compensation sequence contains the compensation signal parameters at each key moment in the entire deviation process. During long-term operation, the stimulation parameters of the operator may deviate from the safe range multiple times, and compensation pulses are generated at each counter-stimulus point of each deviation. These pulses form a continuous reverse compensation sequence to continuously counteract the deviation trend.

[0087] The reverse compensation sequence is used to modulate the phase compensation of the control deviation signal to generate a counter-stimulus waveform. Each compensation pulse in the reverse compensation sequence is superimposed with the control deviation signal at the corresponding moment, and the superimposition process considers the phase relationship and amplitude weight. Since the phase of the compensation pulse is opposite to that of the deviation signal, the superimposition produces a phase cancellation effect, and the amplitude of the deviation signal is weakened or eliminated. The amplitude weight of the compensation pulse is adjusted so that the amplitude of the superimposed signal returns to the safe range. The superimposed signal is smoothed to eliminate possible spikes or mutations in the compensation process. The smoothed signal is the counter-stimulus waveform, which retains the effective components of the original stimulus while suppressing the abnormal components that deviate from the safe range. When the operator is working at high altitude, the counter-stimulus waveform generated by the reverse compensation sequence is superimposed with the original stimulus to keep the actual stimulation intensity acting on the operator within a safe and effective level. When the miner is working underground, if the stimulation frequency deviates from the optimal resonance frequency, the counter-stimulus waveform corrects the frequency back to the target range through inverse phase modulation to ensure the intervention effect.

[0088] The phase compensation execution instruction is generated based on the counter-stimulus waveform. The amplitude and phase information of the counter-stimulus waveform are analyzed to determine the type and intensity of the phase compensation. The type of phase compensation is divided into positive compensation and negative compensation. The positive compensation is used to enhance some insufficient stimulation parameters, and the negative compensation is used to weaken some excessive stimulation parameters. The intensity of the phase compensation is dynamically adjusted according to the severity of the control deviation and the safety requirements of the operation scene. When the deviation is serious, strong compensation is used, and when the deviation is slight, weak compensation is used. The phase matching degree of the compensation waveform and the original deviation signal is evaluated. High matching degree indicates good compensation effect, and low matching degree indicates that the compensation parameters need to be further optimized. The phase compensation execution instruction is generated, which contains parameters such as the frequency, phase, amplitude and action time of the compensation waveform. These parameters jointly determine the accuracy and effectiveness of the compensation. The phase compensation execution instruction is sent to the stimulation output device of the safety helmet. The device modulates the original stimulation signal in real time according to the instruction parameters, and realizes deviation correction through phase superposition. If the worker detects a deviation in the stimulation parameters during high-risk operations, the phase compensation execution instruction is immediately generated to quickly correct the deviation through the counter-stimulus waveform, ensuring that the intervention measures are timely and effective and will not cause discomfort or interference due to excessive stimulation. When a miner encounters an emergency situation underground, the phase compensation mechanism is used to adjust the stimulation parameters in real time, ensuring the awakening effect while avoiding secondary damage caused by excessive stimulation intensity. The entire closed-loop control process starts with the comparison of the variable speed control parameters and the safety threshold, proceeds through the identification of the control deviation signal and the generation of the counter-stimulus waveform, and finally realizes the dynamic adjustment and safety protection of the stimulation parameters through the phase compensation execution instruction, ensuring that workers can receive safe and effective intervention stimulation in various cognitive states.

[0089] In order to implement the above-mentioned method embodiment corresponding to the safety helmet intelligent control method based on electroencephalogram features, the corresponding functions and technical effects are realized. Referring to Figure 2 , Figure 2 The structure block diagram of the safety helmet intelligent control system 200 based on electroencephalogram features provided by the embodiment of the application is shown. For ease of illustration, only the part related to the embodiment is shown. The safety helmet intelligent control system 200 based on electroencephalogram features provided by the embodiment of the application comprises:

[0090] The signal acquisition module 201 is configured to acquire real-time electroencephalogram signals of workers, perform frequency spectrum decomposition on the real-time electroencephalogram signals to generate cognitive load frequency domain features, and establish an electroencephalogram control space based on the cognitive load frequency domain features.

[0091] The frequency band analysis module 202 is configured to scan the electroencephalogram control space to form a multi-frequency band energy distribution matrix, identify a frequency band coupling imbalance point along the multi-frequency band energy distribution matrix, reversely deduce a cooperative intervention opportunity via the frequency band coupling imbalance point, and form a multi-channel adjustment mode based on the cooperative intervention opportunity.

[0092] a resonance matching module 203, configured to acquire a safety helmet stimulation frequency range and a target electroencephalogram frequency band, map the stimulation frequency range to a frequency matching domain generated by mapping the electroencephalogram control space, and establish a resonance intervention window based on resonance analysis of the frequency matching domain based on the target electroencephalogram frequency band;

[0093] a propagation tracking module 204, configured to perform frequency band monitoring in the resonance intervention window using the multi-channel adjustment mode to obtain an energy migration trajectory, construct a risk propagation chain according to the energy migration trajectory to identify a cascading failure node, and perform a blocking priority evaluation on the cascading failure node to form a key blocking anchor point;

[0094] a response scheduling module 205, configured to perform risk acceleration analysis on the key blocking anchor point to form a state deterioration rate, fuse the cognitive load frequency domain feature using the state deterioration rate to determine a response time and generate a dynamic delay tolerance, perform fast-slow grading according to the dynamic delay tolerance to generate a variable speed control parameter;

[0095] a phase control module 206, configured to compare the variable speed control parameter with a preset safety threshold to generate a control deviation signal, perform anti-phase modulation on the control deviation signal to generate a hedging stimulation waveform, and form a phase compensation execution instruction based on the hedging stimulation waveform.

[0096] The safety helmet intelligent control system 200 based on electroencephalogram features described above can implement the safety helmet intelligent control method based on electroencephalogram features of the method embodiment described above. The optional items in the method embodiment described above are also applicable to this embodiment, and will not be described in detail here. The remaining content of the present embodiment can refer to the content of the method embodiment described above, and will not be described in detail in this embodiment.

[0097] The purpose of the above embodiments is to exemplarily reproduce and deduce the technical solutions of the present application, and to completely describe the technical solutions, purposes and effects of the present application. The purpose is to make the public understand the disclosure of the present application more thoroughly and comprehensively, and does not limit the protection scope of the present application.

[0098] The above embodiments are also not an exhaustive enumeration based on the present application. In addition, there can be many other unlisted embodiments. Any substitution and improvement made without violating the concept of the present application is within the protection scope of the present application.

Claims

1. A method for intelligent control of a safety helmet based on electroencephalogram features, characterized in that, The method comprises the following steps: Collecting real-time brain electrical signals of the working personnel, performing frequency spectrum decomposition on the real-time brain electrical signals to generate cognitive load frequency domain features, and establishing a brain electrical control space based on the cognitive load frequency domain features; Scanning the brain electrical control space to form a multi-frequency band energy distribution matrix, identifying a frequency band coupling imbalance point along the multi-frequency band energy distribution matrix, reversely deducing a collaborative intervention time through the frequency band coupling imbalance point, and forming a multi-channel adjustment mode based on the collaborative intervention time; Obtaining a safety helmet stimulation frequency range and a target brain electrical frequency band, mapping the stimulation frequency range to the brain electrical control space to generate a frequency matching domain, and establishing a resonance intervention window based on the target brain electrical frequency band and the frequency matching domain harmonic analysis; Using the multi-channel adjustment mode to perform frequency band monitoring in the resonance intervention window to obtain an energy migration trajectory, constructing a risk propagation chain according to the energy migration trajectory to identify a cascading failure node, performing a blocking priority evaluation on the cascading failure node to form a key blocking anchor point; Performing risk acceleration analysis on the key blocking anchor point to form a state deterioration rate, using the state deterioration rate to fuse the cognitive load frequency domain features to determine a response time and generate a dynamic delay tolerance, and performing fast-slow grading according to the dynamic delay tolerance to generate a variable speed control parameter; Comparing the variable speed control parameter with a preset safety threshold to generate a control deviation signal, performing anti-phase modulation on the control deviation signal to generate a hedging stimulation waveform, and forming a phase compensation execution instruction based on the hedging stimulation waveform.

2. The method of claim 1, wherein, The method for establishing a brain electrical control space based on the cognitive load frequency domain features comprises the following steps: Performing energy spectrum analysis on the cognitive load frequency domain features to extract energy distribution features; Calibrating a boundary transition threshold based on the energy distribution features; Monitoring fluctuation amplitudes of the cognitive load frequency domain features according to the boundary transition threshold to establish a hierarchical mapping rule, wherein the hierarchical mapping rule comprises a normal working layer, a pre-warning transition layer and a loss-of-control intervention layer; Constructing a brain electrical control space by using the hierarchical mapping rule.

3. The method of claim 1, wherein, The method for reversely deducing a collaborative intervention time through the frequency band coupling imbalance point comprises the following steps: Extracting duration data from the frequency band coupling imbalance point; Identifying a deterioration acceleration gradient based on the duration data; Determining a high-risk time period according to a time point at which the deterioration acceleration gradient exceeds a preset threshold; Moving the start point of the high-risk time period forward by a set buffer duration to serve as a collaborative intervention time.

4. The method of claim 1, wherein, The method for establishing a resonance intervention window based on the target brain electrical frequency band and the frequency matching domain harmonic analysis comprises the following steps: Performing phase locking on the frequency matching domain to extract a frequency component synchronized with the target brain electrical frequency band; Performing coupling degree screening processing on the frequency component to generate a high-coupling frequency set; Performing matching degree analysis on the high-coupling frequency set and the target brain electrical frequency band to obtain a peak resonance point; Determining a resonance intervention window according to a time span set by extending to both sides of the peak resonance point.

5. The method of claim 1, wherein, The method for constructing a risk propagation chain according to the energy migration trajectory to identify a cascading failure node comprises the following steps: Performing propagation direction window division on the energy migration trajectory to generate a direction window set; Performing propagation path planning based on the direction window set to obtain a main propagation channel; Performing intensity modulation processing on the main propagation channel to generate a propagation load; Using the propagation load to perform cascading relationship mapping to identify a cascading failure node.

6. The method of claim 1, wherein, The response time determination based on the state deterioration rate and the cognitive load frequency domain feature includes: Performing change trend analysis on the state deterioration rate to identify an acceleration mutation point; Cross-correlation analysis of the acceleration mutation point and the cognitive load frequency domain feature to generate an urgency index; According to the urgency index, divide the tolerance time level; Based on the tolerance time level, form a dynamic delay tolerance.

7. The method of claim 1, wherein, The anti-phase modulation of the control deviation signal to generate a hedging stimulation waveform includes: Phase domain unfolding of the control deviation signal to obtain a dominant phase component; Reverse detection of the dominant phase component to identify a hedging trigger point; Based on the hedging trigger point, construct a reverse compensation sequence; Phase compensation modulation of the control deviation signal using the reverse compensation sequence to generate a hedging stimulation waveform.

8. The method of claim 6, wherein, The cross-correlation analysis of the acceleration mutation point and the cognitive load frequency domain feature to generate an urgency index includes: Time feature extraction of the acceleration mutation point to obtain a mutation time feature; Based on the mutation time feature, synchronously sample the cognitive load frequency domain feature to extract a synchronization feature; Correlation degree evaluation of the mutation time feature and the synchronization feature to generate a coupling response coefficient; According to the coupling response coefficient, form an urgency index.

9. The method of claim 7, wherein, The reverse detection of the dominant phase component to identify a hedging trigger point includes: Phase deviation detection of the dominant phase component as a detection reference to adjacent phase intervals to obtain deviation distribution data; Reverse amplitude analysis of the deviation distribution data to generate an amplitude distribution; Using the amplitude distribution to continuously locate the reverse position to obtain positioning accuracy, when the positioning accuracy meets the preset phase requirement, the position is determined as the hedging trigger point.

10. A safety helmet intelligent control system based on electroencephalogram features, characterized in that, It includes: A signal acquisition module is used to collect real-time electroencephalogram signals of workers, perform frequency spectrum decomposition on the real-time electroencephalogram signals to generate cognitive load frequency domain features, and establish an electroencephalogram control space based on the cognitive load frequency domain features; A frequency band analysis module is used to scan the electroencephalogram control space to form a multi-frequency band energy distribution matrix, identify a frequency band coupling imbalance point along the multi-frequency band energy distribution matrix, deduce a collaborative intervention opportunity in reverse through the frequency band coupling imbalance point, and form a multi-channel adjustment mode based on the collaborative intervention opportunity; A resonance matching module is used to obtain a safety helmet stimulation frequency range and a target electroencephalogram frequency band, map the stimulation frequency range to the electroencephalogram control space to generate a frequency matching domain, and establish a resonance intervention window based on the target electroencephalogram frequency band and harmonic resonance analysis of the frequency matching domain; A propagation tracking module is used to perform frequency band monitoring in the resonance intervention window using the multi-channel adjustment mode to obtain an energy migration trajectory, construct a risk propagation chain according to the energy migration trajectory to identify a cascading failure node, and perform a blocking priority evaluation on the cascading failure node to form a key blocking anchor point; A response scheduling module is configured to perform risk acceleration analysis on the key blocking anchor point to form a state deterioration rate, to fuse the state deterioration rate with the cognitive load frequency domain features to determine a response time and generate a dynamic delay tolerance, and to perform fast-slow grading according to the dynamic delay tolerance to generate a variable speed control parameter; A phase control module is configured to compare the variable speed control parameter with a preset safety threshold to generate a control deviation signal, to perform inverse phase modulation on the control deviation signal to generate a hedging stimulation waveform, and to form a phase compensation execution instruction based on the hedging stimulation waveform.

Images