Method for realizing electroencephalogram input abnormal state recognition and recovery control, electroencephalogram acquisition device, processor and readable storage medium thereof

Abstract

The application relates to a method for realizing electroencephalogram input abnormal state recognition and recovery control, which comprises the following steps: continuously acquiring original data and system state information of each input channel in a sliding time window during electroencephalogram acquisition; performing feature extraction on each sliding time window; performing abnormal state recognition; selecting corresponding recovery actions from a pre-established recovery action library for execution; continuously observing the channel state to recover normal acquisition output. The method for realizing electroencephalogram input abnormal state recognition and recovery control, the electroencephalogram acquisition device, the processor and the computer readable storage medium thereof improve the distinguishing ability of abnormal root causes and can reduce misjudgment caused by high resistance leakage; a recovery observation and stability confirmation mechanism is arranged to avoid mis-triggering of the abnormal state again after the system is recovered, improve the continuous acquisition stability, and does not depend on additional complex hardware, and has good engineering implementability.

Description

Technical Field

[0001] This invention relates to the field of electroencephalogram (EEG) signal acquisition, and more particularly to the field of wearable physiological electrical front-end control technology. Specifically, it relates to a method for recognizing and controlling abnormal EEG input states, an EEG acquisition device, a processor, and a computer-readable storage medium thereof. Background Technology

[0002] Currently, headphone-type or in-ear EEG devices, compared to traditional desktop EEG devices, have several drawbacks, including smaller electrode contact area, larger contact pressure fluctuations, more frequent changes in wearing posture, more noticeable user motion artifacts, and the internal radio frequency and digital circuits being closer to the analog front end. In these devices, in addition to the normal weak EEG signals, the EEG input end is also prone to issues such as poor contact, lead detachment, input polarization, input stage saturation, electrostatic discharge, ESD protection device leakage, board surface contamination, high-resistance leakage due to moisture or flux residue, BIAS / RLD common-mode circuit abnormalities, and narrowband crosstalk from wireless transmission or digital switching.

[0003] Existing real-time EEG calibration solutions for the ear can effectively select channels or trigger recalibration based on impedance detection, flat channel detection, time / frequency domain characteristics, contextual information, and quality index. However, these solutions mainly address whether the EEG channel is effective and whether the signal quality meets application requirements. They do not further distinguish the specific hardware or front-end root causes of input anomalies, nor do they disclose the implementation of differentiated front-end recovery actions based on different anomaly root causes such as ESD device leakage, board contamination leakage, BIAS / RLD common-mode anomalies, and RF / digital / audio crosstalk.

[0004] Many existing methods rely primarily on single-lead dropout detection, simple threshold comparison, or abnormal amplitude determination to decide whether to stop sampling or reset. This approach has the following shortcomings: 1. It is difficult to distinguish true lead detachment, increased contact resistance, micro-leakage of ESD devices, and high-resistance leakage caused by board contamination, moisture residue, or flux residue based solely on the lead-off register status. 2. When there is a large polarization voltage at the input terminal or an abnormal bias circuit, baseline rise, full range, or repeated triggering after recovery may easily occur, leading to frequent false alarms in the system. 3. When wireless transmission, audio operation, or digital switching introduces fixed frequency points or comb-like interference, the single-channel threshold method is difficult to determine whether it is systematic crosstalk or local contact anomaly. 4. Existing recovery methods are usually crude, often involving stopping the entire machine, resetting all channels, or relying on manual re-wearing, and cannot select different recovery actions according to the type of abnormality; 5. The lack of a stable confirmation mechanism after recovery causes the system to re-enter an anomaly as soon as it resumes output, reducing the availability of continuous data acquisition.

[0005] Therefore, there is an urgent need for a multi-feature recognition and hierarchical recovery control method for abnormal EEG input, in order to improve the accuracy of abnormality judgment and reduce false stop sampling and false reset. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method, EEG acquisition device, processor and computer-readable storage medium for recognizing and controlling abnormal EEG input states with high accuracy, small error and wide applicability.

[0007] To achieve the above objectives, the present invention provides a method for recognizing and controlling abnormal EEG input states, an EEG acquisition device, a processor, and a computer-readable storage medium thereof, as follows: The method for recognizing and controlling abnormal EEG input states is characterized by the following steps: (1) During the EEG acquisition process of the EEG acquisition system, the raw data and system status information of each input channel are continuously acquired by sliding time window. The system status information includes lead detachment detection results, contact status information, bias or common mode node status and system context information. (2) Extract multiple features for each sliding time window, wherein the multiple features are at least one or more of the following: DC bias features, saturation features, contact features, spectrum features, channel consistency features, bias loop features, and history and context features; (3) Identify abnormal states based on multiple features; (4) Select the corresponding front-end recovery action according to the abnormal root cause category. The front-end recovery action includes at least one of the following: reducing gain, temporarily blocking abnormal channels, removing abnormal channels from the BIAS / RLD set, bias release, reference point reconstruction, adjusting lead-off detection timing, and channel-level or AFE-level local reset. (5) After the recovery action is executed, the preset stability conditions are checked within multiple consecutive sliding time windows. When the preset stability conditions are met, normal acquisition output is gradually restored. If the recovery fails within a predetermined number of times, the recovery action is upgraded or the fault is recorded.

[0008] Preferably, the abnormal state identification in step (3) specifically includes: Abnormal EEG input can be identified as at least one of the following: lead detachment or poor contact, input polarization, input stage saturation, ESD protection device leakage, high resistance leakage caused by board contamination, moisture residue or flux residue, bias or common mode circuit abnormality, narrowband crosstalk, and motion artifact.

[0009] Preferably, step (3) further includes the following steps: The extracted features are fused and judged using rule thresholds, weighted scoring, state machines, decision trees, or combinations thereof. The system first enters the anomaly candidate state, and then judges whether the anomaly persists or whether the confidence of multi-feature fusion exceeds the threshold within multiple consecutive time windows before entering the anomaly confirmation state.

[0010] Preferably, step (4) further includes the following steps: When an abnormal channel is identified as having poor contact, input polarization, input stage saturation, leakage of ESD protection devices, or high-resistance leakage caused by board contamination, and it is determined that the abnormal channel affects the stability of the bias or common-mode circuit, the controller will temporarily remove the abnormal channel from the BIAS / RLD participation set. After recovery observation and stability confirmation meet the preset conditions, the abnormal channel will be restored to participate in the BIAS / RLD bias or common-mode drive.

[0011] Preferably, the recovery action reselects an effective channel or prompts for recalibration, and also controls the channel output, gain, bias participation set, lead-off detection parameters, reference point, or AFE working state of the EEG acquisition front end. The selection of the corresponding front-end recovery action based on the root cause category of the anomaly is specifically as follows: If the lead is detached or has poor contact, perform a re-contact confirmation, wearing reminder, and / or temporarily block abnormal channel data output; For input polarization state or DC bias accumulation state, perform at least one of the following: bias release wait, reduce the gain of abnormal channel or corresponding AFE, and reference point reconstruction; If the input stage is saturated, the gain of the corresponding channel or the corresponding AFE is reduced and / or a local reset is performed. For narrowband crosstalk caused by radio frequency, digital clock, audio service or power supply coupling, the wireless transmission, audio service and digital clock status are combined for cross-judgment, and the acquisition timing is adjusted, the channel participation relationship is modified or the anomaly judgment threshold is increased. For motion artifacts, the data for the corresponding time period is first marked and kept under observation. Only when the anomaly persists and the contact characteristics deteriorate will the recovery action be upgraded. In cases where the ESD protection device is determined to be in an abnormal leakage state, channel isolation and local reset of the EEG simulation front-end AFE containing the abnormal channel are performed first. For cases where the high resistance leakage is determined to be caused by board surface contamination, moisture residue, or flux residue, the priority should be to implement cleaning instructions, wait for drying, and re-confirm contact.

[0012] Preferably, the recovery action further includes: Adjust the timing, duty cycle, or detection current of the lead detachment detection to reduce the disturbance to the data acquisition during the detection process; For persistent and unrecoverable anomalies, perform a local reset of the corresponding AFE, rebuild the bias circuit, or generate information prompting the user to wear the device again or to intervene manually.

[0013] Preferably, the abnormal leakage state of the ESD protection device is distinguished from the high-resistance leakage state caused by board surface contamination, moisture residue, or flux residue by the following method: If a fixed offset or lead-off code abnormality occurs when the channel is not in place, and the abnormality is concentrated in a single channel or a single input terminal, and cannot be significantly restored after cleaning or drying, or is improved after replacing / isolating the corresponding input protection device, it should be preliminarily judged as an abnormal leakage state of the ESD protection device. If the abnormality worsens after being exposed to moisture, multiple adjacent channels deteriorate simultaneously, or improves after drying or cleaning, it should be primarily identified as a high-resistance leakage condition caused by board surface contamination, moisture residue, or flux residue.

[0014] Preferably, the preset stability conditions in step (5) include at least one of the following: the channel baseline returns to a preset range, the saturation ratio is lower than a threshold, the abnormal spectral line energy decays, the lead dropout and impedance characteristics recover to stability, the bias or common-mode circuit node recovers to the normal fluctuation range, and the abnormal EEG input no longer recurs continuously. When these conditions are met continuously within multiple sliding time windows, the original gain, the original bias participation set, and the normal output strategy are gradually restored.

[0015] Preferably, the EEG acquisition system is a multi-channel headphone-type or in-ear EEG acquisition system; the system status information continuously acquired in step (1) also includes the user wearing status, the inertial measurement unit (IMU) action status, the wireless transmission status, or the audio working status as auxiliary criteria.

[0016] Preferably, for a multi-channel EEG system consisting of multiple EEG simulation front-ends (AFEs), when the abnormality is concentrated in only one channel or one EEG simulation front-end (AFE), step (4) performs local recovery according to the abnormality recovery granularity at the channel level or AFE level, ensuring that the EEG simulation front-end (AFE) and its corresponding channel that are not affected by the abnormality maintain normal acquisition.

[0017] The main feature of this EEG acquisition device is that it includes an electrode input terminal, an EEG simulation front-end (AFE), a processor, and a memory. The EEG simulation front-end AFE includes a configurable gain, a lead-off detection module, and a bias / common-mode driver interface. The processor is used to configure AFE registers, adjust lead-off detection timing, control the output state of abnormal channels, adjust the BIAS / RLD participation set, and perform channel-level or AFE-level local reset. The processor is configured to execute computer-executable instructions; The memory stores one or more computer-executable instructions, which, when executed by the processor, implement the various steps of the method for recognizing and controlling abnormal EEG input states.

[0018] The processor for recognizing and controlling abnormal EEG input states is characterized in that it is configured to execute computer-executable instructions, which, when executed by the processor, implement the various steps of the method for recognizing and controlling abnormal EEG input states.

[0019] The main feature of this computer-readable storage medium is that it stores a computer program thereon, which can be executed by a processor to implement the various steps of the method for recognizing and controlling abnormal EEG input states described above.

[0020] This invention employs a method for recognizing and controlling abnormal EEG input states, an EEG acquisition device, a processor, and its computer-readable storage medium. Through joint determination of DC bias, saturation rate, spectrum, common-mode nodes, channel consistency, and contextual information, it improves the ability to distinguish the root causes of abnormalities. Separating ESD device leakage from board contamination and residual leakage due to moisture reduces misjudgments caused by high-resistance leakage. A tiered recovery strategy prioritizes low-side-effect actions, avoiding triggering a complete system shutdown or full-channel reset upon the appearance of a local abnormality. Temporarily removing abnormal channels from the BIAS / RLD aggregation reduces the drag on overall common-mode suppression capabilities from local abnormalities. A recovery observation and stability confirmation mechanism prevents the system from falsely triggering abnormalities again immediately after recovery, improving continuous acquisition stability. This method can be directly implemented in small EEG devices such as earphones and in-ear monitors without relying on additional complex hardware, demonstrating good engineering feasibility. Attached Figure Description

[0021] Figure 1 This is a system block diagram of the method for recognizing and controlling abnormal EEG input states according to the present invention.

[0022] Figure 2 This is a flowchart illustrating the abnormal identification and recovery control state of the method for realizing abnormal state identification and recovery control of EEG input according to the present invention. Detailed Implementation

[0023] To more clearly describe the technical content of the present invention, the following description is provided in conjunction with specific embodiments.

[0024] The method for recognizing and controlling abnormal EEG input states according to the present invention includes the following steps: (1) During the EEG acquisition process of the EEG acquisition system, the raw data and system status information of each input channel are continuously acquired by sliding time window. The system status information includes lead detachment detection results, contact status information, bias or common mode node status and system context information. (2) Extract multiple features for each sliding time window, wherein the multiple features are at least one or more of the following: DC bias features, saturation features, contact features, spectrum features, channel consistency features, bias loop features, and history and context features; (3) Identify abnormal states based on multiple features; (4) Select the corresponding front-end recovery action according to the abnormal root cause category. The front-end recovery action includes at least one of the following: reducing gain, temporarily blocking abnormal channels, removing abnormal channels from the BIAS / RLD set, bias release, reference point reconstruction, adjusting lead-off detection timing, and channel-level or AFE-level local reset. (5) After the recovery action is executed, the preset stability conditions are checked within multiple consecutive sliding time windows. When the preset stability conditions are met, normal acquisition output is gradually restored. If the recovery fails within a predetermined number of times, the recovery action is upgraded or the fault is recorded.

[0025] In a preferred embodiment of the present invention, step (3) of identifying abnormal states specifically includes: Abnormal EEG input can be identified as at least one of the following: lead detachment or poor contact, input polarization, input stage saturation, ESD protection device leakage, high resistance leakage caused by board contamination, moisture residue or flux residue, bias or common mode circuit abnormality, narrowband crosstalk, and motion artifact.

[0026] In a preferred embodiment of the present invention, step (3) further includes the following steps: The extracted features are fused and judged using rule thresholds, weighted scoring, state machines, decision trees, or combinations thereof. The system first enters the anomaly candidate state, and then judges whether the anomaly persists or whether the confidence of multi-feature fusion exceeds the threshold within multiple consecutive time windows before entering the anomaly confirmation state.

[0027] Preferably, step (4) further includes the following steps: When an abnormal channel is identified as having poor contact, input polarization, input stage saturation, ESD protection device leakage, or high-resistance leakage caused by board contamination, and it is determined that the abnormal channel affects the stability of the bias or common-mode circuit, the controller temporarily removes the abnormal channel from the BIAS / RLD participation set. After recovery observation and stability confirmation meet preset conditions, the abnormal channel is then restored to participate in the BIAS / RLD bias or common-mode drive. Preferably, the recovery action reselects an effective channel or prompts for recalibration, and also controls the channel output, gain, bias participation set, lead-off detection parameters, reference point, or AFE working state of the EEG acquisition front end. The selection of the corresponding front-end recovery action based on the root cause category of the anomaly is specifically as follows: If the lead is detached or has poor contact, perform a re-contact confirmation, wearing reminder, and / or temporarily block abnormal channel data output; For input polarization state or DC bias accumulation state, perform at least one of the following: bias release wait, reduce the gain of abnormal channel or corresponding AFE, and reference point reconstruction; If the input stage is saturated, the gain of the corresponding channel or the corresponding AFE is reduced and / or a local reset is performed. For narrowband crosstalk caused by radio frequency, digital clock, audio service or power supply coupling, the wireless transmission, audio service and digital clock status are combined for cross-judgment, and the acquisition timing is adjusted, the channel participation relationship is modified or the anomaly judgment threshold is increased. For motion artifacts, the data for the corresponding time period is first marked and kept under observation. Only when the anomaly persists and the contact characteristics deteriorate will the recovery action be upgraded. In cases where the ESD protection device is determined to be in an abnormal leakage state, channel isolation and local reset of the EEG simulation front-end AFE containing the abnormal channel are performed first. For cases where the high resistance leakage is determined to be caused by board surface contamination, moisture residue, or flux residue, the priority should be to implement cleaning instructions, wait for drying, and re-confirm contact.

[0028] In a preferred embodiment of the present invention, the recovery action further includes: Adjust the timing, duty cycle, or detection current of the lead detachment detection to reduce the disturbance to the data acquisition during the detection process; For persistent and unrecoverable anomalies, perform a local reset of the corresponding AFE, rebuild the bias circuit, or generate information prompting the user to wear the device again or to intervene manually.

[0029] In a preferred embodiment of the present invention, the abnormal leakage state of the ESD protection device is distinguished from the high-resistance leakage state caused by board surface contamination, moisture residue, or flux residue in the following manner: If a fixed offset or lead-off code abnormality occurs when the channel is not in place, and the abnormality is concentrated in a single channel or a single input terminal, and cannot be significantly restored after cleaning or drying, or is improved after replacing / isolating the corresponding input protection device, it should be preliminarily judged as an abnormal leakage state of the ESD protection device. If the abnormality worsens after being exposed to moisture, multiple adjacent channels deteriorate simultaneously, or improves after drying or cleaning, it should be primarily identified as a high-resistance leakage condition caused by board surface contamination, moisture residue, or flux residue.

[0030] As a preferred embodiment of the present invention, the preset stability conditions in step (5) include at least one of the following: the channel baseline returns to a preset range, the saturation ratio is lower than a threshold, the abnormal spectral line energy decays, the lead detachment and impedance characteristics recover to stability, the bias or common-mode circuit node recovers to the normal fluctuation range, and the abnormal EEG input no longer recurs continuously. When these conditions are met continuously within multiple sliding time windows, the original gain, the original bias participation set, and the normal output strategy are gradually restored.

[0031] As a preferred embodiment of the present invention, the EEG acquisition system is a multi-channel headphone-type or in-ear EEG acquisition system; the system status information continuously acquired in step (1) also includes the user wearing status, the inertial measurement unit (IMU) action status, the wireless transmission status, or the audio working status as auxiliary criteria.

[0032] As a preferred embodiment of the present invention, for a multi-channel EEG system consisting of multiple EEG analog front-ends (AFEs), when the abnormality is concentrated in only one channel or one EEG analog front-end (AFE), step (4) performs local recovery according to the abnormality recovery granularity at the channel level or AFE level, so as to ensure that the EEG analog front-end (AFE) and its corresponding channel that are not affected by the abnormality maintain normal acquisition.

[0033] The electroencephalogram (EEG) acquisition device of the present invention includes an electrode input terminal, an EEG simulation front-end (AFE), a processor, and a memory. The EEG simulation front-end AFE includes a configurable gain, a lead-off detection module, and a bias / common-mode driver interface. The processor is used to configure AFE registers, adjust lead-off detection timing, control the output state of abnormal channels, adjust the BIAS / RLD participation set, and perform channel-level or AFE-level local reset. The processor is configured to execute computer-executable instructions; The memory stores one or more computer-executable instructions, which, when executed by the processor, implement the various steps of the method for recognizing and controlling abnormal EEG input states.

[0034] The processor of the present invention for recognizing and controlling abnormal EEG input states is configured to execute computer-executable instructions, which, when executed by the processor, implement the various steps of the method for recognizing and controlling abnormal EEG input states described above.

[0035] The computer-readable storage medium of the present invention stores a computer program thereon, which can be executed by a processor to implement the various steps of the method for recognizing and controlling abnormal EEG input states as described above.

[0036] This invention belongs to the field of EEG signal acquisition and wearable physiological electrical front-end control technology, specifically relating to a method for identifying, classifying, recovering, and confirming the stability of abnormal states at the EEG input end. This method is applicable to earphone-type, in-ear, headband-type, and other high-input-impedance EEG acquisition devices, and is particularly suitable for dry electrodes, semi-dry electrodes, flexible electrodes, and miniaturized EEG systems integrating wireless communication, audio, IMU, or charging management modules.

[0037] The purpose of this invention is to provide a method for identifying and controlling abnormal EEG input states. Addressing common issues in headphone-type EEG systems such as lead detachment, poor contact, input polarization, input stage saturation, ESD device leakage, board contamination leakage, common-mode circuit abnormalities, RF / digital crosstalk, and motion artifacts, a closed-loop control mechanism is established: "multi-feature identification—abnormality classification—graded recovery—stability confirmation." This method enhances the system's ability to identify the root causes of abnormalities without significantly increasing hardware complexity. While maintaining continuous acquisition of other channels, targeted recovery actions are performed on abnormal channels or abnormal AFEs, thereby reducing misjudgments, shortening recovery waiting time, and increasing the proportion of effective data from continuous EEG acquisition.

[0038] like Figure 1The block diagram of the EEG input abnormal state recognition and recovery control system illustrates the relationship between the electrodes / input terminal, EEG AFE, feature extraction module, abnormal state recognition module, recovery control module, and output module.

[0039] like Figure 2 The flowchart for the anomaly identification-recovery control state machine illustrates the process from normal monitoring, anomaly candidate detection, anomaly confirmation, recovery control, recovery observation, and successful recovery or escalation processing.

[0040] The EEG input abnormality state recognition and recovery control method of the present invention is applied to an EEG acquisition system, comprising: During EEG acquisition, data from each input channel, lead detachment detection results, contact status information, and bias or common-mode node status are obtained. Extract at least one of the following features based on a sliding time window: DC bias feature, saturation feature, contact feature, spectral feature, channel consistency feature, and bias loop feature; Based on the aforementioned characteristics, abnormal EEG input can be identified as at least one of the following: lead detachment or poor contact, input polarization, input stage saturation, ESD protection device leakage, board surface contamination leakage, bias / common mode circuit abnormality, narrowband crosstalk, and motion artifacts. Select the corresponding recovery action based on the anomaly category, and confirm the stability after recovery. Once the stability conditions are met, resume normal data acquisition and output.

[0041] The anomaly identification is achieved using rule-based thresholds, weighted scoring, state machines, decision trees, or combinations thereof, and anomaly candidate states and anomaly confirmation states are set to reduce misjudgments caused by instantaneous fluctuations.

[0042] The recovery actions include at least one of the following: reducing gain, adjusting lead detection timing, temporarily blocking abnormal channels, removing abnormal channels from the bias set, re-performing contact confirmation, bias release waiting, reference reconstruction, local reset of the corresponding AFE, and reporting manual intervention.

[0043] Different identification rules and recovery actions are adopted for the abnormal leakage current state of ESD protection devices and the high resistance leakage current state caused by board surface contamination or moisture residue.

[0044] The stability confirmation includes at least one of the following: baseline regression, saturation ratio reduction, abnormal spectral line attenuation, lead state recovery, and bias / common mode node return to normal range, which must be continuously satisfied for a preset time.

[0045] The EEG acquisition system is an earphone-type or in-ear EEG acquisition system, and the user's wearing status, IMU action status, wireless transmission status or audio working status are used as auxiliary criteria.

[0046] For multichannel EEG systems composed of multiple AFEs, abnormal recovery is performed at the channel level or AFE level to ensure that unaffected channels continue to acquire data.

[0047] In small-area EEG circuits, a unified reference ground plane is used for analog and digital grounds, and abnormal channel isolation and backflow path control are combined to improve the stability of abnormality recognition.

[0048] An electroencephalogram (EEG) acquisition device includes an electrode input terminal, an EEG simulation front end, a processor, and a memory. The memory stores a program, which, when executed by the processor, implements the method described above.

[0049] This invention provides a method for recognizing and controlling abnormal EEG input states, deployed in an EEG acquisition system including an electrode input terminal, an EEG analog front-end (AFE), a processor or controller, an optional bias / common-mode drive circuit, and an output management module. The technical solution includes at least the following steps: S1. Data Collection and Monitoring: During normal EEG acquisition, the controller continuously acquires raw data from each input channel, lead-off detection results, contact impedance estimates, BIAS / RLD node status, AFE saturation flag, gain status, and system context information through a sliding time window. The context information may include wireless transmission status, audio operating status, IMU motion status, charging status, historical recovery results, or user wearing status.

[0050] S2, Feature Extraction: For each time window, extract at least one or more of the following features: (1) DC bias characteristics: including channel mean, baseline drift velocity, and inter-window mean change; (2) Saturation characteristics: including the proportion of full-range sampling points, continuous saturation duration, and positive and negative saturation distribution; (3) Contact characteristics: including lead-off register state, impedance change, and contact recovery duration; (4) Spectral characteristics: including power frequency and harmonics, fixed narrowband interference, comb-spaced interference and abnormal frequency band energy; (5) Channel consistency characteristics: including correlation between left and right ears or adjacent channels, degree of common-mode synchronization change, number of abnormal channels and spatial distribution; (6) Bias loop characteristics: including BIAS / RLD node potential, fluctuation trend, and changes in the set of bias participation channels; (7) Historical and contextual characteristics: including whether the anomaly is associated with wireless transmission, digital service switching, audio prompts, IMU actions, or previous recovery failures.

[0051] S3. Abnormal Status Identification: Abnormal EEG input can be classified into at least one or more of the following categories: (a) Lead detachment or poor contact; (b) Input polarization or DC bias accumulation state; (c) Input stage saturation or overload condition after electrostatic discharge; (d) Abnormal leakage current state of ESD protection devices; (e) High-resistance leakage caused by board surface contamination, moisture residue, or flux residue; (f) Abnormal state of BIAS / RLD common mode circuit; (g) Narrowband crosstalk caused by radio frequency, digital clock, audio or power supply coupling; (h) Impedance jump state caused by motion artifacts or attitude changes.

[0052] During identification, rule-based thresholding, weighted scoring, state machines, decision trees, or combinations thereof can be used for fusion judgment. Preferably, the system first enters the abnormal candidate state, and only enters the abnormal confirmation state when the candidate conditions continuously reach a set window number, or when the confidence of multi-feature fusion exceeds a threshold.

[0053] S4. Graded recovery control: After an anomaly is confirmed, the controller selects a higher-priority action with fewer side effects from the recovery action library based on the anomaly type, severity, number of affected channels, and historical recovery records. If recovery is ineffective, a stronger recovery action is then executed. The recovery actions include at least: (1) Reduce the gain of the corresponding channel or the corresponding AFE; (2) Temporarily block the data output of abnormal channels, but continue to collect data from the remaining channels; (3) Temporarily remove the abnormal channel from the BIAS / RLD participant set to avoid single-point abnormality from damaging the overall common-mode circuit; (4) Adjust the lead-off detection timing, duty cycle or detection current to reduce the disturbance of the detection process itself to the data acquisition; (5) Re-execute the contact confirmation, impedance estimation, or wearing status confirmation process; (6) Perform bias release waiting, reference point reconstruction, or stepwise gain recovery for suspected polarization states; (7) Reduce the judgment weight of suspected ESD leakage or board surface contamination leakage, and perform channel isolation or AFE partial reset if necessary; (8) Jointly determine the status of wireless transmission, audio service and digital clock for suspected crosstalk, and switch the acquisition gap timing or modify the bias participation relationship when necessary; (9) For persistent and unrecoverable anomalies, perform corresponding AFE local reset, bias circuit reconstruction, or report for manual intervention.

[0054] S5. Recovery observation and stabilization confirmation: After the recovery action is executed, the system enters recovery observation mode, checking whether the following conditions are met in multiple consecutive windows: (1) The channel baseline is returned to the preset range; (2) The saturation ratio is below the threshold; (3) The energy of the abnormal spectral lines decreases significantly or disappears; (4) The lead-off and impedance characteristics have recovered to a stable state; (5) The BIAS / RLD node returned to its normal fluctuation range; (6) The abnormality no longer recurs continuously.

[0055] The system will only release the recovery state and gradually restore the original gain, original bias participation set, and normal output strategy after the conditions have been met for a set time.

[0056] S6. Exception escalation and logging: If recovery fails within the predetermined number of attempts, the system will upgrade the anomaly level and perform actions such as resetting the affected AFE, reinitializing the entire device, prompting the user to wear the device again, or outputting a fault log. At the same time, the system will record the anomaly type, duration, recovery action, and recovery result to provide a basis for subsequent parameter adaptation or production quality traceability.

[0057] Preferably, the present invention is applicable to small-area earphone-type EEG circuits, wherein the analog ground and digital ground can adopt a unified reference ground plane and reduce noise through layout, return path control and local isolation strategies, so that the anomaly recognition method can rely more on the input features themselves, rather than on the additional reference error caused by complex grounding.

[0058] The inventive point of this invention is: Multi-feature fusion identification of the root cause of input anomalies; state machine for anomaly candidates and anomaly confirmation; differential and hierarchical recovery control according to anomaly category; local recovery of anomaly channels / AFE; removal of anomaly channels from BIAS / RLD to participate in the set; stability confirmation and failure escalation after recovery.

[0059] 1. Two-level judgment mechanism for anomaly candidates and anomaly confirmation The system does not immediately perform recovery upon detecting an anomaly. Instead, it first enters the anomaly candidate state and determines whether the anomaly persists or whether the confidence level of multi-feature fusion exceeds the threshold within multiple consecutive time windows before entering the anomaly confirmation state.

[0060] This feature can reduce false recovery and false reset caused by instantaneous movement, short-term contact jitter, and instantaneous wireless transmission.

[0061] 2. Multi-feature fusion recognition mechanism The characteristics include at least a number of the following: DC bias characteristics, saturation characteristics, contact characteristics, spectral characteristics, channel consistency characteristics, BIAS / RLD bias loop characteristics, and system context characteristics.

[0062] This feature distinguishes this case from simple impedance testing, simple lead-off testing, or simple signal quality scoring.

[0063] 3. Abnormal root cause classification mechanism The anomalies can be categorized into at least the following types: contact anomalies, polarization / DC bias anomalies, saturation / overload anomalies, ESD device leakage anomalies, board surface contamination / moisture residual leakage anomalies, BIAS / RLD common mode circuit anomalies, RF / digital / audio coupling crosstalk, and motion artifact anomalies.

[0064] In particular, it is recommended to emphasize that "ESD protection device leakage" and "board surface contamination / moisture residual high-resistance leakage" should be identified as different root causes, because they may appear similar, but the recovery strategies are different.

[0065] 4. Perform differentiated recovery actions based on the anomaly category. Different types of anomalies require different recovery actions, rather than a uniform halt to sampling or a uniform reset. For example: Abnormal contact: Re-contact confirmation, wearing reminder, temporary shielding channel; Input polarization: bias release wait, reduce gain, reference reconstruction; Saturation / Overload: Reduce gain, AFE local reset; ESD leakage current: channel isolation, device fault marking, partial reset; Surface contamination / dampness residue: Cleaning / drying reminder, delayed retesting; BIAS / RLD anomaly: Remove the BIAS from the abnormal channel to participate in the set and rebuild the common mode loop; RF / Digital Crosstalk: Adjust the acquisition timing or increase the corresponding threshold based on the wireless / audio / digital service status; Motion artifacts: First mark the data and observe it, then do not reset it immediately.

[0066] This feature is a key creative source that distinguishes this case from existing impedance detection technologies.

[0067] 5. Abnormal channels are removed from BIAS / RLD to participate in the set. This feature is suggested as a key dependent claim. Its function is to prevent a channel from participating in common-mode / bias drive calculations when a contact anomaly, polarization anomaly, or leakage anomaly occurs, thereby hindering other normal channels. The effectiveness of this technology is directly related to the stability of multi-channel continuous acquisition in earphone-type miniature EEG systems.

[0068] 6. Channel-level / AFE-level local recovery When anomalies are concentrated in a specific channel or a single AFE, partial recovery is performed only on that channel or AFE, while data acquisition continues on other channels. This feature highlights the technical effectiveness of this project in reducing overall system downtime and increasing the proportion of effective data.

[0069] 7. Recovery observation and stabilization confirmation mechanism After the recovery action is executed, normal output is not restored immediately. Instead, the baseline, saturation ratio, abnormal spectral lines, lead-off, impedance, and BIAS / RLD node status are checked for stability over several consecutive windows. If these conditions are not met, a higher-level recovery or logging process is initiated.

[0070] This feature indicates that the case involves closed-loop control, rather than a one-time detection or one-time alarm.

[0071] The following are alternative solutions to the present invention: 1. The anomaly detection part can be replaced by a lightweight machine learning classifier instead of a fixed rule threshold. 2. Recovery actions can be implemented at the channel level, AFE level, or whole-machine level; 3. In addition to IMU, contact status auxiliary information can also be obtained from wear detection, capacitive touch, temperature or optical sensor signals; 4. For systems without active BIAS / RLD drivers, the common-mode reference node state can be used instead of the bias drive state for judgment.

[0072] This invention focuses on strengthening protection against ESD device leakage, high-resistance leakage due to board contamination / moisture residue, BIAS / RLD common-mode loop anomalies, and RF / digital / audio coupling crosstalk anomalies. These anomalies are not equivalent to "poor electrode contact" or "poor channel quality." For example, ESD device leakage and board contamination leakage may manifest as baseline drift or false alarms, but are not necessarily electrode contact problems; RF / audio coupling interference may simultaneously appear as narrowband or comb-like spectral lines in multiple channels, and cannot be accurately attributed through impedance detection or flat channel detection.

[0073] In headphone-type or in-ear EEG systems, if a channel has abnormal contact, input polarization, leakage, or saturation, and that channel is still involved in BIAS / RLD common-mode drive calculations, the common-mode circuit may be deflected by the abnormal channel, thereby affecting other normal channels.

[0074] This invention proposes that when an abnormal channel is identified that may affect the BIAS / RLD common-mode circuit, the abnormal channel can be temporarily removed from the BIAS / RLD participation set, and its participation can be gradually restored after recovery observation and stability confirmation.

[0075] This invention establishes a closed-loop control mechanism targeting multiple abnormal root causes at the input side of EEG acquisition front-end, encompassing "abnormal candidate—abnormal confirmation—abnormal root cause classification—graded recovery control—post-recovery stability confirmation—failure escalation." Specifically, this application further distinguishes between front-end abnormal root causes such as ESD protection device leakage, high-resistance leakage due to board contamination / moisture residue, BIAS / RLD common-mode circuit abnormalities, and RF / digital / audio coupling crosstalk, and performs differentiated recovery actions based on different abnormality categories. These actions include removing abnormal channels from BIAS / RLD aggregation, reducing gain, bias release, reference reconstruction, lead-off timing adjustment, and channel-level / AFE-level local reset. These features have not been disclosed or explicitly illustrated in existing technologies.

[0076] In specific embodiments of the present invention, the following examples are provided: Example 1: Anomaly Recognition and Recovery in Headphone-Type EEG Devices An earphone-style EEG system includes left and right ear electrodes, an EEG AFE (Automatic External Wire), a processor, a wireless module, an IMU (Integrated Mutual Module), and a bias / common-mode circuit. The processor continuously reads EEG data at a sampling rate of 500 sps or higher and extracts the mean, saturation rate, lead-off state, specific frequency energy, left and right ear correlation, and BIAS node state for each channel using a sliding time window of 250 ms to 2 s.

[0077] When a channel experiences a lead-off anomaly, a sudden increase in impedance, and the anomaly is primarily concentrated in that single channel, the system identifies it as a candidate state for lead detachment or poor contact. If this state persists for multiple consecutive windows, the anomaly is confirmed, and contact reassessment and temporary removal from the BIAS participation set are performed for recovery. If the channel regains contact, the system maintains a stable state for several windows after recovery observation before gradually restoring its participation in the bias circuit and data output.

[0078] When the mean of a channel continuously shifts, the baseline slowly drifts, and the mean continuously approaches the range boundary in either the forward or reverse direction, but the lead-off state has not completely disengaged, the system identifies this as an input polarization or DC bias accumulation state. In this case, the gain of the corresponding channel or the corresponding AFE is first reduced, and a bias release wait is performed. If the saturation disappears after the bias returns, the original gain is restored; if this is ineffective, the system upgrades to reference reconstruction or a partial reset.

[0079] When a channel exhibits a fixed offset, abnormal lead-off code, and significant differences before and after cleaning while remaining stationary, the system prioritizes identifying it as an ESD protection device leakage anomaly. Conversely, when the anomaly worsens after moisture exposure, deteriorates simultaneously in multiple adjacent channels, and improves after drying or cleaning, the system prioritizes identifying it as a high-resistance leakage anomaly caused by board surface contamination or residual moisture. Although both may superficially cause baseline drift and false alarms, their recovery priorities differ: the former prioritizes channel isolation and partial reset, while the latter prioritizes cleaning prompts, drying wait times, and reconfirmation.

[0080] When the BIAS / RLD node fluctuates abnormally, multiple channels simultaneously experience common-mode rise, while the impedance of a single channel does not deteriorate significantly, the system judges it as an abnormal state of the common-mode loop. In this case, the abnormal channels can be reduced from participating in the bias operation. If necessary, the bias loop can be rebuilt, and then the remaining channels can be observed to see if they recover.

[0081] When fixed narrowband or comb-shaped frequency points related to wireless transmission, digital service handover, or prompt tones appear in the spectrum, and multiple channels simultaneously experience near-synchronous interference, the system classifies it as crosstalk and performs cross-verification in conjunction with wireless / audio service flags. Recovery actions may include adjusting the acquisition and transmission timing, switching detection time slots, modifying channel participation relationships, or temporarily increasing the anomaly detection threshold to avoid mistaking systemic crosstalk for contact anomalies.

[0082] When the IMU detects a significant user movement, and impedance characteristics, waveform amplitude, and low-frequency energy change synchronously, the system can classify this anomaly as a motion artifact. In this case, the system does not immediately perform a hard reset; instead, it first marks the data for that time period as an artifact and continues to observe it. Only when the anomaly persists and is accompanied by a deterioration in contact characteristics will further recovery actions be escalated.

[0083] Example 2: Local Recovery in a Multi-AFE Cascade System For a multi-channel EEG system composed of multiple AFEs, this invention allows the granularity of anomaly recovery to be set at the channel level or the AFE level. If the anomaly is concentrated in one or more input channels under a certain AFE, then only the register reconfiguration or partial reset is performed on that AFE, while other AFEs maintain continuous acquisition. This can reduce the interruption of effective data caused by a system reset.

[0084] Example 3: System without Active RLD For systems that do not use active RLD drivers and only use reference nodes or virtual midpoints, the same anomaly identification and recovery control logic as the present invention can still be achieved through common-mode reference node voltage, the degree of synchronization offset of each channel, and baseline changes before and after recovery.

[0085] The threshold, window length, number of recovery attempts, and upgrade conditions in this invention can all be adjusted according to different electrode types, sampling rates, front-end ranges, and wearing methods. As long as multi-feature fusion is used to identify abnormal EEG input and graded recovery control is performed according to the abnormality category, they all fall within the protection scope of this invention.

[0086] In this invention, the English abbreviations are explained as follows: EEG stands for electroencephalogram (EEG); AFE stands for Analog FrontEnd; PGA stands for Programmable Gain Amplifier; BIAS / RLD stands for Bias / Right Leg Drive or Equivalent Common Mode Drive Circuit; ESD stands for Electrostatic Discharge Protection; IMU stands for Inertial Measurement Unit; and lead-off stands for Lead-Off Detection.

[0087] For the specific implementation scheme of this embodiment, please refer to the relevant descriptions in the above embodiments, which will not be repeated here.

[0088] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0089] It should be noted that in the description of this invention, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means at least two.

[0090] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0091] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution device. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0092] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The corresponding program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0093] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0094] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.

[0095] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0096] This invention employs a method for recognizing and controlling abnormal EEG input states, an EEG acquisition device, a processor, and its computer-readable storage medium. Through joint determination of DC bias, saturation rate, spectrum, common-mode nodes, channel consistency, and contextual information, it improves the ability to distinguish the root causes of abnormalities. Separating ESD device leakage current from residual leakage current caused by board contamination / dampness reduces misjudgments due to high-resistance leakage. A tiered recovery strategy prioritizes low-side-effect actions, avoiding triggering a complete system shutdown or full-channel reset upon the appearance of a local abnormality. Temporarily removing abnormal channels from the BIAS / RLD aggregation reduces the drag on overall common-mode suppression capabilities from local abnormalities. A recovery observation and stability confirmation mechanism prevents the system from falsely triggering abnormalities again immediately after recovery, improving continuous acquisition stability. This method can be directly implemented in small EEG devices such as earphones and in-ear monitors without relying on additional complex hardware, demonstrating good engineering feasibility.

[0097] In this specification, the invention has been described with reference to specific embodiments thereof. However, it will be apparent that various modifications and variations can be made without departing from the spirit and scope of the invention. Therefore, the specification and drawings should be considered illustrative rather than restrictive.

Claims

1. A method for recognizing and controlling abnormal EEG input states, characterized in that, The method includes the following steps: (1) During the EEG acquisition process of the EEG acquisition system, the raw data and system status information of each input channel are continuously acquired by sliding time window. The system status information includes lead detachment detection results, contact status information, bias or common mode node status and system context information. (2) Extract multiple features for each sliding time window, wherein the multiple features are at least one or more of the following: DC bias features, saturation features, contact features, spectrum features, channel consistency features, bias loop features, and history and context features; (3) Identify abnormal states based on multiple features; (4) Select the corresponding front-end recovery action according to the abnormal root cause category. The front-end recovery action includes at least one of the following: reducing gain, temporarily blocking abnormal channels, removing abnormal channels from the BIAS / RLD set, bias release, reference point reconstruction, adjusting lead-off detection timing, and channel-level or AFE-level local reset. (5) After the recovery action is executed, the preset stability conditions are checked within multiple consecutive sliding time windows. When the preset stability conditions are met, normal acquisition output is gradually restored. If the recovery fails within a predetermined number of times, the recovery action is upgraded or the fault is recorded.

2. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, The abnormal state identification in step (3) is specifically as follows: Abnormal EEG input can be identified as at least one of the following: lead detachment or poor contact, input polarization, input stage saturation, ESD protection device leakage, high resistance leakage caused by board contamination, moisture residue or flux residue, bias or common mode circuit abnormality, narrowband crosstalk, and motion artifact.

3. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, Step (3) further includes the following steps: The extracted features are fused and judged using rule thresholds, weighted scoring, state machines, decision trees, or combinations thereof. The system first enters the anomaly candidate state, and then judges whether the anomaly persists or whether the confidence of multi-feature fusion exceeds the threshold within multiple consecutive time windows before entering the anomaly confirmation state.

4. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, Step (4) further includes the following steps: When an abnormal channel is identified as having poor contact, input polarization, input stage saturation, leakage of ESD protection devices, or high-resistance leakage caused by board contamination, and it is determined that the abnormal channel affects the stability of the bias or common-mode circuit, the controller will temporarily remove the abnormal channel from the BIAS / RLD participation set. After recovery observation and stability confirmation meet the preset conditions, the abnormal channel will be restored to participate in the BIAS / RLD bias or common-mode drive.

5. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, The recovery action reselects an effective channel or prompts for recalibration, and also controls the channel output, gain, bias participation set, lead-off detection parameters, reference point, or AFE working state of the EEG acquisition front end; The selection of the corresponding front-end recovery action based on the root cause category of the anomaly is specifically as follows: If the lead is detached or has poor contact, perform a re-contact confirmation, wearing reminder, and / or temporarily block abnormal channel data output; For input polarization state or DC bias accumulation state, perform at least one of the following: bias release wait, reduce the gain of abnormal channel or corresponding AFE, and reference point reconstruction; If the input stage is saturated, the gain of the corresponding channel or the corresponding AFE is reduced and / or a local reset is performed. For narrowband crosstalk caused by radio frequency, digital clock, audio service or power supply coupling, the wireless transmission, audio service and digital clock status are combined for cross-judgment, and the acquisition timing is adjusted, the channel participation relationship is modified or the anomaly judgment threshold is increased. For motion artifacts, the data for the corresponding time period is first marked and kept under observation. Only when the anomaly persists and the contact characteristics deteriorate will the recovery action be upgraded. In cases where the ESD protection device is determined to be in an abnormal leakage state, channel isolation and local reset of the EEG simulation front-end AFE containing the abnormal channel are performed first. For cases where the high resistance leakage is determined to be caused by board surface contamination, moisture residue, or flux residue, the priority should be to implement cleaning instructions, wait for drying, and re-confirm contact.

6. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, The recovery process also includes: Adjust the timing, duty cycle, or detection current of the lead detachment detection to reduce the disturbance to the data acquisition during the detection process; For persistent and unrecoverable anomalies, perform a local reset of the corresponding AFE, rebuild the bias circuit, or generate information prompting the user to re-wear the device or require manual intervention.

7. The method for recognizing and controlling abnormal EEG input states according to claim 2, characterized in that, The following methods can be used to distinguish the abnormal leakage state of the ESD protection device from the high-resistance leakage state caused by board surface contamination, moisture residue, or flux residue: If a fixed offset or lead-off code abnormality occurs when the channel is not in place, and the abnormality is concentrated in a single channel or a single input terminal, and cannot be significantly restored after cleaning or drying, or is improved after replacing / isolating the corresponding input protection device, it should be preliminarily judged as an abnormal leakage state of the ESD protection device. If the abnormality worsens after being exposed to moisture, multiple adjacent channels deteriorate simultaneously, or improves after drying or cleaning, it should be primarily identified as a high-resistance leakage condition caused by board surface contamination, moisture residue, or flux residue.

8. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, The preset stability conditions in step (5) include at least one of the following: the channel baseline returns to the preset range, the saturation ratio is lower than the threshold, the abnormal spectral line energy decays, the lead detachment and impedance characteristics recover to stability, the bias or common-mode circuit node recovers to the normal fluctuation range, and the abnormal EEG input no longer recurs continuously. When these conditions are met continuously within multiple sliding time windows, the original gain, the original bias participation set, and the normal output strategy are gradually restored.

9. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, The EEG acquisition system is a multi-channel headphone-type or in-ear EEG acquisition system; the system status information continuously acquired in step (1) also includes the user wearing status, the inertial measurement unit (IMU) action status, the wireless transmission status, or the audio working status as auxiliary criteria.

10. The method for recognizing and controlling abnormal EEG input states according to claim 1, characterized in that, For a multi-channel EEG system consisting of multiple EEG simulation front-ends (AFEs), when the abnormality is concentrated in only one channel or one EEG simulation front-end (AFE), step (4) performs local recovery according to the abnormality recovery granularity at the channel level or AFE level, ensuring that the EEG simulation front-end (AFE) and its corresponding channel that are not affected by the abnormality maintain normal acquisition.

11. An electroencephalogram (EEG) acquisition device, characterized in that, The aforementioned EEG acquisition device includes electrode input terminals, an EEG simulation front-end (AFE), a processor, and a memory. The EEG simulation front-end AFE includes a configurable gain, a lead-off detection module, and a bias / common-mode driver interface. The processor is used to configure AFE registers, adjust lead-off detection timing, control the output state of abnormal channels, adjust the BIAS / RLD participation set, and perform channel-level or AFE-level local reset. The processor is configured to execute computer-executable instructions; The memory stores one or more computer-executable instructions, which, when executed by the processor, implement the steps of the method for recognizing and controlling abnormal EEG input states as described in any one of claims 1 to 10.

12. A processor for recognizing and controlling abnormal EEG input states, characterized in that, The processor is configured to execute computer-executable instructions, which, when executed by the processor, implement the steps of the method for recognizing and controlling abnormal EEG input states as described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, It stores a computer program that can be executed by a processor to implement the steps of the method for recognizing and controlling abnormal EEG input states as described in any one of claims 1 to 10.

Images