An electroencephalogram signal edge collection, analysis and processing method and system

Abstract

The present application relates to the technical field of edge computing, in particular to an electroencephalogram edge collection, analysis and processing method and system, comprising the following steps: acquiring a multi-channel electroencephalogram sampling sequence, a sampling time record and a contact impedance sequence and cutting electroencephalogram segments, calculating a potential change amplitude cumulative value, a baseline drift proportion and a contact impedance abnormality proportion, generating an activity amplitude lower limit, a drift proportion upper limit and an impedance proportion upper limit according to the sorting position within the same processing period, screening analyzable electroencephalogram segments and forming a channel analyzable segment density, and determining writing, residence and sending content in combination with a cache, a calculation time slot and a sending queue. In the present application, through the linkage of segment quality indicators and edge resource states, noise, drift and contact abnormality segments occupy storage, computing power and transmission links are reduced, analyzable data is preferentially retained and processed, and the continuity of electroencephalogram edge analysis, the resource matching and the result feedback effectiveness are improved.

Description

Technical Field

[0001] This invention relates to the field of edge computing technology, and in particular to a method and system for edge acquisition, analysis and processing of electroencephalogram (EEG) signals. Background Technology

[0002] Edge computing technology involves deploying computing resources close to the data source, such as terminals, gateways, or edge servers, to locally cache, preprocess, assign tasks, and transmit results back to sensor data. Traditional edge acquisition and analysis methods for EEG signals involve acquiring multi-channel electrical signals through an EEG acquisition terminal, with the edge processor running sampling control, cache management, data preprocessing, and computational task scheduling locally. The processed EEG data or analysis results are then sent to a server or cloud platform via a communication interface.

[0003] Traditional edge acquisition and analysis of EEG signals relies on fixed sampling control, local caching, preprocessing, and task scheduling processes. When faced with quality fluctuations in multi-channel EEG segments, it is difficult to simultaneously measure the strength of potential activity, the degree of baseline shift, and abnormal contact impedance. This can easily lead to the inclusion of noisy segments, drifting segments, or unstable contact segments in the analysis process, resulting in an imbalance in edge cache usage, insufficient basis for computation task residency, increased communication queue waiting, and difficulty in matching processing priorities. This affects the continuity of local EEG data analysis and the effectiveness of result feedback. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for edge acquisition, analysis and processing of electroencephalogram (EEG) signals. After acquiring multi-channel EEG signals, the edge side can determine analyzable segments by combining the potential activity state, baseline offset state and contact impedance state of the EEG segments. Based on the density of analyzable segments in the channels and the edge resource state, the original segments, segment statistics, local analysis tasks and communication transmission content are determined, thereby providing a consistent data judgment basis for the edge acquisition, analysis and processing and resource scheduling of EEG signals.

[0005] To achieve the above objectives, the present invention provides a method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals, comprising the following steps: A multi-channel EEG sampling sequence, sampling time record, and contact impedance sequence are acquired, and the EEG segments are segmented according to segment time length. Based on the EEG segments, the absolute value of the potential difference between adjacent sampling points is accumulated to generate a cumulative value of potential change amplitude. A baseline value is generated based on the average potential of the starting sampling point of the segment. The percentage of sampling points that deviate from the baseline value by more than a preset baseline deviation threshold is calculated to generate a baseline drift percentage. The percentage of sampling time records whose contact impedance exceeds a preset channel impedance threshold is calculated to generate a contact impedance abnormality percentage. Based on the sorting position of the cumulative value of potential change amplitude, the baseline drift percentage, and the contact impedance abnormality percentage of the EEG segments within the same processing cycle, a lower limit of activity amplitude, an upper limit of drift percentage, and an upper limit of impedance percentage are generated, respectively. The EEG segments that meet the lower limit of activity amplitude, the upper limit of drift percentage, and the upper limit of impedance percentage are determined as analyzable EEG segments, and a channel analyzable segment density is generated by channel. Based on the channel analyzable segment density, the remaining capacity of the buffer queue, the number of idle computing time slots, and the waiting length of the sending queue, the original segment writing content, the segment statistical writing content, the local analysis task resident content, and the communication sending content are determined.

[0006] Further, when acquiring the multi-channel EEG sampling sequence, the sampling time record, and the contact impedance sequence and segmenting the EEG segments, the channel identifier, potential sampling value, sampling time record, and contact impedance value in the multi-channel EEG acquisition buffer are acquired. The potential sampling value and the contact impedance value are time-correlated according to the channel identifier and the sampling time record to generate a channel-aligned sampling record. The segment start time and segment end time are determined according to the segment time length. The channel-aligned sampling records located between the segment start time and the segment end time are written into the same EEG segment to generate a segment channel identifier and a segment time identifier.

[0007] As a progressive constraint, when determining the start and end times of a segment, the sampling frequency and timestamp counting unit are obtained. A standard sampling interval is determined based on the sampling frequency, and the number of segment sampling points is determined based on the segment duration and the standard sampling interval. Continuity is judged for adjacent sampling time records according to the standard sampling interval, and the number of sampling points is accumulated during continuous sampling. When the accumulated number of sampling points reaches the number of segment sampling points, the end time of the current EEG segment is determined. When the time interval between adjacent sampling time records does not exceed twice the standard sampling interval, the end time of the segment is determined along the continuous sampling sequence. When the time interval between adjacent sampling time records exceeds twice the standard sampling interval, the next sampling time record is determined as the new segment start time.

[0008] Further, when generating the cumulative value of potential change amplitude, the baseline value, the baseline drift ratio, and the contact impedance abnormality ratio, the sampling point sorting result of the same channel within the EEG segment is obtained, the absolute value of the potential difference between adjacent sampling points is calculated according to the sampling time order of the sampling points, and the absolute value of the potential difference is accumulated to generate the cumulative value of potential change amplitude; the set of the segment start sampling points of the EEG segment is obtained, the potential sampling values ​​in the set of the segment start sampling points are averaged to generate the baseline value, and the baseline drift ratio is generated based on the deviation of the sampling points from the baseline value; the contact impedance sequence corresponding to the EEG segment is obtained, the median of the channel impedance in the same channel within the same processing cycle is calculated, and the contact impedance abnormality ratio is generated based on the proportion of sampling time records exceeding the preset channel impedance threshold.

[0009] Preferably, when generating the baseline value, a starting time window is determined based on the sampling time record corresponding to the EEG segment, and sampling points located within the starting time window with continuous sampling time records are selected as starting candidate sampling points; when the number of starting candidate sampling points reaches the preset number of starting points, the maximum potential sampling value and the minimum potential sampling value among the starting candidate sampling points are removed, and the remaining potential sampling values ​​are averaged to generate the baseline value; when the number of starting candidate sampling points does not reach the preset number of starting points, the sampling points within the current EEG segment that can correspond to the channel identifier, sampling time record, and potential sampling value, and can enter the EEG segment are determined as valid sampling points, and the baseline value is generated based on the potential sampling value of the earliest sampling time of the valid sampling point.

[0010] Furthermore, when generating the lower limit of activity amplitude, the upper limit of drift ratio, and the upper limit of impedance ratio, the cumulative value of potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio corresponding to each EEG segment within the same processing cycle are obtained, and a segment feature sorting table is established according to the channel identifier and segment time identifier; the cumulative value of potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio are sorted respectively, and the corresponding value is read according to the preset activity sorting percentile position to generate the lower limit of activity amplitude, the corresponding value is read according to the preset drift sorting percentile position to generate the upper limit of drift ratio, and the corresponding value is read according to the preset impedance sorting percentile position to generate the upper limit of impedance ratio.

[0011] Further, when determining the analyzable EEG segments and generating the channel analyzable segment density, the cumulative value of the potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio corresponding to each EEG segment within the same processing cycle are obtained. The cumulative value of the potential change amplitude is compared with the lower limit of the activity amplitude, the baseline drift ratio is compared with the upper limit of the drift ratio, and the contact impedance abnormality ratio is compared with the upper limit of the impedance ratio to generate a segment determination identifier. When the segment determination identifier indicates that the cumulative value of the potential change amplitude is not less than the lower limit of the activity amplitude, the baseline drift ratio is not greater than the upper limit of the drift ratio, and the contact impedance abnormality ratio is not greater than the upper limit of the impedance ratio, the corresponding EEG segment is determined as the analyzable EEG segment, and the effective ratio of the analyzable EEG segments is statistically analyzed by channel to generate the channel analyzable segment density.

[0012] As an additional optimization feature, when generating the channel analyzable fragment density, the segment duration and effective acquisition duration of the analyzable EEG segment corresponding to the same channel within the same processing cycle are obtained. The segment duration corresponding to the analyzable EEG segment is accumulated to generate the channel analyzable cumulative duration. The channel analyzable cumulative duration is calculated as a ratio to the channel effective acquisition duration. When the calculation result exceeds a preset density upper limit (where the preset density upper limit is less than 1), the calculation result is replaced by the preset density upper limit to generate the channel analyzable fragment density.

[0013] Further, when determining the original fragment writing content, the fragment statistical writing content, the local analysis task resident content, and the communication transmission content, the channel analyzable fragment density, the remaining capacity of the buffer queue, the number of idle computing time slots, and the sending queue waiting length corresponding to each channel within the same processing cycle are obtained. A channel resource status record is generated based on the channel analyzable fragment density, the remaining capacity of the buffer queue, the number of idle computing time slots, and the sending queue waiting length. The channel resource status record is matched with the resource scheduling policy table to determine the fragment retention status, statistical retention status, task resident status, and sending queuing status. The original fragment writing content, the fragment statistical writing content, the local analysis task resident content, and the communication transmission content are determined based on the fragment retention status, the statistical retention status, the task resident status, and the sending queuing status, respectively.

[0014] This invention also provides a system for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals. The system implements the above-mentioned method and includes an EEG segmentation module, a segment feature calculation module, a segment threshold generation module, a segment density determination module, and an edge resource scheduling module. The EEG segmentation module acquires multi-channel EEG sampling sequences, sampling time records, and contact impedance sequences, and segments the EEG into segments according to segment time length. The segment feature calculation module, based on the EEG segments, accumulates the absolute value of the potential difference between adjacent sampling points to generate a cumulative value of potential change amplitude, generates a baseline value based on the average potential of the segment's initial sampling point, calculates the percentage of sampling points deviating from the baseline value exceeding a preset baseline deviation threshold to generate a baseline drift percentage, and calculates the percentage of sampling time records whose contact impedance exceeds a preset channel impedance threshold to generate a contact impedance abnormality percentage. The segment threshold generation module is used to determine the contact impedance density within the same processing cycle... The sorting positions of the cumulative amplitude of potential changes, the baseline drift ratio, and the contact impedance abnormality ratio of the EEG segments are used to generate a lower limit for activity amplitude, an upper limit for drift ratio, and an upper limit for impedance ratio, respectively. The segment density determination module is used to determine the EEG segments that meet the lower limit for activity amplitude, the upper limit for drift ratio, and the upper limit for impedance ratio as analyzable EEG segments, and generate channel-specific analyzable segment densities. The edge resource scheduling module is used to determine the original segment writing content, segment statistical writing content, local analysis task resident content, and communication transmission content based on the channel-specific analyzable segment density, the remaining capacity of the buffer queue, the number of idle computing time slots, and the sending queue waiting length.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, a collaborative discrimination process is established around the quality of EEG segments and the status of marginal resources. Segment segmentation forms a unified processing object. The cumulative value of potential change amplitude, the proportion of baseline drift, and the proportion of abnormal contact impedance are used to jointly characterize the activity, stability, and contact reliability of segments. Dynamic thresholds are generated by the sorting position within the same processing cycle to ensure that the selection of analyzable segments is consistent with the current sampling state. Furthermore, the density of analyzable segments is formed by channel. The distribution of EEG data quality is transformed into a selection criterion for cache writing, statistical retention, local computing residency, and communication transmission. This reduces the crowding out of storage, computing power, and transmission links by low-value segments and improves the matching and processing reliability of marginal EEG analysis resources. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the main process of the EEG signal edge acquisition, analysis and processing method of the present invention; Figure 2 This is a diagram illustrating the data flow of EEG segmentation in this invention. Figure 3 This is a logic diagram for generating the segment feature calculation and sorting threshold of the present invention; Figure 4 This invention provides an analytical method for determining and generating density state diagrams of electroencephalogram (EEG) segments. Figure 5 This is the edge resource scheduling data matching diagram of the present invention; Figure 6 This is an interactive diagram of the EEG signal edge acquisition, analysis and processing system module of the present invention. Detailed Implementation

[0018] The following is combined with Figure 1 The newly added accompanying drawings further illustrate the technical solutions described in this invention. The following embodiments are based on the scope of protection text, the purpose of the invention, the technical solutions, and the contents of the application documents, and are used to explain the acquisition, processing, transmission, and retrieval relationships between multi-channel EEG sampling sequences, sampling time records, contact impedance sequences, EEG segments, channel-aligned sampling records, cumulative potential change amplitude values, baseline values, baseline drift percentage, contact impedance abnormality percentage, lower limit of activity amplitude, upper limit of drift percentage, upper limit of impedance percentage, analyzable EEG segments, channel analyzable segment density, channel resource status records, original segment writing content, segment statistical writing content, local analysis task resident content, and communication transmission content. The names of each data object remain consistent in subsequent embodiments, process descriptions, and accompanying reference text. Specifically, a multi-channel EEG sampling sequence refers to a set of potential sampling values ​​arranged according to channel identifiers and sampling time records; a contact impedance sequence refers to a set of contact impedance values ​​corresponding to the corresponding channel identifier and sampling time record; and an EEG segment refers to a fragmented sampling object formed based on segment time length and sampling continuity. Example 1

[0019] Please see Figures 1 to 5This embodiment provides a method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals. The method uses multi-channel EEG sampling sequences, sampling time records, and contact impedance sequences obtained at the edge as input. First, it forms EEG segments with segment channel identifiers and segment time identifiers. Then, within the same processing cycle, it generates segment features, sorting thresholds, and channel-analyzable segment density. The channel-analyzable segment density, along with the remaining capacity of the buffer queue, the number of idle computing time slots, and the sending queue waiting length, are used together for edge resource scheduling.

[0020] S1: Acquire multi-channel EEG sampling sequences, sampling time records, and contact impedance sequences, and segment the EEG into segments according to the segment time length.

[0021] S111, the edge side retrieves channel identifiers, potential sampling values, sampling time records, and contact impedance values ​​from the multi-channel EEG acquisition buffer. Channel identifiers distinguish different acquisition channels, potential sampling values ​​represent the EEG potentials recorded for the corresponding channel at the corresponding sampling time, and contact impedance values ​​represent the electrode contact state for the corresponding channel at the corresponding sampling time. After obtaining the above data, the potential sampling values ​​and contact impedance values ​​are temporally correlated according to the channel identifiers, ensuring that potential sampling values ​​and contact impedance values ​​belonging to the same channel at the same sampling time are written into the same channel-aligned sampling record. The channel-aligned sampling record must have at least a one-to-one correspondence between the channel identifier, sampling time record, potential sampling value, and contact impedance value, providing a unified input for subsequent segmentation and feature calculation.

[0022] S112, the segment start time and segment end time are determined based on the segment duration. The sampling frequency and timestamp counting unit are obtained at the edge. The standard sampling interval is determined based on the sampling frequency, and the number of sampling points for the segment is determined based on the segment duration. For adjacent sampling time records in the same channel, if the time interval between adjacent sampling time records does not exceed twice the standard sampling interval, the segment end time of the EEG segment is determined by continuing the continuous sampling sequence. If the time interval between adjacent sampling time records exceeds twice the standard sampling interval, the later sampling time record is not merged into the previous continuous segment; instead, the later sampling time record is determined as the new segment start time. Thus, channel-aligned sampling records entering the same EEG segment have a continuity boundary formed based on the sampling time records, and the sampling time interruption position becomes the trigger condition for restarting the segment.

[0023] S113: Channel-aligned sampling records located between the segment start time and segment end time are written into the same EEG segment, and segment channel identifier and segment time identifier are generated. The segment channel identifier is used to indicate the acquisition channel corresponding to the EEG segment, and the segment time identifier is used to indicate the time position of the EEG segment in the processing cycle. If a channel-aligned sampling record lacks a sampling time record for forming the segment boundary, or cannot be time-correlated with the potential sampling value and contact impedance value of the same channel, then the record is not used to form a valid continuous sampling interval for the same EEG segment. The EEG segment, segment channel identifier, and segment time identifier output by S1 are received by S2 and used to calculate segment features.

[0024] S2 is based on the cumulative value of the amplitude of the generated potential change in EEG segments, the baseline value, the percentage of baseline drift, and the percentage of abnormal contact impedance.

[0025] S211: For sampling points within the same channel of an EEG segment, a sampling point sorting result is generated based on the sampling time record. Along the edge of the sampling point sorting result, the potential sampling values ​​of two adjacent sampling points are read sequentially. The absolute value of the potential difference between adjacent sampling points is calculated, and the absolute value of the potential difference is accumulated within the EEG segment to generate a cumulative value of potential change amplitude. The cumulative value of potential change amplitude, together with the segment channel identifier and segment time identifier, is written into the segment feature record to represent the amplitude state of potential activity within the EEG segment.

[0026] S212, S212, Generate baseline values ​​for the same EEG segment. The edge side determines a starting time window based on the sampling time record corresponding to the EEG segment, and selects consecutive sampling points within the starting time window as candidate starting sampling points. When the number of candidate starting sampling points reaches a preset number of starting points, the maximum and minimum potential sampling values ​​among the candidate starting sampling points are removed, and the remaining potential sampling values ​​are averaged to generate the baseline value. When the number of candidate starting sampling points does not reach the preset number of starting points, sampling points with associated channel identifiers, sampling time records, and potential sampling values, and whose sampling time records are located between the segment start time and segment end time of the EEG segment, are determined as valid sampling points, and the baseline value is generated based on the potential sampling value of the earliest valid sampling point. After the baseline value is generated, a correspondence is established between it and the segment channel identifier and segment time identifier of the EEG segment.

[0027] S213, Based on the deviation of sampling points from baseline values, a baseline drift percentage is generated. Sampling points in the EEG segment are read at the edge, and the potential sample value of each sampling point is compared with the baseline value to determine the deviation of the sampling point from the baseline value. Then, the number of sampling points determined to deviate from the baseline value is counted, and the baseline drift percentage is generated using the sampling points involved in the deviation statistics within the EEG segment as the percentage benchmark. The baseline drift percentage and the cumulative potential change amplitude are different feature items of the same EEG segment and are subsequently entered together into the segment feature sorting table.

[0028] S214, Generate the contact impedance abnormality percentage based on the contact impedance sequence. The contact impedance sequence corresponding to the EEG segment is read at the edge, and the median channel impedance of the same channel is calculated within the same processing cycle. For sampling time records within the EEG segment, if the corresponding contact impedance value exceeds the median channel impedance, the sampling time record is included in the contact impedance abnormality statistics. Then, using the sampling time records participating in the impedance statistics within the EEG segment as the percentage benchmark, the contact impedance abnormality percentage is generated. The contact impedance abnormality percentage reflects the abnormal state of the EEG segment at the contact impedance level and, together with the cumulative value of potential change amplitude and the baseline drift percentage, serves as the input to S3.

[0029] S3 generates the lower limit of activity amplitude, the upper limit of drift ratio, and the upper limit of impedance ratio based on the sorting position of the cumulative value of potential change amplitude, the percentage of baseline drift, and the percentage of abnormal contact impedance of the EEG segments within the same processing cycle.

[0030] S311, the edge side acquires the cumulative value of potential change amplitude, baseline drift percentage, and contact impedance abnormality percentage for each EEG segment within the same processing cycle, and establishes a segment feature sorting table according to channel identifier and segment time identifier. Each record in the segment feature sorting table corresponds to an EEG segment, and the record content includes the segment channel identifier, segment time identifier, cumulative value of potential change amplitude, baseline drift percentage, and contact impedance abnormality percentage of the EEG segment. EEG segments that fail to form a complete segment feature record are not considered valid objects for sorting position reading within this processing cycle.

[0031] In S312, the edge side sorts the cumulative value of potential change amplitude, the proportion of baseline drift, and the proportion of abnormal contact impedance, respectively. For the cumulative value of potential change amplitude, the corresponding value is read according to the preset activity sorting percentile position to generate the lower limit of activity amplitude; for the proportion of baseline drift, the corresponding value is read according to the preset drift sorting percentile position to generate the upper limit of drift proportion; for the proportion of abnormal contact impedance, the corresponding value is read according to the preset impedance sorting percentile position to generate the upper limit of impedance proportion. The lower limit of activity amplitude, the upper limit of drift proportion, and the upper limit of impedance proportion are all derived from the segment feature sorting results formed within the same processing cycle, without the need to introduce fixed external evaluation objects outside the processing cycle. The three thresholds output by S3 are received by S4 and used to make each EEG segment analyzable.

[0032] S4 identifies EEG segments that meet the lower limit of activity amplitude, the upper limit of drift ratio, and the upper limit of impedance ratio as analyzable EEG segments, and generates channel-based analyzable segment density by channel.

[0033] S411, the edge side reads the cumulative value of potential change amplitude, baseline drift percentage, and contact impedance abnormality percentage for each EEG segment within the same processing cycle, and performs comparison processing accordingly. The cumulative value of potential change amplitude is compared with the lower limit of activity amplitude, the baseline drift percentage is compared with the upper limit of drift percentage, and the contact impedance abnormality percentage is compared with the upper limit of impedance percentage, generating a segment determination label. The segment determination label records the comparison results of the EEG segment in terms of potential activity, baseline drift, and contact impedance, serving as a direct basis for whether to include it in the set of analyzable EEG segments.

[0034] S412, when the cumulative value of the potential change amplitude indicated by the segment determination identifier is not less than the lower limit of the activity amplitude, the baseline drift ratio is not greater than the upper limit of the drift ratio, and the contact impedance abnormality ratio is not greater than the upper limit of the impedance ratio, the corresponding EEG segment is determined as an analyzable EEG segment. When all three conditions are met, the EEG segment is added to the analyzable EEG segment set; if any condition is not met, the EEG segment is not included in the analyzable EEG segment set, but its segment feature record can still be used as a statistical source within the same processing cycle. The analyzable EEG segment set is grouped according to channel identifiers, enabling subsequent density results to be generated by channel.

[0035] S413, the edge side acquires the segment duration and effective acquisition duration of the analyzable EEG segments corresponding to the same channel within the same processing cycle. The segment duration corresponding to the analyzable EEG segments in that channel is accumulated to generate the channel's cumulative analyzable duration. Subsequently, the ratio of the channel's cumulative analyzable duration to the channel's effective acquisition duration is calculated to obtain the effective proportion of analyzable segments in the same processing cycle for that channel. When the calculation result exceeds a preset density upper limit, the calculation result is replaced with the preset density upper limit to generate the channel's analyzable segment density. A correspondence is established between the channel's analyzable segment density and the channel identifier, which is read by S5. Through this process, the analyzable segment status of the same channel within a processing cycle is converted into density data objects that can be used for resource scheduling.

[0036] S5 determines the original fragment writing content, fragment statistics writing content, local analysis task resident content, and communication sending content based on the channel analyzable fragment density, remaining capacity of the buffer queue, number of idle computing time slots, and sending queue waiting length.

[0037] In step S511, the edge side acquires the analyzable fragment density for each channel within the same processing cycle, and simultaneously acquires the remaining capacity of the buffer queue, the number of idle computation slots, and the sending queue waiting length. The analyzable fragment density is derived from step S4. The remaining buffer queue capacity indicates the remaining state of the edge side's buffer queue, which can continue to receive data objects. The number of idle computation slots indicates the state of computing resources that the local analysis task can reside in or execute. The sending queue waiting length indicates the waiting state of the communication sending object in the sending queue. The edge side writes the above data into the channel resource status record by channel, aligning the analyzable fragment status of each channel with the edge resource status within the same record.

[0038] S512 matches the channel resource status record with the resource scheduling policy table. The resource scheduling policy table receives the channel analyzable segment density, buffer queue remaining capacity, number of idle computing time slots, and transmission queue waiting length from the channel resource status record, and outputs segment retention status, statistical retention status, task residency status, and transmission queuing status. The segment retention status indicates whether the original EEG segment of the corresponding channel is written into the buffer queue as original segment content; the statistical retention status indicates whether the segment features, segment judgment identifiers, or density-related statistical objects of the corresponding channel are written into the statistical storage object as segment statistics content; the task residency status indicates whether the analyzable EEG segment of the corresponding channel or its corresponding task object is resided in the local computing time slot as local analysis task residency content; and the transmission queuing status indicates whether the data object to be transmitted of the corresponding channel is transmitted into the transmission queue as communication transmission content.

[0039] S513, based on the segment retention status, statistical retention status, task dwell status, and transmission queuing status, the original segment writing content, segment statistical writing content, local analysis task dwell content, and communication transmission content are determined respectively. The original segment writing content originates from the EEG segment formed in S1, along with its segment channel identifier and segment time identifier; the segment statistical writing content originates from the cumulative value of potential change amplitude, baseline value, baseline drift percentage, contact impedance abnormality percentage, lower limit of activity amplitude, upper limit of drift percentage, upper limit of impedance percentage, segment determination identifier, and channel analyzable segment density formed in S2 to S4; the local analysis task dwell content originates from the analyzable EEG segments determined in S4 and the task objects that can be received by the local calculation time slot; the communication transmission content originates from the data object to be transmitted as indicated by the transmission queuing status. All four types of content are determined by the matching results of the channel resource status record and the resource scheduling strategy table, avoiding the use of a single segment quality characteristic or a single resource status as the basis for edge-side writing, dwelling, and transmission.

[0040] In the processing chain of this embodiment, the EEG segment output by S1 provides fragmentation input for S2; the cumulative value of potential change amplitude, baseline drift ratio, and contact impedance abnormality ratio output by S2 provide sorting objects for S3; the lower limit of activity amplitude, upper limit of drift ratio, and upper limit of impedance ratio output by S3 provide judgment boundaries for S4; the channel analyzable segment density output by S4 provides scheduling input for S5; the original segment writing content, segment statistical writing content, local analysis task resident content, and communication transmission content output by S5 respectively enter the corresponding buffer, statistics, local analysis, and communication transmission links. Each output of the above data stream has a subsequent receiving position, forming a closed processing process from acquisition, fragmentation, feature calculation, threshold generation, segment judgment, density generation to edge resource scheduling. Example 2

[0041] Please see Figure 6 This embodiment provides a system for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals. The difference between this embodiment and Embodiment 1 is that this embodiment uses the input, output, and collaborative relationships between system modules to carry out the method flow described in Embodiment 1, rather than directly unfolding the edge processing process according to the method steps; its input sources, fragment features, threshold objects, fragment determination results, density results, and scheduling outputs remain consistent with Embodiment 1.

[0042] The system includes an EEG segmentation module, a segment feature calculation module, a segment threshold generation module, a segment density determination module, and a marginal resource scheduling module. The EEG segmentation module receives multi-channel EEG sampling sequences, sampling time records, and contact impedance sequences. It segments the EEG into segments according to segment duration and outputs the EEG segment, segment channel identifier, and segment time identifier. When forming EEG segments, the module performs time correlation between the potential sampling values ​​and contact impedance values ​​according to the channel identifier, generating channel-aligned sampling records. Simultaneously, it determines the segment boundaries based on the sampling frequency, timestamp counting unit, standard sampling interval, and number of segment sampling points. When the time interval between adjacent sampling time records exceeds twice the standard sampling interval, the EEG segmentation module determines the subsequent sampling time record as the new segment start time.

[0043] The segment feature calculation module receives the EEG segment, segment channel identifier, and segment time identifier output by the EEG segmentation module. It processes the sorting results of sampling points within the same channel within the EEG segment to generate a cumulative value of potential change amplitude. It generates a baseline value based on the initial candidate sampling point set and generates a baseline drift percentage based on the deviation of sampling points from the baseline value. It receives the contact impedance sequence corresponding to the EEG segment, calculates the median channel impedance for the same channel within the same processing cycle, and generates a contact impedance abnormality percentage. The segment feature calculation module writes the cumulative value of potential change amplitude, baseline value, baseline drift percentage, and contact impedance abnormality percentage into the segment feature record and outputs the segment feature record to the segment threshold generation module.

[0044] The segment threshold generation module receives segment feature records output by the segment feature calculation module, establishes a segment feature sorting table according to channel identifier and segment time identifier, and sorts the cumulative value of potential change amplitude, baseline drift ratio, and contact impedance anomaly ratio within the same processing cycle. The segment threshold generation module reads the corresponding value at the preset activity sorting percentile position to generate a lower limit for activity amplitude, reads the corresponding value at the preset drift sorting percentile position to generate an upper limit for drift ratio, and reads the corresponding value at the preset impedance sorting percentile position to generate an upper limit for impedance ratio. These three thresholds are then output to the segment density determination module.

[0045] The segment density determination module receives the lower limit of activity amplitude, the upper limit of drift ratio, and the upper limit of impedance ratio output by the segment threshold generation module. Simultaneously, it reads the cumulative value of potential change amplitude, baseline drift ratio, and contact impedance abnormality ratio from the segment feature record, generating a segment determination identifier. When the segment determination identifier indicates that the cumulative value of potential change amplitude is not less than the lower limit of activity amplitude, the baseline drift ratio is not greater than the upper limit of drift ratio, and the contact impedance abnormality ratio is not greater than the upper limit of impedance ratio, the segment density determination module determines the corresponding EEG segment as an analyzable EEG segment. The segment density determination module also calculates the segment time length corresponding to the analyzable EEG segments by channel and generates the channel analyzable segment density by combining it with the effective acquisition duration of the channel; when the calculation result exceeds the preset density upper limit, the calculation result is replaced by the preset density upper limit.

[0046] The edge resource scheduling module receives the channel analyzable segment density output by the segment density determination module, and also receives the remaining capacity of the buffer queue, the number of idle computing slots, and the sending queue waiting length. The edge resource scheduling module writes the channel analyzable segment density, remaining buffer queue capacity, number of idle computing slots, and sending queue waiting length into the channel resource status record. It then matches the channel resource status record with the resource scheduling policy table to obtain the segment retention status, statistical retention status, task residency status, and sending queuing status. Based on these statuses, the edge resource scheduling module determines the original segment writing content, segment statistical writing content, local analysis task residency content, and communication sending content, respectively. Thus, the input and output boundaries of each module in the system correspond to S1 to S5 of the method described in Embodiment 1, enabling the same acquisition, segmentation, feature calculation, threshold generation, segment determination, density generation, and resource scheduling links to be implemented in a module-cooperative manner.

[0047] The processes, field states, processing order, judgment conditions, parameter sources, module collaboration, and data transmission relationships described in the above embodiments are used to explain the possible implementations of the present invention and should not limit the present invention to the specific embodiments listed. Without departing from the technical solutions described in the present invention and the scope supported by the application documents, any equivalent substitutions, equivalent modifications, equivalent combinations, order adjustments, corresponding module replacements, equivalent changes in field names, equivalent acceptance of the executing entity, or equivalent changes in the carrier form that can be conceived by those skilled in the art should fall within the scope of protection of this patent; however, they should not be extended to unclaimed topics, nor should the substantive correspondence of the technical objects be changed through name changes.

Claims

1. A method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals, characterized in that, include: Acquire multi-channel EEG sampling sequences, sampling time records, and contact impedance sequences, and segment EEG segments according to segment time length; Based on the EEG segment, the absolute value of the potential difference between adjacent sampling points is accumulated to generate the cumulative value of potential change amplitude. The average potential of the starting sampling point of the segment is used to generate a baseline value. The percentage of sampling points that deviate from the baseline value by more than a preset baseline deviation threshold is calculated to generate the baseline drift percentage. The percentage of sampling time records that exceed the preset channel impedance threshold is calculated to generate the contact impedance abnormality percentage. Based on the sorting positions of the cumulative value of the potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio of the EEG segments within the same processing cycle, the lower limit of activity amplitude, the upper limit of drift ratio, and the upper limit of impedance ratio are generated respectively. The EEG segments that satisfy the lower limit of the activity amplitude, the upper limit of the drift ratio, and the upper limit of the impedance ratio are identified as analyzable EEG segments, and channel-based analyzable segment density is generated; Based on the channel, the fragment density, remaining capacity of the buffer queue, number of idle computing time slots, and waiting length of the sending queue can be analyzed to determine the original fragment writing content, fragment statistics writing content, local analysis task resident content, and communication sending content.

2. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 1, characterized in that, The steps of acquiring the multi-channel EEG sampling sequence, the sampling time record, and the contact impedance sequence, and segmenting the EEG into segments according to the segment time length, include: The channel identifier, potential sampling value, sampling time record, and contact impedance value in the multi-channel EEG acquisition buffer are obtained. The potential sampling value and contact impedance value are time-correlated according to the channel identifier and the sampling time record to generate a channel-aligned sampling record. The start and end times of a segment are determined based on the segment duration. The channel-aligned sampling records located between the start and end times of the segment are written into the same EEG segment, generating a segment channel identifier and a segment time identifier.

3. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 2, characterized in that, The process of determining the segment start time and segment end time includes: The sampling frequency and timestamp counting unit are obtained. A standard sampling interval is determined based on the sampling frequency. The number of sampling points for a segment is determined based on the segment duration and the standard sampling interval. The continuity of adjacent sampling time records is judged according to the standard sampling interval. The number of sampling points is accumulated during continuous sampling. When the accumulated number of sampling points reaches the number of sampling points for the segment, the segment termination time of the current EEG segment is determined. When the time interval between adjacent sampling time records does not exceed twice the standard sampling interval, the segment termination time is determined along the continuous sampling sequence; when the time interval between adjacent sampling time records exceeds twice the standard sampling interval, the next sampling time record is determined as the new segment start time.

4. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 1, characterized in that, The steps of generating the cumulative value of the potential change amplitude, the baseline value, the baseline drift percentage, and the contact impedance abnormality percentage based on the EEG segments include: Obtain the sorting result of the sampling points in the same channel within the EEG segment, calculate the absolute value of the potential difference between adjacent sampling points according to the sampling time order of the sampling points, and accumulate the absolute value of the potential difference to generate the cumulative value of the potential change amplitude. Obtain the set of starting sampling points of the EEG segment, average the potential sampling values ​​in the set of starting sampling points to generate the baseline value, and generate the baseline drift ratio based on the deviation of the sampling points from the baseline value. The contact impedance sequence corresponding to the EEG segment is obtained, the median channel impedance of the same channel in the same processing cycle is calculated, and the abnormal contact impedance percentage is generated based on the percentage of sampling time records that exceed the preset channel impedance threshold.

5. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 4, characterized in that, The process of generating the baseline values ​​includes: The starting time window is determined based on the sampling time record corresponding to the EEG segment, and sampling points that are located within the starting time window and have continuous sampling time records are selected as starting candidate sampling points. When the number of initial candidate sampling points reaches the preset number of initial points, the maximum and minimum potential sampling values ​​among the initial candidate sampling points are removed, and the remaining potential sampling values ​​are averaged to generate the baseline value; when the number of initial candidate sampling points does not reach the preset number of initial points, the sampling points in the current EEG segment that correspond to the channel identifier, sampling time record, and potential sampling value, and that can enter the EEG segment are determined as valid sampling points, and the baseline value is generated based on the potential sampling value of the earliest valid sampling point.

6. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 1, characterized in that, The steps for generating the lower limit of the activity amplitude, the upper limit of the drift ratio, and the upper limit of the impedance ratio respectively include: Obtain the cumulative value of the potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio for each EEG segment within the same processing cycle, and establish a segment feature sorting table according to channel identifier and segment time identifier; The cumulative value of the potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio are sorted respectively. The corresponding value is read according to the preset activity sorting percentile position to generate the lower limit of the activity amplitude, the corresponding value is read according to the preset drift sorting percentile position to generate the upper limit of the drift ratio, and the corresponding value is read according to the preset impedance sorting percentile position to generate the upper limit of the impedance ratio.

7. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 1, characterized in that, The step of determining the EEG segments that satisfy the lower limit of the activity amplitude, the upper limit of the drift ratio, and the upper limit of the impedance ratio as analyzable EEG segments, and generating the channel-specific analyzable segment density by channel, includes: The cumulative value of the potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio corresponding to each EEG segment within the same processing cycle are obtained. The cumulative value of the potential change amplitude is compared with the lower limit of the activity amplitude, the baseline drift ratio is compared with the upper limit of the drift ratio, and the contact impedance abnormality ratio is compared with the upper limit of the impedance ratio to generate a segment determination identifier. When the segment determination identifier indicates that the cumulative value of the potential change amplitude is not less than the lower limit of the activity amplitude, the baseline drift ratio is not greater than the upper limit of the drift ratio, and the contact impedance abnormality ratio is not greater than the upper limit of the impedance ratio, the corresponding EEG segment is determined as the analyzable EEG segment, and the effective ratio of the analyzable EEG segment is calculated by channel to generate the channel analyzable segment density.

8. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 7, characterized in that, The process of generating the channel-analyzable fragment density includes: The segment time length and effective acquisition duration of the channel corresponding to the analyzable EEG segment within the same processing cycle are obtained. The segment time length corresponding to the analyzable EEG segment is accumulated, and the channel's cumulative analyzable duration is generated. The analyzable cumulative duration of the channel is calculated as a percentage of the effective acquisition duration of the channel. When the calculation result exceeds the preset density limit (where the preset density limit is less than 1), the calculation result is replaced with the preset density limit to generate the analyzable segment density of the channel.

9. The method for edge acquisition, analysis, and processing of electroencephalogram (EEG) signals according to claim 1, characterized in that, The steps for determining the original fragment write content, the fragment statistics write content, the local analysis task resident content, and the communication transmission content include: The analyzable fragment density, remaining capacity of the buffer queue, number of idle computing slots, and waiting length of the sending queue are obtained for each channel within the same processing cycle. A channel resource status record is generated based on the analyzable fragment density, remaining capacity of the buffer queue, number of idle computing slots, and waiting length of the sending queue. The channel resource status record is matched with the resource scheduling policy table to determine the fragment retention status, statistical retention status, task dwell status, and transmission queuing status. Based on the fragment retention status, the statistical retention status, the task dwell status, and the transmission queuing status, the original fragment writing content, the fragment statistical writing content, the local analysis task dwell content, and the communication transmission content are determined respectively.

10. A system for acquiring, analyzing, and processing edge signals from electroencephalograms (EEGs), characterized in that, The system is used to implement the method according to any one of claims 1-9, the system comprising: The EEG segmentation module acquires multi-channel EEG sampling sequences, sampling time records, and contact impedance sequences, and segments EEG segments according to segment time length. The segment feature calculation module, based on the EEG segment, accumulates the absolute value of the potential difference between adjacent sampling points to generate a cumulative value of potential change amplitude, generates a baseline value based on the average potential of the starting sampling point of the segment, calculates the percentage of sampling points that deviate from the baseline value by more than a preset baseline deviation threshold to generate a baseline drift percentage, and calculates the percentage of sampling time records that have contact impedance exceeding a preset channel impedance threshold to generate a contact impedance abnormality percentage. The segment threshold generation module generates a lower limit for activity amplitude, an upper limit for drift ratio, and an upper limit for impedance ratio based on the sorting position of the cumulative value of the potential change amplitude, the baseline drift ratio, and the contact impedance abnormality ratio of the EEG segments within the same processing cycle. The segment density determination module identifies EEG segments that satisfy the lower limit of activity amplitude, the upper limit of drift ratio, and the upper limit of impedance ratio as analyzable EEG segments, and generates channel-specific analyzable segment densities by channel. The edge resource scheduling module determines the original fragment writing content, fragment statistics writing content, local analysis task resident content, and communication sending content based on the channel's analyzable fragment density, remaining cache queue capacity, number of idle computing time slots, and sending queue waiting length.

Images