An electrophysiological intelligent analysis system based on knowledge graph

Through an electrophysiological intelligent analysis system based on knowledge graph, the event human action model and electrophysiological knowledge graph are established, and the accuracy and general availability of electrophysiological data analysis in the existing technology are solved, and the electrophysiological signal is realized High accuracy and widely applicable analysis results.

CN118750238BActive Publication Date: 2024-11-22JILIN UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411206036.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-22
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

The existing electrophysiological analysis technology is inaccurate in character learning when dealing with noise and interference, and there is a problem that predictions tend to frequently appear in categories, resulting in insufficient accuracy and general applicability of electrophysiological data analysis results.

Method used

Using an electrophysiological intelligent analysis system based on knowledge graphs, the electrophysiological signal acquisition module, knowledge graph module and action command module are connected through the cloud computing platform to establish an event human action model and electrophysiological knowledge graph, and use the sound user's myoelectric signal and brain Electric signal preprocessing obtains normal target signal segments, generates robotic arm action instructions, and improves the accuracy and universality of data analysis.

Benefits of technology

It effectively improves the accuracy of electrophysiological data, ensures the generalization of electrophysiological data analysis results, and can more accurately identify and perform the behaviors of disabled users.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118750238B_ABST
    Figure CN118750238B_ABST
Patent Text Reader

Abstract

The present invention discloses an electrophysiological intelligent analysis system based on a knowledge graph, relates to the technical field of electrophysiological analysis, and effectively improves the versatility of electrophysiological data analysis results. The present invention preprocesses the electromyographic signals and electroencephalographic signals of several healthy users under different action events, obtains the normal target electromyographic signal segment and the normal target electroencephalographic signal segment under each action event according to the preprocessing result, and then establishes an event human action model and an electrophysiological knowledge graph according to the normal target electromyographic signal segment, the normal target electroencephalographic signal segment and each action event, inputs the electromyographic signals and electroencephalographic signals of the disabled user in real time into the electrophysiological knowledge graph, and then obtains the corresponding action name and action trajectory from the electrophysiological knowledge graph, and generates a robot arm action instruction, and the robot arm calls the corresponding mechanical joint according to the robot arm action instruction and performs the corresponding behavior action.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of electrophysiological analysis technology, and specifically to an electrophysiological intelligent analysis system based on a knowledge graph. Background Art

[0002] Electrophysiological analysis is the process of analyzing and interpreting bioelectric signals by combining electrophysiological principles with intelligent analysis technology. By recording brain electrical activity, brain-computer interface technology can convert brain signals into instructions for controlling external devices (such as prostheses, wheelchairs, etc.), providing people with disabilities with a better quality of life.

[0003] Existing electrophysiological analysis has the following defects:

[0004] Machine learning algorithms: including support vector machines, artificial neural networks, decision trees, random forests, etc. These algorithms can learn and simulate the patterns of electrophysiological signals for classification, recognition, and prediction. However, electrophysiological data may be affected by noise, interference, or distortion, resulting in inaccurate or misleading features learned by the machine.

[0005] Natural language processing technology: Convert electrophysiological data into natural language descriptions, allowing doctors or researchers to understand and interpret bioelectric signals more intuitively. Some categories of electrophysiological data may be more common than other categories, resulting in a bias towards frequently occurring categories when making predictions.

[0006] Therefore, how to improve the accuracy of electrophysiological data while ensuring the versatility of electrophysiological data analysis results is a difficulty in the existing technology. To this end, an electrophysiological intelligent analysis system based on a knowledge graph is provided. Summary of the invention

[0007] In order to solve the above technical problems, the purpose of the present invention is to provide an electrophysiological intelligent analysis system based on knowledge graph.

[0008] In order to achieve the above object, the present invention provides the following technical solutions:

[0009] An electrophysiological intelligent analysis system based on a knowledge graph comprises a cloud computing platform, wherein the cloud computing platform is communicatively connected with an electrophysiological signal acquisition module, a knowledge graph module and an action instruction module;

[0010] The electrophysiological signal acquisition module is used to collect the electromyographic signals and electroencephalographic signals of healthy users under different action events, and to collect the electromyographic signals and electroencephalographic signals of disabled users in real time;

[0011] The knowledge graph module is used to preprocess the electromyographic signals and electroencephalographic signals of several healthy users under different action events, obtain the normal target electromyographic signal segments and normal target electroencephalographic signal segments under each action event according to the preprocessing results, and then establish an event human action model and an electrophysiological knowledge graph according to the normal target electromyographic signal segments and normal target electroencephalographic signal segments and each action event.

[0012] The action instruction module is provided with an action instruction generating unit and a mechanical arm device;

[0013] The action instruction generation unit is used to input the electromyographic signal and the electroencephalographic signal of the disabled user in real time into the electrophysiological knowledge graph, and then obtain the corresponding action name and action trajectory from the electrophysiological knowledge graph, and generate the action instruction of the robot arm;

[0014] The robotic arm device is composed of a plurality of mechanical joints. The robotic arm device calls corresponding mechanical joints according to the robotic arm action instructions and performs corresponding behavior actions.

[0015] Furthermore, the process of collecting the electromyographic signals and electroencephalographic signals of healthy users under different action events includes:

[0016] A plurality of electrode stickers are respectively pasted on the heads and joints of healthy users and disabled users, and a magnetic field sensor is respectively connected to the electrode stickers at the same joint position and head position, and each magnetic field sensor is numbered;

[0017] Preset Num action events, the action event includes event name, event duration, associated joint number and corresponding joint trajectory, where Num is a natural number greater than 50;

[0018] The electrophysiological signal acquisition module communicates with the magnetic field sensors of each user, and then N healthy users perform corresponding action events according to the event duration, associated joint numbers and corresponding joint trajectories in each action event. During the execution of the action event, the electrodes at the joints and the head respectively collect corresponding electromyographic signals and electroencephalographic signals, where N is a natural number greater than 0;

[0019] The electrophysiological signal acquisition module sets a unique identity number for each healthy user and disabled user, generates corresponding myoelectric signal spectrum and EEG signal spectrum according to the myoelectric signal and EEG signal of each magnetic field sensor, and marks the corresponding magnetic field sensor number on each myoelectric signal spectrum and EEG signal spectrum;

[0020] All the EMG signal spectra and EEG signal spectra generated in the same action event are integrated to generate an action event signal dataset corresponding to the action event, and the identity number of the corresponding healthy user and the action event name are marked.

[0021] Furthermore, the process of collecting the real-time electromyographic signal and the real-time electroencephalographic signal of the disabled user in real time includes:

[0022] The magnetic field sensors at the head and joint positions of the disabled user send the real-time electromyographic signals and real-time electroencephalographic signals collected in real time to the electrophysiological signal acquisition module. The electrophysiological signal acquisition module then generates a corresponding disability signal data set based on the real-time electromyographic signals and real-time electroencephalographic signals, and marks the corresponding disabled user's identity number.

[0023] Furthermore, the myoelectric signal spectrum and the electroencephalographic signal spectrum include electrophysiological signals collected by multiple electrode stickers and a reference electrode sticker, wherein a reference electrode sticker is respectively provided at the head and the joint position.

[0024] Furthermore, the process of preprocessing the myoelectric signal spectrum and the electroencephalographic signal spectrum in the action event signal data set includes:

[0025] Establishing a two-dimensional spatial coordinate system, mapping the myoelectric signal spectrum and the electroencephalogram signal spectrum at the same position in the action event signal data set with different identity numbers but the same action event name to the same two-dimensional spatial coordinate system;

[0026] The myoelectric signal spectrum and the EEG signal spectrum corresponding to the reference electrode patch are recorded as the reference myoelectric signal spectrum and the reference EEG signal spectrum, and the reference myoelectric signal spectrum and the reference EEG signal spectrum with different identity numbers are sequentially subtracted, and the subtraction results of each group are sequentially overlapped and mapped according to the time domain position to obtain the common mode signal at the corresponding position;

[0027] The common mode signal is input into the filter, and the filter frequency is set according to the phase value of the common mode signal at each time domain position. Each EMG signal spectrum and EEG signal spectrum is input into the filter, and the filter performs phase filtering on the EMG signal spectrum and EEG signal spectrum according to the phase value at each time domain position to obtain the target EMG signal spectrum and target EEG signal spectrum.

[0028] Furthermore, the process of acquiring the normal target electromyographic signal segment and the normal target electroencephalographic signal segment includes:

[0029] Establish a human body structure model, set action nodes corresponding to each joint position on the human body structure model, mark each action node with the magnetic field sensor number corresponding to the joint position, establish a three-dimensional coordinate system, and input the human body structure model into the three-dimensional coordinate system;

[0030] Mapping the associated nodes in each action event to the corresponding action node in the human body structure model, and recording the action node as an event action node;

[0031] According to the event duration in the action event, the same event timeline is set for each event action node, and then each joint trajectory in the action event is mapped to the target node, and the corresponding joint trajectory line is planned;

[0032] Divide the event timeline into K time segments, associate the joint trajectory line with the event timeline, and then divide the joint trajectory line into K sub-joint trajectory lines according to the time segment on the event timeline, where K is a natural number greater than 0;

[0033] Divide each target myoelectric signal spectrum and target EEG signal spectrum into K target myoelectric signal segments and target EEG signal segments according to the time segment on the event time axis;

[0034] Overlap and map the target electromyographic signal segments and target electroencephalographic signal segments in the same time segment but with different identity numbers onto the same two-dimensional coordinate system;

[0035] Several time nodes are set on the two-dimensional coordinate system, and each target electromyographic signal segment and target electroencephalographic signal segment are divided into several coordinate points according to the time nodes. The coordinate points on the same time node are normally distributed according to the ordinate value. The center position is taken as the starting point, and the coordinate points corresponding to the ordinate value of 5% of the whole part at the left and right positions of the starting point are selected as normal coordinate points;

[0036] The normal coordinate points at each time node are connected in sequence, and then the normal target electromyographic signal segment and the normal target electroencephalographic signal segment in the corresponding time segment are obtained.

[0037] Furthermore, the process of establishing the electrophysiological knowledge graph includes:

[0038] Bind the normal target electromyogram signal segment and the normal target EEG signal segment in each action event to the corresponding action node, and associate and bind each normal target electromyogram signal segment and the normal target EEG signal segment with the sub-joint trajectory line of the same time segment according to the event timeline on the action node, so as to obtain the event human action model of the corresponding action event;

[0039] An electrophysiological knowledge graph is established, wherein the electrophysiological knowledge graph includes event names of several action events and corresponding event human action models.

[0040] Furthermore, the process of generating the robot arm action instruction includes:

[0041] The action instruction module maps each action node to the corresponding mechanical joint of the disabled user's robotic arm according to the event human action model in the electrophysiological knowledge graph, and marks each mechanical joint with the same magnetic field sensor number;

[0042] The action instruction generation unit extracts the real-time electromyographic signal and the real-time electroencephalogram signal from the disabled signal data set of the disabled user, and inputs the real-time electromyographic signal and the real-time electroencephalogram signal into the event human action model, and then matches the real-time electromyographic signal and the real-time electroencephalogram signal with each normal target electromyographic signal segment and each normal target electroencephalogram signal segment, and obtains the desired behavior event of the corresponding disabled user according to the matching result;

[0043] Extracting sub-joint trajectories of each joint position in the behavior event expected by the disabled user from the electrophysiological knowledge graph, and matching the extraction results with each mechanical joint on the robotic arm of the disabled user according to the magnetic field sensor number, and then the action instruction generation unit generates a robotic arm action instruction according to the matched sub-joint trajectories, and sends the robotic arm action instruction to the robotic arm device;

[0044] The robot arm device sends the robot arm action command to the corresponding mechanical joint, and then each mechanical joint performs a corresponding action according to the robot arm action command.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] The present invention obtains normal target EMG signal segments and normal target EEG signal segments through the EMG signals and EEG signals of several healthy users under different action events, and then establishes corresponding event human action models and electrophysiological knowledge graphs. The EMG signals and EEG signals of disabled users in real time are input into the electrophysiological knowledge graph, and then the corresponding action names and action trajectories are obtained from the electrophysiological knowledge graph, and then the robot arm action instructions are generated and executed, which effectively improves the accuracy of electrophysiological data and ensures the versatility of electrophysiological data analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention.

[0048] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0049] To make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described in detail below. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0050] like Figure 1 As shown, an electrophysiological intelligent analysis system based on a knowledge graph includes a cloud computing platform, wherein the cloud computing platform is communicatively connected with an electrophysiological signal acquisition module, a knowledge graph module, and an action instruction module;

[0051] The electrophysiological signal acquisition module is used to collect the electromyographic signals and electroencephalographic signals of healthy users under different action events, and to collect the electromyographic signals and electroencephalographic signals of disabled users in real time;

[0052] The knowledge graph module is used to preprocess the electromyographic signals and electroencephalographic signals of several healthy users under different action events, obtain the normal target electromyographic signal segments and normal target electroencephalographic signal segments under each action event according to the preprocessing results, and then establish an event human action model and an electrophysiological knowledge graph according to the normal target electromyographic signal segments and normal target electroencephalographic signal segments and each action event.

[0053] The action instruction module is provided with an action instruction generating unit and a mechanical arm device;

[0054] The action instruction generation unit is used to input the electromyographic signal and the electroencephalographic signal of the disabled user in real time into the electrophysiological knowledge graph, and then obtain the corresponding action name and action trajectory from the electrophysiological knowledge graph, and generate the action instruction of the robot arm;

[0055] The robotic arm device is composed of a plurality of mechanical joints. The robotic arm device calls corresponding mechanical joints according to the robotic arm action instructions and performs corresponding behavior actions.

[0056] Further, the working principle of the present invention is described below by way of examples:

[0057] Several electrode stickers are placed on the heads and joints of healthy and disabled users, and a magnetic field sensor is connected to each electrode sticker at the same joint and head position. Each magnetic field sensor is numbered s1, s2, ..., s n , where n is a natural number greater than 0;

[0058] It should be noted that the joint positions include joints of human bones such as knee joints, elbow joints, and finger joints;

[0059] Preset Num action events, the action event includes event name, event duration, associated joint number and corresponding joint trajectory, where Num is a natural number greater than 50;

[0060] The electrophysiological signal acquisition module communicates with the magnetic field sensors of each user, and then N healthy users perform corresponding action events according to the event duration, associated joint numbers and corresponding joint trajectories in each action event. During the execution of the action event, the electrodes at the joints and the head respectively collect corresponding electromyographic signals and electroencephalographic signals, where N is a natural number greater than 0;

[0061] The electrophysiological signal acquisition module sets a unique identity number for each healthy user and disabled user. During the execution of the action event, each magnetic field sensor sends the collected electromyographic signal and electroencephalographic signal to the electrophysiological signal acquisition module;

[0062] When the action event is executed, the electrophysiological signal acquisition module generates corresponding myoelectric signal spectrum and EEG signal spectrum according to the myoelectric signal and EEG signal of each magnetic field sensor, and marks each myoelectric signal spectrum and EEG signal spectrum with the number of the corresponding magnetic field sensor;

[0063] It should be noted that the myoelectric signal spectrum and the EEG signal spectrum include electrophysiological signals collected by multiple electrode stickers and a reference electrode sticker, wherein one reference electrode sticker is provided at the head and one at the joint position respectively;

[0064] Integrate all the myoelectric signal spectra and EEG signal spectra generated in the same action event to generate an action event signal dataset corresponding to the action event, and annotate the identity number of the corresponding healthy user and the action event name;

[0065] The magnetic field sensors at the head and joint positions of the disabled user send the real-time electromyographic signals and real-time electroencephalographic signals collected in real time to the electrophysiological signal acquisition module. The electrophysiological signal acquisition module then generates a corresponding disability signal data set based on the real-time electromyographic signals and real-time electroencephalographic signals, and marks the corresponding disabled user's identity number.

[0066] Furthermore, the electrophysiological signal acquisition module sends all action event signal data sets and action time to the knowledge graph module, and sends the disabled signal data set of the disabled user to the action execution module in real time;

[0067] The knowledge graph module extracts the EMG signal spectrum and EEG signal spectrum of the head position and joint position from the action event signal dataset with different identity numbers but the same action event name;

[0068] First, each EMG signal spectrum and EEG signal spectrum is preprocessed in turn, wherein the preprocessing process includes:

[0069] Establishing a two-dimensional spatial coordinate system, mapping the myoelectric signal spectrum and the electroencephalogram signal spectrum at the same position in the action event signal data set with different identity numbers but the same action event name to the same two-dimensional spatial coordinate system;

[0070] The myoelectric signal spectrum and the EEG signal spectrum corresponding to the reference electrode patch are recorded as the reference myoelectric signal spectrum and the reference EEG signal spectrum, and the reference myoelectric signal spectrum and the reference EEG signal spectrum with different identity numbers are sequentially subtracted, and the subtraction results of each group are sequentially overlapped and mapped according to the time domain position to obtain the common mode signal at the corresponding position;

[0071] The common mode signal is input into the filter, and the filter frequency is set according to the phase value of the common mode signal at each time domain position. Each EMG signal spectrum and EEG signal spectrum is input into the filter, and the filter performs phase filtering on the EMG signal spectrum and EEG signal spectrum according to the preset phase value at each time domain position to obtain the target EMG signal spectrum and target EEG signal spectrum.

[0072] Furthermore, the knowledge graph module establishes a human body structure model, sets action nodes corresponding to each joint position on the human body structure model, and marks each action node with a number corresponding to the joint position;

[0073] The knowledge graph module extracts the corresponding event duration, associated node number and joint trajectory from each action event, establishes a three-dimensional coordinate system, and inputs the human body structure model into the three-dimensional coordinate system;

[0074] Mapping the associated nodes in each action event to the corresponding action node in the human body structure model, and recording the action node as an event action node;

[0075] According to the event duration in the action event, the same event timeline is set for each event action node, and then each joint trajectory in the action event is mapped to the target node, and the corresponding joint trajectory line is planned;

[0076] Divide the event timeline into K time segments, associate the joint trajectory line with the event timeline, and then divide the joint trajectory line into K sub-joint trajectory lines according to the time segment on the event timeline, where K is a natural number greater than 0;

[0077] Divide each target myoelectric signal spectrum and target EEG signal spectrum into K target myoelectric signal segments and target EEG signal segments according to the time segment on the event time axis;

[0078] Overlap and map the target electromyographic signal segments and target electroencephalographic signal segments in the same time segment but with different identity numbers onto the same two-dimensional coordinate system;

[0079] Several time nodes are set on the two-dimensional coordinate system, and each target electromyographic signal segment and target electroencephalographic signal segment are divided into several coordinate points according to the time nodes. The coordinate points on the same time node are normally distributed according to the ordinate value. The center position is taken as the starting point, and the coordinate points corresponding to the ordinate value of 5% of the whole part at the left and right positions of the starting point are selected as normal coordinate points;

[0080] The normal coordinate points at each time node are connected in sequence, and then the normal target electromyographic signal segment and the normal target electroencephalographic signal segment in the corresponding time segment are obtained.

[0081] Furthermore, the normal target electromyographic signal segments and normal target electroencephalographic signal segments in each action event are bound to the corresponding action node, and each normal target electromyographic signal segment and normal target electroencephalographic signal segment are associated and bound with the sub-joint trajectory line of the same time segment according to the event timeline on the action node, thereby obtaining the event human action model of the corresponding action event;

[0082] An electrophysiological knowledge graph is established, wherein the electrophysiological knowledge graph includes event names of several action events and corresponding event human action models.

[0083] Further, the knowledge graph module sends the electrophysiological knowledge graph to the action instruction module;

[0084] The action instruction module maps each action node to the corresponding mechanical joint of the disabled user's robotic arm according to the event human action model in the electrophysiological knowledge graph, and marks each mechanical joint with the same magnetic field sensor number;

[0085] The action instruction generation unit extracts the real-time electromyographic signal and the real-time electroencephalogram signal from the disabled signal data set of the disabled user, and inputs the real-time electromyographic signal and the real-time electroencephalogram signal into the event human action model, and then matches the real-time electromyographic signal and the real-time electroencephalogram signal with each normal target electromyographic signal segment and each normal target electroencephalogram signal segment, and obtains the desired behavior event of the corresponding disabled user according to the matching result;

[0086] The sub-joint trajectories of each joint position in the behavioral event expected by the disabled user are extracted from the electrophysiological knowledge graph, and the extracted results are matched with the various mechanical joints on the disabled user's robotic arm according to the magnetic field sensor number. Then the action instruction generation unit generates a robotic arm action instruction according to the matched sub-joint trajectories and sends the robotic arm action instruction to the robotic arm device.

[0087] The robot arm device sends the robot arm action command to the corresponding mechanical joint, and then each mechanical joint performs a corresponding action according to the robot arm action command.

[0088] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An electrophysiological intelligent analysis system based on knowledge graph, including a cloud computing platform, characterized in that: The cloud computing platform is communicatively connected to an electrophysiological signal acquisition module, a knowledge graph module, and an action instruction module; The electrophysiological signal acquisition module is used to collect the electromyographic signals and electroencephalographic signals of healthy users under different action events, and to collect the electromyographic signals and electroencephalographic signals of disabled users in real time; The process of collecting electromyographic and electroencephalographic signals of healthy users under different action events includes: A plurality of electrode stickers are respectively pasted on the heads and joints of healthy users and disabled users, and a magnetic field sensor is respectively connected to the electrode stickers at the same joint position and head position, and each magnetic field sensor is numbered; Preset Num action events, the action event includes event name, event duration, associated joint number and corresponding joint trajectory, where Num is a natural number greater than 50; The electrophysiological signal acquisition module communicates with the magnetic field sensors of each user, and then N healthy users perform corresponding action events according to the event duration, associated joint numbers and corresponding joint trajectories in each action event. During the execution of the action event, the electrodes at the joints and the head respectively collect corresponding electromyographic signals and electroencephalographic signals, where N is a natural number greater than 0; The electrophysiological signal acquisition module sets a unique identity number for each healthy user and disabled user, generates corresponding electromyographic signal spectrum and electroencephalographic signal spectrum according to the electromyographic signal and electroencephalographic signal of each magnetic field sensor, integrates all electromyographic signal spectrum and electroencephalographic signal spectrum generated in the same action event to generate an action event signal data set of the corresponding action event, and marks the identity number of the corresponding healthy user and the name of the action event; The knowledge graph module is used to preprocess the electromyographic signals and electroencephalographic signals of several healthy users under different action events, obtain the normal target electromyographic signal segment and the normal target electroencephalographic signal segment under each action event according to the preprocessing result, and then establish the event human action model and the electrophysiological knowledge graph according to the normal target electromyographic signal segment, the normal target electroencephalographic signal segment and each action event; The acquisition process of the normal target electromyographic signal segment and the normal target electroencephalographic signal segment includes: establishing a human body structure model, setting action nodes corresponding to each joint position on the human body structure model, marking each action node with a magnetic field sensor number corresponding to the joint position, establishing a three-dimensional coordinate system, and inputting the human body structure model into the three-dimensional coordinate system; Mapping the associated nodes in each action event to the corresponding action node in the human body structure model, and recording the action node as an event action node; According to the event duration in the action event, the same event timeline is set for each event action node, and then each joint trajectory in the action event is mapped to the target node, and the corresponding joint trajectory line is planned; Divide the event timeline into K time segments, associate the joint trajectory line with the event timeline, and then divide the joint trajectory line into K sub-joint trajectory lines according to the time segment on the event timeline, where K is a natural number greater than 0; Divide each target myoelectric signal spectrum and target EEG signal spectrum into K target myoelectric signal segments and target EEG signal segments according to the time segment on the event time axis; Overlap and map the target electromyographic signal segments and target electroencephalographic signal segments in the same time segment but with different identity numbers onto the same two-dimensional coordinate system; A number of time nodes are set on the two-dimensional coordinate system, and each target electromyographic signal segment and target electroencephalographic signal segment are divided into a number of coordinate points according to the time nodes, and the coordinate points on the same time node are normally distributed according to the ordinate values ​​to obtain normal coordinate points; Connect the normal coordinate points at each time node in sequence, and then obtain the normal target electromyographic signal segment and the normal target electroencephalographic signal segment in the corresponding time segment; The process of establishing the electrophysiological knowledge graph includes: Bind the normal target electromyogram signal segment and the normal target EEG signal segment in each action event to the corresponding action node, and associate and bind each normal target electromyogram signal segment and the normal target EEG signal segment with the sub-joint trajectory line of the same time segment according to the event timeline on the action node, so as to obtain the event human action model of the corresponding action event; Establishing an electrophysiological knowledge graph, wherein the electrophysiological knowledge graph includes event names of a plurality of action events and corresponding event human action models; The action instruction module is provided with an action instruction generating unit and a mechanical arm device; The action instruction generation unit is used to input the electromyographic signal and the electroencephalographic signal of the disabled user in real time into the electrophysiological knowledge graph, and then obtain the corresponding action name and action trajectory from the electrophysiological knowledge graph, and generate the action instruction of the robot arm; The robotic arm device is composed of a plurality of mechanical joints. The robotic arm device calls corresponding mechanical joints according to the robotic arm action instructions and performs corresponding behavior actions.

2. According to claim 1, an electrophysiological intelligent analysis system based on knowledge graph is characterized in that: The process of collecting real-time electromyographic signals and real-time electroencephalographic signals of disabled users in real time includes: The magnetic field sensors at the head and joint positions of the disabled user send the real-time electromyographic signals and real-time electroencephalographic signals collected in real time to the electrophysiological signal acquisition module. The electrophysiological signal acquisition module then generates a corresponding disability signal data set based on the real-time electromyographic signals and real-time electroencephalographic signals, and marks the corresponding disabled user's identity number.

3. According to claim 1, the electrophysiological intelligent analysis system based on knowledge graph is characterized in that: The myoelectric signal spectrum and the brain electroencephalographic signal spectrum are composed of electrophysiological signals collected by multiple electrode stickers and a reference electrode sticker, wherein a reference electrode sticker is respectively arranged at the head and the joint position.

4. According to claim 3, an electrophysiological intelligent analysis system based on knowledge graph is characterized in that: The process of preprocessing the myoelectric signal spectrum and the EEG signal spectrum in the action event signal data set includes: A two-dimensional spatial coordinate system is established, and the myoelectric signal spectrum and the electroencephalogram signal spectrum at the same position in the action event signal data set with different identity numbers but the same action event name are mapped to the same two-dimensional spatial coordinate system; The myoelectric signal spectrum and the EEG signal spectrum corresponding to the reference electrode patch are recorded as the reference myoelectric signal spectrum and the reference EEG signal spectrum, and the reference myoelectric signal spectrum and the reference EEG signal spectrum with different identity numbers are sequentially subtracted, and the subtraction results of each group are sequentially overlapped and mapped according to the time domain position to obtain the common mode signal at the corresponding position; The common mode signal is input into the filter, and the filter frequency is set according to the phase value of the common mode signal at each time domain position. Each EMG signal spectrum and EEG signal spectrum is input into the filter, and the filter performs phase filtering on the EMG signal spectrum and EEG signal spectrum according to the preset phase value at each time domain position to obtain the target EMG signal spectrum and target EEG signal spectrum.

5. The electrophysiological intelligent analysis system based on knowledge graph according to claim 4 is characterized in that: The process of generating the robot arm action instruction includes: Map each action node to the corresponding mechanical joint of the disabled user's robotic arm, and label each mechanical joint with the same magnetic field sensor number; The action instruction generation unit extracts the real-time electromyographic signal and the real-time electroencephalogram signal from the disabled signal data set of the disabled user, and inputs the real-time electromyographic signal and the real-time electroencephalogram signal into the event human action model, and then matches the real-time electromyographic signal and the real-time electroencephalogram signal with each normal target electromyographic signal segment and each normal target electroencephalogram signal segment, and obtains the desired behavior event of the corresponding disabled user according to the matching result; Extracting sub-joint trajectories of each joint position in the behavior event expected by the disabled user from the electrophysiological knowledge graph, and matching the extraction results with each mechanical joint on the robotic arm of the disabled user according to the magnetic field sensor number, and then the action instruction generation unit generates a robotic arm action instruction according to the matched sub-joint trajectories, and sends the robotic arm action instruction to the robotic arm device; The robot arm device sends the robot arm action command to the corresponding mechanical joint, and then each mechanical joint performs a corresponding action according to the robot arm action command.