A method and system for analyzing and processing electromyographic signals based on big data
By collecting and personalizing user EMG signals in real time, the problem of low accuracy and efficiency of EMG signals analysis caused by individual differences in users is solved, and more efficient and accurate EMG signals analysis is achieved.
Patent Information
- Application Number
- CN202410539253.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-30
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-04-30
AI Technical Summary
In the prior art, the accuracy and low processing efficiency of electromyography signal analysis due to individual differences in users.
By collecting user EMG signals in real time, conducting personalized differences analysis and adaptive training, and combining environment and feedback data for correction and optimization, improving the accuracy and efficiency of EMG signal analysis.
The accuracy and processing efficiency of electromyography signal analysis are improved, and the targetedness and accuracy of electromyography signal analysis are enhanced through personalized models and noise optimization.
Smart Images

Figure CN118216937B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of information processing technology, and in particular to a method and system for analyzing and processing electromyographic signals based on big data. Background Art
[0002] The electromyographic signal is the superposition of the action potentials of the motor units in numerous muscle fibers in time and space. It is a bioelectrical signal generated in muscle tissue and accompanied by muscle contraction. The electromyographic signal is usually recorded by electromyography, which is a curve reflecting the muscle action potential. The change in potential difference can reflect the activity state of the muscle, and can be used to study the function of the muscle and diagnose diseases. The representation of the electromyographic signal is mainly reflected by the electromyogram, which is a function of time and amplitude, reflecting the potential changes generated by the muscle during the contraction process. Analysis of the electromyographic signal can evaluate the functional state of the muscle, or provide guidance for rehabilitation medicine, sports science and other fields. For example, in rehabilitation medicine, electromyographic signal analysis can be used to evaluate the recovery of muscle injuries; in sports science, it can be used to analyze the muscle activity patterns of athletes to optimize training plans.
[0003] Chinese Patent Publication No.: CN108937926A discloses a method for analyzing surface electromyography signals of big data, which includes obtaining surface electromyography signal data of multiple users; processing the obtained surface electromyography signal data of multiple users, and completing the division operation of training data and test data; establishing a model for the surface electromyography signal data of multiple users based on the training data, generating a first model, and training the first model; and completing the analysis and execution operation of the surface electromyography signal of the target individual based on the trained first model. This method restores the stimulation generated by the surface electromyography signal of the body during movement, stimulates the corresponding neural network, and intelligently determines the required output based on the spatial position information of the sensor. However, this solution does not adjust the individual differences of the users themselves, the accuracy of the electromyography signal analysis is low, and the efficiency of the electromyography signal analysis and processing cannot be improved. Summary of the invention
[0004] To this end, the present invention provides a method and system for myoelectric signal analysis and processing based on big data, so as to overcome the problems in the prior art of low accuracy and efficiency of myoelectric signal analysis and processing caused by differences in myoelectric signals generated by individual users themselves.
[0005] To achieve the above objectives, on the one hand, the present invention provides a method for analyzing and processing myoelectric signals based on big data, comprising:
[0006] Step S1, real-time collection of user electromyographic signals;
[0007] Step S2, analyzing the user's personality differences with the preset sample database, and adaptively training the preset electromyography analysis model according to the analysis results and outputting the user's personality model;
[0008] Step S3, judging the validity of the user adaptability data according to the input environment coefficient, and correcting the user adaptability data according to the judgment result;
[0009] Step S4, identifying the high-frequency specific data according to the amount of invalid data, and correcting the validity judgment result according to the identification result;
[0010] Step S5, inputting the user's electromyographic signal into the user personality model and outputting the interaction result;
[0011] Step S6, providing initial feedback on the output of the interaction result according to the number of times the user actively adjusts the interaction result, providing secondary feedback on the output of the interaction result in combination with the user feedback, and performing optimization judgment based on the initial feedback result and the secondary feedback result;
[0012] Step S7, after determining to perform optimization, perform noise optimization on the user's electromyographic signal acquisition process, and perform feature extraction optimization on the user's personality model.
[0013] Further, in step S2, when performing the user personality difference analysis, the personality electromyographic signal in the user adaptability data is compared with the electromyographic signal in the preset sample database to obtain the vertical difference value A, the vertical difference value A is compared with the preset vertical difference value A0, and the user personality difference analysis is performed according to the comparison result, wherein:
[0014] When A<A0, it is determined that the user's personality difference is normal, and the preset electromyography analysis model is not adaptively trained, and the preset electromyography analysis model is output as the user's personality model;
[0015] When A≥A0, it is determined that the user's personality difference is abnormal, and the user's adaptability data is used as a difference training set to perform adaptive training on the preset electromyography analysis model, and the preset electromyography analysis model after adaptive training is output as the user's personality model.
[0016] Further, in step S3, when judging the validity of the user adaptability data, the recording environment coefficient Q of each individual electromyographic signal is calculated according to the recording time T and the environmental noise intensity D, and Q=0.3×T / T0+0.7×D / D0 is set, T0 is the preset recording time, D0 is the preset environmental noise intensity, and the recording environment coefficient Q is compared with the preset recording environment coefficient Q0, and the validity of the user adaptability data is judged according to the comparison result, wherein:
[0017] When Q≤Q0, it is determined that the individual electromyographic signal in the user adaptability data is valid;
[0018] When Q>Q0, it is determined that the individual electromyographic signal in the user adaptability data is invalid, and the user adaptability data is corrected, and the individual electromyographic signal is deleted from the user adaptability data.
[0019] Furthermore, in step S4, when correcting the validity judgment result, the invalid data amount P is compared with the preset invalid data amount P0, and the high-frequency specific data is identified according to the comparison result, wherein:
[0020] When P<P0, it is determined that the individual electromyographic signal is not high-frequency specific data;
[0021] When P≥P0, the personalized electromyographic signal is determined to be high-frequency specific data, and the validity judgment results of all personalized electromyographic signals with the same interaction results corresponding to the personalized electromyographic signal are corrected to the personalized electromyographic signal being valid, and the preset number of groups of preset interaction results corresponding to the personalized electromyographic signal is increased.
[0022] Furthermore, in step S4, when adding the preset number of groups of preset interaction results corresponding to the personalized electromyographic signal, the increase coefficient a is set to correct the preset number of groups n of the interaction results, the corrected preset number of groups is na, na=n×a is set, and 1.5≤a<5 is set.
[0023] Furthermore, in step S6, when the output of the interaction result is fed back for the first time, the number of active adjustments m by the user is compared with each preset number of active adjustments by the user, and the output of the interaction result is fed back for the first time according to the comparison result, wherein:
[0024] When m>m1, the interaction result output is determined to be abnormal, and optimization is performed;
[0025] When m2<m≤m1, the output of the interaction result is determined to be subject to secondary feedback;
[0026] When m≤m2, the interaction result output is determined to be normal;
[0027] m1 is the first preset number of times of active adjustment by the user, m2 is the second preset number of times of active adjustment by the user, and 0<m2<m1.
[0028] Furthermore, in step S6, when performing secondary feedback on the output of the interaction result, the user satisfaction U is obtained, and the user satisfaction U is compared with the preset user satisfaction U0, and secondary feedback is performed on the output of the interaction result according to the comparison result, wherein:
[0029] When U≤U0, the interaction result output is determined to be abnormal, and optimization is performed;
[0030] When U>U0, it is determined that the interaction result output is normal.
[0031] Further, in step S7, after determining to optimize, when performing noise optimization on the user electromyographic signal acquisition process, the electromyographic signal characteristics of the resting state are obtained, and the signal characteristics of the user electromyographic signal acquired in real time are compared with the electromyographic signal characteristics of the resting state, and the electromyographic signal in the resting state is identified according to the comparison result, wherein:
[0032] When the characteristics of the user's electromyographic signal collected in real time are inconsistent with the characteristics of the electromyographic signal in the resting state, it is determined that the user's electromyographic signal is not in the resting state, and the electromyographic signal is input into a Butterworth bandpass filter for noise optimization to remove low-frequency artifacts and high-frequency noise;
[0033] When the characteristics of the user's electromyographic signal collected in real time are consistent with the characteristics of the electromyographic signal in the resting state, it is determined that the user's electromyographic signal is in the resting state and is deleted instead of being input into the user personality model for analysis and processing.
[0034] Furthermore, in step S7, after determining to perform optimization, when performing feature extraction optimization on the user personality model, obtain the noise-optimized user electromyographic signal within the optimization time interval, and perform standard feature extraction on it to obtain standard data features, use the noise-optimized user electromyographic signal within the optimization time interval - standard data features as the feature extraction optimization data set, and use 70% of the feature extraction optimization data set as the feature extraction optimization training set, and 30% of the feature extraction optimization data set as the feature extraction optimization test set, train the convolutional neural network model according to the feature extraction optimization training set, and test the trained convolutional neural network model according to the feature extraction optimization test set, if the accuracy reaches 95%, output the tested convolutional neural network model as the feature extraction optimization model, if the accuracy does not reach 95%, continue to obtain the noise-optimized user electromyographic signal within the next optimization time interval to train the tested convolutional neural network model until the accuracy reaches 95%, and input the noise-optimized user electromyographic signal into the feature extraction optimization model for feature extraction.
[0035] On the other hand, the present invention also provides a myoelectric signal analysis and processing system based on big data, comprising:
[0036] Signal acquisition module, used to collect user's electromyographic signals in real time;
[0037] The personality difference analysis module is used to analyze the user's personality difference with the preset sample database, and to perform adaptive training on the preset electromyography analysis model according to the analysis results and output the user's personality model;
[0038] The validity judgment module is used to judge the validity of the user adaptability data according to the input environment coefficient, and to correct the user adaptability data according to the judgment result;
[0039] A specific data identification module is used to identify high-frequency specific data according to the amount of invalid data, and to correct the validity judgment result according to the identification result;
[0040] An interaction result output module, used to input the user's electromyographic signal into the user's personality model and output the interaction result;
[0041] The result feedback module is used to provide initial feedback on the output of the interaction result according to the number of active adjustments made by the user, and to provide secondary feedback on the output of the interaction result in combination with the user feedback, and to make optimization judgments based on the initial feedback results and the secondary feedback results;
[0042] The feedback optimization module is used to optimize the noise of the user's electromyographic signal acquisition process and to optimize the feature extraction of the user's personality model after the optimization is determined.
[0043] Compared with the prior art, the beneficial effects of the present invention are that the method collects the user's electromyographic signal in real time through step S1, so as to perform real-time analysis based on the user's electromyographic signal; the method performs user personality difference analysis through step S2, and adaptively trains the preset electromyographic analysis model and outputs the user personality model, so as to adjust the personality difference of the user himself; the method judges the validity of the user's adaptability data through step S3, and corrects the user's adaptability data according to the judgment result to improve the efficiency of electromyographic signal analysis and processing; the method identifies the high-frequency specific data through step S4, and corrects the validity judgment result. The result is corrected to further improve the efficiency of electromyographic signal analysis and processing. The method inputs the user's electromyographic signal into the user's personality model and outputs the interaction result through step S5, and performs primary feedback and secondary feedback on the output of the interaction result through step S6, and performs optimization judgment based on the primary feedback result and the secondary feedback result, thereby improving the accuracy of the user's personality model analysis and further improving the efficiency of electromyographic signal analysis and processing. The method performs noise optimization on the user's electromyographic signal acquisition process through step S7, and performs feature extraction optimization on the user's personality model, thereby further improving the accuracy of the user's personality model analysis and thus improving the efficiency of electromyographic signal analysis and processing. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 Schematic diagram of the process of myoelectric signal analysis and processing method based on big data in this embodiment;
[0045] Figure 2Schematic diagram of the structure of the electromyographic signal analysis and processing system based on big data in this embodiment. DETAILED DESCRIPTION
[0046] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0047] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.
[0048] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0049] See also Figure 1 As shown, it is a flow chart of the electromyographic signal analysis and processing method based on big data in this embodiment, and the method includes:
[0050] Step S1, real-time collection of user electromyographic signals;
[0051] Step S2, analyzing the user's personality differences with the preset sample database, and adaptively training the preset electromyography analysis model according to the analysis results and outputting the user's personality model;
[0052] Step S3, judging the validity of the user adaptability data according to the input environment coefficient, and correcting the user adaptability data according to the judgment result;
[0053] Step S4, identifying the high-frequency specific data according to the amount of invalid data, and correcting the validity judgment result according to the identification result;
[0054] Step S5, inputting the user's electromyographic signal into the user personality model and outputting the interaction result;
[0055] Step S6, providing initial feedback on the output of the interaction result according to the number of times the user actively adjusts the interaction result, providing secondary feedback on the output of the interaction result in combination with the user feedback, and performing optimization judgment based on the initial feedback result and the secondary feedback result;
[0056] Step S7, after determining to perform optimization, perform noise optimization on the user's electromyographic signal acquisition process, and perform feature extraction optimization on the user's personality model.
[0057] Specifically, the method is applied to an electromyographic signal analysis device, and the electromyographic signal is analyzed according to big data, so as to adjust the personality differences of the user himself, improve the accuracy of the electromyographic signal analysis, and improve the efficiency of the electromyographic signal analysis processing. The method collects the user's electromyographic signal in real time through step S1, so as to perform real-time analysis based on the user's electromyographic signal. The method performs user personality difference analysis through step S2, and adaptively trains a preset electromyographic analysis model and outputs a user personality model, so as to adjust the personality differences of the user himself. The method judges the validity of the user's adaptability data through step S3, and corrects the user's adaptability data according to the judgment result, so as to improve the efficiency of the electromyographic signal analysis processing. The method identifies high-frequency specific data through step S4 and corrects the validity judgment result, further improving the efficiency of electromyographic signal analysis and processing. The method inputs the user electromyographic signal into the user personality model and outputs the interaction result through step S5, and performs initial feedback and secondary feedback on the output of the interaction result through step S6, and performs optimization judgment based on the initial feedback result and the secondary feedback result, thereby improving the accuracy of the user personality model analysis and further improving the efficiency of the electromyographic signal analysis and processing. The method performs noise optimization on the user electromyographic signal acquisition process through step S7, and performs feature extraction optimization on the user personality model, further improving the accuracy of the user personality model analysis and thus improving the efficiency of the electromyographic signal analysis and processing.
[0058] Specifically, in step S1, the user's electromyographic signal is collected in real time by electromyography. The user's electromyographic signal refers to the potential difference generated when the user's nervous system sends signals to the muscles. The user refers to the object wearing the electromyographic signal collection device for analysis.
[0059] Specifically, in step S2, when performing the user personality difference analysis, the individual electromyographic signal in the user adaptability data is compared with the electromyographic signal in the preset sample database to obtain the vertical difference value A, the vertical difference value A is compared with the preset vertical difference value A0, and the user personality difference analysis is performed according to the comparison result, wherein:
[0060] When A<A0, it is determined that the user's personality difference is normal, and the preset electromyography analysis model is not adaptively trained, and the preset electromyography analysis model is output as the user's personality model;
[0061] When A≥A0, it is determined that the user's personality difference is abnormal, and the user's adaptability data is used as a difference training set to perform adaptive training on the preset electromyography analysis model, and the preset electromyography analysis model after adaptive training is output as the user's personality model.
[0062] Specifically, the user personality difference analysis is performed through the step S2, and when the vertical difference value is greater than or equal to the preset vertical difference value, the user personality difference is determined to be abnormal, and the user adaptability data is used as a difference training set to perform adaptive training on the preset electromyography analysis model, and the preset electromyography analysis model after the adaptive training is output as the user personality model to facilitate personality difference adjustments for the user himself, thereby improving the accuracy of electromyography signal analysis.
[0063] Specifically, the user adaptability data refers to a preset number of user electromyographic signals collected according to preset interaction results after the user wears the electromyographic signal collection device, such as collecting 10 groups of user electromyographic signals of the user raising hands, raising hands and clasping fists respectively after the user wears the electromyographic signal collection device. The personalized electromyographic signal refers to the user electromyographic signal corresponding to the preset interaction result in the user adaptability data. The preset sample database refers to a preset sample database stored in the form of user electromyographic signal-interaction result. This embodiment does not limit the setting process of the preset sample database, and those skilled in the art can freely set it according to actual needs, such as setting it to sample user electromyographic signals from a cloud database-interaction results. The interaction result data is used as a preset sample database. The cloud database refers to the user electromyographic signal-interaction result database generated in the electromyographic signal analysis and processing process uploaded by the electromyographic signal analysis and processing system. The vertical difference value refers to the characteristic difference value of the electromyographic signal when the preset interaction result in the user adaptability data is the same as the interaction result in the preset sample database. This embodiment does not limit the method for obtaining the vertical difference value, and only needs to meet the need to reflect the differences in user personality. For example, it can be set to pre-process the electromyographic signal when the preset interaction result in the user adaptability data is the same as the interaction result in the preset sample database, including filtering, noise reduction, rectification, etc., and the pre-processed electromyographic signal is obtained. Characteristic values are extracted from the electromyographic signal, including time domain characteristics such as mean, variance, peak value, and frequency domain characteristics such as power spectral density and frequency distribution. Statistical analysis is performed on the extracted characteristic values, such as calculating the mean, standard deviation, coefficient of variation, etc., and charts or images are used to intuitively present the difference values in the electromyogram, such as bar charts, line charts or scatter plots to compare the electromyographic signal characteristics under different conditions. The preset vertical difference value refers to a preset value that reflects the abnormal personality differences of users after considering the influencing factors of the acquisition process such as individual differences, muscle fatigue, and electrode position. For example, a preset vertical difference value can be set. This embodiment does not limit the adaptive training method, and those skilled in the art can freely adjust it according to actual conditions. Settings, such as setting the user's adaptive data as a difference training set to perform adaptive training on the preset electromyography analysis model, until the accuracy of the preset electromyography analysis model reaches 95% and outputs it as the user personality model. The preset electromyography analysis model refers to a random forest algorithm model obtained by training according to a preset sample database. This embodiment does not limit the training process of the preset electromyography analysis model, and technical personnel in this field can freely set it according to actual needs, such as setting the data in the preset sample database to be preprocessed, and performing signal segmentation on the preprocessed data, performing feature extraction on the data after signal segmentation, and performing signal classification on the data after feature extraction.
[0064] Specifically, in step S3, when judging the validity of the user adaptability data, the recording environment coefficient Q of each individual electromyographic signal is calculated according to the recording time T and the environmental noise intensity D, and Q=0.3×T / T0+0.7×D / D0 is set, T0 is the preset recording time, D0 is the preset environmental noise intensity, and the recording environment coefficient Q is compared with the preset recording environment coefficient Q0, and the validity of the user adaptability data is judged according to the comparison result, wherein:
[0065] When Q≤Q0, it is determined that the individual electromyographic signal in the user adaptability data is valid;
[0066] When Q>Q0, it is determined that the individual electromyographic signal in the user adaptability data is invalid, and the user adaptability data is corrected, and the individual electromyographic signal is deleted from the user adaptability data.
[0067] Specifically, the recording time refers to the recording time of the personalized electromyographic signal, the environmental noise intensity refers to the environmental noise when recording the personalized electromyographic signal, the preset recording time refers to the preset value of the recording time that reflects the invalidity of the personalized electromyographic signal, the preset environmental noise intensity refers to the preset value of the environmental noise intensity that reflects the invalidity of the personalized electromyographic signal, and the preset recording environment coefficient refers to the preset value of the recording environment coefficient that reflects the invalidity of the personalized electromyographic signal.
[0068] Specifically, the validity of the user adaptability data is judged through step S3, and when the individual electromyographic signal is invalid, the user adaptability data is corrected and the individual electromyographic signal is deleted from the user adaptability data to improve the validity of the user adaptability data, to improve the accuracy of electromyographic signal analysis, and thus to further improve the efficiency of electromyographic signal analysis and processing.
[0069] Specifically, in step S4, when correcting the validity judgment result, the invalid data amount P is compared with the preset invalid data amount P0, and the high-frequency specific data is identified according to the comparison result, wherein:
[0070] When P<P0, it is determined that the individual electromyographic signal is not high-frequency specific data;
[0071] When P≥P0, the personalized electromyographic signal is determined to be high-frequency specific data, and the validity judgment results of all personalized electromyographic signals with the same interaction results corresponding to the personalized electromyographic signal are corrected to the personalized electromyographic signal being valid, and the preset number of groups of preset interaction results corresponding to the personalized electromyographic signal is increased.
[0072] Specifically, high-frequency specific data is identified through the step S4, and when the personalized electromyographic signal is high-frequency specific data, the validity judgment results of all personalized electromyographic signals with the same interaction results corresponding to the personalized electromyographic signal are corrected to the personalized electromyographic signal being valid, and the preset number of groups of preset interaction results corresponding to the personalized electromyographic signal is increased to adapt to the user's special usage habits, and the data collection intensity of the user's special usage habits is increased to improve the accuracy of electromyographic signal analysis, thereby further improving the efficiency of electromyographic signal analysis and processing.
[0073] Specifically, the invalid data amount refers to the total amount of invalid data of the personalized electromyographic signals in the user adaptability data, and the preset invalid data amount refers to the preset value of the data amount reflecting that the personalized electromyographic signals are high-frequency specific data. For example, the preset invalid data amount can be set to 70% of the total amount of personalized electromyographic signals in the user adaptability data.
[0074] Specifically, in step S4, when adding the preset number of groups of preset interaction results corresponding to the personalized electromyographic signal, the increase coefficient a is set to correct the preset number of groups n of the interaction results, and the corrected preset number of groups is na, na=n×a is set, and 1.5≤a<5 is set.
[0075] Specifically, in step S5, the user's electromyographic signal collected in real time is input into the user's personality model, and the interaction result is output.
[0076] Specifically, the interaction result refers to the user interaction awareness recognition result obtained based on the user's electromyography signal recognition, such as the user raising his hand, raising his hand and clasping his fist.
[0077] Specifically, in step S6, when the output of the interaction result is fed back for the first time, the number of active adjustments m by the user is compared with each preset number of active adjustments by the user, and the output of the interaction result is fed back for the first time according to the comparison result, wherein:
[0078] When m>m1, the interaction result output is determined to be abnormal, and optimization is performed;
[0079] When m2<m≤m1, the output of the interaction result is determined to be subject to secondary feedback;
[0080] When m≤m2, the interaction result output is determined to be normal;
[0081] m1 is the first preset number of times of active adjustment by the user, m2 is the second preset number of times of active adjustment by the user, and 0<m2<m1.
[0082] Specifically, the number of user-initiated adjustments refers to the total number of times the user actively adjusts the interaction results within a preset feedback cycle, such as the total number of times the user actively adjusts the interaction results within 10 days. The first preset number of user-initiated adjustments refers to the preset value of the number of user-initiated adjustments that reflects abnormal interaction result output. The second preset number of user-initiated adjustments refers to the preset value of the number of user-initiated adjustments that reflects normal interaction result output. For example, m1=20 and m2=10 can be set.
[0083] Specifically, in step S6, when performing secondary feedback on the output of the interaction result, the user satisfaction U is obtained, and the user satisfaction U is compared with the preset user satisfaction U0, and secondary feedback on the output of the interaction result is performed according to the comparison result, wherein:
[0084] When U≤U0, the interaction result output is determined to be abnormal, and optimization is performed;
[0085] When U>U0, it is determined that the interaction result output is normal.
[0086] Specifically, the user satisfaction refers to the user's satisfaction with the interaction results. This embodiment does not limit the method for obtaining user satisfaction. Technical personnel in the field can limit it according to actual conditions. Technical personnel in the field can freely set it according to actual conditions, such as by pushing the interaction window to the user terminal and obtaining the user's satisfaction with the interaction results. The preset user satisfaction refers to a preset value of user satisfaction indicating abnormal output of the interaction result.
[0087] Specifically, in step S7, after determining to optimize, when performing noise optimization on the user's electromyographic signal acquisition process, the electromyographic signal characteristics of the resting state are obtained, and the signal characteristics of the user's electromyographic signal acquired in real time are compared with the electromyographic signal characteristics of the resting state, and the electromyographic signal in the resting state is identified according to the comparison result, wherein:
[0088] When the characteristics of the user's electromyographic signal collected in real time are inconsistent with the characteristics of the electromyographic signal in the resting state, it is determined that the user's electromyographic signal is not in the resting state, and the electromyographic signal is input into a Butterworth bandpass filter for noise optimization to remove low-frequency artifacts and high-frequency noise;
[0089] When the characteristics of the user's electromyographic signal collected in real time are consistent with the characteristics of the electromyographic signal in the resting state, it is determined that the user's electromyographic signal is in the resting state and is deleted instead of being input into the user personality model for analysis and processing.
[0090] Specifically, the electromyographic signal in the resting state refers to the electromyographic signal indicating that the user is in a resting state. This embodiment does not limit the method for obtaining the electromyographic signal characteristics in the resting state. Technical personnel in this field can freely set it according to actual conditions. For example, it can be set to manually pre-enter the electromyographic signal of the user in a resting state. The Butterworth bandpass filter refers to a filter that performs noise optimization and removes low-frequency artifacts and high-frequency noise.
[0091] Specifically, in step S7, after determining to perform optimization, when performing feature extraction optimization on the user personality model, obtain the noise-optimized user electromyographic signal within the optimization time interval, and perform standard feature extraction on it to obtain standard data features, use the noise-optimized user electromyographic signal within the optimization time interval - standard data features as the feature extraction optimization data set, and use 70% of the feature extraction optimization data set as the feature extraction optimization training set, and 30% of the feature extraction optimization data set as the feature extraction optimization test set, train the convolutional neural network model according to the feature extraction optimization training set, and test the trained convolutional neural network model according to the feature extraction optimization test set. If the accuracy reaches 95%, output the tested convolutional neural network model as the feature extraction optimization model. If the accuracy does not reach 95%, continue to obtain the noise-optimized user electromyographic signal within the next optimization time interval to train the tested convolutional neural network model until the accuracy reaches 95%, and input the noise-optimized user electromyographic signal into the feature extraction optimization model for feature extraction.
[0092] Specifically, the optimization time interval refers to a preset time interval for optimization, such as the optimization time interval can be set to 5 days. The standard feature extraction refers to a feature extraction method that is pre-set for feature extraction. This embodiment does not limit the parameter setting of the convolutional neural network model. Technical personnel in this field can freely set it according to actual needs, as long as the recognition requirements for feature extraction are met.
[0093] See also Figure 2 As shown, it is a structural diagram of the electromyographic signal analysis and processing system based on big data in this embodiment, and the system includes:
[0094] Signal acquisition module, used to collect user's electromyographic signals in real time;
[0095] The personality difference analysis module is used to analyze the user's personality difference with the preset sample database, and to perform adaptive training on the preset electromyography analysis model according to the analysis results and output the user's personality model;
[0096] The validity judgment module is used to judge the validity of the user adaptability data according to the input environment coefficient, and to correct the user adaptability data according to the judgment result;
[0097] A specific data identification module is used to identify high-frequency specific data according to the amount of invalid data, and to correct the validity judgment result according to the identification result;
[0098] An interaction result output module, used to input the user's electromyographic signal into the user's personality model and output the interaction result;
[0099] The result feedback module is used to provide initial feedback on the output of the interaction result according to the number of active adjustments made by the user, and to provide secondary feedback on the output of the interaction result in combination with the user feedback, and to make optimization judgments based on the initial feedback results and the secondary feedback results;
[0100] The feedback optimization module is used to optimize the noise of the user's electromyographic signal acquisition process and to optimize the feature extraction of the user's personality model after the optimization is determined.
[0101] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
Claims
1. A method for analyzing and processing electromyographic signals based on big data, characterized in that: include: Step S1, real-time collection of user electromyographic signals; Step S2, analyzing the user's personality differences with the preset sample database, and adaptively training the preset electromyography analysis model according to the analysis results and outputting the user's personality model; Step S3, judging the validity of the user adaptability data according to the input environment coefficient, and correcting the user adaptability data according to the judgment result; Step S4, identifying the high-frequency specific data according to the amount of invalid data, and correcting the validity judgment result according to the identification result; Step S5, inputting the user's electromyographic signal into the user personality model and outputting the interaction result; Step S6, providing initial feedback on the output of the interaction result according to the number of times the user actively adjusts the interaction result, providing secondary feedback on the output of the interaction result in combination with the user feedback, and performing optimization judgment based on the initial feedback result and the secondary feedback result; Step S7, after determining to optimize, performing noise optimization on the user's electromyographic signal acquisition process, and performing feature extraction optimization on the user's personality model; In step S3, when judging the validity of the user adaptability data, the recording environment coefficient Q of each individual electromyographic signal is calculated according to the recording time T and the environmental noise intensity D, and Q=0.3×T / T0+0.7×D / D0 is set, T0 is the preset recording time, D0 is the preset environmental noise intensity, and the recording environment coefficient Q is compared with the preset recording environment coefficient Q0, and the validity of the user adaptability data is judged according to the comparison result, wherein: When Q≤Q0, it is determined that the individual electromyographic signal in the user adaptability data is valid; When Q>Q0, the personalized electromyographic signal in the user adaptability data is determined to be invalid, and the user adaptability data is corrected, and the personalized electromyographic signal is deleted from the user adaptability data; When correcting the validity judgment result, the invalid data amount P is compared with the preset invalid data amount P0, and the high-frequency specific data is identified according to the comparison result, wherein: When P<P0, it is determined that the individual electromyographic signal is not high-frequency specific data; When P≥P0, the personalized electromyographic signal is determined to be high-frequency specific data, and the validity judgment results of all personalized electromyographic signals with the same interaction results as the personalized electromyographic signal are corrected to be valid personalized electromyographic signals, and the preset number of groups of preset interaction results corresponding to the personalized electromyographic signal is increased; In step S4, when increasing the preset number of groups of preset interaction results corresponding to the personalized electromyographic signal, the increase coefficient a is set to correct the preset number of groups n of interaction results, the corrected preset number of groups is na, na=n×a is set, and 1.5≤a<5 is set.
2. The method for analyzing and processing myoelectric signals based on big data according to claim 1, characterized in that: In step S2, when performing the user personality difference analysis, the individual electromyographic signal in the user adaptability data is compared with the electromyographic signal in the preset sample database to obtain the vertical difference value A, the vertical difference value A is compared with the preset vertical difference value A0, and the user personality difference analysis is performed according to the comparison result, wherein: When A<A0, it is determined that the user's personality difference is normal, and the preset electromyography analysis model is not adaptively trained, and the preset electromyography analysis model is output as the user's personality model; When A≥A0, it is determined that the user's personality difference is abnormal, and the user's adaptability data is used as a difference training set to perform adaptive training on the preset electromyography analysis model, and the preset electromyography analysis model after adaptive training is output as the user's personality model.
3. The method for analyzing and processing myoelectric signals based on big data according to claim 1, characterized in that: In step S6, when the output of the interaction result is fed back for the first time, the number of active adjustments m by the user is compared with each preset number of active adjustments by the user, and the output of the interaction result is fed back for the first time according to the comparison result, wherein: When m>m1, the interaction result output is determined to be abnormal, and optimization is performed; When m2<m≤m1, the output of the interaction result is determined to be subject to secondary feedback; When m≤m2, the interaction result output is determined to be normal; m1 is the first preset number of times of active adjustment by the user, m2 is the second preset number of times of active adjustment by the user, and 0<m2<m1.
4. The method for analyzing and processing myoelectric signals based on big data according to claim 3 is characterized in that: In step S6, when performing secondary feedback on the output of the interaction result, the user satisfaction U is obtained, and the user satisfaction U is compared with the preset user satisfaction U0, and secondary feedback on the output of the interaction result is performed according to the comparison result, wherein: When U≤U0, the interaction result output is determined to be abnormal, and optimization is performed; When U>U0, it is determined that the interaction result output is normal.
5. The method for analyzing and processing myoelectric signals based on big data according to claim 4 is characterized in that: In step S7, after determining to optimize, when performing noise optimization on the user's electromyographic signal acquisition process, the electromyographic signal characteristics of the resting state are obtained, and the signal characteristics of the user's electromyographic signal acquired in real time are compared with the electromyographic signal characteristics of the resting state, and the electromyographic signal in the resting state is identified according to the comparison result, wherein: When the characteristics of the user's electromyographic signal collected in real time are inconsistent with the characteristics of the electromyographic signal in the resting state, it is determined that the user's electromyographic signal is not in the resting state, and the electromyographic signal is input into a Butterworth bandpass filter for noise optimization to remove low-frequency artifacts and high-frequency noise; When the characteristics of the user's electromyographic signal collected in real time are consistent with the characteristics of the electromyographic signal in the resting state, it is determined that the user's electromyographic signal is in the resting state and is deleted instead of being input into the user personality model for analysis and processing.
6. The method for analyzing and processing myoelectric signals based on big data according to claim 5, characterized in that: In the step S7, after determining to perform optimization, when performing feature extraction optimization on the user personality model, obtain the noise-optimized user electromyographic signal within the optimization time interval, perform standard feature extraction on it to obtain standard data features, use the noise-optimized user electromyographic signal within the optimization time interval - standard data features as the feature extraction optimization data set, and use 70% of the feature extraction optimization data set as the feature extraction optimization training set, and 30% of the feature extraction optimization data set as the feature extraction optimization test set, train the convolutional neural network model according to the feature extraction optimization training set, and test the trained convolutional neural network model according to the feature extraction optimization test set. If the accuracy reaches 95%, output the tested convolutional neural network model as the feature extraction optimization model. If the accuracy does not reach 95%, continue to obtain the noise-optimized user electromyographic signal within the next optimization time interval to train the tested convolutional neural network model until the accuracy reaches 95%, and input the noise-optimized user electromyographic signal into the feature extraction optimization model for feature extraction.
7. A system using the method for analyzing and processing myoelectric signals based on big data as described in any one of claims 1 to 6, characterized in that: include: Signal acquisition module, used to collect user's electromyographic signals in real time; The personality difference analysis module is used to analyze the user's personality difference with the preset sample database, and to perform adaptive training on the preset electromyography analysis model according to the analysis results and output the user's personality model; The validity judgment module is used to judge the validity of the user adaptability data according to the input environment coefficient, and to correct the user adaptability data according to the judgment result; A specific data identification module is used to identify high-frequency specific data according to the amount of invalid data, and to correct the validity judgment result according to the identification result; An interaction result output module, used to input the user's electromyographic signal into the user's personality model and output the interaction result; The result feedback module is used to provide initial feedback on the output of the interaction result according to the number of active adjustments made by the user, and to provide secondary feedback on the output of the interaction result in combination with the user feedback, and to make optimization judgments based on the initial feedback results and the secondary feedback results; The feedback optimization module is used to optimize the noise of the user's electromyographic signal acquisition process and to optimize the feature extraction of the user's personality model after the optimization is determined.
