CN114035679B - Brain nerve signal parallel decoding method and device capable of dynamically recombining - Google Patents
Brain nerve signal parallel decoding method and device capable of dynamically recombining Download PDFInfo
- Publication number
- CN114035679B CN114035679B CN202111246696.2A CN202111246696A CN114035679B CN 114035679 B CN114035679 B CN 114035679B CN 202111246696 A CN202111246696 A CN 202111246696A CN 114035679 B CN114035679 B CN 114035679B
- Authority
- CN
- China
- Prior art keywords
- decoding
- signal
- parallel
- subsystem
- brain nerve
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 210000005036 nerve Anatomy 0.000 title claims abstract description 91
- 210000004556 brain Anatomy 0.000 title claims abstract description 82
- 238000000034 method Methods 0.000 title claims abstract description 18
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 106
- 210000003792 cranial nerve Anatomy 0.000 claims abstract description 48
- 238000004364 calculation method Methods 0.000 claims abstract description 30
- 238000004458 analytical method Methods 0.000 claims abstract description 27
- 230000000638 stimulation Effects 0.000 claims abstract description 11
- 230000003993 interaction Effects 0.000 claims abstract description 9
- 238000004891 communication Methods 0.000 claims description 24
- 230000002452 interceptive effect Effects 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 238000004806 packaging method and process Methods 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000007635 classification algorithm Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 4
- 230000001537 neural effect Effects 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 241000414697 Tegra Species 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000006798 recombination Effects 0.000 description 2
- 238000005215 recombination Methods 0.000 description 2
- HPTJABJPZMULFH-UHFFFAOYSA-N 12-[(Cyclohexylcarbamoyl)amino]dodecanoic acid Chemical compound OC(=O)CCCCCCCCCCCNC(=O)NC1CCCCC1 HPTJABJPZMULFH-UHFFFAOYSA-N 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002490 cerebral effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Neurosurgery (AREA)
- General Health & Medical Sciences (AREA)
- Neurology (AREA)
- Health & Medical Sciences (AREA)
- Dermatology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a brain nerve signal parallel decoding device and method capable of dynamically recombining, comprising the following steps: the interaction control platform issues control commands and configuration parameters, and receives a cranial nerve signal decoding result fed back by the calculation main control subsystem; the calculation main control subsystem analyzes the cranial nerve signals according to the control command and the configuration parameters, generates a decoding task according to the analysis result, and dynamically distributes the decoding task to the parallel calculation subsystem according to the load condition of the parallel calculation subsystem; receiving a cranial nerve signal decoding result fed back by the parallel computing subsystem; the parallel computing subsystem dynamically reorganizes to the computing main control subsystem, generates a decoding algorithm chain adapted to the analysis task according to the decoding task, and executes the decoding task by using the decoding algorithm chain to obtain a brain nerve signal decoding result; the nerve feedback device converts the brain nerve signal decoding result obtained from the calculation main control subsystem into a stimulation signal and feeds back the stimulation signal to act on the acquisition object, so that the high-performance parallel decoding of the brain nerve signal is realized.
Description
Technical Field
The invention relates to the field of brain-computer interfaces, in particular to a brain nerve signal parallel decoding method and device capable of dynamically recombining.
Background
Brain-computer interface technology is a bridge for brain science and information science research, wherein signal analysis and output control of brain nerve signals are an indispensable part of the brain-computer interface technology. However, the following problems are generally existed in the cranial nerve signal decoding and control feedback system on the market at present:
(1) The type of the brain electrical signals which can be decoded by a single system is single, and a plurality of different brain nerve signals can not be processed simultaneously by the same system; (2) The brain nerve signal decoding capability of a single system is limited, and the performance bottleneck of the maximum channel number of the nerve signal exists; (3) The decoding system is difficult to update iteration and can not realize dynamic updating of the decoding algorithm library; (4) The decoding configuration is complicated, and a large number of parameters need to be modified when the signal data source is switched or the signal data type is modified; (5) The decoding algorithm has poor flexibility, and the decoding process cannot be flexibly stretched according to specific requirements.
The prior patent document CN102426661a discloses a neural decoding device based on FPGA, comprising: the method is based on a parallel processing calculation mode of the FPGA, so that the decoding speed of the nerve signals is improved, but the common problems of the cerebral nerve signal decoding and control feedback system on the market are still faced.
The prior patent document CN103338368A discloses a JPEG parallel decoding device based on an FPGA, which comprises a data buffer unit, a data preprocessing unit and a parallel decoding unit, wherein the data buffer unit comprises an input data buffer module for receiving external JPEG signals and an output data buffer module for outputting decoded JPEG signals, the communication end of the input data buffer module is connected with the communication end of the data preprocessing unit, the communication end of the data preprocessing unit is connected with the signal input end of the parallel decoding unit, the signal output end of the parallel decoding unit is connected with the signal input end of the output data buffer module, RSTi (reset mark) and APPn (comment field) in the JPEG standard are fully utilized, the decoding of JPEG is realized, and parallel decoding is supported, but the device still faces the common problems of the above-mentioned cranial nerve signal decoding and control feedback system on the market.
Disclosure of Invention
In view of the above, the present invention aims to provide a method and an apparatus for parallel decoding of brain nerve signals capable of dynamic recombination, which achieve high performance decoding of brain nerve signals and rapid updating and switching of decoding algorithms by a design of a distributed decoding part of dynamic recombination.
In order to achieve the above purpose, the technical scheme provided by the invention is as follows:
in a first aspect, an embodiment provides a brain nerve signal parallel decoding device capable of dynamically recombining, which includes an interactive control platform, a computing main control subsystem, a plurality of parallel computing subsystems, and a nerve feedback device;
the interactive control platform is used for issuing control commands and configuration parameters to the calculation main control subsystem and receiving a cranial nerve signal decoding result fed back by the calculation main control subsystem;
the computing main control subsystem is used for analyzing the received cranial nerve signals according to the control command and the configuration parameters, generating decoding tasks according to the analysis results, and dynamically distributing the decoding tasks to the parallel computing subsystem according to the load conditions of the parallel computing subsystem; the brain nerve signal decoding result fed back by the parallel computing subsystem is also received;
the parallel computing subsystem is dynamically recombined to the computing main control subsystem and is used for generating a decoding algorithm chain matched with the analysis task according to the decoding task, and executing the decoding task by using the decoding algorithm chain to obtain a brain nerve signal decoding result;
the nerve feedback device is used for converting the brain nerve signal decoding result into a stimulation signal after the brain nerve signal decoding result is obtained from the calculation main control subsystem and feeding back the stimulation signal to the acquisition object.
In one embodiment, the computing main control subsystem comprises a data receiving module, an intelligent distribution module, a first communication module and a command processing module;
the data receiving module is used for receiving on-line or off-line brain nerve signals and analyzing the on-line or off-line brain nerve signals to obtain analysis results, wherein the analysis results comprise signal types and signal channel numbers;
the intelligent distribution module is used for constructing a decoding task for the analysis result, and dynamically distributing the decoding task to the matched and relatively idle parallel computing subsystem according to the respective load conditions of the current online parallel computing subsystems;
the first communication module is used for being responsible for being communicated with the interaction control platform, the parallel computing subsystem and the nerve feedback device network respectively;
the command processing module is used for processing the received control command and configuration parameters so as to maintain the work of the calculation main control subsystem and the parallel calculation subsystem.
In one embodiment, the data receiving module, upon receiving the cranial nerve signal, time stamps the cranial nerve signal according to the current system time; and when the signal type is not obtained by analyzing the cranial nerve signal, obtaining the manually marked signal type from the interactive control platform.
In one embodiment, the data receiving module dynamically adjusts the analysis method according to the data source type when analyzing the cranial nerve signals, wherein the analysis is performed by adopting a related cranial nerve signal classification algorithm aiming at the cranial nerve signals collected on line in real time; aiming at the local off-line brain nerve signals, an analysis result is obtained from the brain nerve signal storage file according to the file format, the file content and the file header information.
In one embodiment, the dynamic allocation procedure of the decoding task implemented by the intelligent distribution module includes:
firstly, determining a decoding algorithm to be adopted according to a signal type, and packaging a cranial nerve signal, the signal type, the decoding algorithm and a time stamp into a decoding task;
then, according to the decoding algorithm, the number of signal channels and the respective load conditions of the on-line parallel computing subsystems, searching the parallel computing subsystems which can be matched and executed with the decoding algorithm and are relatively idle and distributing and sending decoding tasks.
In one embodiment, the parallel computing subsystem comprises a second communication module, a parameter configuration module, a signal decoding module and a decoding algorithm library module;
the second communication module is used for carrying out network communication with the computing main control subsystem to realize command interaction and data transmission, and the data comprises a decoding task, configuration parameters and a decoding result;
the parameter configuration module is used for configuring system parameters and decoding algorithm parameters of the parallel computing subsystem according to the configuration parameters acquired from the computing main control subsystem;
the decoding algorithm library module is used for maintaining decoding algorithms of various brain nerve signals and realizing real-time switching and updating of the decoding algorithms;
and the signal decoding module is used for executing the decoding task by adopting a decoding algorithm chain corresponding to the decoding algorithm and the signal type according to the signal type and the decoding algorithm contained in the decoding task after analyzing the decoding task according to the acquired configuration parameters, so as to finish the parallel decoding work of the brain nerve signals.
In one embodiment, in the signal decoding module, after the decoding task is parsed to obtain the signal type and the decoding algorithm, whether a decoding algorithm chain matched with the signal type and the decoding algorithm exists is judged, if yes, the decoding task is executed by using the existing decoding algorithm chain; if not, acquiring corresponding decoding algorithms from the decoding algorithm library to load a plurality of brain nerve signal decoding executors to form a new decoding algorithm chain, and executing decoding tasks by using the new decoding algorithm chain to finish chain decoding work of brain nerve signal data.
In one embodiment, in the decoding algorithm library module, each decoding algorithm is packaged as a dynamic library with a unified interface, and the decoding algorithm library is dynamically loaded, so that real-time switching and updating of the decoding algorithm are realized.
In one embodiment, the decoding result obtained by the parallel computing subsystem executing the decoding task is fed back to the computing main control subsystem, and the computing main control subsystem updates the load condition of the parallel computing subsystem and removes the decoding task record after storing the decoding result.
In a second aspect, an embodiment further provides a method for parallel decoding of brain nerve signals capable of being dynamically recombined, wherein the method for parallel decoding of brain nerve signals is implemented by adopting the device for parallel decoding of brain nerve signals capable of being dynamically recombined.
Compared with the prior art, the beneficial effects that have include at least:
the invention applies the distributed computation to the decoding of the cranial nerve signals, realizes a cranial nerve signal decoding system with high performance, high utilization rate and dynamic adjustment of calculation force through intelligent distribution regulation and control of decoding tasks, realizes closed-loop control and real-time monitoring of the cranial nerve signals through a stimulation feedback device and an interactive control platform, and simultaneously adopts the chain design of the dynamic loading and decoding algorithm of a decoding algorithm library to improve the flexibility of the updating of the decoding algorithm and the decoding process of the cranial nerve signals.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic structural diagram of a brain nerve signal parallel decoding device capable of dynamically recombining according to an embodiment;
FIG. 2 is a schematic diagram of a computing master subsystem according to an embodiment;
FIG. 3 is a flow chart of intelligent distribution of decoding tasks provided by an embodiment;
FIG. 4 is a schematic diagram of decoding task construction and allocation according to an embodiment;
FIG. 5 is a schematic diagram of a parallel computing subsystem according to an embodiment;
fig. 6 is a flowchart of cranial nerve signal decoding according to an embodiment.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the detailed description is presented by way of example only and is not intended to limit the scope of the invention.
Fig. 1 is a schematic structural diagram of a brain nerve signal parallel decoding device capable of dynamically recombining according to an embodiment. As shown in fig. 1, the cranial nerve signal parallel decoding device provided by the embodiment includes four devices including an interactive control platform, a computing main control subsystem, a plurality of parallel computing subsystems and a nerve feedback device, and the devices are communicated by adopting an information exchange network composed of wired routers.
The interaction control platform is used for issuing control commands and configuration parameters to the calculation main control subsystem and receiving a cranial nerve signal decoding result fed back by the calculation main control subsystem; the computing main control subsystem is used for analyzing the received brain nerve signals according to the control command and the configuration parameters, generating a decoding task according to the analysis result, and dynamically distributing the decoding task to the parallel computing subsystem according to the load condition of the parallel computing subsystem; the brain nerve signal decoding result fed back by the parallel computing subsystem is also received; the parallel computing subsystem is dynamically recombined to the computing main control subsystem and is used for generating a decoding algorithm chain matched with the analysis task according to the decoding task, and executing the decoding task by using the decoding algorithm chain to obtain a brain nerve signal decoding result; the nerve feedback device is used for converting the brain nerve signal decoding result into a stimulation signal after the brain nerve signal decoding result is obtained from the calculation main control subsystem and feeding back the stimulation signal to act on the acquisition object. In this way, the brain nerve signal closed-loop control of acquisition-decoding-feedback is formed by the calculation main control subsystem, the parallel calculation subsystem and the stimulation feedback device, meanwhile, the interactive control platform adopts a desktop computer of a Windows system, and can carry out parameter configuration on the calculation main control subsystem according to requirements, receive a decoding result sent by the calculation main control subsystem, display the decoding result on a software interface, and monitor and control the decoding process.
Fig. 2 is a schematic structural diagram of a computing master subsystem according to an embodiment. As shown in fig. 2, the computing master control subsystem provided in the embodiment may be implemented based on a GPU Tegra xavir platform, and includes a data receiving module, an intelligent distributing module, a first communication module, and a command processing module.
The data receiving module is used for receiving the on-line or off-line cranial nerve signals and analyzing to obtain analysis results; the intelligent distribution module is used for constructing a decoding task for the analysis result, and dynamically distributing the decoding task to the matched and relatively idle parallel computing subsystem according to the respective load conditions of the current online parallel computing subsystems; the first communication module is used for being in charge of communicating with the interaction control platform, the parallel computing subsystem and the neural feedback device network respectively; the command processing module is used for processing the received control command and the configuration parameter command and making corresponding reactions so as to maintain the work of the calculation main control subsystem and the parallel calculation subsystem.
To ensure reliability of communication, all network communications are based on TCP reliable communications. The first network communication module continuously monitors whether a new parallel computing subsystem is on line, and when one new parallel computing subsystem logs in the computing main control subsystem, the computing main control subsystem regards the new parallel computing subsystem as a resource for parallel decoding.
In the embodiment, the data receiving module can receive the online real-time cranial nerve signals acquired by the cranial nerve signal acquisition system, and can also select to receive the offline cranial nerve signals stored in the local storage file through the control command by the interactive control platform, based on the offline cranial nerve signals, when the data receiving module analyzes the cranial nerve signals, the data receiving module dynamically adjusts the analysis method according to the data source type, analyzes the real-time online acquired cranial nerve signals by adopting a related cranial nerve signal classification algorithm, specifically, extracts and analyzes the characteristics of the cranial nerve signals by adopting a Principal Component Analysis (PCA) algorithm, and automatically clusters the cranial nerve potentials by adopting a K-means (K-means), thereby classifying and analyzing the cranial nerve signals. And aiming at the local off-line brain nerve signals, acquiring the data information of the stored signals from the brain nerve signal storage file according to file formats, file contents, file header information and the like so as to obtain analysis results.
In an embodiment, the analysis results include signal type, number of signal channels. The signal types are used as the brain nerve signal type distinction and are used for guiding the selection of the corresponding analysis algorithm. The number of signal channels characterizes the data size and is used for guiding the allocation problem of the decoding task relative to the decoding executor and the decoding algorithm chain.
In an embodiment, when receiving a cranial nerve signal, the data receiving module marks a timestamp for the cranial nerve signal according to the current system time; and when the signal type is not obtained by analyzing the cranial nerve signal, obtaining the manually marked signal type from the interactive control platform.
In an embodiment, based on the above timestamp marking and signal type marking, the dynamic allocation flow of the decoding task implemented by the intelligent distribution module includes:
firstly, determining a decoding algorithm to be adopted according to a signal type, and packaging a cranial nerve signal, the signal type, the decoding algorithm and a time stamp into a decoding task;
then, according to the decoding algorithm, the number of signal channels and the respective load conditions of the on-line parallel computing subsystems, searching the parallel computing subsystems which can be matched and executed with the decoding algorithm and are relatively idle and distributing and sending decoding tasks.
The intelligent distribution module realizes six neural decoding algorithms based on GPU acceleration, namely Monte Carlo Point Passing (MCPP), linear point process (Linear PP), exponential point process (expPP), kalman (Kalman), generalized Regression Neural Network (GRNN) and Support Vector Regression (SVR). The user can adopt one or more decoding algorithms to analyze according to the actual requirements.
After the decoding task is finished, the decoding result obtained by the parallel computing subsystem executing the decoding task is fed back to the computing main control subsystem, and the computing main control subsystem updates the load condition of the parallel computing subsystem and removes the decoding task record after storing the decoding result.
In the embodiment, the relative idleness is a relative concept, and the parallel computing subsystem is considered to have less load, that is, the decoding task is less, and the parallel computing subsystem with more decoding tasks is idle. The decoding algorithm is corresponding to the signal type, and the decoding algorithms existing in each parallel computing subsystem may be different, so that a process of matching the decoding algorithm included in one decoding task with the decoding algorithm included in the parallel computing subsystem is also required when the decoding task is allocated, and the rapid and effective allocation of the decoding task is realized.
Fig. 3 is a flowchart of intelligent distribution of decoding tasks according to an embodiment. As shown in fig. 3, the computing main control subsystem records the current assigned decoding task number of each parallel computing subsystem, when the computing main control subsystem receives a group of complete cranial nerve signal data, firstly, determining the type of the cranial nerve signal in a proper mode according to the receiving source mode, marking a timestamp for the data according to the current system time, then, determining the available parallel computing subsystem with the minimum current decoding task number according to the type of the cranial nerve signal and the decoding algorithm flow required to be adopted, and then, packaging the cranial nerve signal data, the type, the algorithm flow and the timestamp into decoding tasks and sending the decoding tasks to the parallel computing subsystem. Finally, the relevant information of the task is added in the assigned task table of the parallel computing subsystem, and the process can be represented by fig. 4. After a certain parallel computing subsystem completes a certain decoding task, the decoding result is added with a time stamp, an algorithm flow and an original data type and returned to the computing main control subsystem, the computing main control subsystem feeds back the decoding result to the stimulation feedback device and the interaction control platform, and the decoding task is removed from a corresponding task table, so that the parallel computing subsystem completes the decoding task.
Fig. 5 is a schematic structural diagram of a parallel computing subsystem according to an embodiment. As shown in fig. 5, the parallel computing subsystem provided in the embodiment may be implemented based on an inflight GPU Tegra xakier platform, and includes a second communication module, a parameter configuration module, a signal decoding module, and a decoding algorithm library module.
The second communication module is used for carrying out network communication with the computing main control subsystem, namely establishing TCP reliable communication, realizing command interaction and data transmission, and the data comprises decoding tasks, configuration parameters, decoding results and the like; when the parallel computing subsystem is started, connection is actively established with the computing main control subsystem, and information such as decoding algorithm supported by the computing main control subsystem is provided for the computing main control subsystem.
The parameter configuration module is used for configuring system parameters and decoding algorithm parameters of the parallel computing subsystem according to the configuration parameters acquired from the computing main control subsystem. The decoding algorithm library module is used for maintaining decoding algorithms of various brain nerve signals and realizing real-time switching and updating of the decoding algorithms. The brain nerve signal type comprises surface brain electrical signal (EEG), spike potential signal (Spike) and local field potential model (LFP). In the embodiment, each decoding algorithm is packaged into a dynamic library with a unified interface, and the dynamic library is dynamically loaded, so that the real-time switching and updating of the decoding algorithm are realized.
The signal decoding module utilizes the CUDA technology to load a plurality of brain nerve signal decoding algorithms to form a decoding algorithm chain according to the configuration parameters of the main control subsystem, and the high-performance decoding work of brain nerve signals is completed. In the embodiment, after the signal decoding module analyzes the decoding task according to the acquired configuration parameters, the decoding task is executed by adopting a decoding algorithm chain corresponding to the decoding algorithm and the signal type according to the signal type and the decoding algorithm contained in the decoding task, so as to complete the parallel decoding work of the brain nerve signals. In the embodiment, after the decoding task is analyzed to obtain the signal type and the decoding algorithm, judging whether a decoding algorithm chain matched with the signal type and the decoding algorithm exists at present, and if so, executing the decoding task by using the existing decoding algorithm chain; if not, acquiring corresponding decoding algorithms from the decoding algorithm library to load a plurality of brain nerve signal decoding executors to form a new decoding algorithm chain, and executing decoding tasks by using the new decoding algorithm chain to finish chain decoding work of brain nerve signal data.
Fig. 6 is a flowchart of cranial nerve signal decoding according to an embodiment. As shown in fig. 6, the computing main control subsystem packages the cranial nerve signals, the signal types, the algorithm types and the like into decoding tasks and sends the decoding tasks to the parallel computing subsystem, after the parallel computing subsystem receives the signals, the types of the cranial nerve signals in the decoding tasks and the algorithm flow needed to be used are firstly analyzed, then whether the signal decoding module has constructed corresponding decoding executors or not is determined, if not, the missing decoding executors are constructed according to the decoding task information, a decoding algorithm chain is formed, and then the data of the cranial nerve signals are decoded by using the decoding algorithm chain. After the decoding result is obtained, the decoding result is fed back to the calculation main control subsystem through the network communication module.
The device for parallel decoding of the brain nerve signals capable of being dynamically recombined provided by the embodiment adopts a distributed technology, the decoding algorithm of the brain nerve signals is respectively deployed in each parallel computing subsystem, the computing main control subsystem is only responsible for identifying the type and the channel number of the received brain nerve signals and then distributing and receiving the decoding result, each parallel computing subsystem receives the brain nerve signals and carries out hierarchical analysis on the brain nerve signals in a chained decoding mode, the decoding algorithm of the parallel computing subsystem can be quickly updated through a decoding algorithm library, the number of the parallel computing subsystems in the system can be dynamically adjusted and recombined according to specific requirements, so that the problem of computation bottleneck when the number of the brain nerve signal channels is large is solved, and the problem of calculation redundancy of the brain nerve signal channels in hours caused by single system specification is avoided.
The foregoing detailed description of the preferred embodiments and advantages of the invention will be appreciated that the foregoing description is merely illustrative of the presently preferred embodiments of the invention, and that no changes, additions, substitutions and equivalents of those embodiments are intended to be included within the scope of the invention.
Claims (8)
1. The brain nerve signal parallel decoding device capable of dynamically recombining is characterized by comprising an interactive control platform, a calculation main control subsystem, a plurality of parallel calculation subsystems and a nerve feedback device;
the interactive control platform is used for issuing control commands and configuration parameters to the calculation main control subsystem and receiving a cranial nerve signal decoding result fed back by the calculation main control subsystem;
the computing main control subsystem is used for analyzing the received cranial nerve signals according to the control command and the configuration parameters, generating decoding tasks according to the analysis results, and dynamically distributing the decoding tasks to the parallel computing subsystem according to the load conditions of the parallel computing subsystem; the brain nerve signal decoding result fed back by the parallel computing subsystem is also received;
the parallel computing subsystem is dynamically recombined to the computing main control subsystem and is used for generating a decoding algorithm chain matched with the analysis task according to the decoding task, and executing the decoding task by using the decoding algorithm chain to obtain a brain nerve signal decoding result;
the nerve feedback device is used for converting the brain nerve signal decoding result into a stimulation signal after the brain nerve signal decoding result is obtained from the calculation main control subsystem and feeding back the stimulation signal to act on an acquisition object;
the parallel computing subsystem comprises a second communication module, a parameter configuration module, a signal decoding module and a decoding algorithm library module; the second communication module is used for carrying out network communication with the computing main control subsystem to realize command interaction and data transmission, and the data comprises a decoding task, configuration parameters and a decoding result; the parameter configuration module is used for configuring system parameters and decoding algorithm parameters of the parallel computing subsystem according to the configuration parameters acquired from the computing main control subsystem; the decoding algorithm library module is used for maintaining decoding algorithms of various brain nerve signals and realizing real-time switching and updating of the decoding algorithms; the signal decoding module is used for executing a decoding task by adopting a decoding algorithm chain corresponding to the decoding algorithm and the signal type according to the signal type and the decoding algorithm contained in the decoding task after analyzing the decoding task according to the acquired configuration parameters, so as to complete the parallel decoding work of the brain nerve signals;
in the signal decoding module, after analyzing the decoding task to obtain a signal type and a decoding algorithm, judging whether a decoding algorithm chain matched with the signal type and the decoding algorithm exists at present, and if so, executing the decoding task by using the existing decoding algorithm chain; if not, acquiring corresponding decoding algorithms from the decoding algorithm library to load a plurality of brain nerve signal decoding executors to form a new decoding algorithm chain, and executing decoding tasks by using the new decoding algorithm chain to finish chain decoding work of brain nerve signal data.
2. The device for parallel decoding of brain nerve signals capable of being dynamically recombined according to claim 1, wherein the computing main control subsystem comprises a data receiving module, an intelligent distribution module, a first communication module and a command processing module;
the data receiving module is used for receiving on-line or off-line brain nerve signals and analyzing the on-line or off-line brain nerve signals to obtain analysis results, wherein the analysis results comprise signal types and signal channel numbers;
the intelligent distribution module is used for constructing a decoding task for the analysis result, and dynamically distributing the decoding task to the matched and relatively idle parallel computing subsystem according to the respective load conditions of the current online parallel computing subsystems;
the first communication module is used for being responsible for being communicated with the interaction control platform, the parallel computing subsystem and the nerve feedback device network respectively;
the command processing module is used for processing the received control command and configuration parameters so as to maintain the work of the calculation main control subsystem and the parallel calculation subsystem.
3. The dynamically reconfigurable brain nerve signal parallel decoding device according to claim 2, wherein the data receiving module, when receiving the brain nerve signal, marks a time stamp for the brain nerve signal according to the current system time; and when the signal type is not obtained by analyzing the cranial nerve signal, obtaining the manually marked signal type from the interactive control platform.
4. The brain nerve signal parallel decoding device capable of dynamically recombining according to claim 2, wherein the data receiving module dynamically adjusts an analysis method according to a data source type when analyzing brain nerve signals, wherein the brain nerve signals collected on line in real time are analyzed by adopting a related brain nerve signal classification algorithm; aiming at the local off-line brain nerve signals, an analysis result is obtained from the brain nerve signal storage file according to the file format, the file content and the file header information.
5. The brain nerve signal parallel decoding device capable of being dynamically recombined according to claim 2, wherein the dynamic allocation flow of the decoding task realized by the intelligent distribution module comprises:
firstly, determining a decoding algorithm to be adopted according to a signal type, and packaging a cranial nerve signal, the signal type, the decoding algorithm and a time stamp into a decoding task;
then, according to the decoding algorithm, the number of signal channels and the respective load conditions of the on-line parallel computing subsystems, searching the parallel computing subsystems which can be matched and executed with the decoding algorithm and are relatively idle and distributing and sending decoding tasks.
6. The device for parallel decoding of brain nerve signals capable of being dynamically recombined according to claim 1, wherein each decoding algorithm in the decoding algorithm library module is packaged into a dynamic library with a unified interface, and the decoding algorithm library is dynamically loaded, so that real-time switching and updating of the decoding algorithm are realized.
7. The brain nerve signal parallel decoding device capable of dynamically reorganizing according to claim 1, wherein a decoding result obtained by the parallel computing subsystem executing a decoding task is fed back to the computing main control subsystem, and the computing main control subsystem updates a load condition of the parallel computing subsystem and removes a decoding task record after storing the decoding result.
8. A method for parallel decoding of brain nerve signals capable of being dynamically recombined, which is characterized in that the method adopts the device for parallel decoding of brain nerve signals capable of being dynamically recombined as claimed in any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111246696.2A CN114035679B (en) | 2021-10-26 | 2021-10-26 | Brain nerve signal parallel decoding method and device capable of dynamically recombining |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111246696.2A CN114035679B (en) | 2021-10-26 | 2021-10-26 | Brain nerve signal parallel decoding method and device capable of dynamically recombining |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114035679A CN114035679A (en) | 2022-02-11 |
CN114035679B true CN114035679B (en) | 2023-11-17 |
Family
ID=80135421
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111246696.2A Active CN114035679B (en) | 2021-10-26 | 2021-10-26 | Brain nerve signal parallel decoding method and device capable of dynamically recombining |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114035679B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004289745A (en) * | 2003-03-25 | 2004-10-14 | Toshiba Corp | Moving picture decoding method and device |
CN105488566A (en) * | 2015-12-10 | 2016-04-13 | 浙江大学 | VPX bus based brain neural signal real-time parallel processing system |
CN105491377A (en) * | 2015-12-15 | 2016-04-13 | 华中科技大学 | Video decoding macro-block-grade parallel scheduling method for perceiving calculation complexity |
CN112083707A (en) * | 2020-08-05 | 2020-12-15 | 深圳市永达电子信息股份有限公司 | Industrial control physical signal processing method, controller and processing system |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP4100816A4 (en) * | 2020-02-04 | 2023-07-05 | HRL Laboratories, LLC | System and method for asynchronous brain control of one or more tasks |
-
2021
- 2021-10-26 CN CN202111246696.2A patent/CN114035679B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004289745A (en) * | 2003-03-25 | 2004-10-14 | Toshiba Corp | Moving picture decoding method and device |
CN105488566A (en) * | 2015-12-10 | 2016-04-13 | 浙江大学 | VPX bus based brain neural signal real-time parallel processing system |
CN105491377A (en) * | 2015-12-15 | 2016-04-13 | 华中科技大学 | Video decoding macro-block-grade parallel scheduling method for perceiving calculation complexity |
CN112083707A (en) * | 2020-08-05 | 2020-12-15 | 深圳市永达电子信息股份有限公司 | Industrial control physical signal processing method, controller and processing system |
Non-Patent Citations (1)
Title |
---|
基于异构多核嵌入式平台的视频监控系统;尹雷;卿粼波;滕奇志;付雄;;微电子学与计算机(第05期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN114035679A (en) | 2022-02-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109933306B (en) | Self-adaptive hybrid cloud computing framework generation method based on operation type recognition | |
CN112039987B (en) | Method and device for processing blocks in block chain, node equipment and storage medium | |
CN107239675A (en) | Biological information analysis system based on cloud platform | |
CN108989358B (en) | Method for acquiring running data of textile machine based on TCP/IP protocol | |
CN103684898A (en) | Method and device for monitoring operation of user request in distributed system | |
CA2386272A1 (en) | Collaborative design | |
CN112527848B (en) | Report data query method, device and system based on multiple data sources and storage medium | |
CN108416657B (en) | Order generation method and equipment based on consultation service | |
CN106168993B (en) | Real-time simulation analysis system for power grid | |
CN105096181A (en) | E-commerce transaction method and E-commerce transaction system for big data | |
CN111445135A (en) | Intelligent maintenance method and device, computer equipment and storage medium | |
CN115511086B (en) | Distributed reasoning deployment system for oversized model | |
CN111444158A (en) | Long-short term user portrait generation method, device, equipment and readable storage medium | |
CN115470915A (en) | Server system of quantum computer and implementation method thereof | |
CN114035679B (en) | Brain nerve signal parallel decoding method and device capable of dynamically recombining | |
CN106817262A (en) | A kind of log analysis device | |
CN109800090A (en) | A kind of data integrated system and method | |
CN110320837A (en) | It is a kind of can edge calculations industrial computer | |
CN112346980B (en) | Software performance testing method, system and readable storage medium | |
CN111562922A (en) | Method, system and electronic equipment for modularizing command line program and cloud-end method and system | |
CN116821114A (en) | Computer for big data statistics | |
US11687377B1 (en) | High performance computer with a control board, modular compute boards and resource boards that can be allocated to the modular compute boards | |
CN112883110A (en) | Terminal big data distribution method, storage medium and system based on NIFI | |
CN113032522A (en) | Text parallel data mining system and method | |
CN107766497A (en) | The method and terminal of Data Collection based on container |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |