Vestibular illusion training method, device and electronic equipment

Abstract

The present invention relates to the field of vestibular illusion training technology, and in particular to a vestibular illusion training method, device and electronic device. Obtain preset test control instructions corresponding to preset vestibular illusion training types and preset vestibular illusion training intensity; perform vestibular illusion test on a subject based on the preset test control instructions, and obtain motion parameters and physiological parameters corresponding to the subject; identify the motion parameters to determine the target vestibular illusion recognition type corresponding to the subject; identify the physiological parameters to determine the target vestibular illusion recognition intensity corresponding to the subject; compare the target vestibular illusion recognition type with the preset vestibular illusion training type, and compare the target vestibular illusion recognition intensity with the preset vestibular illusion training intensity; and adjust the preset test control instructions according to the comparison results. Effectively improve the spatial orientation ability of the trainees and enhance the intelligent level of the vestibular illusion training process.

Description

Technical Field

[0001] The invention relates to the technical field of vestibular illusion training, and in particular to a vestibular illusion training method, a device and electronic equipment. Background Art

[0002] Spatial disorientation is one of the problems that almost all pilots have encountered during flight. Among them, vestibular illusion is caused by the physiological limitations of the human vestibular system and is also one of the types of spatial disorientation that pilots find difficult to deal with.

[0003] The existing vestibular illusion training equipment, such as the spatial orientation training system composed of a swivel chair, mainly uses motion stimulation to induce vestibular illusion and apply motion stimulation to the trainees in an open-loop control manner. However, since the trainees' limb movements and physiological states are not taken into account, this open-loop control method is difficult to accurately apply the preset vestibular illusion to the trainees, and the training effect needs to be further improved.

[0004] Therefore, how to improve the effect of vestibular illusion training has become an urgent problem to be solved. Summary of the invention

[0005] In view of this, the present invention provides a vestibular illusion training method, device and electronic equipment to solve the problem of how to improve the vestibular illusion training effect.

[0006] In a first aspect, the present invention provides a vestibular illusion training method, the method comprising:

[0007] Obtaining a preset test control instruction corresponding to a preset vestibular illusion training type and a preset vestibular illusion training intensity;

[0008] Conducting a vestibular illusion test on the subject based on preset test control instructions, and obtaining corresponding motion parameters and physiological parameters of the subject;

[0009] Identify the motion parameters and determine the target vestibular illusion identification type corresponding to the subject;

[0010] Identify physiological parameters and determine the corresponding target vestibular illusion recognition strength of the subjects;

[0011] The target vestibular illusion recognition type was compared with the preset vestibular illusion training type, and the target vestibular illusion recognition intensity was compared with the preset vestibular illusion training intensity;

[0012] According to the comparison results, the preset test control instructions are adjusted.

[0013] The vestibular illusion training method provided in the embodiment of the present application obtains the preset test control instructions corresponding to the preset vestibular illusion training type and the preset vestibular illusion training intensity; based on the preset test control instructions, the vestibular illusion test is performed on the subject, and the motion parameters and physiological parameters corresponding to the subject are obtained, thereby ensuring the accuracy of the acquired motion parameters and physiological parameters corresponding to the subject. The motion parameters are identified to determine the target vestibular illusion recognition type corresponding to the subject, thereby ensuring the accuracy of the target vestibular illusion recognition type corresponding to the determined subject. The physiological parameters are identified to determine the target vestibular illusion recognition intensity corresponding to the subject, thereby ensuring the accuracy of the target vestibular illusion recognition intensity corresponding to the determined subject. The target vestibular illusion recognition type is compared with the preset vestibular illusion training type, and the target vestibular illusion recognition intensity is compared with the preset vestibular illusion training intensity; according to the comparison result, the preset test control instructions are adjusted, thereby ensuring the accuracy of the adjusted preset test control instructions, thereby ensuring the accuracy of the vestibular illusion training for the subject. The above-mentioned vestibular illusion training method based on closed-loop feedback of the trainee's state helps to effectively improve the trainee's spatial orientation ability. While ensuring the effectiveness of the training, it greatly improves the intelligence level of the vestibular illusion training process, marking the further development of vestibular illusion training technology. This method will show great application prospects for the spatial ability training of trainees.

[0014] In an optional implementation of the present application, the motion parameters include a head motion parameter sequence and a body motion parameter sequence, and the motion parameters are identified to determine the target vestibular illusion identification type corresponding to the subject, including:

[0015] Performing data conversion on the head motion parameter sequence to generate a head grayscale image, and performing data conversion on the body motion parameter sequence to generate a body grayscale image;

[0016] Convert the head grayscale image into a head color image, and convert the body grayscale image into a body color image;

[0017] The head color image and the body color image are spliced ​​to generate a target color image;

[0018] The target color image is identified to determine the target vestibular illusion recognition type corresponding to the subject.

[0019] The vestibular illusion training method provided in the embodiment of the present application ensures the accuracy of the generated head grayscale image and body grayscale image, and ensures the accuracy of the converted head color image and body color image. The head color image and the body color image are spliced ​​to generate a target color image, the target color image is identified, and the target vestibular illusion recognition type corresponding to the subject is determined, thereby ensuring the accuracy of the vestibular illusion recognition class corresponding to the determined subject.

[0020] In an optional implementation of the present application, data conversion is performed on the head motion parameter sequence to generate a head grayscale image, including:

[0021] Splitting the head motion parameters to generate multiple non-intersecting head motion parameter subsequences;

[0022] Calculating the correlation integral between the head motion parameter subsequences according to the relationship between the head motion parameter subsequences;

[0023] Calculating the autocorrelation characteristics between the head motion parameter subsequences according to the correlation integral between the head motion parameter subsequences;

[0024] Determine the delay time and embedding dimension corresponding to the head motion parameter sequence according to the autocorrelation characteristics between each head motion parameter subsequence;

[0025] Based on the delay time and embedding dimension, the head phase space corresponding to the head motion parameter sequence is generated;

[0026] A head grayscale image is generated according to the distance between each head vector in the head phase space.

[0027] The vestibular illusion training method provided in the embodiment of the present application ensures the accuracy of the head phase space corresponding to the generated head motion parameter sequence. A head grayscale image is generated according to the distance between each head vector in the head phase space, thereby ensuring the accuracy of the generated head grayscale image.

[0028] In an optional embodiment of the present application, the head motion parameter sequence includes the linear acceleration motion parameter sequence of the head in the X, Y, and Z axes and the angular acceleration motion parameter sequence of the head in the X, Y, and Z axes respectively; the head grayscale image includes the head acceleration grayscale image in the X axis line, the head acceleration grayscale image in the Y axis line, the head acceleration grayscale image in the Z axis line, the head angular acceleration grayscale image in the X axis, the head angular acceleration grayscale image in the Y axis, and the head angular acceleration grayscale image in the Z axis; the head color image includes the head linear acceleration color image and the head angular acceleration color image; wherein, converting the head grayscale image into the head color image includes:

[0029] Determine a first red channel from the grayscale image of the acceleration of the head on the X axis, the grayscale image of the acceleration of the head on the Y axis, and the grayscale image of the acceleration of the head on the Z axis, determine a first green channel from the two grayscale images except the first red channel, and determine the last grayscale image as the first blue channel;

[0030] Generate a head linear acceleration color image according to the determined first red channel, first green channel and first blue channel;

[0031] Determine a second red channel from the grayscale image of the angular acceleration of the head on the X-axis, the grayscale image of the angular acceleration of the head on the Y-axis, and the grayscale image of the angular acceleration of the head on the Z-axis, determine a second green channel from the two grayscale images except the second red channel, and determine the last grayscale image as the second blue channel;

[0032] A head angular acceleration color image is generated according to the determined second red channel, second green channel, and second blue channel.

[0033] The vestibular illusion training method provided in the embodiment of the present application generates a head linear acceleration color image, which ensures that the generated head linear acceleration color image can represent the linear acceleration motion parameter sequence of the head in the X, Y, and Z axes. Generates a head angular acceleration color image. It ensures that the generated head angular acceleration color image can represent the angular acceleration motion parameter sequence of the head in the X, Y, and Z axes.

[0034] In an optional implementation of the present application, identifying the target color image and determining the target vestibular illusion recognition type corresponding to the subject includes:

[0035] The target color image is input into a preset image recognition model, which performs feature recognition and feature extraction on the target color image, and outputs the target vestibular illusion recognition type based on the extracted features; wherein the preset image recognition model includes a multi-layer perceptron layer, a window multi-head self-attention layer, a shift window-based multi-head self-attention layer, and a normalization layer.

[0036] The vestibular illusion training method provided in the embodiment of the present application inputs a target color image into a preset image recognition model, which performs feature recognition and feature extraction on the target color image, and outputs a target vestibular illusion recognition type based on the extracted features, thereby ensuring the accuracy of the output target vestibular illusion recognition type.

[0037] In an optional implementation of the present application, identifying physiological parameters to determine the target vestibular illusion recognition strength corresponding to the subject includes:

[0038] Obtain an initial classification model;

[0039] The physiological parameters are identified, and the corresponding target vestibular illusion recognition strength of the subjects is determined based on the initial classification model.

[0040] The vestibular illusion training method provided in the embodiment of the present application obtains an initial classification model; identifies physiological parameters, and determines the target vestibular illusion recognition strength corresponding to the subject based on the initial classification model, thereby ensuring the accuracy of the determined target vestibular illusion recognition strength corresponding to the subject.

[0041] In an optional implementation of the present application, obtaining an initial classification model includes:

[0042] Acquire an initial training data set, the initial training data set including a plurality of initial training data, each of which includes a training physiological parameter and a training vestibular illusion intensity corresponding to the training physiological parameter;

[0043] The initial training data set is input into the initial classification network, the initial classification network recognizes each piece of initial training data, generates a plurality of initial micro-clusters according to the recognition result, and determines the training vestibular illusion intensity corresponding to each initial micro-cluster;

[0044] Based on each initial micro-cluster, an initial classification model is generated.

[0045] The vestibular illusion training method provided in the embodiment of the present application obtains an initial training data set, inputs the initial training data set into an initial classification network, and the initial classification network recognizes each piece of initial training data, generates multiple initial micro-clusters based on the recognition results, and determines the training vestibular illusion strength corresponding to each initial micro-cluster, thereby ensuring the accuracy of the determined training vestibular illusion strength corresponding to each initial micro-cluster. Based on each initial micro-cluster, an initial classification model is generated, thereby ensuring the accuracy of the generated initial classification model.

[0046] In an optional embodiment of the present application, the physiological parameters are identified, and the target vestibular illusion recognition strength corresponding to the subject is determined based on the initial classification model, including:

[0047] Obtaining the number of preset classifiers and the number of preset microclusters corresponding to each preset classifier;

[0048] For each preset classifier, calculating a first similarity between the physiological parameter and each initial micro-cluster in the initial classification model;

[0049] Selecting a preset number of target microclusters from each of the initial microclusters according to the first similarity;

[0050] According to the training vestibular illusion strengths corresponding to the target micro-clusters, the training vestibular illusion strength with the largest number is determined as the backup vestibular illusion recognition strength corresponding to the preset classifier;

[0051] Determining the accuracy corresponding to each preset classifier according to the alternate vestibular illusion recognition strength corresponding to each preset classifier;

[0052] The backup vestibular illusion recognition strength corresponding to the preset classifier with the greatest accuracy is determined as the target vestibular illusion recognition strength.

[0053] The vestibular illusion training method provided in the embodiment of the present application selects a preset number of target microclusters from each initial microcluster according to the first similarity, thereby ensuring the accuracy of the determined target microclusters. The training vestibular illusion strength with the largest number is determined as the backup vestibular illusion recognition strength corresponding to the preset classifier, thereby ensuring the accuracy of the backup vestibular illusion recognition strength corresponding to each preset classifier. According to the backup vestibular illusion recognition strength corresponding to each preset classifier, the accuracy corresponding to each preset classifier is determined, thereby ensuring the accuracy corresponding to the determined preset classifier. The backup vestibular illusion recognition strength corresponding to the preset classifier with the highest accuracy is determined as the target vestibular illusion recognition strength, thereby ensuring the accuracy of the determined target vestibular illusion recognition strength and ensuring the robustness of the algorithm.

[0054] In an optional implementation of the present application, the backup vestibular illusion recognition strength corresponding to the preset classifier with the highest accuracy is determined as the target vestibular illusion recognition strength, and the method further includes:

[0055] Obtain each target micro-cluster corresponding to the preset classifier with the highest accuracy;

[0056] According to the relationship between the training vestibular illusion strength corresponding to each target micro-cluster and the target vestibular illusion recognition strength, each target micro-cluster is updated to generate an updated micro-cluster;

[0057] The initial classification model is updated according to each updated micro-cluster.

[0058] The vestibular illusion training method provided in the embodiment of the present application obtains each target micro-cluster corresponding to the preset classifier with the highest accuracy; updates each target micro-cluster according to the relationship between the training vestibular illusion strength corresponding to each target micro-cluster and the target vestibular illusion recognition strength, thereby ensuring the accuracy of the generated updated micro-cluster. The initial classification model is updated according to each updated micro-cluster, thereby ensuring the accuracy of the update of the initial classification model, so that the target vestibular illusion recognition strength corresponding to the physiological parameters generated later can be determined based on the updated initial classification model.

[0059] In an optional embodiment of the present application, the method further includes:

[0060] For each preset classifier, if a preset number of target microclusters is not selected from each initial microclusters according to the first similarity, then generating new microclusters based on the physiological parameters;

[0061] Obtain the total number of microclusters in the initial classification model; the total number of microclusters includes the number of initial microclusters and the number of newly added microclusters;

[0062] If the total number of micro-clusters is greater than a preset total number of micro-clusters threshold, then calculating a second similarity between the newly added micro-clusters and each initial micro-cluster;

[0063] According to each second similarity, merging the newly added micro-cluster with the initial micro-cluster with the largest second similarity to generate a merged first merged micro-cluster;

[0064] If there is no newly added micro-cluster, the current number of micro-clusters corresponding to each training vestibular illusion intensity in the initial classification model is obtained;

[0065] Calculate the Euclidean distance between each current micro-cluster corresponding to the training vestibular illusion strength with the largest number of current micro-clusters;

[0066] Merge the current micro-clusters with the closest Euclidean distance to generate a second merged micro-cluster;

[0067] The initial classification model is updated according to the second merged micro-clusters.

[0068] The vestibular illusion training method provided in the embodiment of the present application ensures that the newly added micro-clusters can characterize physiological parameters. The newly added micro-cluster is merged with the initial micro-cluster with the second largest similarity to generate a first merged micro-cluster after the merger, so as to reduce the total number of micro-clusters in the initial classification model, thereby reducing the occupied resources and speeding up the calculation speed. The Euclidean distance between each current micro-cluster corresponding to the training vestibular illusion intensity with the largest number of current micro-clusters is calculated to ensure the accuracy of the calculated Euclidean distance; the current micro-clusters with the closest Euclidean distance are merged to generate a second merged micro-cluster, which reduces the total number of micro-clusters in the initial classification model and ensures that the number of current micro-clusters corresponding to each training vestibular illusion intensity is not too small. According to the second merged micro-cluster, the initial classification model is updated to ensure the accuracy of the update of the initial classification model. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0070] Figure 1 is a flow chart of a vestibular illusion training method according to an embodiment of the present invention;

[0071] Figure 2is a flow chart of another vestibular illusion training method according to an embodiment of the present invention;

[0072] Figure 3 is a schematic diagram of a process of converting motion parameters to generate a target color image according to an embodiment of the present invention;

[0073] Figure 4 is a schematic diagram of a flow chart of identifying a target color image and determining a target vestibular illusion recognition type according to an embodiment of the present invention;

[0074] Figure 5 is a schematic diagram of a Swin Transformer network structure according to an embodiment of the present invention;

[0075] Figure 6 It is a schematic diagram of a process of identifying physiological parameters and determining a target vestibular illusion recognition strength according to an embodiment of the present invention;

[0076] Figure 7 4 is a flow chart of another vestibular illusion training method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0078] It should be noted that the vestibular illusion training method provided in the embodiment of the present application, its execution subject can be a vestibular illusion training device, and the vestibular illusion training device can be implemented as part or all of a computer device through software, hardware, or a combination of software and hardware, wherein the computer device can be a server or a terminal, wherein the server in the embodiment of the present application can be a single server or a server cluster composed of multiple servers, and the terminal in the embodiment of the present application can be a smart phone, a personal computer, a tablet computer, a wearable device, and other intelligent hardware devices such as an intelligent robot. In the following method embodiments, the execution subject is an electronic device as an example for explanation.

[0079] According to an embodiment of the present invention, an embodiment of a vestibular illusion training method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0080] In this embodiment, a vestibular illusion training method is provided, which can be used in the above-mentioned electronic device. Figure 1 is a flow chart of a vestibular illusion training method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0081] Step S101, obtaining a preset test control instruction corresponding to a preset vestibular illusion training type and a preset vestibular illusion training intensity.

[0082] Specifically, the electronic device can receive a preset vestibular illusion training type and a preset vestibular illusion training intensity input by a user, and then determine a preset test control instruction corresponding to the preset vestibular illusion training type and the preset vestibular illusion training intensity based on the corresponding relationship between the preset vestibular illusion training type and the preset vestibular illusion training intensity and the preset test control instructions.

[0083] Among them, the preset vestibular illusion training types are the names of different types of illusions, such as Coriolis-backward roll illusion, Coriolis-forward roll illusion, Coriolis-rightward roll illusion, body gravity illusion, body rotation illusion, etc. The preset vestibular illusion training intensity is the level corresponding to the preset vestibular illusion training type. The preset vestibular illusion training intensity can be divided into three levels, the number "1" represents the preset vestibular illusion training intensity level is level 1, the number "2" represents the preset vestibular illusion training intensity level is level 2, and the number "3" represents the preset vestibular illusion training intensity level is level 3, where the larger the number, the greater the intensity.

[0084] Among them, different preset vestibular illusion training types and preset vestibular illusion training intensities correspond to different preset test control instructions, such as setting the rotation axis, rotation direction, rotation speed, movement time, linear movement mode, etc.

[0085] For example, taking the triggering of the Coriolis-backward roll illusion as an example, the Coriolis illusion-backward roll illusion refers to the trainee rotating the body z b The axis keeps rotating clockwise at a constant speed. If the body is suddenly rotated at the same time b After the axis rotation, the trainee will produce the body is rotating around the body y b The feeling of the axis rotating backward. Therefore, the preset test control instruction field includes {(axis), (direction), (speed), (duration), (start rotation time) (axis), (direction), (speed), (duration)}. For example, the preset test control instruction corresponds to {(z b ), (clockwise), (60° / s), (60s), (30s) (x b ), (clockwise), (45° / s), (1s)}.

[0086] Step S102, performing a vestibular illusion test on the subject based on a preset test control instruction, and obtaining corresponding motion parameters and physiological parameters of the subject.

[0087] Specifically, the electronic device can drive the six-degree-of-freedom motion platform to perform linear and rotational motions to test the vestibular illusion of the subject based on preset test control instructions. The electronic device can obtain the corresponding motion parameters of the subject based on the motion sensor worn on the subject, and obtain the corresponding physiological parameters of the subject through the physiological sensor worn on the subject.

[0088] Among them, the six-degree-of-freedom motion platform can be moved along the six-degree-of-freedom motion platform coordinate system x p ,y p 、z p The three axes can perform rotation and linear motion, and can provide complex six-degree-of-freedom comprehensive motion acceleration stimulation. The origin O of the six-degree-of-freedom motion platform coordinate system is specified. p Located at the mass center of the six-degree-of-freedom motion platform, the six-degree-of-freedom motion platform coordinate system x p The axis is located within the symmetry plane of the six-degree-of-freedom motion platform, with the forward direction as positive; the six-degree-of-freedom motion platform coordinate system y p Axis perpendicular to x p The symmetry plane of the six-degree-of-freedom motion platform where the axis is located is positive when it points to the right; the coordinate system z of the six-degree-of-freedom motion platform p Axis perpendicular to x p O p y p plane, and determine z using the right-hand rule p The positive direction of the axis is downward.

[0089] Step S103, identifying the motion parameters to determine the target vestibular illusion recognition type corresponding to the subject.

[0090] Specifically, the electronic device can input the motion parameters into a preset motion parameter recognition model to determine the target vestibular illusion recognition type corresponding to the subject.

[0091] Among them, the preset motion parameter recognition model can be any one of a radial basis function (RBF) network, a feedforward neural network (FFNN), a convolutional neural network (CNN), a deconvolutional neural network (DN), a deep convolutional inverse graphics network (DCIGN), a generative adversarial network (GAN), a recurrent neural network (RNN), a long short-term memory network (LSTM), a deep residual network (DRN), and an extreme learning machine (ELM). The embodiment of the present application does not limit the preset motion parameter recognition model.

[0092] This step will be described in detail below.

[0093] Step S104, identifying the physiological parameters to determine the target vestibular illusion recognition strength corresponding to the subject.

[0094] Specifically, the electronic device can input the physiological parameters into a preset physiological parameter recognition model to determine the target vestibular illusion recognition strength corresponding to the subject.

[0095] Among them, the preset physiological parameter recognition model can be any one of a radial basis function (RBF) network, a feedforward neural network (FFNN), a convolutional neural network (CNN), a deconvolutional neural network (DN), a deep convolutional inverse graphics network (DCIGN), a generative adversarial network (GAN), a recurrent neural network (RNN), a long short-term memory network (LSTM), a deep residual network (DRN), and an extreme learning machine (ELM). The embodiment of the present application does not limit the preset physiological parameter recognition model.

[0096] This step will be described in detail below.

[0097] Step S105 , comparing the target vestibular illusion recognition type with the preset vestibular illusion training type, and comparing the target vestibular illusion recognition intensity with the preset vestibular illusion training intensity.

[0098] Specifically, the electronic device may calculate the similarity between the target vestibular illusion recognition type and the preset vestibular illusion training type, and the similarity between the target vestibular illusion recognition intensity and the preset vestibular illusion training intensity.

[0099] Step S106: adjusting the preset test control instruction according to the comparison result.

[0100] Specifically, if the similarity between the target vestibular illusion recognition type and the preset vestibular illusion training type is less than a first preset similarity threshold, and the similarity between the target vestibular illusion recognition intensity and the preset vestibular illusion training intensity is less than a second preset similarity threshold, the electronic device can adjust the preset test control instruction.

[0101] In an optional embodiment of the present application, the electronic device can obtain multiple subjects, perform vestibular illusion tests according to preset test control instructions corresponding to preset vestibular illusion training types and preset vestibular illusion training intensities, and obtain target vestibular illusion recognition types and target vestibular illusion recognition intensities.

[0102] When the similarity between the target vestibular illusion recognition type and the preset vestibular illusion training type corresponding to more than a preset number of subjects among multiple subjects is less than a first preset similarity threshold, and the similarity between the target vestibular illusion recognition intensity and the preset vestibular illusion training intensity is less than a second preset similarity threshold, the electronic device can adjust the preset test control instructions.

[0103] The preset number may be 50% or 60% of the total subjects. The embodiment of the present application does not specifically limit the preset number.

[0104] Among them, the electronic device can adjust at least one parameter among {(axial direction), (direction), (rotation speed), (duration), (start rotation time) (axial direction), (direction), (rotation speed), (duration)} corresponding to the preset test control instruction.

[0105] The vestibular illusion training method provided in the embodiment of the present application obtains the preset test control instructions corresponding to the preset vestibular illusion training type and the preset vestibular illusion training intensity; based on the preset test control instructions, the vestibular illusion test is performed on the subject, and the motion parameters and physiological parameters corresponding to the subject are obtained, thereby ensuring the accuracy of the acquired motion parameters and physiological parameters corresponding to the subject. The motion parameters are identified to determine the target vestibular illusion recognition type corresponding to the subject, thereby ensuring the accuracy of the target vestibular illusion recognition type corresponding to the determined subject. The physiological parameters are identified to determine the target vestibular illusion recognition intensity corresponding to the subject, thereby ensuring the accuracy of the target vestibular illusion recognition intensity corresponding to the determined subject. The target vestibular illusion recognition type is compared with the preset vestibular illusion training type, and the target vestibular illusion recognition intensity is compared with the preset vestibular illusion training intensity; according to the comparison result, the preset test control instructions are adjusted, thereby ensuring the accuracy of the adjusted preset test control instructions, thereby ensuring the accuracy of the vestibular illusion training for the subject. The above-mentioned vestibular illusion training method based on closed-loop feedback of the trainee's state helps to effectively improve the trainee's spatial orientation ability. While ensuring the effectiveness of the training, it greatly improves the intelligence level of the vestibular illusion training process, marking the further development of vestibular illusion training technology. This method will show great application prospects for the spatial ability training of trainees.

[0106] In this embodiment, a vestibular illusion training method is provided, which can be used in the above-mentioned electronic device. Figure 2 is a flow chart of a vestibular illusion training method according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0107] Step S201, obtaining a preset test control instruction corresponding to a preset vestibular illusion training type and a preset vestibular illusion training intensity.

[0108] For details about this step, please refer to the above description of step S101, which will not be elaborated here.

[0109] Step S202: Perform a vestibular illusion test on the subject based on a preset test control instruction, and obtain corresponding motion parameters and physiological parameters of the subject.

[0110] For details about this step, please refer to the above description of step S102, which will not be elaborated here.

[0111] Step S203, identifying the motion parameters to determine the target vestibular illusion recognition type corresponding to the subject.

[0112] In an optional implementation of the present application, before identifying the motion parameters, the electronic device may unify the sampling frequency of the motion parameters, then perform noise reduction processing, and finally normalize the data using the minimum-maximum method.

[0113] In an optional embodiment of the present application, the motion parameters include a head motion parameter sequence and a body motion parameter sequence. The motion sensor includes a head-mounted motion sensor and a body motion sensor. The head-mounted motion sensor can collect the head motion parameter sequence, and the body motion sensor can collect the body motion parameter sequence.

[0114] The above step S203 may include the following steps:

[0115] Step S2031, performing data conversion on the head motion parameter sequence to generate a head grayscale image, and performing data conversion on the body motion parameter sequence to generate a body grayscale image.

[0116] Specifically, the electronic device can use phase space reconstruction and recursive graph technology to perform data conversion on the head motion parameter sequence to generate a head grayscale image, and perform data conversion on the body motion parameter sequence to generate a body grayscale image.

[0117] Specifically, the step S2031 of “converting the head motion parameter sequence into data to generate a head grayscale image” may include the following steps:

[0118] Step a1, split the head motion parameters to generate multiple non-intersecting head motion parameter subsequences.

[0119] Step a2: Calculate the correlation integral between each head motion parameter subsequence according to the relationship between each head motion parameter subsequence.

[0120] Step a3, calculating the autocorrelation characteristics between the head motion parameter subsequences according to the correlation integral between the head motion parameter subsequences.

[0121] Step a4: Determine the delay time and embedding dimension corresponding to the head motion parameter sequence according to the autocorrelation characteristics between each head motion parameter subsequence.

[0122] Step a5: Generate a head phase space corresponding to the head motion parameter sequence based on the delay time and the embedding dimension.

[0123] Step a6, generating a head grayscale image according to the distance between each head vector in the head phase space.

[0124] Similarly, the step S2031 of “converting the body motion parameter sequence to generate a body grayscale image” may include the following steps:

[0125] Step a1, splitting the body motion parameters to generate multiple non-intersecting body motion parameter subsequences.

[0126] Step a2, calculating the correlation integral between each body motion parameter subsequence according to the relationship between each body motion parameter subsequence.

[0127] Step a3, calculating the autocorrelation characteristics between the body motion parameter subsequences according to the correlation integral between the body motion parameter subsequences.

[0128] Step a4: Determine the delay time and embedding dimension corresponding to the body motion parameter sequence according to the autocorrelation characteristics between the body motion parameter subsequences.

[0129] Step a5, generating a body phase space corresponding to the body motion parameter sequence based on the delay time and the embedding dimension.

[0130] Step a6, generating a body grayscale image according to the distance between each body vector in the body phase space.

[0131] Specifically, two key parameters of phase space reconstruction are the delay time and the embedding dimension. , a given time series A can represent a head motion parameter sequence or a body motion parameter sequence, that is, a time series data of a segment of motion parameters.

[0132] Phase space reconstruction can be done using phase space B, as shown in formula (1).

[0133] (1)

[0134] Where p is the embedding dimension, τ is the delay time, is the number of phase points in the phase space. By properly choosing the embedding dimension and the delay time, the reconstructed phase space becomes a diffeomorphism of the original system.

[0135] The electronic device can use the CC method to calculate the delay time τ and the embedding dimension p. The specific process is as follows:

[0136] First, split the time series A into τ non-intersecting subsequences, each with a length of . The value of τ can start from 1, and the maximum value is the smallest number between 40 and (N2 / 16-1), where N is the number of data points in time series A. As shown in formula (2).

[0137] (2)

[0138] The sampling interval of time series A is , is the optimal delay time of the time series, is the delay time window, It can be calculated according to formula (3).

[0139] (3)

[0140] The correlation integral is a cumulative distribution function that represents the probability that the distance between any two points in the phase space is less than a given search radius r. Distance refers to the distance between points in a high-dimensional space measured by the infinite norm of the difference between the vectors of the two points. Therefore, the correlation integral that defines the reconstructed phase space B is calculated according to formula (4).

[0141] (4)

[0142] Where r is the search radius, , is the infinite norm between two subsequences in the reconstructed phase space B, , is the number of phase points in the phase space. t is

[0143] Defining the test statistic , calculated according to formula (5).

[0144] (5)

[0145] The actual time series is of finite length and there is a certain correlation between the elements, resulting in the actual detection statistics Usually not zero. Test statistic Reflects the autocorrelation characteristics of each sub-time series, the optimal time delay Desirable The first zero of For the time point at which all radii r differ from each other the least, the points in the reconstructed phase space are closest to uniform distribution, and the orbit of the reconstructed attractor is fully expanded in the phase space. Select the two radii r with the largest and smallest corresponding values, and define the difference calculated according to formula (6). This time point is either the first zero point of a function or the time point at which all radii r differ from each other the least. At this time point, the points in the reconstructed phase space are closest to uniform distribution, and the orbit of the reconstructed attractor is fully expanded in the phase space.

[0146] (6)

[0147] In formula (6), The detection statistic is measured The maximum deviation degree for all search radius r. So the local maximum time t should be The zero and The minimum value, optimal delay time corresponds to the first of these local maximum times t.

[0148] Then, the electronic device uses the CC method to calculate the delay time and embedding dimension. For example, N=320, p=2,3,4,5, , is the standard deviation of the time series, where i=1,2,3,4. Calculated according to formulas (7)-(9) , and .

[0149] (7)

[0150] (8)

[0151] (9)

[0152] In formulas (7)-(9), the CC method takes The first zero point or The first minimum value of , calculate the time delay τ = . And comprehensive and ,Pick The global minimum of the time series is taken as the time window length Finally, according to formula (3), the embedding dimension p is calculated. Then, according to formula (3), the delay time is calculated using the embedding dimension, and the phase space is calculated based on the embedding dimension and the delay time.

[0153] The recurrence graph is a two-dimensional graph consisting of black points, white points, and two time axes. After reconstructing the time series in the phase space, the phase space is obtained. .

[0154] Among them, Representation vector With vector If and The distance is less than a certain threshold ,but and are considered to be close to each other, and the corresponding position N on the recurrence graph ij is marked as 1 (indicated by a black dot); otherwise, it is marked as 0 (indicated by a white dot), as shown in formula (10).

[0155] (10)

[0156] In formula (10), , is the distance threshold, is the Heaviside function, calculated according to formula (11).

[0157] (11)

[0158] N ij Calculate according to formula (12).

[0159] (12)

[0160] Step S2032, converting the head grayscale image into a head color image, and converting the body grayscale image into a body color image.

[0161] In an optional embodiment, the head-mounted motion sensor is used to detect the head movement of the trainee, and record the three-direction linear acceleration (hx, hy, hz) and angular acceleration (hrx, hry, hrz) of the trainee's head. The three-direction linear acceleration (hx, hy, hz) and angular acceleration (hrx, hry, hrz) of the head are relative to the head coordinate system, and it is stipulated that the origin Oh of the head coordinate system is located at the mass center of the head, the xh axis of the head coordinate system is located within the head symmetry plane, and the direction of the nose is positive; the yh axis of the head coordinate system is perpendicular to the head symmetry plane where the xh axis is located, and the direction to the right is positive; the zh axis of the head coordinate system is perpendicular to the xhOhyh plane, and the positive direction of the zh axis is determined to be downward according to the right-hand rule.

[0162] The body motion sensor is used to detect the motion parameters of the motion platform and record the three-dimensional linear acceleration (bx, by, bz) and three-dimensional angular position (bRx, bRy, bRz) of the trainee's body. The three-dimensional linear acceleration (bx, by, bz) and three-dimensional angular position (bRx, bRy, bRz) of the body are relative to the body coordinate system. It is stipulated that the origin Ob of the body coordinate system is located at the mass center of the head, the xb axis of the body coordinate system is located within the body symmetry plane, and the chest direction is positive; the yb axis of the body coordinate system is perpendicular to the body symmetry plane where the xb axis is located, and the right direction is positive; the zb axis of the body coordinate system is perpendicular to the xbObyb plane, and the positive direction of the zb axis is determined to be downward according to the right-hand rule.

[0163] Therefore, the head motion parameter sequence includes the linear acceleration motion parameter sequence of the head in the X, Y, and Z axes and the angular acceleration motion parameter sequence of the head in the X, Y, and Z axes respectively; the head grayscale image includes the head acceleration grayscale image in the X axis, the head acceleration grayscale image in the Y axis, the head acceleration grayscale image in the Z axis, the head angular acceleration grayscale image in the X axis, the head angular acceleration grayscale image in the Y axis, and the head angular acceleration grayscale image in the Z axis; the head color image includes the head linear acceleration color image and the head angular acceleration color image. The "converting the head grayscale image into the head color image" in the above step S2032 may include:

[0164] Step b1, determine the first red channel from the grayscale image of the head acceleration on the X-axis, the grayscale image of the head acceleration on the Y-axis, and the grayscale image of the head acceleration on the Z-axis, determine the first green channel from the two grayscale images except the first red channel, and determine the last grayscale image as the first blue channel.

[0165] Step b2: generating a head linear acceleration color image according to the determined first red channel, first green channel and first blue channel.

[0166] Step b3, determine the second red channel from the grayscale image of the head angular acceleration on the X-axis, the grayscale image of the head angular acceleration on the Y-axis, and the grayscale image of the head angular acceleration on the Z-axis, determine the second green channel from the two grayscale images except the second red channel, and determine the last grayscale image as the second blue channel.

[0167] Step b4, generating a head angular acceleration color image according to the determined second red channel, second green channel and second blue channel.

[0168] Specifically, the electronic device can use the grayscale image of the head acceleration on the X-axis, the grayscale image of the head acceleration on the Y-axis, and the grayscale image of the head acceleration on the Z-axis as the red, green, and blue channels of the head linear acceleration color image, respectively, to obtain a head linear acceleration color image.

[0169] The electronic device can use the grayscale image of the head's angular acceleration on the X-axis, the grayscale image of the head's angular acceleration on the Y-axis, and the grayscale image of the head's angular acceleration on the Z-axis as red, green, and blue channels of a color image, respectively, to obtain a color image of the head's angular acceleration.

[0170] It should be noted that the embodiment of the present application does not specifically limit the correspondence between the red, green and blue channels of the head X-axis angular acceleration grayscale image, the head Y-axis angular acceleration grayscale image, and the head Z-axis angular acceleration grayscale image as color images, respectively.

[0171] Similarly, the body motion parameter sequence includes the linear acceleration motion parameter sequence of the body in the X, Y, and Z axes and the angular acceleration motion parameter sequence of the body in the X, Y, and Z axes respectively; the body grayscale image includes the body acceleration grayscale image in the X axis, the body acceleration grayscale image in the Y axis, the body acceleration grayscale image in the Z axis, the body angular acceleration grayscale image in the X axis, the body angular acceleration grayscale image in the Y axis, and the body angular acceleration grayscale image in the Z axis; the body color image includes the body linear acceleration color image and the body angular acceleration color image. The "converting the body grayscale image into the body color image" in the above step S2032 may include:

[0172] Step b5, determine the third red channel from the grayscale image of the body's acceleration on the X-axis, the grayscale image of the body's acceleration on the Y-axis, and the grayscale image of the body's acceleration on the Z-axis, determine the third green channel from the two grayscale images except the third red channel, and determine the last grayscale image as the third blue channel.

[0173] Step b6, generating a body linear acceleration color image according to the determined third red channel, third green channel and third blue channel.

[0174] Step b7, determine the fourth red channel from the grayscale image of the body's angular acceleration on the X-axis, the grayscale image of the body's angular acceleration on the Y-axis, and the grayscale image of the body's angular acceleration on the Z-axis, determine the fourth green channel from the two grayscale images other than the fourth red channel, and determine the last grayscale image as the fourth blue channel.

[0175] Step b8, generating a body angular acceleration color image according to the determined fourth red channel, fourth green channel and fourth blue channel.

[0176] Specifically, the electronic device can use the grayscale image of the body's acceleration on the X-axis, the grayscale image of the body's acceleration on the Y-axis, and the grayscale image of the body's acceleration on the Z-axis as red, green, and blue channels of a color image, respectively, to obtain a body linear acceleration color image.

[0177] The electronic device can use the grayscale image of the body's angular acceleration on the X-axis, the grayscale image of the body's angular acceleration on the Y-axis, and the grayscale image of the body's angular acceleration on the Z-axis as red, green, and blue channels of a color image, respectively, to obtain a body angular acceleration color image.

[0178] It should be noted that the embodiment of the present application does not specifically limit the correspondence between the red, green and blue channels of the body's X-axis angular acceleration grayscale image, the body's Y-axis angular acceleration grayscale image, and the body's Z-axis angular acceleration grayscale image as color images, respectively.

[0179] Step S2033, splicing the head color image and the body color image to generate a target color image.

[0180] Specifically, the electronic device splices the head linear acceleration color image, the head angular acceleration color image, the body linear acceleration color image, and the body angular acceleration color image to generate a target color image.

[0181] For example, Figure 3 As shown, the electronic device converts the linear acceleration motion parameter sequence of the head in the X, Y, and Z axes and the angular acceleration motion parameter sequence of the head in the X, Y, and Z axes to generate the head acceleration grayscale image in the X axis, the head acceleration grayscale image in the Y axis, the head acceleration grayscale image in the Z axis, the head angular acceleration grayscale image in the X axis, the head angular acceleration grayscale image in the Y axis, and the head angular acceleration grayscale image in the Z axis. The electronic device converts the linear acceleration motion parameter sequence of the body in the X, Y, and Z axes and the angular acceleration motion parameter sequence of the body in the X, Y, and Z axes to generate the body acceleration grayscale image in the X axis, the body acceleration grayscale image in the Y axis, the body acceleration grayscale image in the Z axis, the body angular acceleration grayscale image in the X axis, the body angular acceleration grayscale image in the Y axis, and the body angular acceleration grayscale image in the Z axis.

[0182] Then, the electronic device transforms the grayscale image of the head acceleration on the X-axis, the grayscale image of the head acceleration on the Y-axis, and the grayscale image of the head acceleration on the Z-axis to generate a head linear acceleration color image; transforms the grayscale image of the head angular acceleration on the X-axis, the grayscale image of the head angular acceleration on the Y-axis, and the grayscale image of the head acceleration on the Z-axis to generate a head angular acceleration color image; transforms the grayscale image of the body acceleration on the X-axis, the grayscale image of the body acceleration on the Y-axis, and the grayscale image of the body acceleration on the Z-axis to generate a body linear acceleration color image; transforms the grayscale image of the body angular acceleration on the X-axis, the grayscale image of the body angular acceleration on the Y-axis, and the grayscale image of the body acceleration on the Z-axis to generate a body angular acceleration color image.

[0183] The electronic device splices the head linear acceleration color image, the head angular acceleration color image, the body linear acceleration color image and the body angular acceleration color image to generate a target color image.

[0184] Step S2034, identifying the target color image to determine the target vestibular illusion recognition type corresponding to the subject.

[0185] Specifically, the above step S2034 may include the following steps:

[0186] Step c1, inputting the target color image into a preset image recognition model, the preset image recognition model performs feature recognition and feature extraction on the target color image, and outputs the target vestibular illusion recognition type based on the extracted features.

[0187] Among them, the preset image recognition model includes a multi-layer perceptron layer, a window multi-head self-attention layer, a shift window-based multi-head self-attention layer and a normalization layer.

[0188] Specifically, the architecture of the preset image recognition model is as shown in the attached Figure 4 The entire preset image recognition model includes a block partitioning layer, a Swin Transformer layer, a pooling layer, a fully connected layer, and a Softmax layer.

[0189] The size of the target color image obtained by phase space reconstruction, recursive graph and image stitching is W × H ×3, where W and H are the width and length of the target color image, respectively, and "3" represents the three color channels of the target color image. First, the target color image passes through the block partitioning layer, where the target color image is divided into "blocks" and blocks are used as the smallest unit instead of pixels. Each block corresponds to a "marker", and the number of marks is the same as the number of blocks, and the output forms a H / 4× W / 4 three-dimensional matrix.

[0190] Then, it passes through the Swin Transformer layer. The Swin Transformer layer has four stages. In stage 1, the number of tags is kept constant. W / 4× H / 4, the size of the output feature is set to D , stage 1 is repeated twice; in stage 2, the number of markers is reduced to 1 / 4 of the previous one, becoming W / 8× H / 8, the output size is 2 D , stage 2 is repeated twice; stages 3 and 4 follow a similar pattern, and the number of tokens they output isW / 16× H / 16 and W / 32× H / 32, the output size is 4 D and 8 D , Stage 3 and Stage 4 were repeated 6 times and 2 times respectively.

[0191] For example, as shown in the attached Figure 4 As shown in the figure, the four stages include 2, 2, 6, and 2 Swin Transformers respectively. The number of channels in the four stages are 96, 192, 384, and 768 respectively. The number of attention heads included in each stage is 3, 6, 12, and 24 respectively.

[0192] The function of the block fusion layer in the Swin Transformer layer is to downsample, reduce resolution, and adjust the number of channels to establish a hierarchical structure. The features of each block group are concatenated in the first block fusion layer. The concatenated features are then input into the linear embedding layer for processing. The linear embedding layer performs a linear transformation on the channel data of each pixel and projects the channel data to an arbitrary dimension. The dimension after projection is recorded as " D ”, followed by feature processing in the Swin Transformer module.

[0193] Finally, through the pooling layer, the fully connected layer and the Softmax layer, the target color image is classified into different types of vestibular illusions, and the illusion type is output.

[0194] The Swin Transformer consists of a multi-layer perceptron (MLP), a window multi-head self-attention (W-MSA), a shifted window-based multi-head self-attention (SW-MSA), and a layer normalization (LN). The architecture of the Swin Transformer is shown in the attached figure. Figure 5 shown.

[0195] As attached Figure 5As shown in Figure 1, the feature parameters are first input to the LN layer of the Swin Transformer network. Then, they are passed through the W-MSA layer for feature learning, followed by residual calculation, and the LN of this layer is obtained. The structure of the SW-MSA layer is similar to that of the W-MSA layer, except that the feature calculation part involves a sliding window operation. The calculations of each part of the network structure are shown in formulas (13)-(16).

[0196] (13)

[0197] (14)

[0198] (15)

[0199] (16)

[0200] in, and They represent the output features of W-MSA and MLP for l, respectively, where l is the index of the partition.

[0201] The use of MSA improves accuracy and improves loss, thereby enhancing the generalization ability of the model. However, MSA has high computational complexity. By adopting W-MSA, the computational complexity can be reduced. The standard MSA calculation is shown in formulas (17)-(20), assuming that there are i heads and the number of blocks is w×h.

[0202] (17)

[0203] (18)

[0204] (19)

[0205] (20)

[0206] In formula (17)-(20), is the input matrix, are query matrix, key matrix and value matrix respectively, They are query parameter matrix, key parameter matrix and value parameter matrix respectively. is the transformation matrix, s is the dimension of the query matrix and the key matrix, Represents the connectivity matrix.

[0207] The self-attention calculation of each window is shown in formula (21).

[0208] (twenty one)

[0209] In formula (21) are the query matrix, key matrix and value matrix respectively, where is the number of blocks in the current window, and s is the dimension of the query matrix and the key matrix. F is from the bias matrix Obtained.

[0210] The calculation of each window of MSA and W-MSA is shown in formulas (22)-(24).

[0211] (twenty two)

[0212] (twenty three)

[0213] (twenty four)

[0214] In formulas (22)-(24), D represents the dimension of the feature. P represents the length and width of each window (both length and width are P).

[0215] Step S204, identifying the physiological parameters to determine the target vestibular illusion recognition strength corresponding to the subject.

[0216] For details about this step, please refer to the above description of step S104, which will not be elaborated here.

[0217] Step S205 , comparing the target vestibular illusion recognition type with the preset vestibular illusion training type, and comparing the target vestibular illusion recognition intensity with the preset vestibular illusion training intensity.

[0218] For details about this step, please refer to the above description of step S105, which will not be elaborated here.

[0219] Step S206: adjusting the preset test control instruction according to the comparison result.

[0220] For details about this step, please refer to the above description of step S106, which will not be elaborated here.

[0221] The vestibular illusion training method provided in the embodiment of the present application splits the head motion parameters to generate multiple non-intersecting head motion parameter subsequences; according to the relationship between the head motion parameter subsequences, the correlation integral between the head motion parameter subsequences is calculated, and the accuracy of the correlation integral between the calculated head motion parameter subsequences is ensured. According to the correlation integral between the head motion parameter subsequences, the autocorrelation characteristics between the head motion parameter subsequences are calculated, and the accuracy of the autocorrelation characteristics between the calculated head motion parameter subsequences is ensured. According to the autocorrelation characteristics between the head motion parameter subsequences, the delay time and embedding dimension corresponding to the head motion parameter sequence are determined, and the accuracy of the delay time and embedding dimension corresponding to the determined head motion parameter sequence is ensured. Based on the delay time and embedding dimension, the head phase space corresponding to the head motion parameter sequence is generated, and the accuracy of the head phase space corresponding to the generated head motion parameter sequence is ensured. According to the distance between the head vectors in the head phase space, a head grayscale image is generated, and the accuracy of the generated head grayscale image is ensured. Similarly, the body motion parameter sequence is converted to generate a body grayscale image, and the accuracy of the generated head grayscale image and body grayscale image is ensured.

[0222] The first red channel is determined from the grayscale image of the acceleration of the head on the X axis, the grayscale image of the acceleration of the head on the Y axis, and the grayscale image of the acceleration of the head on the Z axis, the first green channel is determined from the two grayscale images except the first red channel, and the last grayscale image is determined as the first blue channel; based on the determined first red channel, first green channel, and first blue channel, a head linear acceleration color image is generated, ensuring that the generated head linear acceleration color image can represent the linear acceleration motion parameter sequence of the head on the X, Y, and Z axes. The second red channel is determined from the grayscale image of the angular acceleration of the head on the X axis, the grayscale image of the angular acceleration of the head on the Y axis, and the grayscale image of the angular acceleration of the head on the Z axis, the second green channel is determined from the two grayscale images except the second red channel, and the last grayscale image is determined as the second blue channel; based on the determined second red channel, second green channel, and second blue channel, a head angular acceleration color image is generated. It is ensured that the generated head angular acceleration color image can represent the angular acceleration motion parameter sequence of the head on the X, Y, and Z axes. Similarly, the body grayscale image is converted into a body color image, which ensures the accuracy of the converted head color image and body color image. The head color image and the body color image are spliced ​​to generate a target color image, so that the generated target color image can accurately characterize the characteristics of the head motion parameter sequence and the body motion parameter sequence. The target color image is input into a preset image recognition model, which performs feature recognition and feature extraction on the target color image, and outputs the target vestibular illusion recognition type based on the extracted features, thereby ensuring the accuracy of the output target vestibular illusion recognition type.

[0223] In this embodiment, a vestibular illusion training method is provided, which can be used in the above-mentioned electronic device. Figure 6 is a flow chart of a vestibular illusion training method according to an embodiment of the present invention. Figure 6 As shown, the process includes the following steps:

[0224] Step S301, obtaining a preset test control instruction corresponding to a preset vestibular illusion training type and a preset vestibular illusion training intensity.

[0225] For details about this step, please refer to the introduction of step S201 above, which will not be described in detail here.

[0226] Step S302: Perform a vestibular illusion test on the subject based on a preset test control instruction, and obtain corresponding motion parameters and physiological parameters of the subject.

[0227] For details about this step, please refer to the introduction of step S202 above, which will not be described in detail here.

[0228] Step S303: Identify the motion parameters to determine the target vestibular illusion identification type corresponding to the subject.

[0229] For details about this step, please refer to the introduction of step S203 above, which will not be described in detail here.

[0230] Step S304, identifying the physiological parameters to determine the target vestibular illusion recognition strength corresponding to the subject.

[0231] Specifically, the physiological sensor may include an electrocardiogram sensor, a pulse sensor, a respiration sensor, etc., and the physiological parameters include an electrocardiogram signal (ECG), a pulse signal (PPG), and a respiration signal (RESP) corresponding to the trainee. Optionally, the electronic device may take the electrocardiogram data (ECG), the pulse data (PPG), and the respiration data (RESP) within a 5-second time window and combine them into physiological parameter instance data.

[0232] Specifically, the above step S304 may include the following steps:

[0233] Step S3041, obtaining an initial classification model.

[0234] Specifically, the above step S3041 may include the following steps:

[0235] Step d1, obtaining the initial training data set.

[0236] The initial training data set includes multiple pieces of initial training data, and each piece of initial training data includes a training physiological parameter and a training vestibular illusion intensity corresponding to the training physiological parameter.

[0237] Specifically, the electronic device may receive an initial training data set input by a user, may receive an initial training data set sent by other devices, or may receive an initial training data set sent by other devices.

[0238] The embodiment of the present application does not specifically limit the manner in which the electronic device obtains the initial training data set.

[0239] Step d2, inputting the initial training data set into the initial classification network, the initial classification network recognizes each piece of initial training data, generates a plurality of initial micro-clusters according to the recognition results, and determines the training vestibular illusion intensity corresponding to each initial micro-cluster.

[0240] Specifically, the electronic device can input the initial training data set into the initial classification network, and the initial classification network recognizes each piece of initial training data, generates multiple initial micro-clusters according to the recognition results, and determines the training vestibular illusion intensity corresponding to each initial micro-cluster.

[0241] The initial classification network may be a k-means clustering algorithm or other networks, and the embodiment of the present application does not specifically limit the initial classification network.

[0242] For example, the k-means clustering algorithm is used as an example. Specifically, the electronic device can input the initial training data set into the k-means clustering algorithm, and use the k-means clustering algorithm to divide the initial training data into k clusters. Among them:

[0243] The k-means clustering strategy selects the best partitioning function for the initial training data set by minimizing the loss function C* Optimal partitioning function C* As shown below:

[0244] (25)

[0245] The loss function is defined as the first similarity between a sample and the center of its category, as shown in formula (26).

[0246] (26)

[0247] In formula (26) It is the first l The mean or center of the clusters, , is a classification function. Q Describes the distance between samples in a cluster and the cluster center. Q The smaller the value, the higher the similarity of samples in the cluster.

[0248] Then, the electronic device calculates the feature information of each cluster and stores it in the initial micro-cluster in the form of cluster features. The cluster features contain seven pieces of feature information as shown in formula (27).

[0249] (27)

[0250] In formula (27), , are the linear sum and square sum of all initial training data in the initial micro-cluster, respectively and ; N Represents the number of all initial training data in the initial micro-cluster; Z is the reliability of the initial micro-cluster at the current time, and its initial value is set to 1; T is the update time of the initial micro-cluster, the initial value is 0; M L is the label of the initial microcluster; R is the radius of the initial micro-cluster, R is defined as the standard deviation of the distance between the initial training data points and the centroid in the cluster, calculated according to formula (28).

[0251] (28)

[0252] The electronic device obtains the training vestibular illusion strength corresponding to each initial training data in the initial micro-cluster, and determines the training vestibular illusion strength corresponding to the largest number of initial training data as the training vestibular illusion strength corresponding to each initial micro-cluster.

[0253] Step d3: generating an initial classification model based on each initial micro-cluster.

[0254] Specifically, the electronic device generates an initial classification model according to each initial micro-cluster.

[0255] Step S3042, identifying the physiological parameters, and determining the target vestibular illusion recognition strength corresponding to the subject based on the initial classification model.

[0256] Specifically, the above step S3042 may include the following steps:

[0257] Step e1, obtaining the number of preset classifiers and the number of preset micro-clusters corresponding to each preset classifier.

[0258] Specifically, the electronic device can receive the number of preset classifiers input by the user, or obtain the number of preset classifiers sent by other devices. In addition, the electronic device can also obtain the number of preset microclusters corresponding to each preset classifier, or receive the number of preset microclusters corresponding to each preset classifier sent by other devices.

[0259] The embodiment of the present application does not specifically limit the manner in which the electronic device obtains the number of preset classifiers and the number of preset micro-clusters corresponding to each preset classifier.

[0260] For example, it is assumed that the number of preset classifiers is θ ,Right now is used, K represents the number of microclusters corresponding to the preset classifier. For example, when θ = 1, there is only one K-NN classifier (i.e., 1-NN, i.e., the number of microclusters corresponding to the preset classifier is 1, or 3-NN, i.e., the number of microclusters corresponding to the preset classifier is 3); θ = 2 represents the integration strategy of 2 × K-NN classifiers (i.e., 1-NN and 3-NN). Similarly, when θ = 4, 4 × K-NN classifiers are used (i.e., , that is, the preset numbers of microclusters corresponding to the preset classifiers are 1, 3, 5, and 7 respectively).

[0261] Step e2: for each preset classifier, calculating the first similarity between the physiological parameter and each initial micro-cluster in the initial classification model.

[0262] Specifically, for each preset classifier, the electronic device can calculate the similarity between the physiological parameter and each initial training data corresponding to each initial micro-cluster in the initial classification model. Then, for each initial micro-cluster, based on the similarity between each initial training data in each initial micro-cluster and the physiological parameter, the average similarity between each initial training data and the physiological parameter is calculated, and the average similarity corresponding to each initial micro-cluster is determined as the first similarity between the physiological parameter and each initial micro-cluster in the initial classification model.

[0263] The similarity may be Euclidean distance, cosine distance, or other parameters. The embodiment of the present application does not specifically limit the similarity.

[0264] Step e3: selecting a preset number of target microclusters from each of the initial microclusters according to the first similarity.

[0265] Specifically, the electronic device compares the physiological parameters with the first similarities between the initial microclusters in the initial classification model, arranges the initial microclusters in descending order of the first similarities, and determines the initial microclusters with a preset number of microclusters arranged in front as target microclusters.

[0266] Exemplarily, if θ = 1, and there is only one K-NN classifier, that is, 1-NN, the preset number of micro-clusters is determined to be 1, and the electronic device determines the first initial micro-cluster as the target micro-cluster.

[0267] For example, if θ = 2, and there are two K-NN classifiers, i.e., 1-NN and 3-NN. For 1-NN, the electronic device determines that the preset number of microclusters is 1, and the electronic device determines the first initial microcluster as the target microcluster. For 3-NN, the electronic device determines that the preset number of microclusters is 3, and the electronic device determines the first three initial microclusters as the target microclusters.

[0268] Step e4, according to the training vestibular illusion strengths corresponding to the target micro-clusters, determining the training vestibular illusion strength with the largest number as the backup vestibular illusion recognition strength corresponding to the preset classifier.

[0269] Specifically, for each preset classifier, the electronic device determines the trained vestibular illusion strength corresponding to each target micro-cluster in the preset classifier, and then compares the trained vestibular illusion strength corresponding to each target micro-cluster, and determines that the trained vestibular illusion strength with the largest number is the backup vestibular illusion strength corresponding to the preset classifier.

[0270] Exemplarily, for 1-NN, the electronic device determines the training vestibular illusion strength corresponding to the target micro-cluster in 1-NN as the backup vestibular illusion strength corresponding to 1-NN.

[0271] For 3-NN, the electronic device compares the training vestibular illusion strengths corresponding to the three target microclusters in the 3-NN, and determines the training vestibular illusion strength with the largest number as the backup vestibular illusion strength corresponding to the 3-NN.

[0272] Exemplarily, the electronic device finds a preset number of target microclusters closest to the physiological parameter χ in each initial microcluster in the initial classification model according to the Euclidean distance, forming a set denoted as DK(χ); in DK(χ), the majority voting rule is used to determine the category y of χ, as shown in formula (29).

[0273] (29)

[0274] Where I in formula (29) is the indicator function. When , I is 1, otherwise I is 0. is the class of physiological parameters, .

[0275] Step e5, determining the accuracy corresponding to each preset classifier according to the backup vestibular illusion recognition strength corresponding to each preset classifier.

[0276] Specifically, the electronic device can calculate the weight corresponding to each preset classifier according to the standby vestibular illusion recognition strength corresponding to each preset classifier, and then determine the accuracy corresponding to the preset classifier according to the weight corresponding to each preset classifier.

[0277] Step e6, determining the backup vestibular illusion recognition strength corresponding to the preset classifier with the highest accuracy as the target vestibular illusion recognition strength.

[0278] Specifically, the electronic device may determine the backup vestibular illusion recognition strength corresponding to the preset classifier with the greatest accuracy as the target vestibular illusion recognition strength.

[0279] Exemplarily, each preset classifier is t i The weight calculation is shown in formula (30).

[0280] (30)

[0281] In formula (30), ζ represents the weight matrix of each preset classifier, c represents the number of recently labeled instances, where an instance is used to represent a set of physiological parameters, represents the time step, For classification , is the true class label. Among them, is the time step The logical function of an instance of .

[0282] If the recognition result is consistent with the actual result, the sum value is increased by 1, otherwise it is increased by 0. Divide the result by c , and the average weight of the classifier is obtained. Therefore, the weight value represents the average effective classification accuracy of the corresponding classifier.

[0283] Finally, the best classifier is the one with the largest weight, which is used to identify the category label of the current instance, as shown in Formula (31) and Formula (32).

[0284] (31)

[0285] (32)

[0286] Among them, in formula (31) and formula (32), K yes K -Important parameters in NN classification algorithm and ensemble strategy θ The choice of value is related.

[0287] In an optional embodiment of the present application, the vestibular illusion training method, after step e6 "determining the standby vestibular illusion recognition strength corresponding to the preset classifier with the highest accuracy as the target vestibular illusion recognition strength", the method may further include the following steps:

[0288] Step e7, obtaining each target micro-cluster corresponding to the preset classifier with the highest accuracy.

[0289] Specifically, the electronic device may obtain each target micro-cluster corresponding to the preset classifier with the highest accuracy.

[0290] Step e8, updating each target micro-cluster according to the relationship between the training vestibular illusion strength corresponding to each target micro-cluster and the target vestibular illusion recognition strength, to generate an updated micro-cluster.

[0291] Specifically, the electronic device may update the features corresponding to each target micro-cluster according to the relationship between the training vestibular illusion strength corresponding to each target micro-cluster and the target vestibular illusion recognition strength, and generate an updated micro-cluster.

[0292] Exemplarily, the electronic device may use an exponential decay function to calculate the change in the target micro-cluster weight over time, as shown in formula (33).

[0293] (33)

[0294] In formula (33), λ is the decay rate, Eis the time that has passed since the target microcluster was last updated.

[0295] Therefore, the reliability of the target micro-cluster gradually decreases over time. However, if the target micro-cluster is selected to make a correct classification, the reliability of the target micro-cluster increases by 1.

[0296] After eliminating obsolete target microclusters, physiological parameters are transmitted in real time χ Incrementally added to the nearest target micro-cluster, the target micro-cluster is updated according to formulas (34)-(37).

[0297] (34)

[0298] (35)

[0299] (36)

[0300] (37)

[0301] Among them, in formulas (34)-(37) is the current time.

[0302] Step e9: updating the initial classification model according to each updated micro-cluster.

[0303] Specifically, the electronic device updates the initial classification model according to each updated micro-cluster.

[0304] In an optional implementation of the present application, the vestibular illusion training method may further include the following steps:

[0305] Step e10: for each preset classifier, if a preset number of target microclusters is not selected from each initial microclusters according to the first similarity, then newly added microclusters are generated based on the physiological parameters.

[0306] Specifically, for each preset classifier, if the first similarity between the physiological parameter and each of the initial microclusters in the initial classification model is less than the preset similarity threshold, that is, there is no target microcluster in each of the initial microclusters that is relatively similar to the physiological parameter, then it is determined that there is no target microcluster in each of the initial microclusters. That is, the electronic device does not select a preset number of target microclusters from each of the initial microclusters, then the electronic device can generate a new microcluster based on the physiological parameter, and the new microcluster includes the physiological parameter.

[0307] Step e11: calculating the second similarity between the newly added micro-cluster and each initial micro-cluster.

[0308] Specifically, the electronic device may calculate the second similarity between the newly added feature and each of the initial features based on the newly added feature corresponding to the newly added microcluster and the initial features corresponding to the initial microcluster, thereby obtaining the second similarity between the newly added microcluster and each of the initial microcluster.

[0309] The second similarity may be Euclidean distance, cosine distance, or other parameters. The embodiment of the present application does not specifically limit the second similarity.

[0310] Step e12: according to each of the second similarities, determining the training vestibular illusion intensity corresponding to the initial micro-cluster with the largest second similarity as the target vestibular illusion intensity.

[0311] Specifically, the electronic device may compare the second similarities between the newly added micro-cluster and each of the initial micro-clusters, and then determine the training vestibular illusion intensity corresponding to the initial micro-cluster with the largest second similarity as the target vestibular illusion intensity.

[0312] Step e13, obtaining the first total number of micro-clusters currently corresponding to the initial classification model.

[0313] The first total number of micro-clusters includes the initial number of micro-clusters currently included in the initial classification model and the newly added number of micro-clusters.

[0314] Specifically, the electronic device counts the initial micro-clusters and newly added micro-clusters in the initial classification model to determine a first total number of micro-clusters currently corresponding to the initial classification model.

[0315] Step e14: comparing the first total number of micro-clusters with a preset total number of micro-clusters threshold.

[0316] Specifically, the electronic device may receive a preset total micro-cluster number threshold value input by a user, may receive a preset total micro-cluster number threshold value sent by other devices, or may set the preset total micro-cluster number threshold value according to a memory of the electronic device.

[0317] The embodiment of the present application does not specifically limit the manner in which the electronic device obtains the preset total micro-cluster number threshold.

[0318] Then, the electronic device compares the first total number of micro-clusters with a preset total number of micro-clusters threshold.

[0319] Step e15: if the first total number of micro-clusters is greater than the preset total number of micro-clusters threshold, merging the newly added micro-cluster with the initial micro-cluster with the second largest similarity to generate the first merged micro-cluster.

[0320] Specifically, the electronic device may compare the second similarities between the newly added micro-cluster and each of the initial micro-clusters, and then merge the newly added micro-cluster with the initial micro-cluster with the largest second similarity to generate the first merged micro-cluster.

[0321] In an optional implementation manner of the present application, after the above step b11, the following steps may also be included:

[0322] Step e16, obtaining the second total number of micro-clusters currently corresponding to the initial classification model.

[0323] Specifically, the electronic device may count the current micro-clusters in the initial classification model to obtain the second total number of micro-clusters currently corresponding to the initial classification model.

[0324] Step e17: if the second total number of microclusters is greater than the preset total number of microclusters threshold, and the newly added microclusters do not exist in the initial classification model, then the current number of microclusters corresponding to each of the training vestibular illusion intensities in the initial classification model is obtained.

[0325] Specifically, the electronic device may compare the second total number of microclusters with the preset total number of microclusters threshold. If the second total number of microclusters is greater than the preset total number of microclusters threshold, the electronic device may detect whether there is a newly added microcluster in the initial classification model. If the newly added microcluster does not exist in the initial classification model, the electronic device counts the training vestibular illusion strengths corresponding to each current microcluster in the initial classification model, thereby determining the current number of microclusters corresponding to each training vestibular illusion strength.

[0326] Step e18, calculating the third similarity between the current micro-clusters corresponding to the training vestibular illusion strength with the largest number of current micro-clusters.

[0327] Specifically, the electronic device may calculate, based on a preset similarity algorithm, the third similarity between the current micro-clusters corresponding to the training vestibular illusion strength having the largest number of current micro-clusters.

[0328] Among them, the preset similarity algorithm can be any one of the algorithms such as the Euclidean distance algorithm, the cosine similarity algorithm, the Pearson correlation coefficient algorithm, etc. The embodiment of the present application does not specifically limit the preset similarity algorithm.

[0329] Step e19: merging the current micro-clusters with the closest third similarity to generate a second merged micro-cluster.

[0330] Specifically, the electronic device may merge the current micro-clusters with the closest third similarity to generate a second merged micro-cluster.

[0331] Step e20: updating the initial classification model according to the first merged micro-cluster and the second merged micro-cluster to generate a target classification model.

[0332] Specifically, the electronic device may update the initial classification model according to the first merged micro-cluster and the second merged micro-cluster to generate a target classification model.

[0333] Specifically, in order to better introduce the vestibular illusion training method provided by the application embodiment, as Figure 7 As shown, the embodiment of the present application provides a flow chart for identifying physiological parameters and determining the target vestibular illusion recognition strength corresponding to the subject. Figure 7 As shown, the electronic device can obtain physiological parameters, which may include electrocardiogram (ECG), pulse signal (PPG), and respiration signal (RESP). Then the electronic device performs data preprocessing on the physiological parameters, and divides the physiological parameters using a sliding time window to generate a sliding time window data stream. The electronic device determines the label information corresponding to the physiological parameters based on the K-NN integrated classifier. The K-NN integrated classifier is generated based on the initial model and the preset classifier. The label information is the target vestibular illusion recognition strength corresponding to the physiological parameters. If the electronic device determines the label information corresponding to the physiological parameters, a representative learning method based on error driving is performed to update the target micro-cluster in the initial classification model, thereby updating the initial classification model. If the electronic device fails to determine the label information corresponding to the physiological parameters, an exponential decay function is used to discard the micro-clusters with negative confidence / low negative confidence. Then, it is determined whether the physiological parameters are within the radius R of the nearest micro-cluster. If so, the micro-cluster is updated; if not, a new micro-cluster is created to update the initial classification model. The initial classification model is obtained by training the k-means clustering algorithm based on the initial training set, and the initial classification model includes multiple initial micro-clusters.

[0334] Step S305 , comparing the target vestibular illusion recognition type with the preset vestibular illusion training type, and comparing the target vestibular illusion recognition intensity with the preset vestibular illusion training intensity.

[0335] For details about this step, please refer to the above description of step S205, which will not be elaborated here.

[0336] Step S306: adjusting the preset test control instruction according to the comparison result.

[0337] For details about this step, please refer to the above description of step S206, which will not be elaborated here.

[0338] The vestibular illusion training method provided in the embodiment of the present application obtains an initial training data set, inputs the initial training data set into an initial classification network, and the initial classification network recognizes each piece of initial training data, generates multiple initial micro-clusters based on the recognition results, and determines the training vestibular illusion strength corresponding to each initial micro-cluster, thereby ensuring the accuracy of the determined training vestibular illusion strength corresponding to each initial micro-cluster. Based on each initial micro-cluster, an initial classification model is generated, thereby ensuring the accuracy of the generated initial classification model.

[0339] The number of preset classifiers and the number of preset microclusters corresponding to each preset classifier are obtained; for each preset classifier, the first similarity between the physiological parameter and each initial microcluster in the initial classification model is calculated to ensure the accuracy of the calculated first similarity. According to the first similarity, a target microcluster with a preset number of microclusters is selected from each initial microcluster to ensure the accuracy of the determined target microcluster. According to the training vestibular illusion strength corresponding to each target microcluster, the training vestibular illusion strength with the largest number is determined as the backup vestibular illusion recognition strength corresponding to the preset classifier, to ensure the accuracy of the backup vestibular illusion recognition strength corresponding to each preset classifier, and then the target vestibular illusion recognition strength can be determined according to each backup vestibular illusion recognition strength. According to the backup vestibular illusion recognition strength corresponding to each preset classifier, the accuracy corresponding to each preset classifier is determined to ensure the accuracy corresponding to the determined preset classifier. The backup vestibular illusion recognition strength corresponding to the preset classifier with the highest accuracy is determined as the target vestibular illusion recognition strength, to ensure the accuracy of the determined target vestibular illusion recognition strength, and to ensure the robustness of the algorithm.

[0340] Obtain each target micro-cluster corresponding to the preset classifier with the highest accuracy; update each target micro-cluster according to the relationship between the training vestibular illusion strength and the target vestibular illusion recognition strength corresponding to each target micro-cluster, and generate an updated micro-cluster, thereby ensuring the accuracy of the generated updated micro-cluster. Update the initial classification model according to each updated micro-cluster, thereby ensuring the accuracy of the update of the initial classification model, so that the target vestibular illusion recognition strength corresponding to the physiological parameters generated later can be determined based on the updated initial classification model.

[0341] For each preset classifier, if the target micro-cluster with the preset number of micro-clusters is not selected from each initial micro-cluster according to the first similarity, a new micro-cluster is generated based on the physiological parameters, ensuring that the new micro-cluster can represent the physiological parameters. The total number of micro-clusters in the initial classification model is obtained; if the total number of micro-clusters is greater than the preset total number of micro-clusters threshold, the second similarity between the new micro-cluster and each initial micro-cluster is calculated, and the new micro-cluster and the initial micro-cluster with the largest second similarity are merged according to each second similarity to generate a merged first merged micro-cluster, thereby reducing the total number of micro-clusters in the initial classification model, thereby reducing the occupied resources and speeding up the calculation speed. If there is no new micro-cluster, the current number of micro-clusters corresponding to each training vestibular illusion intensity in the initial classification model is obtained; the Euclidean distance between each current micro-cluster corresponding to the training vestibular illusion intensity with the largest current number of micro-clusters is calculated to ensure the accuracy of the calculated Euclidean distance; the current micro-clusters with the closest Euclidean distance are merged to generate a second merged micro-cluster, that is, the total number of micro-clusters in the initial classification model is reduced, and the current number of micro-clusters corresponding to each training vestibular illusion intensity is ensured not to be too small. The initial classification model is updated according to the second merged micro-cluster, thereby ensuring the accuracy of updating the initial classification model.

[0342] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A vestibular illusion training method, characterized in that: The method comprises: Obtaining a preset test control instruction corresponding to a preset vestibular illusion training type and a preset vestibular illusion training intensity; Performing a vestibular illusion test on the subject based on the preset test control instruction, and obtaining corresponding motion parameters and physiological parameters of the subject; Identifying the motion parameters to determine the target vestibular illusion identification type corresponding to the subject; Identifying the physiological parameters to determine the target vestibular illusion recognition intensity corresponding to the subject; Comparing the target vestibular illusion recognition type with the preset vestibular illusion training type, and comparing the target vestibular illusion recognition intensity with the preset vestibular illusion training intensity; According to the comparison result, the preset test control instruction is adjusted; The motion parameters include a head motion parameter sequence and a body motion parameter sequence, and the identifying of the motion parameters to determine the target vestibular illusion identification type corresponding to the subject includes: Performing data conversion on the head motion parameter sequence to generate a head grayscale image, and performing data conversion on the body motion parameter sequence to generate a body grayscale image; Converting the head grayscale image into a head color image, and converting the body grayscale image into a body color image; splicing the head color image and the body color image to generate a target color image; The target color image is identified to determine the target vestibular illusion identification type corresponding to the subject.

2. The method according to claim 1, characterized in that The step of performing data conversion on the head motion parameter sequence to generate a head grayscale image includes: Splitting the head motion parameter to generate a plurality of mutually non-intersecting head motion parameter subsequences; Calculating the correlation integral between the head motion parameter subsequences according to the relationship between the head motion parameter subsequences; Calculating the autocorrelation characteristics between the head motion parameter subsequences according to the correlation integral between the head motion parameter subsequences; Determine the delay time and embedding dimension corresponding to the head motion parameter sequence according to the autocorrelation characteristics between each of the head motion parameter subsequences; Generate a head phase space corresponding to the head motion parameter sequence based on the delay time and the embedding dimension; The head grayscale image is generated according to the distances between the head vectors in the head phase space.

3. The method according to claim 1, characterized in that The head motion parameter sequence includes a linear acceleration motion parameter sequence of the head in the X, Y, and Z axes and an angular acceleration motion parameter sequence of the head in the X, Y, and Z axes respectively; the head grayscale image includes an acceleration grayscale image of the head in the X axis, a grayscale image of the head in the Y axis, a grayscale image of the head in the Z axis, a grayscale image of the head angular acceleration in the X axis, a grayscale image of the head angular acceleration in the Y axis, and a grayscale image of the head angular acceleration in the Z axis; the head color image includes a head linear acceleration color image and a head angular acceleration color image; wherein, converting the head grayscale image into a head color image includes: Determine a first red channel from the grayscale image of the acceleration of the head on the X axis, the grayscale image of the acceleration of the head on the Y axis, and the grayscale image of the acceleration of the head on the Z axis, determine a first green channel from the two grayscale images except the first red channel, and determine the last grayscale image as a first blue channel; generating the head linear acceleration color image according to the determined first red channel, the first green channel and the first blue channel; Determine a second red channel from the grayscale image of the angular acceleration of the head on the X-axis, the grayscale image of the angular acceleration of the head on the Y-axis, and the grayscale image of the angular acceleration of the head on the Z-axis, determine a second green channel from the two grayscale images except the second red channel, and determine the last grayscale image as the second blue channel; The head angular acceleration color image is generated according to the determined second red channel, the second green channel, and the second blue channel.

4. The method according to claim 1, characterized in that: The step of identifying the target color image and determining the target vestibular illusion identification type corresponding to the subject includes: The target color image is input into a preset image recognition model, and the preset image recognition model performs feature recognition and feature extraction on the target color image, and outputs the target vestibular illusion recognition type based on the extracted features; wherein the preset image recognition model includes a multi-layer perceptron layer, a window multi-head self-attention layer, a shift window-based multi-head self-attention layer, and a normalization layer.

5. The method according to claim 1, characterized in that The identifying of the physiological parameters to determine the target vestibular illusion recognition strength corresponding to the subject includes: Obtain an initial classification model; The physiological parameters are identified, and the target vestibular illusion recognition strength corresponding to the subject is determined based on the initial classification model.

6. The method according to claim 5, characterized in that The obtaining of the initial classification model comprises: Acquire an initial training data set, wherein the initial training data set includes a plurality of initial training data, each of which includes a training physiological parameter and a training vestibular illusion intensity corresponding to the training physiological parameter; Inputting the initial training data set into an initial classification network, the initial classification network recognizes each piece of the initial training data, generates a plurality of initial micro-clusters according to the recognition result, and determines the training vestibular illusion intensity corresponding to each of the initial micro-clusters; The initial classification model is generated according to each of the initial micro-clusters.

7. The method according to claim 6, characterized in that The step of identifying the physiological parameters and determining the target vestibular illusion recognition strength corresponding to the subject based on the initial classification model includes: Obtaining the number of preset classifiers and the number of preset microclusters corresponding to each of the preset classifiers; For each of the preset classifiers, calculating a first similarity between the physiological parameter and each of the initial micro-clusters in the initial classification model; Selecting the preset number of target micro-clusters from each of the initial micro-clusters according to the first similarity; According to the training vestibular illusion strengths corresponding to the target micro-clusters, determining the training vestibular illusion strength with the largest number as the backup vestibular illusion recognition strength corresponding to the preset classifier; Determining the accuracy corresponding to each of the preset classifiers according to the backup vestibular illusion recognition strength corresponding to each of the preset classifiers; The backup vestibular illusion recognition strength corresponding to the preset classifier with the highest accuracy is determined as the target vestibular illusion recognition strength.

Images