Customer risk early warning system based on behavior analysis

Abstract

The application discloses a customer risk early warning system based on behavior analysis and relates to the technical field of behavior analysis.The system comprises a motion capture module, a human body model module, a virtual monitoring module, a motion recognition module, a first motion judgment module, a second motion judgment module and a risk early warning module.The motion capture module is used for being worn on the customer's body, monitoring the customer's motion and generating motion data.The customer risk early warning system based on behavior analysis can calculate the proportion of the user's physical and mental state in the negative state in the overall judgment, judge whether the negative state will cause adverse effects on the customer's physical and mental health, so as to timely perform early warning, remind the user to adjust the state or perform guidance and ensure the customer's physical and mental health and safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of behavior analysis, in particular to a customer risk early warning system based on behavior analysis. BACKGROUND

[0002] Human behavior reflects the psychological state and physical state of a person to some extent, and in order to ensure the physical and mental health and safety of a person, timely detection of abnormal conditions and prevention of possible risks, it is necessary to regularly detect the body and mind.

[0003] The prior art disclosed in Publication No. CN117216650A discloses a target object risk behavior early warning method and related equipment. In the specific method, first, emotion analysis is performed based on a preset emotion analysis model and a plurality of historical information published by the target object, and then it is determined whether the target object is emotionally abnormal. Then, a preset risk behavior analysis model is used to determine the risk behavior type that the target object is most likely to have. Finally, information mining is performed on the historical information corresponding to the risk behavior type, so as to clearly present various key information related to the risk behavior type. Finally, based on the key information related to the risk behavior, risk behavior early warning of the target object is realized. It can perform deep risk behavior information mining on the emotionally abnormal object and perform early warning based on the mined risk behavior information, thereby improving the accuracy of risk behavior early warning of the target object.

[0004] The above-mentioned prior art mainly analyzes the remarks published by the customer on the network to determine whether the customer's emotion is abnormal. However, on the one hand, the customer's emotional state is determined only from the remarks published by the customer on the network, which has certain limitations, so that the determination result may have certain deviation. Moreover, for customers who rarely comment on the network, the collected information will be too little, which further reduces the accuracy of the determination result, and even the sample may be too small to determine. SUMMARY

[0005] The purpose of the present application is to provide a customer risk early warning system based on behavior analysis to solve the above-mentioned deficiencies in the prior art.

[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical solution: a customer risk early warning system based on behavior analysis, comprising a motion capture module, a human body model module, a virtual monitoring module, a motion recognition module, a first motion judgment module, a second motion judgment module, and a risk early warning module.

[0007] The motion capture module is used to be worn on the customer's body to monitor the customer's motion and generate motion data. The motion capture module can use a portable motion capture device, which is convenient for the customer to wear and reduces the impact on the customer's normal life.

[0008] The human body model module is configured to establish a human body skeleton model, acquire the action data, and control the human body skeleton model to make actions corresponding to the action data; wherein the human body skeleton model is a human three-dimensional model after being skinned with a skeleton, so that the human three-dimensional model can imitate the corresponding active joints of the human body to make various actions by controlling the activities of the skeleton;

[0009] The virtual monitoring module is configured to monitor the human body skeleton model to obtain a model action video composed of continuous image frames;

[0010] The action recognition module is configured to set key action tags, and mark key action images in historical model action videos using the key action tags; wherein a plurality of key action tags can be set according to different key actions, and the marking can be achieved by associating the key action images with the corresponding key action tags;

[0011] The action recognition module is further configured to train a first neural network model based on the model action video, the key action images, and the key action tags to obtain a key action recognition model for identifying key action images from an input model action video and outputting corresponding key action tags; wherein the model action video is decomposed into continuous image frames during training, and the key action recognition model also decomposes the input model action video into continuous image frames, identifies key action images from the image frames, and outputs key action tags corresponding to the identified key action images; preferably, other action tags can also be set during training and associated with non-key action images, so that the key action recognition model outputs other action tags when the image frames identified during the identification process are not key action images; and the present application does not limit the specific neural network model, and a neural network model can be selected according to actual needs and training conditions during model training, so that the accuracy of the trained neural network model meets the set standard; as a reference, a CNN model can be selected;

[0012] The first action judgment module is configured to set frequency thresholds for each key action tag, acquire the number of times each key action tag appears within a set first time length to obtain a first frequency, and judge the customer state based on the relationship between the first frequency of the key action tag and the corresponding frequency threshold; wherein the customer state includes a positive state and a negative state;

[0013] The second action judgment module is configured to set a plurality of action time length thresholds corresponding to the key action tags one by one;

[0014] The second action judgment module is further configured to extract corresponding key action images with the same key action label and continuity in the model action video from the model action video, obtain corresponding key action steps, and acquire key action step durations, judge the customer state based on a size relationship between the key action step durations corresponding to the same key action label and the action duration threshold.

[0015] The risk warning module is configured to acquire the judged customer state, calculate the proportion of each state, and judge whether the customer's physical and mental health will be affected based on the proportion.

[0016] Further, the system further comprises a third judgment module configured to mark image frames of non-key action images in the model action video as other action images, and judge the customer state based on analysis of feature correlation between the current other action image and historical other action images in a positive state.

[0017] Further, the third judgment module judges the customer state based on analysis of feature correlation between the current other action image and historical other action images in a positive state, and comprises the following steps:

[0018] The third judgment module is configured to acquire historical normal other action images by acquiring historical other action images in which both the key action images before and after the historical other action images are in a positive state from historical model action videos; for example, a group of historical other action images is acquired if there is a group of key action images in a positive state before the group of historical other action images and a group of key action images in a flat state after the group of historical other action images; and the group of historical other action images is not acquired if there is a group of key action images in a positive state before the group of historical other action images and a group of key action images in a negative state after the group of historical other action images.

[0019] The third judgment module is configured to associate the corresponding other action distinguishing label with the normal historical other action images with the same type of other action based on the type of other action corresponding to the normal historical other action images.

[0020] The third judgment module is configured to train a second neural network model based on the historical other action images, the normal historical other action images, and the other action distinguishing labels and other action abnormal labels corresponding to the normal historical other action images, obtain an abnormality recognition model, identify and select normal historical other action images and corresponding other action distinguishing labels based on input other action images, associate the remaining other action images with other action abnormal labels, and then output the other action abnormal labels and the other action images associated with the other action abnormal labels, wherein the abnormality recognition model keeps the normal historical other action images and the corresponding other action distinguishing labels in a database without directly displaying them, and a worker can check the database to view the normal historical other action images and the corresponding other action distinguishing labels identified and selected.

[0021] If the abnormality recognition model does not output, it is determined that the customer is in a positive state.

[0022] Further, the third determination module determines the customer state based on analysis of the feature relevance between the current other action image and the historical other action image in the positive state, and further comprises the following steps:

[0023] An other action image associated with the other action abnormality label is obtained to obtain a to-be-determined other action image, and a similarity between the to-be-determined other action image and a normal historical other action image corresponding to each other action difference label is calculated;

[0024] A similarity determination threshold is set to determine whether the highest similarity calculated is greater than the similarity determination threshold;

[0025] If yes, the to-be-determined other action image and the other action corresponding to the other action difference label corresponding to the highest similarity are of the same type, i.e., the type of the other action corresponding to the to-be-determined other action image is determined, and it is determined that the customer is in a negative state;

[0026] If no, the to-be-determined other action image is marked as an unrecognizable image.

[0027] Further, the first action determination module is configured to set a frequency threshold of each key action label, obtain a number of times each key action label appears within a set first time length to obtain a first frequency, and determine the customer state based on the relationship between the first frequency of the key action label and the corresponding frequency threshold, comprising the following steps:

[0028] The frequency threshold of each key action label is set;

[0029] The key action recognition model outputs a key action label and its output time;

[0030] The key action label is marked on the time axis;

[0031] A sliding box with a first time length is set on the time axis, and the sliding box is made to slide on the time axis, and the right end of the sliding box lags behind the current time by a second time length set by the current time. The number of the same key action labels in the current sliding box is the current first frequency of the corresponding key action label, i.e., the first frequency of each key action label is updated in real time with the sliding of the sliding box. The second time length can be self-defined according to the delay time of the key action recognition model output, so that the second time length is slightly greater than the output delay time of the key action recognition model;

[0032] It is determined whether the first frequency of each key action label is greater than the corresponding frequency threshold;

[0033] If yes, the customer state of the time period in which the sliding box is located is determined as a negative state, and if no, the customer state of the time period in which the sliding box is located is determined as a positive state.

[0034] Further, the second action judgment module is further configured to extract, from the model action video, key action images corresponding to the same key action label and being continuous in the model action video, to obtain corresponding key action steps, and to obtain key action step durations, and to judge the customer state based on a size relationship between the key action step durations corresponding to the same key action label and the action duration threshold, including the following steps:

[0035] extracting, from the model action video, key action images corresponding to the same key action label and being continuous in the model action video, to obtain corresponding key action steps;

[0036] searching whether there are images that cannot be recognized adjacent to each key action step;

[0037] if yes, updating the images that cannot be recognized to a corresponding position in the key action step, wherein updating to the key action step means adding the images that cannot be recognized to the key action step and marking the images that cannot be recognized as the corresponding key action step, and updating to the corresponding position means that if the images that cannot be recognized are in front of the key action images, the images that cannot be recognized are still in front of the key action images after being updated to the key action step;

[0038] calculating key action step durations, and judging whether the key action step durations are greater than corresponding key action duration thresholds;

[0039] if yes, determining that the customer is in a negative state, and if no, determining that the customer is in a positive state.

[0040] Further, the risk warning module is configured to obtain the customer state determined by the first action judgment module, the second action judgment module and the third action judgment module, to calculate the proportion of each state, and to judge whether the state will affect the physical and mental health of the customer based on the proportion, including the following steps:

[0041] obtaining the number of negative states and positive states of the customer determined by the first action judgment module, the second action judgment module and the third action judgment module;

[0042] setting a state judgment threshold, and judging whether the proportion of the negative state is greater than the state judgment threshold;

[0043] if yes, performing a warning, determining that the negative state will adversely affect the physical and mental health of the customer, and needing to adjust the state in time to ensure the physical and mental health and safety, and if no, the negative state is within a normal range and will not affect the physical and mental health of the customer.

[0044] 1. Compared with the prior art, the customer risk early warning system based on behavior analysis provided by the application can capture and record the actions of customers in real time through the setting of the action capture module, the human body model module, the virtual monitoring module, the action recognition module, the first action judgment module, the second action judgment module and the risk early warning module, and can judge the physical and mental state of the user based on the frequency, type and duration of the action, so as to improve the judgment accuracy.

[0045] 2. Compared with the prior art, the customer risk early warning system based on behavior analysis provided by the application can calculate the proportion of the user's physical and mental state in the negative state in the total judgment, judge whether the negative state will cause adverse effects on the physical and mental health of the customer, so as to timely perform early warning, remind the user to adjust the state or perform guidance, and ensure the physical and mental health and safety of the customer. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0047] Figure 1 The system structure block diagram provided for the embodiments of the present application;

[0048] Figure 2 The first action judgment module judgment step diagram provided for the embodiments of the present application;

[0049] Figure 3 The second action judgment module judgment step diagram provided for the embodiments of the present application;

[0050] Figure 4 The third action judgment module judgment step diagram provided for the embodiments of the present application;

[0051] Figure 5 The system work overall step diagram provided for the embodiments of the present application. DETAILED DESCRIPTION

[0052] In order to make those skilled in the art better understand the technical solutions of the present application, the present application will be further described in detail with reference to the drawings.

[0053] Embodiments described herein can be described with reference to plan views and / or cross-sectional views by virtue of the idealized schematic nature of the drawings. In other words, the illustrations presented herein are not necessarily drawn to scale. Accordingly, the embodiments are not limited to the illustrations presented herein but include what is practiced in the art of manufacturing and / or design. Thus, the zones illustrated in the drawings have a schematic nature and the shapes of the zones illustrated in the drawings are not intended to be limiting, but rather are intended to illustrate the specific shape of the zones of the elements. Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.

[0054] Referring to Figures 1-5 , the customer risk early warning system based on behavior analysis comprises an action capture module, a human body model module, a virtual monitoring module, an action recognition module, a first action judgment module, a second action judgment module, and a risk early warning module.

[0055] The action capture module is used to monitor the action of the customer and generate action data, and the action capture module can be a portable action capture device to facilitate the customer to wear and reduce the impact on the normal life of the customer.

[0056] The human body model module is used to establish a human body skeleton model, obtain action data, and control the human body skeleton model to make actions corresponding to the action data. The human body skeleton model is a human three-dimensional model after skinning with a skeleton, so that the human three-dimensional model can imitate the corresponding active joints of the human body by controlling the activity of the skeleton to drive the human three-dimensional model to make various actions.

[0057] The virtual monitoring module is used to monitor the human body skeleton model to obtain a model action video, and the model action video is composed of continuous image frames.

[0058] The action recognition module is used to set a key action label and mark key action images in historical model action videos using the key action label. The key action label can be set in multiple according to different key actions, and the marking can be realized by associating the key action images with the corresponding key action label.

[0059] The action recognition module is further configured to train a first neural network model based on the model action video, the key action image and the key action label, to obtain a key action recognition model for recognizing the key action image from the input model action video and outputting the corresponding key action label. During the training, the model action video is decomposed into continuous image frames. The key action recognition model also decomposes the input model action video into continuous image frames, recognizes the key action image from the image frames, and outputs the key action label corresponding to the recognized key action image. Preferably, other action labels can also be set during the training to be associated with non-key action images, so that when the key action recognition model identifies an image frame that is not a key action image during the recognition process, it outputs the other action label. The present application does not limit the specific neural network model. During the training of the model, a neural network model can be selected according to the actual needs and training conditions, so that the accuracy of the trained neural network model meets the set standard. As a reference, a CNN model can be selected.

[0060] The first action judgment module is configured to set a frequency threshold for each key action label, obtain the number of times each key action label appears within a set first time length, obtain a first frequency, and judge the customer state based on the relationship between the first frequency of the key action label and the corresponding frequency threshold. The customer state includes a positive state and a negative state, and includes the following steps:

[0061] A1, set a frequency threshold for each key action label;

[0062] A2, obtain the key action label output by the key action recognition model and its output time;

[0063] A3, mark the key action label on the time axis;

[0064] A4, set a sliding box with a first time length on the time axis, and make the sliding box slide on the time axis. When the sliding box slides, the right end of the sliding box lags behind the second time length set by the current time. The number of the same key action labels in the current sliding box is the current first frequency of the corresponding key action label, that is, the first frequency of each key action label is updated in real time with the sliding of the sliding box. The second time length can be self-defined according to the delay time of the key action recognition model output, so that the second time length is slightly greater than the delay time of the key action recognition model output;

[0065] A5, judge whether the first frequency of each key action label is greater than the corresponding frequency threshold;

[0066] A6, if yes, the customer state of the time period where the sliding box is located at this time is judged to be a negative state, and if no, the customer state of the time period where the sliding box is located at this time is judged to be a positive state.

[0067] The second action judgment module is configured to set a plurality of action duration thresholds corresponding to key action labels one by one; the second action judgment module is further configured to extract key action images corresponding to the same key action label and continuous in the model action video from the model action video, obtain corresponding key action steps, and acquire key action step durations, judge the customer state based on the size relationship between the key action step durations corresponding to the same key action label and the action duration thresholds, and include the following steps:

[0068] B1. Extracting key action images corresponding to the same key action label and continuous in the model action video from the model action video to obtain corresponding key action steps;

[0069] B2. Retrieving whether there are adjacent unrecognizable images before and after each key action step;

[0070] B3. If yes, updating the unrecognizable images to the corresponding positions in the key action steps, wherein updating to the key action steps means adding the unrecognizable images to the key action steps and marking them as the corresponding key action steps, and updating to the corresponding positions means that if the unrecognizable images start before the key action images, they are still updated before the key action images after being updated to the key action steps;

[0071] B4. Calculating the key action step durations and judging whether the key action step durations are greater than the corresponding key action duration thresholds;

[0072] B5. If yes, judging that the customer is in a negative state, and if not, judging that the customer is in a positive state.

[0073] The system further includes a third judgment module configured to mark image frames of non-key action images in the model action video as other action images and judge the customer state based on analysis of the feature correlation between the current other action images and historical other action images in a positive state. The third judgment module judges the customer state based on analysis of the feature correlation between the current other action images and historical other action images in a positive state, and includes the following steps:

[0074] C1. Obtaining historical normal other action images by acquiring historical other action images with both the key action images before and after being adjacent in the historical model action video being in a positive state; for example, a group of historical other action images, there are a group of key action images in an active state before them and a group of key action images in a flat state after them, so the group of historical other action images are acquired; a group of historical other action images, there are a group of key action images in an active state before them and a group of key action images in a negative state after them, so the group of historical other action images are not acquired;

[0075] C2, according to the category of the other action corresponding to the normal historical other action image, setting the corresponding other action distinguishing label is associated with the normal historical other action image corresponding to the category of the other action;

[0076] C3, based on the historical other action image, the normal historical other action image and the corresponding other action distinguishing label, the other action abnormal label, training the second neural network model, obtaining the abnormal recognition model, for identifying the normal historical other action image and the corresponding other action distinguishing label in the input other action image, and associating the remaining other action image with the other action abnormal label, and then outputting the other action abnormal label and the other action image associated with it, wherein the abnormal recognition model identifies the normal historical other action image and the corresponding other action distinguishing label, and does not directly display it in the database, and the staff can check the database to view the normal historical other action image and the corresponding other action distinguishing label identified;

[0077] C4, if the abnormal recognition model does not output, it is judged that the customer is in a positive state.

[0078] C5, obtaining the other action image associated with the other action abnormal label, obtaining the other action image to be judged, and calculating the similarity between the other action image to be judged and the normal historical other action image corresponding to each other action distinguishing label;

[0079] C6, setting a similarity judgment threshold, judging whether the highest similarity calculated is greater than the similarity judgment threshold;

[0080] C7, if yes, the other action image to be judged is the same as the other action distinguishing label corresponding to the highest similarity, that is, the category of the other action corresponding to the other action image to be judged can be determined, and the customer is judged to be in a negative state;

[0081] C8, if not, the other action image to be judged is marked as an unidentifiable image.

[0082] The risk warning module is used to obtain the customer state, calculate the proportion of each state, and judge whether it will affect the physical and mental health of the customer based on the proportion, including the following steps:

[0083] D1, obtaining the number of negative states and positive states of the customer judged by the first action judgment module, the second action judgment module and the third action judgment module;

[0084] D2, setting a state judgment threshold, judging whether the proportion of negative state is greater than the state judgment threshold;

[0085] D3, if yes, a warning is given, it is judged that the negative state will cause adverse effects on the physical and mental health of the customer, and the state needs to be adjusted in time to ensure the physical and mental health safety, if no, it is indicated that the negative state is within a normal range and will not affect the physical and mental health of the customer.

[0086] The foregoing merely describes some exemplary embodiments of the present application by way of illustration, and it is needless to say that the described embodiments can be modified in various ways without departing from the spirit and scope of the present application for those skilled in the art. Therefore, the above drawings and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the present application.

Claims

1. A customer risk early warning system based on behavioral analysis, characterized in that: The system comprises an action capture module, a human body model module, a virtual monitoring module, an action recognition module, a first action judgment module, a second action judgment module, and a risk warning module. The action capture module is used to monitor the action of the customer and generate action data. The human body model module is used to establish a human body skeleton model, obtain the action data, and control the human body skeleton model to make actions corresponding to the action data. The virtual monitoring module is used to monitor the human body skeleton model and obtain a model action video composed of continuous image frames. The action recognition module is used to set a key action label and mark key action images in historical model action videos using the key action label. The action recognition module is further used to train a first neural network model based on the model action video, the key action image, and the key action label, obtain a key action recognition model, identify key action images from an input model action video, and output corresponding key action labels. The first action judgment module is used to set a frequency threshold for each key action label, obtain the number of times each key action label appears within a set first time length, obtain a first frequency, and judge the customer state based on the relationship between the first frequency of the key action label and the corresponding frequency threshold. The customer state includes a positive state and a negative state. The second action judgment module is used to set a plurality of action time length thresholds corresponding to the key action labels. The second action judgment module is further used to extract key action images corresponding to the same key action label and continuous in the model action video from the model action video, obtain corresponding key action steps, obtain key action step time lengths, and judge the customer state based on the size relationship between the key action step time lengths corresponding to the same key action label and the action time length thresholds. The risk warning module is used to obtain the judged customer state, calculate the proportion of each state, and determine whether the customer's physical and mental health will be affected based on the proportion. The system further comprises a third judgment module, which is used to mark image frames of non-key action images in the model action video as other action images, and judge the customer state based on the analysis of the feature correlation between the current other action image and the historical other action image in the positive state. The third judgment module judges the customer state based on the analysis of the feature correlation between the current other action image and the historical other action image in the positive state, comprising the following steps: Obtain historical normal other action images from historical model action videos, wherein the historical normal other action images are historical other action images in which the previous and subsequent adjacent key action images are in the positive state. According to the type of the other action corresponding to the normal historical other action image, set an other action distinguishing label corresponding to the type of the other action and associate the other action distinguishing label with the normal historical other action image corresponding to the same type of the other action. The second neural network model is trained based on historical other action images, normal historical other action images, and corresponding other action distinguishing labels, other action abnormal labels, to obtain an abnormality recognition model, which is used to identify normal historical other action images and corresponding other action distinguishing labels from input other action images, associate the remaining other action images with other action abnormal labels, and then output the other action abnormal labels and the other action images associated therewith. If the abnormality recognition model does not output, it is determined that the customer is in a positive state.

2. The behavioral analytics based client risk alert system as claimed in claim 1, wherein: The third determination module determines the customer state based on the analysis of the feature correlation between the current other action image and the historical other action image in the positive state, and further comprises the following steps: The other action images associated with the other action abnormal labels are obtained to obtain the other action images to be determined, and the similarity between the other action images to be determined and the normal historical other action images corresponding to each other action distinguishing label is calculated. A similarity determination threshold is set to determine whether the highest similarity calculated is greater than the similarity determination threshold. If yes, the other action of the other action distinguishing label corresponding to the highest similarity is the same as the other action of the other action distinguishing label corresponding to the highest similarity, and it is determined that the customer is in a negative state. If no, the other action image to be determined is marked as an unrecognizable image.

3. The behavioral analytics based client risk alerting system as claimed in claim 1, wherein: The first action determination module is used to set the frequency threshold of each key action label, obtain the number of occurrences of each key action label within a set first time length to obtain a first frequency, and determine the customer state based on the relationship between the first frequency of the key action label and the corresponding frequency threshold, comprising the following steps: Set the frequency threshold of each key action label. Obtain the key action label and its output time output by the key action recognition model. Mark the key action label on the time axis. Set a sliding box with a first time length on the time axis, and make the sliding box slide on the time axis, with the right end of the sliding box lagging behind the current time by a second time length set. Determine whether the first frequency of each key action label is greater than the corresponding frequency threshold. If yes, it is determined that the customer state in the time period where the sliding box is located at this time is negative, and if no, it is determined that the customer state in the time period where the sliding box is located at this time is positive.

4. The behavioral analytics based client risk alerting system as claimed in claim 2, wherein: The second action determination module is further used to extract key action images corresponding to the same key action label and continuous in the model action video from the model action video to obtain corresponding key action steps, and obtain the key action step length, and determine the customer state based on the size relationship between the key action step length corresponding to the same key action label and the action length threshold, comprising the following steps: Extract key action images corresponding to the same key action label and continuous in the model action video from the model action video to obtain corresponding key action steps. Determine whether there are unrecognizable images adjacent to each key action step before and after the key action step. If yes, update the unrecognizable images to the corresponding position in the key action step. The key action step duration is calculated, and it is judged whether the key action step duration is greater than a corresponding key action duration threshold value; If yes, the customer is judged as a negative state, and if no, the customer is judged as a positive state.

5. The behavioral analytics based client risk alerting system as claimed in claim 1, wherein: The risk warning module is used to obtain the judged customer state, calculate the proportion of each state, and judge whether it will affect the physical and mental health of the customer based on the proportion, including the following steps: The number of negative states and positive states of the customer judged by the first action judgment module, the second action judgment module, and the third action judgment module is obtained; A state judgment threshold value is set, and it is judged whether the proportion of negative states is greater than the state judgment threshold value; If yes, a warning is given.

Images