A method for initial activation window (IAW) dynamic detection and mood emergency procedure preloading based on a TCM time constraint model

Abstract

This invention discloses a method for dynamic detection of the initial activation window and pre-installation of emotional emergency procedures based on the TCM time constraint model, relating to the fields of artificial intelligence and digital healthcare. The method includes: real-time acquisition of multimodal interaction data, calculation of speech rate mutation index, semantic aggression entropy value, and word frequency repetition density, and construction of a comprehensive discriminant function F_IAW to dynamically determine the initial activation window state where the user enters a logically offline state; once this state is determined, the semantic empathy thread is immediately suspended, and a third-person instruction protocol is activated within 200 milliseconds, attaching emotional attributes to a virtual object role; simultaneously, during the calm window period, the user is guided to pre-install embodied action sequences into a high-priority cache, which are then directly invoked during the outburst period to achieve millisecond-level response. This invention can accurately capture the precursors to emotional outbursts, bypass damaged logical areas through physiologically matched interaction protocols, significantly reduce cognitive load, and achieve immediate intervention.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence, human-computer interaction and digital healthcare, specifically to a dynamic detection algorithm for the premonitory signs (initial activation window, IAW) of emotional outbursts in children based on the Time Constraint Model (TCM), and a control system and method for forcibly switching the interaction protocol and calling a pre-installed emergency program when the window is detected. Background Technology

[0002] In human-computer interaction scenarios for children's emotional intervention, existing intelligent companion systems typically employ a "semantic empathy" strategy. This involves analyzing the child's speech through natural language processing and generating a second-person response with empathy (such as "I know you're angry"). However, this existing technology suffers from the following serious drawbacks:

[0003] First, the timing of intervention is delayed. Existing systems often only identify and intervene after the emotion has fully erupted (such as crying or aggressive behavior), missing the "Initial Activation Window" (IAW) before the emotion escalates. During the IAW, the child's prefrontal cortex gradually loses its logical processing capacity due to the reallocation of metabolic resources. Continuing to input complex semantic empathic information at this time is not only ineffective but may also exacerbate the cognitive load.

[0004] Second, there is a mismatch in the interaction protocols. During the IAW and subsequent emotional outburst, the child's logical brain regions are in a "physical blackout" state, unable to process second-person logical attributions. Current technology fails to recognize this physiological state and still forcibly uses dialogue protocols based on prefrontal cortex cognition, leading to intervention failure.

[0005] Third, the response latency is too high. Existing systems typically require large cloud-based models to perform real-time inference and generate responses after detecting emotional abnormalities, which takes a considerable amount of time (on the order of seconds). However, in emotional emergency rescue scenarios, millisecond-level conditioned reflex responses are needed, which the existing real-time inference architecture cannot meet.

[0006] Fourth, there is a lack of pre-installed mechanisms. The existing system lacks a mechanism to pre-install "emotional emergency procedures" during calm periods, which makes it impossible to invoke proven and effective embodied action instructions during crises.

[0007] Therefore, there is an urgent need for a technical solution that can accurately detect the IAW status, force a switch to the bypass interaction protocol in that status, and simultaneously invoke pre-installed emergency procedures. Summary of the Invention

[0008] This invention aims to address the problems of delayed emotional intervention timing, mismatched interaction protocols, and high response latency in existing technologies. By introducing the TCM time constraint model, establishing the IAW dynamic detection algorithm, and designing an emergency control mechanism based on "pre-installation during calm periods and direct connection during outbreak periods."

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] In a first aspect, the present invention provides a dynamic detection method for the Initial Activation Window (IAW) based on the TCM time constraint model, executed by a computer device, comprising the following steps: acquiring a multimodal interaction data stream of a target user, the data stream including an acoustic feature stream, a text semantic stream, and an optional physiological feature stream; extracting features from the data stream through a dynamic sliding window, calculating the speech rate mutation index (S_delta), semantic aggression entropy value (E_agg), and word frequency repetition density (D_rep); constructing a comprehensive discriminant function F_IAW, the calculation formula being: F_IAW = w1 × S_delta + w2 × E_agg + w3 × D_rep, where w1, w2, and w3 are weight coefficients, and w1+w2+w3=1; dynamically determining whether the user has entered the IAW state: when the value of the comprehensive discriminant function F_IAW exceeds a preset threshold Theta_IAW, the user is determined to have entered the IAW state, the IAW state being defined as a physical blackout period in which the prefrontal cortex of the brain temporarily shuts down its higher logical processing functions due to the redistribution of metabolic resources.

[0011] Furthermore, the preset threshold Theta_IAW is a physical constraint parameter that is dynamically adjusted based on the target user's historical baseline data, age group, and individual differences. The parameter represents the metabolic threshold at which the user's logical functions go offline. The weight coefficients w1, w2, and w3 are dynamically adjusted according to the real-time scenario, but their sum remains at 1.

[0012] Secondly, this invention provides a human-computer emotional interaction protocol switching method based on the TCM time constraint model, comprising: executing the above-mentioned IAW state dynamic detection method to determine that the user has entered the IAW state; forcibly terminating the current second-person semantic empathy protocol based on prefrontal cognitive logic; instantaneously activating a third-person instruction protocol based on visual bypass and limbic system access, wherein the instruction protocol bypasses the user's damaged logical processing area through a preset symbolic interaction, and the execution delay is less than 200 milliseconds; performing a subject-object topology switch, stripping the emotional attribute from the user's ontology identifier and attaching it to a preset virtual object role identifier, wherein the virtual object role identifier and the user's ontology identifier have no reference relationship at the data level.

[0013] Furthermore, the virtual object role identifier is an instantiation mapping based on at least one of the six emotion instruction prototypes preset by the TCM model. Each prototype is bound to a specific embodied action-driven feature code and a narrative logic label. The narrative logic label mandates that the natural language generation model only outputs third-person declarative sentences and prohibits the output of second-person emotion attribution sentences.

[0014] Thirdly, this invention provides a method for pre-installing an emotion emergency program based on the TCM C5 prior installation law, comprising: a monitoring system determining that the user is currently in a non-IAW state, defined as a calm window period; within the calm window period, actively initiating an emotion emergency program pre-installation training interaction to guide the user to learn and execute an embodied action sequence bound to a specific virtual object role; writing the feature code of the embodied action sequence successfully executed by the user into a high-priority cache of the user behavior model and marking it as an emergency ready state; when the user is subsequently detected to have entered an IAW state or an emotional outburst period, directly calling the action instructions in the high-priority cache to drive the terminal to execute feedback, skipping the real-time action planning and reasoning process, and achieving millisecond-level response.

[0015] Furthermore, the embodied action sequence includes specific physical movements, breathing rhythms, or tactile interaction patterns that correspond one-to-one with the six major emotional archetypes; the high-priority cache is stored in the local neural processing unit (NPU) or embedded memory to ensure that it can still be accessed in milliseconds under network outage or high load conditions.

[0016] Fourthly, this invention provides a cross-validation method for a multimodal emotion labeling protocol, comprising: acquiring machine-recognized IAW state markers and user-inputted role naming tags; when the machine recognition result is inconsistent with the user-inputted role naming tags, the system enters a confidence arbitration mode: if the voice energy features or facial expression features of the user-inputted role naming tags conform to the physical mapping rules of a specific role prototype, then the user-inputted role naming tags are adopted first; the machine's IAW detection weight parameters and emotion recognition model are corrected in reverse to form a two-way closed-loop calibration mechanism of "self-labeling + machine recognition".

[0017] Fifthly, the present invention provides an emotion intelligence system based on the TCM time constraint model, comprising: a perception acquisition module, an IAW detection engine, a mode switching controller, a pre-installed training module, a multimodal output module, and a cross-validation module, which are respectively used to execute the above-mentioned method steps.

[0018] In a sixth aspect, the present invention provides an electronic device and a computer-readable storage medium for implementing the above-described method.

[0019] The beneficial effects of this invention are as follows:

[0020] 1. Precisely capture the intervention window: Through a multimodal feature fusion algorithm, the IAW state can be accurately identified 3-5 seconds before the emotional outburst, seizing the opportunity for intervention.

[0021] 2. Physiologically Matched Interaction Protocol: Forced switching to third-person objectified narrative aligns with the neurophysiological characteristics of children during emotional outbursts, significantly reducing cognitive resistance.

[0022] 3. Millisecond-level emergency response: Through pre-installation during quiet periods and a direct connection cache mechanism, the response latency is reduced from seconds to milliseconds, ensuring the immediacy of intervention.

[0023] 4. Scientific closed-loop evaluation: Introduce a high-weighted biological signal (HRV) evaluation system to ensure the objectivity and scientific nature of the intervention effect. Attached Figure Description

[0024] Figure 1 The flowchart of IAW dynamic detection and protocol switching based on the TCM model is provided for embodiments of the present invention.

[0025] Figure 2 This is a schematic diagram of the execution logic of the pre-installed emotional emergency program (C5 prior installation law) provided in an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram illustrating the relationship between the comprehensive discriminant function F_IAW and the dynamic threshold Theta_IAW provided in this embodiment of the invention.

[0027] Figure 4 The flowchart shows the logic of the multimodal collaborative emotion attribution evaluation algorithm provided in this embodiment of the invention. Detailed Implementation

[0028] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0029] Example 1: IAW Dynamic Detection and Forced Language Mode Switching Mechanism

[0030] This embodiment details how the IAW dynamic detection algorithm identifies premonitory signs of emotional outbursts and how the system performs hard switching of language modes. Figure 1 As shown, this method mainly includes four stages: data acquisition, feature calculation, state determination, and protocol switching.

[0031] First, the system performs data acquisition and feature calculation. The system acquires children's speech (acoustic feature stream) in real time through a microphone array and analyzes the children's text semantic stream in real time through a natural language processing module. The system continuously calculates the following three key indices:

[0032] (1) Speech rate change index (S_delta): Calculates the first derivative of the syllable rate per unit time. For example, if the rate of change of the syllable rate is detected to exceed 0.8 within 300 ms, then S_delta is recorded as a high value.

[0033] (2) Semantic aggression entropy value (E_agg): Based on the sentiment lexicon, the density and information entropy of high-arousal negative words within a sliding window (e.g., 1 second) are calculated. For example, if the proportion of negative words suddenly increases from 5% to 30%, E_agg will increase significantly.

[0034] (3) Word frequency repetition density (D_rep): detects the unnatural repetition frequency of specific pronouns or emotional words (such as "no", "don't") within a short period of time (such as 5 seconds).

[0035] Secondly, the IAW state determination algorithm is executed. The system constructs a comprehensive discriminant function F_IAW, the calculation formula of which is as follows:

[0036] F_IAW = w1 × S_delta + w2 × E_agg + w3 × D_rep

[0037] Wherein, w1, w2, and w3 are the weighting coefficients of the speech rate mutation index, the semantic radicality entropy value, and the word frequency repetition density, respectively. In this embodiment, w1=0.4, w2=0.4, and w3=0.2 are preferably set, which can be adjusted according to the actual scenario, and w1+w2+w3=1.

[0038] When the calculated F_IAW value exceeds the dynamically adjusted preset threshold Theta_IAW (for example, a threshold set to 0.75), the system determines that the user has entered the Initial Activation Window (IAW) state. The preset threshold Theta_IAW is a physical constraint parameter dynamically adjusted based on the target user's historical baseline data, age group, and individual differences, representing the metabolic critical point at which the user's logical functions become offline.

[0039] Finally, a forced language mode switch is executed. Once the IAW state is determined, the system kernel immediately suspends the current "semantic empathy generation thread" and activates the "role naming generation thread" within 200 milliseconds. This thread forces the output of third-person declarative sentences with an independent virtual object character (such as "touching the fire dragon") as the subject (such as "Look, the fire dragon has run out"), strictly prohibiting the output of sentences containing second-person emotional attributions (such as "You are very angry"), thereby bypassing the damaged prefrontal cortex logic processing area.

[0040] Example 2: Cooling-off Period Pre-installation Instruction Set Method Based on TCM C5 Prior Installation Law

[0041] This embodiment describes how to pre-install contingency plans during calm periods to achieve millisecond-level response during outbreaks. For example... Figure 2 As shown, this method includes four steps: quiet period monitoring, pre-installation training, cache writing, and burst period invocation.

[0042] The first step is monitoring the calm window period. The system monitors the F_IAW value in real time. When it detects that F_IAW is lower than Theta_IAW for a certain period of time (e.g., 3 minutes), it determines that the user is in a "calm window period".

[0043] The second step is to pre-install training interactions. The system proactively initiates gamified interactions to guide users to learn and execute embodied action sequences (such as "taking three deep breaths" and "crossing arms over shoulders") that are bound to a specific virtual object character (such as "calm cat").

[0044] The third step is writing to the high-priority cache. After the user successfully executes an action, the system writes the action sequence's signature (Action_Code) to the high-priority cache of the user behavior model and marks it as "ready for emergency response".

[0045] The fourth step is direct connection during the burst phase. When the system detects that a user has entered the IAW state, it directly reads the action instructions from the high-priority cache and drives the terminal to execute feedback (such as flashing lights or robotic arm movements), completely skipping the real-time motion planning and large model inference process in the cloud, achieving millisecond-level response.

[0046] Preferably, the embodied action sequence includes specific limb movements, breathing rhythms, or tactile interaction patterns that correspond one-to-one with the six major emotional archetypes; the high-priority cache is stored in a local neural processing unit (NPU) or embedded memory to ensure that it can still be accessed in milliseconds under network outage or high load conditions.

[0047] Example 3: Multimodal Collaborative Emotion Orientation Evaluation Algorithm

[0048] If the system needs to quantify the intervention effect, a reversion scoring function Q_return can be constructed, and its logical flow is as follows: Figure 4 As shown. The calculation formula is as follows:

[0049] Q_return = alpha × Norm(S_bio) + beta × Norm(S_audio) + gamma ×Sync(S_visual) + delta × Decay(S_time)

[0050] Wherein, Norm(S_bio) is the normalized score of biological signals (such as HRV index); Norm(S_audio) is the normalized score of acoustic features (such as the slope of fundamental frequency change); Sync(S_visual) is the visual interaction synchronization rate (the phase matching degree between user actions and device light rhythm); Decay(S_time) is the time decay factor (the inverse normalized time interval from the start of intervention to the first appearance of positive semantics).

[0051] The weighting coefficients alpha, beta, gamma, and delta satisfy alpha + beta + gamma + delta = 1. In this embodiment, the biological signal weight alpha is specifically specified to be ≥ 0.5 to reflect the core role of physiological indicators in emotion assessment.

[0052] When Q_return exceeds the preset safety threshold Q_safe for N consecutive sampling periods (e.g., N=5), the system determines that the emotion has been restored, automatically exits the emergency mode, and resumes normal interaction.

[0053] Example 4: Cross-validation of multimodal emotion annotation protocols

[0054] To improve system robustness, this embodiment introduces a cross-validation mechanism. The system acquires the machine-recognized IAW status markers and the user-inputted role naming tags.

[0055] When the two are inconsistent, the system enters the confidence arbitration mode: if the voice energy features or facial expression features input by the user conform to the physical mapping law of a specific character prototype (for example, the user shouts "I am a fire dragon" and the facial muscles show specific tension), then the user input is adopted first, and the machine's IAW detection weight parameters are corrected in reverse (such as lowering the weight of E_agg), forming a two-way closed-loop calibration mechanism of "self-labeling + machine recognition".

[0056] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic detection method for the initial activation window (IAW) based on the TCM time constraint model, characterized in that, Performed by a computer device, including the following steps: Acquire multimodal interaction data streams of target users, the data streams including acoustic feature streams, text semantic streams, and optional physiological feature streams; Feature extraction is performed on the data stream using a dynamic sliding window to calculate the speech rate mutation index (S_delta), semantic radicality entropy (E_agg), and word frequency repetition density (D_rep). Construct the comprehensive discriminant function F_IAW, and calculate it using the following formula: F_IAW = w1 × S_delta + w2 × E_agg + w3 × D_rep Where w1, w2, and w3 are weighting coefficients, and w1 + w2 + w3 = 1; Dynamically determine if a user enters the IAW state: When the value of the comprehensive discrimination function F_IAW exceeds the preset threshold Theta_IAW, it is determined that the user has entered the IAW state. The IAW state is defined as a physical blackout period in which the high-level logic processing functions of the prefrontal cortex of the brain are temporarily offline due to the redistribution of metabolic resources.

2. The method according to claim 1, characterized in that: The preset threshold Theta_IAW is a physical constraint parameter that is dynamically adjusted based on the target user's historical baseline data, age group, and individual differences. The parameter represents the metabolic critical point at which the user's logical functions go offline. The weighting coefficients w1, w2, and w3 are dynamically adjusted according to the real-time scenario, but their sum remains at 1.

3. A method for switching human-computer emotional interaction protocols based on the TCM time constraint model, characterized in that, Performed by a computer device, including the following steps: The IAW state dynamic detection method described in claim 1 or 2 is executed to determine that the user has entered the IAW state. Forcefully terminate the current second-person semantic empathy protocol based on prefrontal cognitive logic; Instantly activates a third-person command protocol based on visual bypass and edge system access. The command protocol bypasses the user's impaired logical processing area through preset symbolic interaction, and the execution latency is less than 200 milliseconds. The subject-object topology switching is performed, which separates the emotional attribute from the user ontology identifier and attaches it to a preset virtual object role identifier. The virtual object role identifier and the user ontology identifier have no reference relationship at the data level.

4. The method according to claim 3, characterized in that: The virtual object role identifier is an instantiation mapping based on at least one of the six emotional instruction prototypes preset by the TCM model. Each prototype is bound to a specific embodied action-driven feature code and narrative logic label. The narrative logic label mandates that the natural language generation model only outputs third-person declarative sentences and prohibits the output of second-person emotional attribution sentences.

5. A method for pre-installing emotional emergency procedures based on the TCM C5 prior installation law, characterized in that, Performed by a computer device, including the following steps: The monitoring system determines that the user is currently in a non-IAW state, which is defined as a calm window period; During the calm window period, the pre-installed training interaction of the emotional emergency program is proactively initiated to guide users to learn and execute embodied action sequences bound to a specific virtual object character; Write the feature code of the embodied action sequence successfully executed by the user into the high-priority cache of the user behavior model and mark it as an emergency ready state; When the system detects that a user has entered the IAW state or is in an emotional outburst, it directly calls the action instructions in the high-priority cache to drive the terminal to execute feedback, skipping the real-time action planning and reasoning process, and achieving millisecond-level response.

6. The method according to claim 5, characterized in that: The embodied action sequence includes specific physical movements, breathing rhythms, or tactile interaction patterns that correspond one-to-one with the six emotional archetypes. The high-priority cache is stored in the local neural processing unit (NPU) or embedded memory to ensure millisecond-level access even under network outages or high load conditions.

7. A cross-validation method for a multimodal sentiment annotation protocol, characterized in that, The system applied to the human-computer emotional interaction protocol switching system of claim 3 includes the following steps: Obtain the machine-recognized IAW status markers and the user-inputted role naming tags; When the machine recognition result is inconsistent with the role name tag entered by the user, the system enters the confidence arbitration mode: if the voice energy feature or facial expression feature of the role name tag entered by the user conforms to the physical mapping law of a specific role prototype, the role name tag entered by the user shall be adopted first. The machine's IAW detection weight parameters and emotion recognition model are corrected in reverse to form a two-way closed-loop calibration mechanism of "self-labeling + machine recognition".

8. An emotion intelligence system based on the TCM time constraint model, characterized in that, include: The sensing and acquisition module is used to collect multimodal data streams of user interactions in real time; The IAW detection engine is used to execute the IAW state dynamic detection method as described in claim 1 or 2, and its internal configuration includes a comprehensive discrimination function F_IAW = w1 × S_delta + w2 × E_agg + w3 × D_rep. A mode switching controller is used to execute the human-computer emotional interaction protocol switching method as described in claims 3 to 4 when an IAW state is detected. A pre-installed training module is used to execute the pre-installation method of the emotional emergency procedure as described in claims 5 to 6 during a calm window period; A multimodal output module is used to execute the text output of the instruction protocol and the corresponding hardware control instructions; The cross-validation module is used to perform the cross-validation method for the multimodal sentiment annotation protocol as described in claim 7.

9. An electronic device comprising a processor, a memory, and a computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 7.

Images