A system state transition control method and system based on situation adaptive adjustment

Abstract

The application discloses a system state conversion control method and system based on situation adaptive adjustment, and belongs to the technical field of artificial intelligence and cognitive computing. The system comprises a processor, a memory and a sensor interface. The processor realizes the following modules by executing program instructions in the memory: a state perception module for acquiring a current state and context parameters of the system; a target determination module for determining a target state according to the current state and the context parameters; a convergence rate adaptive calculation module for adaptively calculating a convergence rate parameter of state conversion according to the context parameters and the internal state of the system; a dynamic adjustment module for dynamically adjusting the convergence rate parameter in the state conversion process; a rhythm control module for monitoring an energy change rate and adjusting the speed of each stage of the conversion path; a self-learning module for recording the rhythm parameters and conversion effects of each conversion and optimizing subsequent conversion strategies; and a safety enforcement module for forcibly converging to a conservative state quickly in a high-uncertainty state. The application realizes adaptive rhythm control of 'flexible and tough' in the system state conversion process, makes the system behavior more natural, efficient and safe, and is suitable for application scenarios such as robots, autonomous driving, intelligent agent control and the like which require smooth state conversion.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence and cognitive computing technology, specifically relating to a system state transition control method and system based on situational adaptive adjustment, which can be used in application scenarios that require smooth state transitions, such as robots, autonomous driving, and intelligent decision-making systems. Background Technology

[0002] Existing intelligent agent systems have the following technical problems in state transition control: (1) The state transition rhythm is monotonous and lacks adaptability. Existing systems usually use a fixed transition rate during state transitions, which cannot dynamically adjust the transition speed according to the urgency of the situation and the internal state of the system. In safety-critical scenarios, a fixed rate may result in insufficient response; in normal scenarios, a fixed rate may appear too abrupt and lack smoothness.

[0003] (2) Lack of fine-grained control over each stage of the transition path. The transition process from the current state to the target state of the system usually includes multiple stages such as startup, acceleration, stabilization, and deceleration. The requirements for the transition speed are different in different stages. The existing system lacks the ability to independently control each stage, resulting in the transition process being either too fast (appearing abrupt) or too slow (appearing sluggish).

[0004] (3) Lack of rhythm optimization capability based on historical experience. The existing system uses the same rhythm strategy for each state transition, and cannot learn the optimal rhythm control parameters from historical transition experience, resulting in a lack of personalized and continuous optimization capability of system behavior.

[0005] (4) It is difficult to achieve a balance between security constraints and conversion efficiency. In high uncertainty scenarios, the system needs to converge to a safe state quickly; in normal scenarios, the system needs to transition smoothly and naturally. Existing systems struggle to achieve an adaptive balance between "fast" and "slow".

[0006] Prior to the filing date, the applicant's earlier patent applications (such as CN202610506445.X, CN202610601934.3, etc.) have disclosed a state representation method and energy evolution mechanism based on a six-dimensional energy vector. However, the aforementioned earlier applications mainly focus on "how the system converges to the target state" and have not yet provided a complete technical solution for "how the system adaptively adjusts the transition rhythm according to the situation during the state transition process, how to finely control the speed of each stage of the transition path, and how to learn the optimal rhythm control strategy from historical experience". Summary of the Invention

[0007] (a) Purpose of the invention The purpose of this invention is to provide a system state transition control method and system based on situational adaptive adjustment, so as to solve the technical problems in the prior art, such as the single state transition rhythm, lack of fine control at each stage of the path, lack of rhythm optimization based on historical experience, and difficulty in balancing safety constraints and transition efficiency.

[0008] Specific objectives include: (1) A method for adaptively calculating the convergence rate based on the context and the internal state of the system is provided, so that the system can dynamically adjust the speed of the transition according to the urgency and uncertainty.

[0009] (2) A refined speed control method is provided for each stage of the conversion path, so that the system adopts different speed change rates in different stages such as start-up, acceleration, stabilization and deceleration.

[0010] (3) Provide a rhythm self-learning method based on historical experience, so that the system can continuously optimize the rhythm control strategy from the experience of each transition.

[0011] (4) Provide a safe and forced fast convergence mechanism so that the system can jump to a conservative state in one step under high uncertainty.

[0012] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: I. System Architecture A system state transition control system based on situational adaptive adjustment includes a processor, a memory, and a sensor interface connected to the processor. The processor is configured to implement the following modules by executing program instructions stored in the memory: (1) State awareness module The state awareness module is used to acquire sensor input data through the sensor interface to obtain the current state and context parameters of the system.

[0013] The current state of the system is represented by a six-dimensional energy vector e = (e1, e2, e3, e4, e5, e6), where e_i ∈ [0, 1]. The value of each dimension represents the system's tendency in that dimension: the closer the value is to 1, the more positive the state; the closer the value is to 0, the more negative the state. The initial state of the system is a neutral state where all dimensions are 0.5.

[0014] The context parameters include, but are not limited to: task urgency level, environmental complexity, resource availability, security level, and task type.

[0015] (2) Target Determination Module The target determination module is used to determine the target state based on the current state and the context parameters.

[0016] The target state is a six-dimensional binary encoded vector y_target∈{0,1}^6, which belongs to a complete state space consisting of 2^6=64 basic state types.

[0017] The methods for determining the target state include, but are not limited to: directly determining it based on a preset mapping relationship, determining it through reasoning from a state transition rule base, and determining it through multi-dimensional candidate evaluation.

[0018] (3) Convergence rate adaptive calculation module The convergence rate adaptive calculation module is used to adaptively calculate the convergence rate parameter α of the state transition based on the scenario parameters and the internal state of the system. The value range of the convergence rate parameter α is (0, 1).

[0019] The α value determines the speed of state transition: the closer the α value is to 1, the faster the transition ("rigid"); the closer the α value is to 0, the slower the transition ("soft").

[0020] The adaptive calculation method for the convergence rate parameter α is: α = α_base × f_urgency × f_uncertainty × f_experience.

[0021] in: - α_base is the baseline convergence rate, which is the system's default base value; - f_urgency is an urgency adjustment factor calculated based on the urgency level in the context parameters; - f_uncertainty is a certainty adjustment factor calculated based on the uncertainty of the internal state of the system; - f_experience is a preference adjustment factor calculated based on historical transformation experience.

[0022] The urgency adjustment factor f_urgency is calculated as follows: - When the urgency level is the highest level (such as "emergency avoidance" or "fault response"), f_urgency = 1.0, causing the system to converge at the fastest speed; - When the urgency level is medium (such as "task switching" or "mode switching"), f_urgency = 0.6-0.8, causing the system to converge at a moderate speed; - When the urgency level is the lowest level (such as "routine fine-tuning" or "idle transition"), f_urgency = 0.3-0.5, so that the system transitions smoothly at the slowest speed.

[0023] The certainty adjustment factor f_uncertainty is calculated as follows: f_uncertainty = 1 + min(U / U_threshold, 1.0).

[0024] Where U is the uncertainty of the internal state of the system (defined as the statistical variance of the six-dimensional energy vector), and U_threshold is a preset uncertainty threshold. When the uncertainty increases, f_uncertainty increases, accelerating the convergence rate and speeding up convergence to a definite state. The formula for calculating the uncertainty U is: U = (1 / 6)∑(e_i - ē)^2 Where ē is the mean of the six-dimensional energy vector.

[0025] The preference adjustment factor f_experience is calculated as follows: f_experience = 1 + P_experience.

[0026] Where P_experience is a rhythm preference score generated based on historical transition experience, with a value range of [-0.3, 0.3]. When historical experience shows that a certain rhythm strategy is effective, P_experience is positive, strengthening the tendency of that rhythm strategy; conversely, it is negative, weakening the tendency of that rhythm strategy.

[0027] (4) Dynamic adjustment module The dynamic adjustment module is used to dynamically adjust the convergence rate parameter α during the state transition process.

[0028] State transitions are achieved through continuous interpolation: e_new = e_current + α × (e_target - e_current).

[0029] Where e_current is the current six-dimensional energy vector, and e_target is the six-dimensional energy vector corresponding to the target state.

[0030] The dynamic adjustment module continuously monitors changes in system status and context parameters during the conversion process, and adjusts α in real time when the following conditions are detected: - When the urgency of the situation changes, adjust f_urgency accordingly and change the α value. - When the system uncertainty U changes, f_uncertainty is adjusted accordingly, changing the value of α; - As the transition approaches the target state, gradually decrease the value of α to achieve smooth convergence.

[0031] (5) Rhythm control module The rhythm control module is used to monitor the rate of energy change during the state transition process and adjust the speed of each stage of the transition path according to the rate of energy change.

[0032] In the energy change rate ΔE / Δt, ΔE is defined as the Euclidean distance between two adjacent steps of the six-dimensional energy vector, i.e., ΔE=√(∑(e_new_i-e_current_i)^2); Δt is defined as the time interval between two adjacent steps or a preset unit evolution step size.

[0033] The state transition path consists of multiple stages, including: - Start-up phase: The initial phase of transition from the current state, with the speed gradually increasing from zero; - Acceleration Phase: The phase in which the conversion speed increases rapidly; - Stable phase: The phase in which the conversion rate remains constant; - Deceleration phase: The phase in which the conversion speed gradually decreases to zero; - Detour Phase (Optional): When the path needs to bypass obstacles or change direction, the velocity direction changes but the magnitude remains the same.

[0034] The rhythm control module dynamically determines the rate of change of speed for each stage based on the energy change rate ΔE / Δt and the target speed requirements for each stage. - During the initial phase, the conversion is initiated at the first rate of change. - During the acceleration phase, propulsion accelerates at the second rate of change of speed; - During the steady phase, the current state is maintained at the third rate of change; - During the deceleration phase, it converges with the fourth rate of change of velocity; - During the orbital phase, the direction changes with the fifth rate of change of velocity but the velocity magnitude remains unchanged.

[0035] The first rate of change of velocity, the second rate of change of velocity, the third rate of change of velocity, the fourth rate of change of velocity, and the fifth rate of change of velocity are dynamically determined based on the energy change rate and the situation parameters.

[0036] (6) Self-learning module The self-learning module records the rhythm parameters and effects of each transition, and optimizes the rhythm control strategy for subsequent transitions based on historical records.

[0037] After each state transition, the self-learning module records the following information: - Rhythm parameters of the transformation: convergence rate α, duration of each stage, and energy change curve; - Contextual parameters for the transformation: urgency level, uncertainty, and task type; - Effects of the transformation: Change in uncertainty ΔU after transformation, transformation time T, and whether a safety event is triggered.

[0038] The self-learning module evaluates the rhythm parameters based on the conversion effect and generates a rhythm preference score: - When the uncertainty decreases after the conversion (ΔU<0) and the conversion time T is less than the preset time threshold, it is judged as a positive rhythm and the rhythm preference score is positive; - When the uncertainty increases after the conversion (ΔU>0) or the conversion time T is greater than the preset time threshold, it is judged as a negative rhythm and the rhythm preference score is negative.

[0039] The rhythm preference score P_experience is calculated as follows: P_experience = (number of positive feedbacks - number of negative feedbacks) / total number of feedbacks × R.

[0040] Where R is the preset preference weight range, preferably 0.3. The value range of P_experience is [-0.3, 0.3].

[0041] The self-learning module superimposes the rhythm preference score into the calculation of the convergence rate parameter of subsequent transformations (via f_experience), enabling the system to gradually learn the optimal rhythm control strategy.

[0042] (7) Safety enforcement module The safety enforcement module is used to force the convergence rate parameter α to be set to 1.0 when the system enters a high uncertainty state, so that the system jumps to a preset conservative state in one step.

[0043] The high uncertainty state is defined as: the system uncertainty U exceeds the preset safety threshold U_safety (preferably 0.40).

[0044] The set value of U_safety is strictly greater than that of U_threshold, forming a dual safety protection mechanism: U_threshold is used for proactive acceleration of convergence in normal situations, and U_safety is used for physical forced convergence in extreme situations.

[0045] The conservative state includes at least one of the following: all negative state, cessation of observation state, and humble and upright state.

[0046] The triggering of the safety enforcement module does not depend on external instructions; it is an architectural-level hard constraint of system dynamics.

[0047] 2. Method and Flow A system state transition control method based on situational adaptive adjustment includes the following steps: Step S1: State perception and target determination.

[0048] Sensor input data is acquired through the sensor interface to obtain the current system state e_current and the context parameter P_context. The current system state is represented as a six-dimensional energy vector e=(e1,e2,e3,e4,e5,e6), where e_i∈[0,1].

[0049] Based on the current state and the context parameters, the target state e_target is determined. The target state is a six-dimensional binary encoded vector y_target∈{0,1}^6.

[0050] Step S2: Adaptive calculation of convergence rate.

[0051] Calculate the urgency adjustment factor f_urgency based on the urgency level in the context parameters: - When the urgency level is the highest, f_urgency=1.0; - When the urgency level is medium, f_urgency = 0.6-0.8; - When the urgency level is the lowest, f_urgency = 0.3-0.5.

[0052] The certainty adjustment factor f_uncertainty is calculated based on the uncertainty U of the system's internal state: f_uncertainty = 1 + min(U / U_threshold, 1.0).

[0053] Where U is the uncertainty of the internal state of the system (the statistical variance of the six-dimensional energy vector), and U_threshold is the preset uncertainty threshold.

[0054] Obtain the historical rhythm preference score P_experience from the self-learning module, and calculate the preference adjustment factor f_experience=1+P_experience.

[0055] Calculate the convergence rate parameter α: α = α_base × f_urgency × f_uncertainty × f_experience.

[0056] Step S3: Dynamic adjustment of state transition.

[0057] State transitions are driven by continuous interpolation: e_new = e_current + α × (e_target - e_current).

[0058] During the transition, the changes in scenario parameters and system state are continuously monitored, and α is adjusted in real time.

[0059] As the transition approaches the target state, the value of α is gradually decreased to achieve smooth convergence.

[0060] Step S4: Change path rhythm control.

[0061] Monitor the energy change rate ΔE / Δt during the conversion process. ΔE is defined as the Euclidean distance between two adjacent steps of the six-dimensional energy vector, i.e., ΔE = √(∑(e_new_i - e_current_i)^2); Δt is defined as the time interval between two adjacent steps or a preset unit evolution step size.

[0062] Based on the energy change rate and scenario parameters, determine the velocity change rate at each stage of the conversion path.

[0063] The system starts with the first rate of change of speed during the initial phase, accelerates with the second rate of change of speed during the acceleration phase, maintains stability with the third rate of change of speed during the steady phase, and decelerates with the fourth rate of change of speed during the deceleration phase.

[0064] Step S5: Recording and self-learning of conversion results.

[0065] After the conversion is complete, record the rhythm parameters and conversion effect of this conversion.

[0066] A rhythm preference score is generated based on the conversion effect (change in uncertainty, conversion time).

[0067] The rhythm preference score is stored in the experience database for rhythm optimization in subsequent transformations.

[0068] Step S6: Enforced security protection.

[0069] Continuously monitor the system uncertainty U.

[0070] When U exceeds the preset safety threshold U_safety, α is forcibly set to 1.0, causing the system to jump directly to the preset conservative state. The set value of U_safety is strictly greater than the U_threshold, forming a dual safety protection mechanism.

[0071] (III) Beneficial Effects Compared with the prior art, the present invention has the following beneficial effects: (1) Adaptive adjustment of state transition rhythm is achieved. This invention enables the system to respond quickly in safety-critical scenarios and transition smoothly in normal scenarios by adaptively calculating the convergence rate based on scenario parameters and internal system states, thus achieving "rigid and flexible" intelligent rhythm control.

[0072] (2) It achieves refined speed control at each stage of the transition path. By controlling the rate of change of speed at each stage such as start-up, acceleration, stabilization and deceleration, the present invention makes the state transition process more natural and smooth, avoiding the defects of being too fast or too slow overall.

[0073] (3) It realizes rhythm self-learning based on historical experience. This invention records the rhythm parameters and transformation effect of each transformation, and automatically generates feedback signals based on the changes in internal uncertainty of the system, so that the system can continuously optimize the rhythm control strategy from historical experience without the need for external annotation or reward signals.

[0074] (4) A safe and forced fast convergence mechanism is provided. This invention forces a one-step transition to a conservative state under high uncertainty, ensuring the safety of the system under extreme conditions. Moreover, this mechanism is an architecture-level hard constraint that cannot be bypassed by external instructions.

[0075] (5) Forming a complete transition control system with existing patented technologies. As the rhythm control layer of existing patented technologies (state perception, energy evolution, state transition), this invention together constitutes a complete state transition control link of "target determination → rhythm calculation → smooth transition → effect learning". Attached Figure Description

[0076] Figure 1 The overall flowchart of the system state transition control method based on situational adaptive adjustment described in this invention.

[0077] Figure 2 : Schematic diagram of the adaptive calculation logic for the convergence rate parameter α.

[0078] Figure 3 : Schematic diagram of speed change rate control at each stage of the conversion path.

[0079] Figure 4 : Rhythm preference learning flowchart for the self-learning module.

[0080] Figure 5 : Schematic diagram of the triggering logic of the safety-forced fast convergence mechanism.

[0081] Figure 6 : Module architecture diagram of the system described in this invention. Detailed Implementation

[0082] It should be understood that the following embodiments are used to illustrate the technical solutions of the present invention and not to limit them. Those skilled in the art can implement the complete technical solutions of the present invention based on the teachings of this specification.

[0083] The system can be deployed in an edge computing device, which includes an in-vehicle domain controller, a robot main control chip, or an embedded intelligent processing unit. The state transition rule base and preference matrix are stored in the non-volatile memory of the edge computing device and loaded into the processor's running memory for use when the system starts.

[0084] In a preferred embodiment of the present invention, at least a portion of the causal mapping in the state transition rules is configured based on a traversal rule of the completeness of a six-dimensional binary vector space. This six-dimensional binary vector space contains 2^6 = 64 non-repeating state codes, with a Hamming distance of one between adjacent state codes, forming a structured state transition network. This configuration utilizes the completeness and symmetry of the state space, providing a deterministic and rule-based source of state transition rules for the present invention.

[0085] Example 1: Adaptive calculation of convergence rate parameter This embodiment demonstrates how the system adaptively calculates the convergence rate parameter α based on the context and internal state.

[0086] Scenario 1: Safety-Critical Scenario (Emergency Evacuation) The system is currently in "driving state" [1,1,1,0,1,1], with the following scenario parameters: urgency level = highest, and uncertainty U = 0.45 (exceeding the threshold).

[0087] Calculation process: - α_base=0.35; - f_urgency=1.0 (maximum urgency); - f_uncertainty=1+min(0.45 / 0.30,1.0)=1+1.0=2.0; - f_experience=1.0 (First conversion, no historical experience); α=0.35×1.0×2.0×1.0=0.70.

[0088] The system converges rapidly to a conservative state (stop observation state) with α=0.70, and the response time is significantly shortened.

[0089] Scenario 2: Standard Scenario (Minor Adjustments to Daily Status) The system is currently in "standby" state [0,0,0,0,0,0], with the following scenario parameters: urgency level = lowest and uncertainty U = 0.10.

[0090] Calculation process: - α_base=0.35; - f_urgency=0.35 (minimum urgency); - f_uncertainty=1+min(0.10 / 0.30,1.0)=1+0.33=1.33; - f_experience=1.0; α=0.35×0.35×1.33×1.0≈0.16.

[0091] The system transitions to the target state slowly and smoothly with α=0.16, and the transition process is natural and smooth.

[0092] Example 2: Rhythm control at each stage of the transition path This embodiment demonstrates the system's speed control at each stage of the conversion path.

[0093] The system transitions from "standby state" [0,0,0,0,0,0] to "working state" [1,1,1,1,1,1], with a path length of 6 steps.

[0094] The rhythm control module divides the transition path into four stages: - Start-up phase (step 1): Rate of change of speed = 0.3, smooth start-up; - Acceleration phase (steps 2-3): Rate of change of speed = 0.8, rapid propulsion; - Stable phase (steps 4-5): Rate of change of velocity = 0.5, moving forward at a constant speed; - Deceleration phase (step 6): Rate of change of velocity = 0.2, converges smoothly.

[0095] Actual α values ​​for each step: - Step 1: α = 0.20 (Start); - Step 2: α = 0.45 (acceleration); - Step 3: α = 0.55 (acceleration); - Step 4: α = 0.50 (stable); - Step 5: α = 0.45 (stable); - Step 6: α = 0.25 (deceleration).

[0096] The energy change curve exhibits an S-shaped characteristic of "slow start-rapid rise-stable operation-slow decline", and the conversion process is natural and smooth.

[0097] Example 3: Real-time adjustment in dynamic environments This embodiment demonstrates how the system adjusts α in real time according to environmental changes during the conversion process.

[0098] The system is transitioning from "working state" to "standby state" and has completed 3 steps. The current α=0.35.

[0099] During the transition, the system detected an emergency event (“obstacle appeared”), and the urgency level increased from “lowest” to “highest”.

[0100] The dynamic adjustment module responds in real time: - Original α = 0.35; - New f_urgency=1.0 (originally 0.35); - New α = 0.35 × (1.0 / 0.35) = 1.0.

[0101] As α jumps from 0.35 to 1.0, the system converges directly to the conservative "stop observing state" in the next step, achieving rapid risk avoidance.

[0102] Example 4: Rhythm Preference Self-Learning This embodiment demonstrates how the system optimizes its rhythm control strategy through self-learning.

[0103] Initial state: The system has no historical experience, and all conversions use the default rhythm.

[0104] First transition: The system transitions from state A to state B using α=0.35, taking 2.5 seconds. After the transition, the uncertainty decreases from 0.30 to 0.15 (ΔU=-0.15). The self-learning module determines it to be a positive rhythm and records the preference score +0.1.

[0105] Second transition: The system transitions from state A to state C with α=0.50, taking 1.8 seconds. After the transition, the uncertainty decreases from 0.30 to 0.10 (ΔU=-0.20). The self-learning module determines it to be a positive rhythm and records a preference score of +0.15.

[0106] The third transition: The system transitions from state A to state D with α=0.20, taking 4.0 seconds. After the transition, the uncertainty increases from 0.30 to 0.35 (ΔU=+0.05). The self-learning module determines it to be a negative rhythm and records a preference score of -0.1.

[0107] Cumulative preference: Starting from state A, there are 2 positive rhythms and 1 negative rhythm. The rhythm preference score P_experience = (2-1) / 3 × 0.3 = 0.10.

[0108] When the system transitions from state A again, the preference adjustment factor f_experience=1.10 increases α by 10% overall, and the system tends to adopt a faster transition rhythm.

[0109] Example 5: Safe Forced Fast Convergence This embodiment demonstrates the system's safe forced convergence under high uncertainty conditions.

[0110] The system handles multiple conflict events in a complex environment, and the uncertainty U continues to rise to 0.52, exceeding the preset safety threshold U_safety (0.40).

[0111] Security enforcement module trigger: - Force α to be set to 1.0; - The system jumps from the current state to a preset conservative state (stop observation state [0,0,1,0,0,1]) in one step. - The conversion is completed in a single step, with a response time of <1ms; - The security enforcement process does not rely on any external instructions and cannot be bypassed.

[0112] The system converges quickly to a safe state under a high uncertainty of 0.52, avoiding erroneous decisions caused by uncertainty.

[0113] Example 6: Application Example with Autonomous Systems This embodiment demonstrates the application of the rhythm control method of the present invention in an autonomous system.

[0114] In robot navigation scenarios, the robot needs to move from its current position (state A) to its target position (state B).

[0115] Under normal circumstances: The system adopts a smooth rhythm of α=0.30-0.40. The robot starts slowly during the startup phase, gradually accelerates during the acceleration phase, moves at a constant speed during the stabilization phase, and comes to a smooth stop during the deceleration phase. The whole process is smooth and natural, just like human driving.

[0116] Emergency Situation: The system detects a sudden obstacle ahead, raising the emergency level to the highest. The system adjusts α to 0.70-0.80 in real time, and the robot quickly brakes and steers to avoid the obstacle.

[0117] Learning optimization: After multiple navigation attempts, the system learned that a faster transition pace (α=0.50) can be used in "open environments" and a slower transition pace (α=0.25) should be used in "narrow environments". The system's behavior gradually changed from "mechanical" to "sophisticated".

[0118] The rhythm control method of this invention serves as the core of the underlying rhythm control of an autonomous system, enabling the system state transition process to adaptively adjust according to the situation, thus ensuring both response speed and natural behavior.

Claims

1. A system state transition control system based on situational adaptive adjustment, characterized in that, The processor includes a processor, a memory, and a sensor interface connected to the processor. The processor is configured to implement the following modules by executing program instructions stored in the memory: The state awareness module is used to acquire sensor input data through the sensor interface to obtain the current state and context parameters of the system. The target determination module is used to determine the target state based on the current state and the context parameters; The convergence rate adaptive calculation module is used to adaptively calculate the convergence rate parameter α of the state transition based on the scenario parameters and the internal state of the system. The dynamic adjustment module is used to dynamically adjust the convergence rate parameter α during the state transition process, so that the system converges quickly at the first convergence rate in safety-critical scenarios and smoothly transitions at the second convergence rate in normal scenarios. The rhythm control module is used to monitor the rate of energy change during the state transition process and adjust the speed of each stage of the transition path according to the rate of energy change. The self-learning module records the rhythm parameters and effects of each transition, and optimizes the rhythm control strategy for subsequent transitions based on historical records.

2. The system according to claim 1, characterized in that, The adaptive calculation method for the convergence rate parameter α is as follows: α=α_base×f_urgency×f_uncertainty×f_experience Where α_base is the baseline convergence rate, f_urgency is the urgency adjustment factor calculated based on the urgency level in the scenario parameters, f_uncertainty is the certainty adjustment factor calculated based on the uncertainty of the internal state of the system, and f_experience is the preference adjustment factor calculated based on historical transformation experience. The urgency adjustment factor f_urgency is calculated as follows: when the urgency level is the highest, f_urgency = 1.0; when the urgency level is medium, f_urgency = 0.6-0.8; and when the urgency level is the lowest, f_urgency = 0.3-0.

5. The certainty adjustment factor f_uncertainty is calculated as follows: f_uncertainty = 1 + min(U / U_threshold, 1.0), where U is the uncertainty of the internal state of the system, and U_threshold is a preset uncertainty threshold. When U exceeds U_threshold, the convergence rate parameter α increases, accelerating the convergence to a conservative state.

3. The system according to claim 1, characterized in that, The state transition path consists of multiple stages, including a startup stage, an acceleration stage, a stabilization stage, and a deceleration stage. The rhythm control module controls the speed change rate of each stage according to the energy change rate: in the startup stage, the transition is initiated with a first speed change rate; in the acceleration stage, the transition is accelerated with a second speed change rate; in the stabilization stage, the current state is maintained with a third speed change rate; and in the deceleration stage, the transition is decelerated and converged with a fourth speed change rate. The first, second, third, and fourth speed change rates are dynamically determined based on the energy change rate and the context parameters. In the energy change rate ΔE / Δt, ΔE is defined as the Euclidean distance between two adjacent steps of the six-dimensional energy vector, i.e., ΔE=√(∑(e_new_i - e_current_i)^2); Δt is defined as the time interval between two adjacent steps or a preset unit evolution step size.

4. The system according to claim 1, characterized in that, The self-learning module records the rhythm parameters and conversion effect of each conversion, evaluates the rhythm parameters based on the conversion effect, generates a rhythm preference score, and adds the rhythm preference score to the calculation of the convergence rate parameter of subsequent conversions. The evaluation criteria for the conversion effect are as follows: when the uncertainty decreases after conversion and the conversion time is less than a preset time threshold, it is determined to be a positive rhythm, and the rhythm preference score is positive; when the uncertainty increases after conversion or the conversion time is greater than the preset time threshold, it is determined to be a negative rhythm, and the rhythm preference score is negative.

5. The system according to claim 1, characterized in that, It also includes a safety enforcement module, which is used to force the convergence rate parameter α to be set to 1.0 when the system enters a high uncertainty state, so that the system jumps to a preset conservative state in one step; the high uncertainty state is defined as the system uncertainty U exceeding the preset safety threshold U_safety; the set value of U_safety is strictly greater than the U_threshold, which constitutes a dual safety protection mechanism.

6. The system according to claim 1, characterized in that, The target state is a six-dimensional binary encoded vector y_target∈{0,1}^6, which belongs to a complete state space consisting of 2^6=64 basic state types.

7. The system according to claim 1, characterized in that, The context parameters include at least one of the following: task urgency level, environmental complexity, resource availability, security level, and task type.

8. The system according to claim 3, characterized in that, When the conversion path includes a detour phase, the rhythm control module changes direction with a fifth rate of change of speed during the detour phase but does not change the speed modulus.

9. A system state transition control method based on situational adaptive adjustment, characterized in that, The method is performed by the system according to any one of claims 1 to 8, and includes the following steps: Step S1: Obtain sensor input data through the sensor interface, obtain the current system state and context parameters, and determine the target state based on the current state and context parameters; Step S2: Based on the scenario parameters and the internal state of the system, adaptively calculate the convergence rate parameter α for the state transition. The value range of the convergence rate parameter α is [0, 1]. Step S3: During the transition from the current state to the target state, the convergence rate parameter α is dynamically adjusted so that the system converges rapidly at a first convergence rate in safety-critical scenarios and smoothly transitions at a second convergence rate in normal scenarios. Step S4: Monitor the energy change rate during the state transition process, and adjust the speed of each stage of the transition path according to the energy change rate; Step S5: Record the rhythm parameters and transition effect of each transition, and optimize the rhythm control strategy for subsequent transitions through a self-learning mechanism based on historical records; Step S6: When the system enters a high uncertainty state, the convergence rate parameter α is forcibly set to 1.0, so that the system jumps to the preset conservative state in one step.

10. The method according to claim 9, characterized in that, The adaptive calculation method for the convergence rate parameter α in step S2 is as follows: α = α_base × f_urgency × f_uncertainty × f_experience, where α_base is the baseline convergence rate, f_urgency is the urgency adjustment factor calculated based on the urgency level in the scenario parameters, f_uncertainty is the certainty adjustment factor calculated based on the uncertainty of the internal state of the system, and f_experience is the preference adjustment factor calculated based on historical transformation experience.

Images