Star flash connection intelligent switching decision-making system and method based on equipment state prediction

Abstract

The invention discloses a satellite flash connection intelligent switching decision-making system and method based on equipment state prediction, and particularly relates to the technical field of satellite flash connection. In order to solve the problem of star-flash network linkage failure caused by multi-mechanical-arm cooperative movement and electromagnetic disturbance coupling in an industrial cooperative robot system, a prediction model fusing space attitude cooperative change characteristics and transient electromagnetic disturbance amplitude gradient characteristics is constructed, and high-sensitivity recognition and risk early warning of the coupling state of a star-flash connection network are achieved. A communication channel and communication parameters are dynamically adjusted through a fuzzy logic decision mechanism, and link stability and interference robustness are effectively improved; and meanwhile, a model adaptive optimization mechanism driven by switching effect feedback is introduced, so that the prediction accuracy and the system self-evolution capability are remarkably improved, and the continuity, the safety and the high reliability of multi-node cooperation in an industrial communication scene are guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of star flash connection technology, and in particular to a star flash connection intelligent switching decision system and method based on device state prediction. Background Art

[0002] SparkLink intelligent switching decision means that in devices using SparkLink communication technology, the system can intelligently judge and automatically select the optimal connection method or node based on the current network status, device status and application requirements, thereby achieving efficient, stable and low-latency communication.

[0003] The prior art has the following deficiencies: In an industrial collaborative robot system, when multiple robotic arms are simultaneously adjusted in displacement and the surrounding electromagnetic environment fluctuates violently, the originally stable star flash connection network will cause a chain failure due to a slight coupling imbalance, causing multiple nodes to be disconnected at the same time. Since such problems do not have obvious precursor characteristics and there is a strong nonlinear coupling between states, the existing prediction mechanism cannot provide early warning, which seriously affects the continuity of industrial production and equipment safety. Summary of the invention

[0004] The purpose of the present invention is to provide a star flash connection intelligent switching decision system and method based on device status prediction to solve the shortcomings of the background technology.

[0005] In order to achieve the above object, the present invention provides the following technical solution: a method for intelligent switching decision of star flash connection based on device state prediction, comprising: Obtain the operating status parameters of multiple devices with Star Flash communication function, including device displacement, connection signal strength and network delay; Performing real-time analysis on the operating state parameters to predict the connection quality change trend of the communication link, and extracting the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics in the operating state parameters; The spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are integrated and input into a pre-built equipment state prediction model to determine the coupling state of the star flash connection network and obtain the coupling imbalance risk result; If the risk result is greater than the preset risk threshold, an early warning signal is immediately generated, and the Star Flash connection strategy is adjusted according to the predicted level of coupling imbalance, including switching the communication channel or adjusting the communication parameter configuration; According to the switching effect feedback data, the weight parameters in the device state prediction model are updated to optimize the accuracy of subsequent predictions and decisions.

[0006] Preferably, the device displacement is collected by an inertial measurement unit, a gyroscope, an accelerometer or an external visual positioning system integrated in each device, and the spatial position and posture changes of the device per unit time are recorded in the form of a three-dimensional vector.

[0007] Preferably, the specific calculation method of the spatial posture coordinated change value is: for each device i, at time points t and Get its displacement or attitude change vector : ;in, , represents the position coordinates of device i at time t, and T is the vector transpose; calculate the average displacement vector of all devices in the current time window, and for each device i, calculate the square of the Euclidean distance difference between it and the average change vector, and take the weighted average of the collaborative difference values ​​of all devices to obtain the spatial posture collaborative change value JSVD at the current time point.

[0008] Preferably, the calculation method of the transient electromagnetic disturbance amplitude gradient is: collecting the electromagnetic disturbance signal received by the device per unit time, dividing the signal into multiple overlapping time windows, performing Fourier transform on each window, and calculating the frequency domain amplitude energy of the STFT spectrum under each window: ; Where: E(s) represents the total spectrum energy at time point s, fmin and fmax are the frequency ranges for analysis; is the instantaneous amplitude at frequency f. The first-order difference of the time domain energy change is performed through the sliding window to calculate the transient electromagnetic disturbance amplitude gradient. The expression is: ; Where: Δs is the sliding step length, is the transient electromagnetic disturbance amplitude gradient.

[0009] Preferably, the spatial attitude coordinated change characteristics and the transient electromagnetic disturbance amplitude gradient characteristics are fused and input into a pre-built device state prediction model to determine the coupling state of the star flash connection network, specifically including: The acquired space attitude coordinated change value and transient electromagnetic disturbance amplitude gradient are used as the input items of fuzzy logic, and the star flash connection network coupling state risk value is used as the output item of fuzzy logic; Set corresponding membership functions for each input and output language variable; formulate several fuzzy inference rules and calculate their membership values ​​under each language variable; Using fuzzy reasoning method, the input variables are matched with fuzzy rules to obtain the activation degree of each rule on the output variables. The reasoning result of each rule will generate a fuzzy output subset, and the outputs of multiple rules will be fused; The synthesized fuzzy output results are defuzzified to convert them into clear numerical results, and finally the coupling state risk value is output.

[0010] Preferably, the output value is compared with a preset risk threshold. If the coupling state risk value exceeds the preset threshold, it is judged as high risk. The system will immediately issue an early warning and start the connection switching decision process, including switching the communication channel or adjusting the communication parameter configuration.

[0011] Preferably, the original equipment status prediction model is set , the MAML structure is introduced, and after each communication cycle, the model parameters are fine-tuned according to the switching feedback data, specifically: Collecting the switching feedback data set ;in: is the state parameter before switching, Evaluate tags for actual switching effects; Under the current model parameters θ, Execute training to obtain temporarily updated model parameters: ; where α is the inner loop learning rate, is the loss function of the feedback data, is the gradient vector; Model Based on the performance on the main task, the original model parameters are updated, and the updated parameters θ replace the original model weights to generate an enhanced state prediction model for star flash connection switching decision.

[0012] The present invention also provides a star flash connection intelligent switching decision system based on device state prediction, including a data acquisition module, a feature extraction module, a risk prediction module, a star flash connection strategy adjustment module and an accuracy optimization module; Data acquisition module: obtains the operating status parameters of multiple devices with Star Flash communication function, including device displacement, connection signal strength and network delay; Feature extraction module: performs real-time analysis on the operating state parameters to predict the connection quality change trend of the communication link, and extracts the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics in the operating state parameters; Risk prediction module: the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are integrated and input into a pre-built equipment state prediction model, the coupling state of the star flash connection network is determined, and the coupling imbalance risk result is obtained; Star Flash connection strategy adjustment module: if the risk result is greater than the preset risk threshold, an early warning signal is immediately generated, and the Star Flash connection strategy is adjusted according to the predicted level of coupling imbalance, including switching the communication channel or adjusting the communication parameter configuration; Accuracy optimization module: based on the switching effect feedback data, update the weight parameters in the device state prediction model to optimize the accuracy of subsequent predictions and decisions.

[0013] In the above technical solution, the technical effects and advantages provided by the present invention are: 1. Aiming at the problem that simultaneous displacement of multiple manipulators and strong electromagnetic disturbances in industrial collaborative robot systems easily lead to chain failure of the Star Flash network, the present invention introduces the space posture coordinated change feature (JSVD) and the transient electromagnetic disturbance amplitude gradient feature (TEMG) to achieve high sensitivity perception of the equipment coordination state and environmental interference. The risk level of the Star Flash connection network coupling state is judged through the fuzzy logic reasoning mechanism, and the communication channel switching or parameter adjustment strategy is intelligently triggered under high-risk conditions, which significantly improves the stability and anti-interference of the communication link.

[0014] 2. The present invention further introduces a meta-learning model based on MAML, takes the communication switching feedback data as the basis for model training, and realizes the adaptive update and optimization of the equipment state prediction model. Compared with the traditional fixed rule system, this method has continuous learning ability and environmental adaptability, and can maintain efficient and reliable connection decision-making performance in a dynamic and multi-disturbance industrial environment, thereby effectively ensuring the continuity, security and system robustness of multi-device collaborative communication in the industrial production process. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] Embodiment 1, the method for intelligent switching decision of a star flash connection based on device state prediction described in this embodiment includes: Obtain the operating status parameters of multiple devices with Star Flash communication function, including device displacement, connection signal strength and network delay; Performing real-time analysis on the operating state parameters to predict the connection quality change trend of the communication link, and extracting the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics in the operating state parameters; The spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are integrated and input into a pre-built equipment state prediction model to determine the coupling state of the star flash connection network and obtain the coupling imbalance risk result; If the risk result is greater than the preset risk threshold, an early warning signal is immediately generated, and the Star Flash connection strategy is adjusted according to the predicted level of coupling imbalance, including switching the communication channel or adjusting the communication parameter configuration; According to the switching effect feedback data, the weight parameters in the device state prediction model are updated to optimize the accuracy of subsequent predictions and decisions.

[0017] Comprehensively collect the operating status of all devices in the system that have SparkLink communication functions, so as to provide basic data support for subsequent status prediction and intelligent switching decisions. The status parameters obtained include but are not limited to the following: Device displacement: acquired in real time through the high-precision inertial measurement unit (IMU), gyroscope, accelerometer or external vision / laser positioning system (such as VSLAM, lidar) integrated inside the device; indicates the relative position change and movement speed of the device in three-dimensional space, including translation displacement (x, y, z axis) and attitude angle change (pitch, roll, yaw); can be recorded in the form of time series to determine whether the device is in a high-speed dynamic state or a coordinated displacement change occurs, and is a basic indicator for determining potential changes in network topology.

[0018] Connection signal strength: Through the interface provided by the Star Flash communication protocol stack, the received signal strength (RSSI), link quality index (LQI) or signal-to-noise ratio (SNR) of each communication link can be obtained in real time; the acquisition frequency can be adjusted according to the communication density, generally in the range of 5ms~50ms; a dynamic communication topology map can be further constructed to reflect the stability and spatial attenuation model of the network connection; in the presence of obstruction, interference or long distance, the decrease in signal strength can reflect the risk of connection quality decline in advance.

[0019] Network delay: By periodically sending probe packets or communication heartbeat packets between devices, point-to-point delay or round-trip time (RTT) is measured. If the delay fluctuates frequently or the average value increases, it may indicate that the link is about to become unstable, which is suitable as an early signal for intelligent switching decisions. For industrial collaboration scenarios, time synchronization error (TimeSync Drift) can also be combined to improve the accuracy of system-level communication prediction.

[0020] The continuously collected operating status parameters (device displacement, signal strength, network delay, etc.) are divided into sliding windows in chronological order; a multi-dimensional state vector sequence is formed in each window to describe the evolution process of the operating status of the current system.

[0021] Construct a multivariate timing analysis model for link quality prediction, such as one based on LSTM neural network, Bayesian timing model or Kalman filter; output the connection quality trend of the link at several future moments, such as signal attenuation rate, delay change trend, connection loss probability, etc.; set the prediction time step Δt, for example, if the link quality drops by more than a threshold within the next 0.5s or 1s, the early warning mechanism will be triggered.

[0022] In order to identify non-explicit interference that may cause chain connection failure in complex collaborative scenarios, the spatial posture coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are extracted: Spatial posture coordinated change feature (JSVD): The spatial posture change of each device in unit time is expressed as a three-dimensional motion vector group; the spatial posture coordinated change value is calculated by calculating the cosine of the angle between any two posture change vectors in the device group or the vector difference variance; For example, the specific calculation method of the spatial posture coordinated change value is: For each device i, at time t and Get its displacement or attitude change vector : ;in, , represents the position coordinates of device i at time t, T is the vector transpose; calculate the average displacement vector of all devices in the current time window , for each device i, calculate the square of the Euclidean distance difference between it and the average change vector , the expression is: ; Take the weighted average of the collaborative difference values ​​of all devices to obtain the spatial posture collaborative change value JSVD at the current time point.

[0023] When JSVD is small, it means that the movement direction and amplitude between devices are close, the degree of coordination is high, and the network topology is stable; when JSVD is large, it means that there are devices that deviate sharply from the group movement trend, and there may be problems such as occlusion, interference, and link interruption. In the scenarios of industrial manipulators and collaborative robots, a sudden increase in JSVD usually indicates a sudden change in the state of the manipulator group, which can easily lead to the need for rapid reconstruction of the Star Flash link.

[0024] Transient electromagnetic disturbance amplitude gradient characteristics (TEMG): Through the electromagnetic sensor module on the device, the electromagnetic field strength values ​​EM(t) at multiple time points are collected; the first-order difference of the electromagnetic disturbance signal at continuous moments is performed to obtain the disturbance change gradient: or in a multi-device environment, the gradient distribution is calculated by collecting EMi(t) at different spatial points to reflect the disturbance fluctuation range and directionality.

[0025] For example, the calculation method of the transient electromagnetic disturbance amplitude gradient is: the electromagnetic disturbance signal received by the acquisition device per unit time is expressed as: EM(s), s∈[0,T]; where EM(s) represents the electromagnetic field intensity at time s (which can be the electric field amplitude, voltage or power spectrum density, etc.). The signal is divided into multiple overlapping time windows, and Fourier transform is performed on each window: ;in: is a window function (such as a Hamming window or a Gaussian window) with a length of Lw; f is a frequency variable, and STFT(s,f) is the spectrum at time s in the time-frequency domain. is the kernel function of Fourier transform, EM( ) indicates the time The electromagnetic field intensity at . The frequency domain amplitude energy is calculated for the STFT spectrum in each window: ; Where: E(s) represents the total spectrum energy (or energy density) at time point s, fmin and fmax are the frequency range of analysis (e.g. 10kHz~500kHz); is the instantaneous amplitude at frequency f. The first-order difference of the time domain energy change is performed through the sliding window to calculate the transient electromagnetic disturbance amplitude gradient, and the expression is: ; Where: Δs is the sliding step size (such as 5ms, 10ms), It is the transient electromagnetic disturbance amplitude gradient. A sudden increase in TEMG value indicates that there is a short-term strong interference source in the environment (such as high-power equipment startup, electric welding, EMI coupling), which can easily cause the star flash link to become unstable instantly.

[0026] The spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are fused and input into the pre-built equipment state prediction model to determine the coupling state of the star flash connection network and obtain the coupling imbalance risk results, including: The acquired space attitude coordinated change value and transient electromagnetic disturbance amplitude gradient are used as the input items of fuzzy logic, and the star flash connection network coupling state risk value is used as the output item of fuzzy logic; Set the corresponding membership function for each input and output linguistic variable. Common membership functions include triangular function, trapezoidal function or Gaussian function. For example, set the following membership function for JSVD: "low synergy change" indicates high consistency of system movement; "medium synergy change" indicates slight synergy deviation; "high synergy change" indicates violent inconsistent movement. The fuzzy level and membership curve are also set in the same way for TEMG and output items.

[0027] According to the understanding of system behavior, formulate several fuzzy reasoning rules (IF-THEN rules), for example: if JSVD is "high" and TEMG is "high", the coupling risk is "high"; if JSVD is "medium" and TEMG is "high", the coupling risk is "medium-high"; if JSVD is "low" and TEMG is "low", the coupling risk is "low"; if JSVD is "medium" and TEMG is "low", the coupling risk is "medium", etc. The rule base should cover all possible combinations of input variables to ensure that the system fully responds to various operating states.

[0028] When the actual system collects the specific values ​​of JSVD and TEMG, the membership values ​​under each language variable are calculated respectively. Then, the fuzzy reasoning method (such as the Mamdani model) is used to match the input variables with the fuzzy rules to obtain the activation degree of each rule on the output variable. The reasoning result of each rule will generate a fuzzy output subset, and the outputs of multiple rules are fused through the "maximum membership synthesis" or "weighted average" method.

[0029] The synthesized fuzzy output result is defuzzified to convert it into a clear numerical result. Common methods include the centroid method or the maximum membership method. The final output coupling state risk value is a continuous value (such as a decimal between 0 and 1), which can be used to determine whether the system has entered a high-risk state and decide whether to trigger a connection switch or early warning strategy.

[0030] The output value is compared with the preset risk threshold. If the coupling state risk value exceeds the preset threshold (such as ≥0.7), it is judged as high risk. The system will immediately issue an early warning and start the connection switching decision process, including switching communication channels or adjusting communication parameter configurations.

[0031] When the current communication channel is determined to be interfered or unstable, the system will perform the following operations: Communication frequency band reselection: Scan available Star Flash communication frequency bands (such as those working in the 2.4GHz or 5.8GHz frequency bands); conduct interference assessment on each candidate channel (signal-to-noise ratio, packet loss rate, occupancy rate, etc.); select the channel with the least interference and the best link quality for switching according to priority or policy weight; during the switching process, ensure that the master device and the slave device hop synchronously to avoid connection interruption.

[0032] Fast spectrum sensing and dynamic channel assessment: Use the embedded spectrum analysis module to perform short-time spectrum scanning; evaluate the channel idle rate, bandwidth availability and adjacent signal interference level of the current channel in real time; and quickly select the switching target based on the channel quality scoring model.

[0033] Channel blacklist management mechanism: If a channel fails to communicate multiple times in a short period of time, it will be temporarily added to the blacklist to prevent it from being selected again; the blacklist channel can be automatically restored after a set cooling time.

[0034] When there is no need to switch frequency bands or channels but the current link anti-interference performance needs to be optimized, the communication parameters can be adjusted in the following ways: Adjust the modulation and coding scheme (MCS): Automatically select a more robust modulation scheme (such as reducing from 64-QAM to 16-QAM or QPSK) based on link quality; reduce the bit error rate and improve transmission stability, but may sacrifice throughput; automatically restore to a high-rate configuration after interference is alleviated.

[0035] Dynamically adjust the transmit power: If there is instantaneous attenuation or obstruction in the channel, the transmit power (TXPower) can be temporarily increased; to avoid adjacent channel interference caused by excessive power, a power cap needs to be set; in the absence of interference, the power can be gradually reduced to save energy.

[0036] Enable adaptive retransmission mechanism (ARQ / HARQ): For environments with high packet loss rates, enable the automatic retransmission request mechanism; set parameters such as the maximum number of retransmissions, waiting interval, and ACK confirmation mechanism; combine with the fast retransmission channel in Star Flash to improve stability in high interference environments.

[0037] Increase link redundancy or multi-path backup: Establish a primary-backup connection mechanism so that key devices can quickly switch to the backup channel when the primary channel fails; support multi-link concurrent transmission (such as link aggregation or load balancing) to reduce the risk of single point failure.

[0038] Optimize the MAC layer scheduling strategy: prioritize time slices or transmission windows for key devices; temporarily limit the data rate or transmission cycle of low-priority devices when network resources are tight; enable the time slot rescheduling mechanism to avoid time periods with concentrated interference.

[0039] In the Star Flash connection switching system based on device state prediction, the initial state prediction model is often established based on fixed rules or training samples, which is difficult to cope with complex environmental changes during operation. To this end, it is necessary to introduce a model adaptive optimization mechanism to continuously adjust the model weights using the effect feedback after switching, so as to improve its prediction accuracy and generalization ability in actual scenarios.

[0040] After each connection switch, the system will collect a set of feedback data for evaluating the switching effect, mainly including: the change in communication signal strength before and after the switch (ΔRSSI); the change in network delay before and after the switch (ΔRTT); whether the switch is successful (success rate / failure mark); the stable duration of the network after the switch; performance indicators such as bit error rate and packet loss rate after the switch. These feedback data can be used to measure the decision accuracy and reliability of the current prediction model.

[0041] Traditional model parameter update methods such as back propagation or reinforcement learning are often used in static or semi-dynamic systems. However, in star flash systems, state transitions are frequent and disturbances are complex, so a small sample learning model that can quickly adapt to new environmental changes is required. The MAML meta-learning method can be used to quickly adjust model parameters so that it can achieve better performance on a small number of new samples (i.e. feedback data).

[0042] Setting the original equipment status prediction model , which is used to predict the risk level of the communication link; by introducing the MAML structure, the model parameters are fine-tuned according to the switching feedback data after each communication cycle, specifically: Collecting the switching feedback data set ;in: are the status parameters before switching (such as JSVD, TEMG, signal strength, delay, etc.), Evaluate labels for actual switching effects (e.g., success / failure, network quality improvement rate); Under the current model parameters θ, use Execute training to obtain temporarily updated model parameters: ; where α is the inner loop learning rate, is the loss function of the feedback data (such as prediction error, classification loss, etc.), It is a gradient vector, which represents the set of partial derivatives of a loss function (or objective function) with respect to the model parameters θ.

[0043] Model Based on the performance on the main task (such as future state prediction), update the original model parameters: ; Where: β is the learning rate of the outer loop, It is used to measure the generalization ability of the prediction model in the real environment; the optimization goal of this step is to enable the model to quickly adapt to environmental changes in any future feedback samples. The updated parameters θ replace the original model weights to generate an enhanced state prediction model for the next Star Flash connection switching decision.

[0044] Embodiment 2, the star flash connection intelligent switching decision system based on device state prediction described in this embodiment includes a data acquisition module, a feature extraction module, a risk prediction module, a star flash connection strategy adjustment module and an accuracy optimization module; Data acquisition module: obtains the operating status parameters of multiple devices with Star Flash communication function, including device displacement, connection signal strength and network delay; Feature extraction module: performs real-time analysis on the operating state parameters to predict the connection quality change trend of the communication link, and extracts the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics in the operating state parameters; Risk prediction module: the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are integrated and input into a pre-built equipment state prediction model, the coupling state of the star flash connection network is determined, and the coupling imbalance risk result is obtained; Star Flash connection strategy adjustment module: if the risk result is greater than the preset risk threshold, an early warning signal is immediately generated, and the Star Flash connection strategy is adjusted according to the predicted level of coupling imbalance, including switching the communication channel or adjusting the communication parameter configuration; Accuracy optimization module: based on the switching effect feedback data, update the weight parameters in the device state prediction model to optimize the accuracy of subsequent predictions and decisions.

[0045] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0046] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0047] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0048] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for intelligent switching decision of star flash connection based on device status prediction, characterized by: include: Obtain the operating status parameters of multiple devices with Star Flash communication function, including device displacement, connection signal strength and network delay; Performing real-time analysis on the operating state parameters to predict the connection quality change trend of the communication link, and extracting the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics in the operating state parameters; The spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are integrated and input into a pre-built equipment state prediction model to determine the coupling state of the star flash connection network and obtain the coupling imbalance risk result; If the risk result is greater than the preset risk threshold, an early warning signal is immediately generated, and the Star Flash connection strategy is adjusted according to the predicted level of coupling imbalance, including switching the communication channel or adjusting the communication parameter configuration; According to the switching effect feedback data, the weight parameters in the device state prediction model are updated to optimize the accuracy of subsequent predictions and decisions.

2. The method for intelligent switching decision of star flash connection based on device status prediction according to claim 1 is characterized in that: The device displacement is collected by an inertial measurement unit, gyroscope, accelerometer or external visual positioning system integrated in each device, and the spatial position and posture changes of the device per unit time are recorded in the form of a three-dimensional vector.

3. The method for intelligent switching decision of star flash connection based on device status prediction according to claim 2 is characterized in that: The specific calculation method of the space posture coordinated change value is: For each device i, at time t and Get its displacement or attitude change vector : ;in, , represents the position coordinates of device i at time t, and T is the vector transpose; calculate the average displacement vector of all devices in the current time window, and for each device i, calculate the square of the Euclidean distance difference between it and the average change vector, and take the weighted average of the collaborative difference values ​​of all devices to obtain the spatial posture collaborative change value JSVD at the current time point.

4. The method for intelligent switching decision of star flash connection based on device status prediction according to claim 3 is characterized in that: The calculation method of the transient electromagnetic disturbance amplitude gradient is as follows: collect the electromagnetic disturbance signal received by the device per unit time, divide the signal into multiple overlapping time windows, perform Fourier transform on each window, and calculate the frequency domain amplitude energy of the STFT spectrum under each window: ; Where: E(s) represents the total spectrum energy at time point s, fmin and fmax are the frequency ranges for analysis; is the instantaneous amplitude at frequency f. The first-order difference of the time domain energy change is performed through the sliding window to calculate the transient electromagnetic disturbance amplitude gradient. The expression is: ; Middle: Δs is the sliding step length, is the transient electromagnetic disturbance amplitude gradient.

5. The method for intelligent switching decision of star flash connection based on device status prediction according to claim 4 is characterized in that: The spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are fused and input into a pre-built device state prediction model to determine the coupling state of the star flash connection network, specifically including: The acquired space attitude coordinated change value and transient electromagnetic disturbance amplitude gradient are used as the input items of fuzzy logic, and the star flash connection network coupling state risk value is used as the output item of fuzzy logic; Set corresponding membership functions for each input and output language variable; formulate several fuzzy inference rules and calculate their membership values ​​under each language variable; Using fuzzy reasoning method, the input variables are matched with fuzzy rules to obtain the activation degree of each rule on the output variables. The reasoning result of each rule will generate a fuzzy output subset, and the outputs of multiple rules will be fused; The synthesized fuzzy output results are defuzzified to convert them into clear numerical results, and finally the coupling state risk value is output.

6. The method for intelligent switching decision of star flash connection based on device status prediction according to claim 5 is characterized in that: The output value is compared with the preset risk threshold. If the coupling state risk value exceeds the preset threshold, it is judged as high risk. The system will immediately issue an early warning and start the connection switching decision process, including switching the communication channel or adjusting the communication parameter configuration.

7. The method for intelligent switching decision of star flash connection based on device status prediction according to claim 6 is characterized in that: Setting the original equipment status prediction model , the MAML structure is introduced, and after each communication cycle, the model parameters are fine-tuned according to the switching feedback data, specifically: Collecting the switching feedback data set ;in: is the state parameter before switching, is the actual switching effect evaluation label; under the current model parameter θ, use Execute training to obtain temporarily updated model parameters: ; where α is the inner loop learning rate, is the loss function of the feedback data, is the gradient vector; Model Based on the performance on the main task, the original model parameters are updated, and the updated parameters θ replace the original model weights to generate an enhanced state prediction model for star flash connection switching decision.

8. A Star Flash connection intelligent switching decision system based on device state prediction, used to implement the Star Flash connection intelligent switching decision method based on device state prediction according to any one of claims 1 to 7, characterized in that: It includes data acquisition module, feature extraction module, risk prediction module, star flash connection strategy adjustment module and accuracy optimization module; Data acquisition module: obtains the operating status parameters of multiple devices with Star Flash communication function, including device displacement, connection signal strength and network delay; Feature extraction module: performs real-time analysis on the operating state parameters to predict the connection quality change trend of the communication link, and extracts the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics in the operating state parameters; Risk prediction module: the spatial attitude coordinated change characteristics and transient electromagnetic disturbance amplitude gradient characteristics are integrated and input into a pre-built equipment state prediction model, the coupling state of the star flash connection network is determined, and the coupling imbalance risk result is obtained; Star Flash connection strategy adjustment module: if the risk result is greater than the preset risk threshold, an early warning signal is immediately generated, and the Star Flash connection strategy is adjusted according to the predicted level of coupling imbalance, including switching the communication channel or adjusting the communication parameter configuration; Accuracy optimization module: based on the switching effect feedback data, update the weight parameters in the device state prediction model to optimize the accuracy of subsequent predictions and decisions.