Search

CN-122020509-A - Ripple seat signal fusion system

CN122020509ACN 122020509 ACN122020509 ACN 122020509ACN-122020509-A

Abstract

The invention relates to the technical field of seat safety control and discloses a ripple seat signal fusion system. The system comprises a multi-source signal perception module, a signal feature extraction module, a seat state dynamic modeling module, a multi-target strategy optimization module and a comprehensive safety decision module. The multi-target strategy optimization module performs strategy optimization calculation on the tensor based on a multi-layer strategy strengthening framework to output a safety strategy vector, and the comprehensive safety decision module generates a comprehensive safety regulation instruction according to the safety strategy vector. The system improves the comprehensiveness of seat signal perception, the dynamic property of state modeling and the scientificity of safety decision, and is suitable for the field of seat safety control.

Inventors

  • YAO JIACHAO
  • HU HAIYUN
  • KUAI QI

Assignees

  • 安徽通宇电子股份有限公司

Dates

Publication Date
20260512
Application Date
20251222

Claims (10)

  1. 1. A ripple seat signal fusion system, comprising: the multi-source signal sensing module is used for executing signal sub-term acquisition operation to obtain an original multi-mode sensing signal; The signal characteristic extraction module is used for executing characteristic enhancement conversion operation on the original multi-mode sensing signals to obtain an optimized characteristic matrix; The seat state dynamic modeling module is used for processing the optimized feature matrix by adopting a covariance matrix-driven dynamic structure reconstruction method to generate a seat state association tensor; the multi-target strategy optimization module is used for executing strategy optimization calculation on the seat state association tensor based on a multi-layer strategy reinforcement framework and outputting a safety strategy vector; And the comprehensive safety decision module generates a comprehensive safety regulation instruction according to the safety strategy vector.
  2. 2. The system of claim 1, wherein the multi-source signal sensing module comprises a physiological signal acquisition unit, a pressure distribution acquisition unit, and an environmental disturbance acquisition unit; the physiological signal acquisition unit acquires an occupant electrocardiographic waveform signal and a myoelectric activity signal; the pressure distribution acquisition unit acquires pressure distribution gradient signals of the surface of the seat through the pressure sensitive sensing array; the environment disturbance acquisition unit acquires a vehicle vibration spectrum signal and a temperature and humidity change signal; The signal sub-term acquisition operation integrates the passenger electrocardiographic waveform signal, the myoelectric activity signal, the seat surface pressure distribution gradient signal, the vehicle vibration spectrum signal and the temperature and humidity change signal to form the original multi-mode sensing signal.
  3. 3. The system of claim 2, wherein the feature enhancement conversion operations performed by the signal feature extraction module include a physiological feature reconstruction sub-operation, a pressure feature enhancer operation, and an environmental feature noise reduction sub-operation; The physiological characteristic reconstruction sub-operation performs wavelet packet decomposition on the passenger electrocardio waveform signals and extracts heart rate variability characteristics, and simultaneously performs time-frequency transformation on the myoelectric activity signals to generate muscle fatigue indexes; The pressure characteristic enhancer operation is used for carrying out sliding window characteristic extraction on the seat surface pressure distribution gradient signal to generate pressure distribution entropy characteristics and gravity center offset track characteristics; And the environmental characteristic noise reduction sub-operation performs self-adaptive filtering processing on the vehicle vibration spectrum signal to generate vibration energy spectrum characteristics, and performs trend decomposition on the temperature and humidity change signal to generate a temperature and humidity sensitive factor.
  4. 4. The system of claim 3, wherein the signal feature extraction module further comprises a feature fusion normalization sub-operation; the characteristic fusion standardization sub-operation integrates the heart rate variability characteristic, the muscle fatigue index, the pressure distribution entropy characteristic, the gravity center deviation track characteristic, the vibration energy spectrum characteristic and the temperature and humidity sensitivity factor, and the optimized characteristic matrix comprising optimized physiological characteristics, optimized pressure distribution characteristics and optimized environment disturbance characteristics is formed through a missing value filling operation, an abnormal value cutting operation and a characteristic dimension compression operation.
  5. 5. The system of claim 4, wherein the covariance matrix-driven dynamic structure reconstruction method of the seat state dynamic modeling module comprises a dynamic heterogeneous topology building operation, a spatiotemporal state updating operation, and a risk conduction modeling operation; The dynamic heterogeneous topology construction operation defines three node types of passenger nodes, seat support nodes and environment interference nodes, and constructs a node attribute set and an edge relation set based on the optimized feature matrix; The time-space state updating operation updates node state parameters according to a preset time interval, and rebuilds a topological structure evolution process through a Kalman filtering algorithm; The risk conduction modeling operation calculates fatigue risk propagation probability by adopting a three-layer attention mechanism, and generates the seat state association tensor comprising an occupant fatigue risk assessment tensor and a posture imbalance risk matrix.
  6. 6. The system of claim 5, wherein the three-layer attention mechanism comprises a cross-domain feature attention layer, a time-varying state attention layer, and a risk cascade attention layer; The cross-domain feature attention layer fuses the optimized physiological feature and the optimized pressure distribution feature to generate inter-domain association weights; The time-varying state attention layer introduces a time decay function to calculate the cumulative effect of environmental interference; The risk cascade attention layer integrates the inter-domain correlation weight and the cumulative effect of the environmental interference, and outputs the fatigue risk propagation probability through an activation function.
  7. 7. The system of claim 1, wherein the multi-layer policy reinforcement architecture of the multi-objective policy optimization module comprises a meta-policy generation layer, a dynamic tuning layer, and an execution control layer; The meta-policy generation layer processes a long-term security policy framework and generates a base regulation policy; The dynamic parameter adjusting layer adopts a dual-delay strategy gradient algorithm to optimize a seat support parameter dynamic adjustment sequence; the execution control layer calculates an air bag pressure distribution instruction and a damping coefficient adjustment instruction in real time; And the strategy optimizing calculation integrates the basic regulation strategy, the seat support parameter dynamic regulation sequence, the air bag pressure distribution instruction and the damping coefficient regulation instruction to form the safety strategy vector.
  8. 8. The system of claim 7, wherein the multi-objective strategy optimization module further comprises a disturbance injection operation and a multi-objective rewards calculation operation; The multi-objective rewarding calculation operation defines a comfort index, a stability index and a safety redundancy index, and a strategy evaluation function is generated through weighted summation; And the strategy optimizing calculation optimizes the safety strategy vector according to the strategy evaluation function under the simulated event environment generated by the disturbance injection operation.
  9. 9. The system of claim 8, wherein the staged optimization operation of the multi-objective strategy optimization module includes a steady state optimization stage, an immunity optimization stage, and a safety reinforcement stage; the steady-state optimization stage generates an initial security policy based on a standard policy gradient algorithm; the anti-interference optimization stage injects the sudden vibration event and the temperature and humidity abrupt change event into the initial safety strategy; and the security reinforcement stage performs policy iteration by combining the comfort index, the stability index and the security redundancy index, and outputs the security policy vector of a final version.
  10. 10. The system of claim 1, wherein the integrated security decision module performs a multidimensional policy evaluation operation and a security decision mapping operation; The multidimensional strategy evaluation operation calculates physiological adaptation scores, pressure balance scores and disturbance suppression scores corresponding to the safety strategy vectors; and the safety decision mapping operation generates an airbag partition regulation and control parameter, a lumbar support height parameter and a damping rigidity grade parameter according to the physiological adaptation score, the pressure balance score and the disturbance suppression score to jointly form the comprehensive safety regulation and control instruction.

Description

Ripple seat signal fusion system Technical Field The invention relates to the technical field of seat safety control, in particular to a ripple seat signal fusion system. Background With the rapid development of transportation means and intelligent seat systems, seats are used as important carriers for man-machine interaction, and the accuracy and real-time requirements of state monitoring and safety regulation are increasingly improved. Currently, signal sensing of seat systems mostly relies on a single sensor or a limited modality of signal acquisition, such as detecting the weight distribution of the occupant by pressure sensors alone, or capturing the vibration state of the seat by vibration sensors alone. The single signal source is difficult to comprehensively reflect the complex dynamic characteristics of the seat, and misjudgment on the state of the seat is easy to be caused by signal one-sidedness. In the signal processing link, the traditional method often adopts a simple filtering or characteristic screening means, and lacks of deep mining of the inherent correlation between the multi-mode signals. The signals of different types (such as pressure, vibration, temperature and the like) have complex coupling relations in time and space dimensions, and a large amount of key information can be lost through simple item division processing, so that the subsequent state analysis and decision-making lack of reliable basis. The existing seat state modeling method is mostly based on static or semi-static models, and is difficult to adapt to dynamic changes of the seat under different working conditions. For example, factors such as jolt in the running process of a vehicle and posture adjustment of an occupant can cause continuous change of a seat state, and a static model cannot update state parameters in real time, so that timeliness and accuracy of regulation and control are affected. In addition, in the aspect of safety strategy optimization, the traditional method mostly adopts a single target optimization strategy, so that the multi-dimensional requirements of comfort, safety and the like of the seat cannot be balanced, and an optimal regulation and control scheme is difficult to provide in a complex scene. Disclosure of Invention The present invention is directed to a system for fusing signals of a seat with ripple waves, so as to solve the above-mentioned problems in the prior art. To achieve the above object, the present invention provides a ripple seat signal fusion system, the system comprising: the multi-source signal sensing module is used for executing signal sub-term acquisition operation to obtain an original multi-mode sensing signal; The signal characteristic extraction module is used for executing characteristic enhancement conversion operation on the original multi-mode sensing signals to obtain an optimized characteristic matrix; The seat state dynamic modeling module is used for processing the optimized feature matrix by adopting a covariance matrix-driven dynamic structure reconstruction method to generate a seat state association tensor; the multi-target strategy optimization module is used for executing strategy optimization calculation on the seat state association tensor based on a multi-layer strategy reinforcement framework and outputting a safety strategy vector; And the comprehensive safety decision module generates a comprehensive safety regulation instruction according to the safety strategy vector. Preferably, the multi-source signal sensing module comprises a physiological signal acquisition unit, a pressure distribution acquisition unit and an environmental disturbance acquisition unit; the physiological signal acquisition unit acquires an occupant electrocardiographic waveform signal and a myoelectric activity signal; the pressure distribution acquisition unit acquires pressure distribution gradient signals of the surface of the seat through the pressure sensitive sensing array; the environment disturbance acquisition unit acquires a vehicle vibration spectrum signal and a temperature and humidity change signal; The signal sub-term acquisition operation integrates the passenger electrocardiographic waveform signal, the myoelectric activity signal, the seat surface pressure distribution gradient signal, the vehicle vibration spectrum signal and the temperature and humidity change signal to form the original multi-mode sensing signal. Preferably, the feature enhancement conversion operation performed by the signal feature extraction module includes a physiological feature reconstruction sub-operation, a pressure feature enhancer operation and an environmental feature noise reduction sub-operation; The physiological characteristic reconstruction sub-operation performs wavelet packet decomposition on the passenger electrocardio waveform signals and extracts heart rate variability characteristics, and simultaneously performs time-frequency transformation on the myoelectric act