Multi-dimensional perception and active intervention system and method for intelligent safety cabin of new energy vehicles

Abstract

This invention discloses a multi-dimensional perception and active intervention system for an intelligent safety cabin in new energy vehicles, belonging to the field of new energy vehicle safety technology. The system includes modules for multi-dimensional perception, data fusion processing, risk level assessment, active intervention execution, and cloud collaboration, with each module working in tandem. The multi-dimensional perception module employs four-dimensional redundant perception from inside the cabin, outside the cabin, the battery compartment, and the vehicle network; the data fusion processing module achieves data noise reduction and fusion through improved algorithms; the risk level assessment module achieves multi-dimensional coupled assessment and collision damage-battery leakage coupled prediction; the active intervention execution module intervenes according to risk level, adapting to the CTB architecture; and the cloud collaboration module enables four-dimensional interaction and model optimization. This system achieves multi-dimensional perception, accurate assessment, active intervention, and collaborative protection, improving driving safety, adapting to CTB architecture requirements, and mitigating secondary safety risks.

Description

Technical Field

[0001] This invention belongs to the field of new energy vehicle safety technology, specifically, it relates to a multi-dimensional perception and active intervention system and method for a new energy vehicle intelligent safety cabin. Background Technology

[0002] With the rapid popularization of new energy vehicles, especially the widespread application of CTB (Battery-Body Integration) architecture, the demand for automotive safety protection has shifted from traditional passive collision protection to a full-chain active safety upgrade encompassing "perception-assessment-intervention-protection." Existing new energy vehicle safety systems mostly employ single-dimensional perception designs, only capable of monitoring parameters within or outside the cabin, failing to address the multi-dimensional safety needs of occupants, the external environment, battery compartment status, and vehicle-to-everything (V2X) connectivity. Simultaneously, data fusion algorithms suffer from poor adaptability, struggling to eliminate redundancy and errors in multi-source heterogeneous data, resulting in insufficient risk assessment accuracy. Risk assessment models often employ fixed threshold scoring methods, unable to dynamically adjust based on real-time operating conditions, and can only predict occupant collision injuries, neglecting secondary safety risks from battery leakage under the CTB architecture. Active intervention operations lack deep integration with the battery safety system, resulting in weakly targeted intervention strategies and an inability to achieve full-chain closed-loop protection. Cloud collaboration only enables basic data storage, failing to establish four-dimensional interaction between the vehicle, cloud, roadside, and surrounding vehicles, hindering multi-vehicle collaborative early warning and full battery lifecycle management.

[0003] Therefore, in order to solve the above problems, an intelligent security system that can adapt to the CTB architecture and achieve multi-dimensional perception, accurate assessment, proactive intervention and collaborative protection is proposed. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a multi-dimensional perception and active intervention system for intelligent safety cabins of new energy vehicles. This system enables multi-dimensional state perception, accurate risk assessment, hierarchical active intervention, and full-chain collaborative protection of the safety cabin of new energy vehicles, thereby improving the driving safety of new energy vehicles, adapting to the battery protection requirements of CTB architecture, and avoiding secondary risks of occupant injury and battery leakage.

[0005] To achieve the aforementioned objectives, the technical solution adopted by this invention includes: a multi-dimensional perception and active intervention system for a smart safety cabin of a new energy vehicle, comprising a multi-dimensional perception module, a data fusion processing module, a risk level assessment module, an active intervention execution module, and a cloud collaboration module, wherein each module is electrically connected in sequence, and the cloud collaboration module is bidirectionally electrically connected to the multi-dimensional perception module and the data fusion processing module respectively.

[0006] The multi-dimensional perception module adopts a four-dimensional redundant perception design, including in-cabin, out-of-cabin, battery compartment, and vehicle-to-everything (V2X) network. At least three sets of distributed perception units are set up inside the cabin, and a multi-sensor fusion perception unit is set up outside the cabin. The battery compartment adds a CTB architecture-adaptive perception unit. The V2X unit accesses roadside equipment and surrounding vehicle data to collect multi-dimensional parameters of the new energy vehicle's safety cabin, battery compartment, and V2X network. These multi-dimensional parameters include physiological and behavioral parameters of occupants inside the cabin, parameters of the cabin environment, parameters of the road environment and traffic participants outside the cabin, vehicle's own operating parameters, battery compartment status parameters, and V2X network early warning parameters.

[0007] The data fusion processing module adopts an improved adaptive weighted fusion algorithm. Combined with the data characteristics of the CTB architecture battery compartment, it first performs noise reduction processing on the raw data of a single sensor through a Kalman-particle filter hybrid algorithm, and then eliminates the redundancy and error of multi-source heterogeneous data through dynamic weight allocation, outputting standardized feature data.

[0008] The risk level assessment module is based on standardized feature data output by the data fusion processing module. It adopts a multi-dimensional coupling + dynamic threshold adaptive risk assessment model to determine the safety risk level in real time. The risk level is divided into four levels: no risk, low risk, medium risk, and high risk. It also has a built-in collision damage-battery leakage coupling prediction submodule, which can predict the degree of occupant injury and battery leakage risk in collision scenarios in real time.

[0009] The active intervention execution module, based on the risk level, intervention instructions, and coupling prediction results output by the risk level assessment module, coordinates with the vehicle power system, braking system, steering system, battery safety system, cabin protection system, and vehicle network early warning system to execute corresponding levels of active intervention operations, thereby achieving a closed-loop chain of early warning-intervention-protection-coordination.

[0010] The cloud-based collaborative module adopts a four-dimensional interactive architecture encompassing the vehicle, cloud, roadside, and surrounding vehicles. It incorporates a big data analysis submodule and a CTB architecture battery safety model. Through an improved gradient descent algorithm, it optimizes the weight coefficients and dynamic thresholds of the risk assessment model, while simultaneously enabling multi-vehicle data collaboration, risk tracing, and closed-loop management of the entire battery lifecycle.

[0011] Furthermore, the four-dimensional sensing unit of the multi-dimensional sensing module specifically includes: The cabin sensing unit includes an infrared physiological sensor, a flexible pressure sensor, a gas sensor, and an AI high-definition camera. The infrared physiological sensor is used to collect parameters such as occupant heart rate, respiratory rate, body surface temperature, and blood oxygen saturation at a frequency of 10-20Hz. The flexible pressure sensor is embedded in the seat, seat belt, and steering wheel to collect occupant posture, seat belt wearing status, force parameters, and driving operation force. The gas sensor is used to collect parameters such as CO, CO2, VOC, smoke concentration, and characteristic gases (HF, CO) of battery leakage in the cabin. The AI ​​high-definition camera is used to identify occupant facial states (fatigue, distraction, emotions), foreign objects in the cabin, and children / pets left behind. The facial state recognition uses a deep learning convolutional neural network algorithm with an accuracy rate of ≥98%.

[0012] The external sensing unit includes an ultra-long-range lidar, a 4D imaging radar, an 8-megapixel high-definition camera, and a millimeter-wave radar. The ultra-long-range lidar has a detection range of no less than 500m and is used to identify distant obstacles and adverse weather conditions (heavy rain, heavy fog). The 4D imaging radar is used to collect parameters such as the speed, distance, movement trajectory, and behavioral intent of surrounding vehicles and pedestrians. The high-definition camera enables 360° environmental scanning to identify traffic signs, lane lines, irregularly shaped obstacles, road damage, and water accumulation. The millimeter-wave radar is used for short-range obstacle detection, following distance monitoring, and blind spot warning.

[0013] The battery compartment CTB-adaptive sensing unit includes a battery status sensor, a temperature sensor, a pressure sensor, and a gas sensor. It is adapted to the three-layer structure of the CTB "vehicle floor-blade battery-battery tray" and is embedded at the connection between the battery tray and the vehicle floor. It is used to collect battery voltage / temperature / internal resistance, battery compartment pressure, characteristic gases of battery leakage, and vehicle floor deformation parameters, with a sampling frequency of ≥15Hz.

[0014] The vehicle-to-everything (V2X) sensing unit uses a 5G-V2X module to access roadside units (RSUs) and vehicle data within a 500m radius. It collects roadside traffic warnings, road conditions, traffic congestion, and risk status parameters of surrounding vehicles to achieve vehicle-road and vehicle-to-vehicle cooperative early warning systems.

[0015] Furthermore, the improved adaptive weighted fusion algorithm and Kalman-particle filter hybrid noise reduction algorithm used in the data fusion processing module are specifically calculated using the following formulas: Kalman-Particle Filter Hybrid Noise Recursive Formula: in, Let A be the predicted state value at time k, and let A be the state transition matrix. Let B be the state estimate at time k-1, and let B be the control input matrix. For the control input at time k-1, Let Q be the prediction error covariance at time k, and let Q be the process noise covariance. H is the Kalman gain, and H is the observation matrix. Let R be the observation value at time k, R be the observation noise covariance, and I be the identity matrix. The state estimate at time k. Let k be the time-time estimate of the error covariance. The data is the denoised data after particle filtering optimization, where N is the number of particles. Let be the weight of the i-th particle. Let be the state value of the i-th particle.

[0016] Improved adaptive weighted fusion algorithm formula: in, The data consists of standardized feature data after fusion, where n is the number of sensors involved in the fusion. Let i be the dynamic weighting coefficient of the i-th sensor. The sensor data is denoised using a Kalman-particle filter hybrid algorithm. Let V be the variance of the data collected by the i-th sensor. To adapt correction factors for the CTB architecture, the battery compartment sensors The remaining sensors .

[0017] Furthermore, the multi-dimensional coupling + dynamic threshold adaptive risk assessment model of the risk level assessment module adopts a multi-feature coupling weighted scoring method, combined with a dynamic threshold adjustment mechanism. The specific calculation formula is as follows: Where R is the total risk assessment score, ranging from 0 to 100. The five evaluation dimensions are: occupant condition, cabin environment condition, external road environment, vehicle condition, and battery compartment condition, which serve as the base weighting coefficients for each evaluation dimension. Among them, the weighting coefficient of the occupant status inside the cabin Weighting coefficient of cabin environment status External road environment weighting coefficient Vehicle's own state weighting coefficient Battery compartment status weighting coefficient ; The scores for each evaluation dimension range from 0 to 100. This is the dimensional coupling correction coefficient, used to correct the weight deviation after risk coupling in different dimensions, with a value range of 0.9-1.1; The dynamic threshold for risk level at time t. The risk level threshold at time t-1 This is the threshold adjustment coefficient, with a value ranging from 0.02 to 0.05. This refers to the amount of change in standardized feature data.

[0018] Dynamic risk level assessment criteria: based on Dynamic adjustment, R < ×0.5 indicates no risk. ×0.5≤R< A score of ×0.8 indicates low risk. ×0.8≤R< ×1.2 is classified as medium risk, R≥ ×1.2 is considered high risk.

[0019] The collision injury-battery leakage coupling prediction submodule, based on the correlation between collision impulse, battery compartment deformation, occupant injury, and battery leakage, adopts a planar two-degree-of-freedom rigid body collision-battery deformation coupling model to predict the change in occupant velocity, injury level, and battery leakage risk after the collision. The specific calculation formula is as follows: Where P is the collision impulse, e is the vehicle recovery coefficient, and C is a coefficient related to the collision direction. , These are the velocity components of the vehicle itself and the object it collides with in the direction of impulse, respectively. , These are the distances from the center of mass of the vehicle and the object being collided to the line of action of the impulse, respectively. , These are the yaw rates of the vehicle itself and the object it collided with, respectively. , The masses of the vehicle itself and the object it collided with are respectively. , These are the moments of inertia of the vehicle itself and the object it collides with, respectively. The value represents the change in vehicle speed after a collision; D represents the battery leakage risk value; and k represents the collision impulse correction factor. For the battery compartment protection factor, For the battery compartment protection area; according to The size of the battery is combined with the AIS damage scoring standard to determine the occupant injury level, and the battery leakage risk is determined according to the D value (D<0 means no leakage risk, 0≤D<5 means low leakage risk, and D≥5 means high leakage risk).

[0020] Furthermore, the intervention operations of the active intervention execution module are divided into four levels according to the risk level and the coupling result of battery leakage risk, and are designed in conjunction with the battery protection requirements of the CTB architecture, as follows: (1) No risk (R < ×0.5 points, D<0): No active intervention is performed, only real-time monitoring of various parameters is carried out, data is synchronized to the cloud collaboration module, and the structural status of the battery compartment CTB is monitored routinely. (2) Low risk ×0.5≤R< ×0.8 points, D<0): Implement early warning interventions, including in-cabin audio and visual warnings, instrument panel text prompts, seat vibration warnings, while adjusting the in-cabin air conditioning airflow and air purification system power to optimize the in-cabin environment and simultaneously push vehicle network warning information to the driver. (3) Medium risk ×0.8≤R< ×1.2 points, D<0 or 0≤D<5): Assistance control interventions are implemented, including automatic adjustment of following distance, lane departure warning correction, and battery system pre-cooling / preventive protection; the following distance is based on millimeter-wave radar + vehicle-to-everything (V2X) data, meeting… d is the safe following distance, v is the current vehicle speed, t is the safe reaction time (t is taken as 1.5-2s), and a is the maximum braking acceleration. The road correction distance is based on vehicle network data; the battery system pre-cooling is to activate the CTB-adaptive full-bonding liquid cooling system when the battery temperature exceeds 45℃, with a cooling rate ≥2℃ / min; the battery protection is to activate the battery compartment pressure relief valve when D≥0. (4) High risk (R≥ ×1.2 points, D≥5): Execute emergency control interventions, including automatic emergency braking, emergency steering to avoid hazards, cutting off the high-voltage battery circuit, pretensioning seat belts, graded airbag deployment, emergency ventilation in the cabin, and fire and explosion protection of the battery compartment. Simultaneously, send emergency warning information to the cloud, roadside equipment, and surrounding vehicles, and coordinate with the rescue system; the braking acceleration of the automatic emergency braking is ≥8 m / s². 2The emergency steering response time is ≤0.1s, the seatbelt pretension force is ≥2000N, the airbags are deployed in stages based on collision injury predictions, inflating airbags with varying degrees of force, and the cabin emergency ventilation rate is ≥0.5m. 3 / min, the fireproof and explosion-proof battery compartment is designed to activate the in-cabin fire extinguishing system and provide thermal insulation protection for the CTB structure, cutting off the heat conduction path between the battery compartment and the crew compartment.

[0021] Furthermore, the cloud-based collaborative module adopts a four-dimensional interactive architecture encompassing the vehicle, cloud, roadside, and surrounding vehicles. It incorporates a big data analysis submodule and a CTB-based battery safety model, and optimizes the weight coefficients of the risk assessment model using an improved gradient descent algorithm. and dynamic threshold The optimized formula is as follows: in, The weight coefficients are the weights after the (k+1)th iteration. The weight coefficients are after the k-th iteration. The dynamic threshold after the (k+1)th iteration. Let be the dynamic threshold after the k-th iteration, η be the learning rate (ranging from 0.01 to 0.05), and L be the loss function. The weight coefficients of the loss function for the k-th iteration The partial derivatives, The loss function is the dynamic threshold for the k-th iteration. The partial derivatives, To optimize the correction coefficients for the dimensions, This is the threshold adjustment coefficient.

[0022] The cloud-based collaborative module stores data including raw data collected by the multi-dimensional perception module, standardized data after data fusion processing, risk level assessment results, active intervention operation logs, collision accident data, battery life cycle data, and vehicle network collaborative data. The storage period is no less than 5 years, supporting data traceability, accident review, and multi-vehicle collaborative risk warning. At the same time, through the CTB architecture battery safety model, it updates battery protection parameters in real time and optimizes intervention strategies.

[0023] Furthermore, the active intervention execution module is linked with the battery safety system, and is designed with a six-fold closed-loop protection logic of pre-, guided, structured, isolated, diverted, and connected for the CTB architecture. Specifically, it includes: using high heat-resistant and stable battery cells at the cell level; using mesh nanoporous heat insulation material and a high-temperature resistant upper shell (temperature resistance ≥1400℃) for the battery compartment, which is compatible with the CTB three-layer structure; equipping a fully bonded liquid cooling rapid cooling system, which improves heat dissipation efficiency by ≥30%; collecting battery parameters more than 10 times per second through the battery management system, and immediately initiating cooling and high-voltage cut-off operations in case of abnormality; adding a battery compartment pressure release channel and a fireproof and explosion-proof layer to block heat conduction; and linking with the vehicle network and rescue system to simultaneously push rescue information when the battery leaks.

[0024] Furthermore, the multi-dimensional perception module adopts an aerospace-grade redundant design, with dual backups for the in-cabin, out-of-cabin, battery compartment, and vehicle-to-everything (V2X) perception units. When any perception unit fails, the backup unit can take over within milliseconds, ensuring uninterrupted perception data. The data fusion processing module and risk level assessment module both adopt a dual-chip redundant architecture with a total computing power of no less than 1000 TOPS, ensuring the real-time performance of data processing and risk assessment, with a processing latency of ≤50ms. At the same time, a CTB architecture adaptability detection unit is added to monitor the compatibility status between the perception unit and the CTB structure in real time, avoiding perception errors caused by structural compatibility issues.

[0025] Furthermore, it also includes a customized occupant-battery risk collaborative protection submodule. Based on occupant height, weight, posture, and physiological parameters collected by the in-cabin sensing unit, and combined with the collision injury-battery leakage coupling prediction results, it adjusts the seat belt pretension, airbag inflation volume, and in-cabin protection posture using the following formula: Where F represents the seatbelt pretension force, m represents the occupant's weight, h represents the occupant's height, s represents the occupant's sitting posture offset, and D represents the battery leakage risk value. , , , V is the preload correction factor; V is the airbag inflation volume. This represents the change in vehicle speed after the collision. , , This is the inflation volume correction factor; Adjust the angle for seat protection posture. , This is the attitude adjustment coefficient.

[0026] Compared with the prior art, the advantages of the present invention include: (1) Adopting a four-dimensional redundant perception design of in-cabin, out-of-cabin, battery compartment and vehicle network, adapting to the requirements of CTB architecture, realizing comprehensive collection of multi-dimensional parameters, solving the limitations of existing single-dimensional perception, and improving the comprehensiveness and reliability of perception.

[0027] (2) The improved adaptive weighted fusion algorithm combines Kalman-particle filtering for noise reduction and introduces CTB architecture to adapt and correct the coefficients, thereby improving the accuracy of data fusion and eliminating redundancy and errors in multi-source heterogeneous data.

[0028] (3) A multi-dimensional coupling + dynamic threshold adaptive risk assessment model is adopted, which combines collision damage-battery leakage coupling prediction to achieve accurate prediction of dual risks, dynamically adjust the risk threshold, and improve the pertinence and accuracy of risk assessment.

[0029] (4) The proactive intervention operation is designed according to the risk level and deeply linked to the battery safety system. The protection strategy is optimized for the CTB architecture to realize the closed loop of the whole chain of early warning-intervention-protection-linkage, effectively avoiding the secondary risks of occupant injury and battery leakage.

[0030] (5) The cloud collaboration adopts a four-dimensional interactive architecture to realize multi-vehicle collaboration, battery life cycle management and risk assessment model dynamic optimization, thereby improving the system's collaboration and adaptability.

[0031] (6) The combination of redundant design and customized protection ensures the stability of system operation and can adapt to the protection needs of different occupants, further improving the safety protection effect. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0034] In view of the shortcomings of the prior art, the inventors of this invention, through long-term research and extensive practice, have proposed the technical solution of this invention. The technical solution, its implementation process, and principles will be further explained below with reference to the accompanying drawings and specific implementation examples in the embodiments of this application.

[0035] It should be noted that the embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, the present invention covers any substitutions, modifications, equivalent methods and solutions made on the spirit, principles and scope of the present invention as defined by the claims. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] In the description of this application, the terms "first," "second," "third," and similar words do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms "a" or "one," and similar words, do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms "comprising" or "including," and similar words, mean that the elements or objects preceding "comprising" or "including" encompass the elements or objects listed following "comprising" or "including," and their equivalents, but do not exclude other elements or objects. The terms "connected" or "linked," and similar words, are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0037] In the description of this application, the terms "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used solely for the convenience of describing this application and for simplification, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, when using positional terms such as "both sides," "outer side," and "upper and lower," it should be understood that they are used only for ease of understanding and description, taking into account that the structure may be oriented to other positions.

[0038] In the description of this application, unless otherwise expressly specified and limited, the technical or scientific terms used shall have the ordinary meaning understood by a person with ordinary skills in the art to which this application pertains. Terms such as “installation,” “connection,” and “joining” shall be interpreted broadly, for example, as fixed connection, detachable connection, mating connection, or integral connection. For a person skilled in the art, the specific meaning of the above terms in this application can be understood according to the specific circumstances.

[0039] Example 1, please refer to Figure 1This embodiment provides a multi-dimensional perception and active intervention system for a smart safety cabin of a new energy vehicle, including a multi-dimensional perception module, a data fusion processing module, a risk level assessment module, an active intervention execution module, and a cloud collaboration module. Each module is electrically connected in sequence. The cloud collaboration module is bidirectionally electrically connected to the multi-dimensional perception module and the data fusion processing module. Each module is integrated into the body and battery compartment of the new energy vehicle and is adapted to the three-layer structure of CTB: "body floor - blade battery - battery tray".

[0040] The multi-dimensional perception module adopts a four-dimensional redundant perception design, including in-cabin, out-of-cabin, battery compartment, and vehicle-to-everything (V2X) connectivity. The specific settings are as follows: In-cabin sensing units: One distributed sensing unit is installed in each of the driver's seat, front passenger seat, and rear seats, including an infrared physiological sensor, a flexible pressure sensor, a gas sensor, and an AI high-definition camera; the infrared physiological sensor is installed in the seat headrest, with a sampling frequency of 15Hz, to collect occupant heart rate, respiratory rate, body surface temperature, and blood oxygen saturation in real time; the flexible pressure sensor is embedded in the seat cushion, seat belt webbing, and steering wheel grip, respectively, to collect occupant sitting posture, seat belt tightness, seat force, and steering wheel operation force; the gas sensor is installed in the center console and rear roof to collect the concentration of CO, CO2, VOC, smoke concentration, and the concentration of HF and CO, characteristic gases of battery leakage; the AI ​​high-definition camera is installed in the rearview mirror, using a deep learning convolutional neural network algorithm, with a recognition accuracy of 98.5%, to identify occupant facial status (fatigue, distraction, emotion), foreign objects in the cabin, and children / pets left behind in real time.

[0041] External Sensing Unit: An ultra-long-range lidar is installed at the front grille, with a detection range of 550m, capable of identifying distant obstacles and adverse weather conditions such as heavy rain and fog; a 4D imaging radar is installed at the front and rear bumpers and left and right rearview mirrors to collect data on the speed, distance, movement trajectory, and behavioral intentions of surrounding vehicles and pedestrians; an 8-megapixel high-definition camera is installed around the vehicle body to achieve 360° environmental scanning, identifying traffic signs, lane lines, irregular obstacles, road damage, and water accumulation; a millimeter-wave radar is installed at the front and rear bumpers for close-range obstacle detection, following distance monitoring, and blind spot warning.

[0042] Battery compartment CTB-compatible sensing unit: Embedded at the connection between the battery tray and the vehicle floor, including a battery status sensor, a temperature sensor, a pressure sensor, and a gas sensor; the battery status sensor collects battery voltage, temperature, and internal resistance parameters in real time, the temperature sensor collects the ambient temperature inside the battery compartment, the pressure sensor collects the pressure inside the battery compartment, and the gas sensor collects the concentration of characteristic gases leaking from the battery. The sampling frequency is set to 18Hz to adapt to the installation requirements of the CTB three-layer structure.

[0043] Vehicle-to-everything (V2X) sensing unit: It adopts a 5G-V2X module and is installed on the roof antenna to access roadside units (RSU) and vehicle data within 500m. It collects roadside traffic warnings, road conditions, traffic congestion and surrounding vehicle risk status parameters in real time to realize vehicle-road cooperation and vehicle-to-vehicle cooperation warnings.

[0044] The data fusion processing module employs an improved adaptive weighted fusion algorithm, combined with the data characteristics of the CTB architecture battery compartment. The working process is as follows: The first step involves denoising the raw data from a single sensor using a hybrid Kalman-particle filter algorithm. Kalman filtering is used for initial denoising, while particle filtering optimizes the denoising accuracy. The number of particles, N, is set to 1000. The denoised data for each sensor is then calculated recursively. ; The second step involves using an improved adaptive weighted fusion algorithm to calculate the dynamic weight coefficients of each sensor. The CTB architecture adaptation correction coefficient of the battery compartment sensor The remaining sensors Through formula The fused standardized feature data is calculated to eliminate redundancy and errors in multi-source heterogeneous data, and then output to the risk level assessment module.

[0045] The risk level assessment module is based on standardized feature data and employs a multi-dimensional coupled + dynamic threshold adaptive risk assessment model. Its workflow is as follows: The first step is to determine the basic weighting coefficients for the five evaluation dimensions. The status of the occupants inside the cabin Cabin environment status External road environment Vehicle's own condition Battery compartment status Dimensional Coupling Correction Coefficient The value ranges from 0.9 to 1.1, and is dynamically adjusted according to real-time operating conditions. The second step is to use the formula. Calculate the total risk assessment score R using the formula. Calculate dynamic threshold Threshold adjustment coefficient Set it to 0.03; Third step, based on R and Relationship risk level assessment: R < ×0.5 indicates no risk. ×0.5≤R< A score of ×0.8 indicates low risk. ×0.8≤R< ×1.2 is classified as medium risk, R≥ ×1.2 is considered high risk; The fourth step involves using the collision damage-battery leakage coupling prediction submodule, employing a planar two-degree-of-freedom rigid body collision-battery deformation coupling model, to calculate the collision impulse P and the change in vehicle velocity. The battery leakage risk value D is used to determine the occupant injury level based on the AIS injury scoring standard. The battery leakage risk is determined based on the D value, and the risk level, occupant injury level and battery leakage risk value are output to the active intervention execution module.

[0046] The proactive intervention module executes intervention actions at the corresponding level based on the risk level and the coupling result of battery leakage risk, as follows: (1) No risk (R < ×0.5 points, D<0): No active intervention is performed, only real-time monitoring of various parameters is carried out, data is synchronized to the cloud collaboration module, and the structural status of the battery compartment CTB is monitored routinely. (2) Low risk ×0.5≤R< ×0.8 points, D<0): Activate cabin sound and light warning, instrument panel text prompts, seat vibration warning, adjust cabin air conditioning air volume and air purification system power to optimize cabin environment, and simultaneously push vehicle network warning information to the driver; (3) Medium risk ×0.8≤R< ×1.2 points, D<0 or 0≤D<5): Automatically adjust following distance; in the following distance formula, t is taken as 1.8s. Adjustments are made in real time based on road condition data from the vehicle network; lane departure assist correction is activated to prevent the vehicle from deviating from its lane; when the battery temperature exceeds 45℃, the CTB-adaptive full-bonding liquid cooling system is activated, with the cooling rate controlled at 2.5℃ / min; when D≥0, the battery compartment pressure relief valve is activated. (4) High risk (R≥ ×1.2 points, D≥5): Activate automatic emergency braking, setting the braking acceleration to 8.5 m / s². 2 ; Initiate emergency steering to avoid collisions, with a steering response time controlled within 0.08s; Disconnect the battery high-voltage circuit, activate seatbelt pretensioning (pretension force set to 2200N), and trigger airbag inflation according to the collision injury prediction results; Activate emergency ventilation inside the cabin, with a ventilation rate set to 0.6m. 3 / min; activate the battery compartment fireproof and explosion-proof system, activate the in-cabin fire extinguishing system and CTB structural heat insulation protection, cut off the heat conduction path between the battery compartment and the passenger compartment; at the same time, send emergency warning information to the cloud, roadside equipment and surrounding vehicles, and coordinate with the rescue system.

[0047] The cloud-based collaboration module adopts a four-dimensional interactive architecture encompassing the vehicle, cloud, roadside, and surrounding vehicles. Its operation process is as follows: The first step is to receive multi-dimensional perception raw data, standardized feature data, risk level assessment results, active intervention operation logs, and other data sent by the vehicle terminal, with a storage period of 6 years, supporting data traceability and accident review; The second step is to optimize the weight coefficients of the risk assessment model using an improved gradient descent algorithm. and dynamic threshold The learning rate η is set to 0.03. The optimized parameters are obtained through iterative calculation and then pushed to the vehicle-side risk level assessment module. The third step is to use the CTB architecture battery safety model to update battery protection parameters in real time and optimize active intervention strategies. The fourth step is to enable data interaction between the vehicle, cloud, roadside, and surrounding vehicles, push multi-vehicle collaborative risk warning information, and achieve multi-vehicle collaborative protection.

[0048] The multi-dimensional perception module adopts an aerospace-grade redundancy design, with dual backups for the in-cabin, out-of-cabin, battery compartment, and vehicle-to-everything (V2X) perception units. If any perception unit fails, the backup unit will take over within 50ms to ensure uninterrupted perception data. The data fusion processing module and risk level assessment module adopt a dual-chip redundancy architecture with a total computing power of 1200 TOPS and a processing latency of 45ms. A CTB architecture adaptability detection unit is added to monitor the compatibility status of the perception units with the CTB structure in real time.

[0049] The occupant customization-battery risk collaborative protection submodule, based on occupant height, weight, posture, and physiological parameters collected by the in-cabin sensing unit, combined with collision damage-battery leakage coupling prediction results, adjusts seat belt pretension, airbag inflation volume, and seat protection posture through formulas, including a pretension correction coefficient. , , , Inflation volume correction factor , , Attitude adjustment coefficient , This allows for customized protection for occupants of different body types and physiological states, while also preventing secondary injuries caused by battery leakage.

[0050] It should be understood that the above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. It should not be considered that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, several simple deductions or substitutions can be made without departing from the concept of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the protection scope of the present invention.

Claims

1. A multi-dimensional perception and active intervention system for an intelligent safety cabin in new energy vehicles, characterized in that, It includes a multi-dimensional perception module, a data fusion processing module, a risk level assessment module, an active intervention execution module, and a cloud collaboration module. Each module is electrically connected in sequence, and the cloud collaboration module is bidirectionally electrically connected to the multi-dimensional perception module and the data fusion processing module. The multi-dimensional perception module adopts a four-dimensional redundant perception design, including in-cabin, out-of-cabin, battery compartment, and vehicle-to-everything (V2X) network. At least three sets of distributed perception units are set inside the cabin, and a multi-sensor fusion perception unit is set outside the cabin. The battery compartment adds a CTB architecture-adaptive perception unit. The V2X unit connects to roadside equipment and surrounding vehicle data to collect multi-dimensional parameters of the new energy vehicle's safety cabin, battery compartment, and V2X network. These multi-dimensional parameters include physiological and behavioral parameters of occupants inside the cabin, environmental parameters inside the cabin, road environment and traffic participant parameters outside the cabin, vehicle's own operating parameters, battery compartment status parameters, and V2X network early warning parameters. The data fusion processing module adopts an improved adaptive weighted fusion algorithm. Combined with the data characteristics of the CTB architecture battery compartment, it first performs noise reduction processing on the raw data of a single sensor through a Kalman-particle filter hybrid algorithm, and then eliminates the redundancy and error of multi-source heterogeneous data through dynamic weight allocation, and outputs standardized feature data. The risk level assessment module is based on the standardized feature data output by the data fusion processing module. It adopts a multi-dimensional coupling + dynamic threshold adaptive risk assessment model to determine the safety risk level in real time. The risk level is divided into four levels: no risk, low risk, medium risk and high risk. It also has a built-in collision damage-battery leakage coupling prediction sub-module, which can predict the degree of occupant injury and battery leakage risk in collision scenarios in real time. The active intervention execution module, based on the risk level, intervention instructions, and coupling prediction results output by the risk level assessment module, coordinates with the vehicle power system, braking system, steering system, battery safety system, cabin protection system, and vehicle network early warning system to execute corresponding levels of active intervention operations, thereby achieving a closed-loop chain of early warning-intervention-protection-coordination. The cloud-based collaborative module adopts a four-dimensional interactive architecture encompassing the vehicle, cloud, roadside, and surrounding vehicles. It incorporates a big data analysis submodule and a CTB architecture battery safety model. Through an improved gradient descent algorithm, it optimizes the weight coefficients and dynamic thresholds of the risk assessment model, while simultaneously enabling multi-vehicle data collaboration, risk tracing, and closed-loop management of the entire battery lifecycle.

2. The system according to claim 1, characterized in that, The four-dimensional sensing unit of the multi-dimensional sensing module specifically includes: The cabin sensing unit includes an infrared physiological sensor, a flexible pressure sensor, a gas sensor, and an AI high-definition camera. The infrared physiological sensor collects occupant heart rate, respiratory rate, body surface temperature, and blood oxygen saturation parameters at a frequency of 10-20Hz. The flexible pressure sensor is embedded in the seat, seat belt, and steering wheel to collect occupant posture, seat belt wearing status, force parameters, and driving operation force. The gas sensor collects parameters for CO, CO2, VOC, smoke concentration, and characteristic gases (HF, CO) from battery leaks within the cabin. The AI ​​high-definition camera identifies occupant facial states (fatigue, distraction, emotions), foreign objects in the cabin, and items left by children / pets. The facial recognition uses a deep learning convolutional neural network algorithm with an accuracy rate of ≥98%. The external sensing unit includes an ultra-long-range lidar, a 4D imaging radar, an 8-megapixel high-definition camera, and a millimeter-wave radar. The ultra-long-range lidar has a detection range of no less than 500m and is used to identify distant obstacles and adverse weather conditions (heavy rain, dense fog). The 4D imaging radar is used to collect parameters such as the speed, distance, movement trajectory, and behavioral intent of surrounding vehicles and pedestrians. The high-definition camera performs 360° environmental scanning to identify traffic signs, lane lines, irregularly shaped obstacles, road damage, and water accumulation. The millimeter-wave radar is used for short-range obstacle detection, following distance monitoring, and blind spot warning. The battery compartment CTB-adaptive sensing unit includes a battery status sensor, a temperature sensor, a pressure sensor, and a gas sensor. It is adapted to the three-layer structure of the CTB "vehicle floor - blade battery - battery tray" and is embedded at the connection between the battery tray and the vehicle floor. It is used to collect battery voltage / temperature / internal resistance, battery compartment pressure, characteristic gas of battery leakage, and vehicle floor deformation parameters, with a collection frequency ≥15Hz. The vehicle-to-everything (V2X) sensing unit uses a 5G-V2X module to access roadside units (RSUs) and vehicle data within a 500m radius. It collects roadside traffic warnings, road conditions, traffic congestion, and risk status parameters of surrounding vehicles to achieve vehicle-road and vehicle-to-vehicle cooperative early warning systems.

3. The system according to claim 1, characterized in that, The data fusion processing module employs an improved adaptive weighted fusion algorithm and a Kalman-particle filter hybrid noise reduction algorithm, with the specific calculation formulas as follows: Kalman-Particle Filter Hybrid Noise Recursive Formula: in, Let A be the predicted state value at time k, and let A be the state transition matrix. Let B be the state estimate at time k-1, and let B be the control input matrix. For the control input at time k-1, Let Q be the prediction error covariance at time k, and let Q be the process noise covariance. H is the Kalman gain, and H is the observation matrix. Let R be the observation value at time k, R be the observation noise covariance, and I be the identity matrix. The state estimate at time k. Let k be the time-time estimate of the error covariance. The data is the denoised data after particle filtering optimization, where N is the number of particles. Let be the weight of the i-th particle. Let be the state value of the i-th particle; Improved adaptive weighted fusion algorithm formula: in, The data consists of standardized feature data after fusion, where n is the number of sensors involved in the fusion. Let i be the dynamic weighting coefficient of the i-th sensor. The sensor data is denoised using a Kalman-particle filter hybrid algorithm. Let V be the variance of the data collected by the i-th sensor. To adapt correction factors for the CTB architecture, the battery compartment sensors The remaining sensors .

4. The system according to claim 1, characterized in that, The risk level assessment module employs a multi-dimensional coupling + dynamic threshold adaptive risk assessment model, which uses a multi-feature coupling weighted scoring method combined with a dynamic threshold adjustment mechanism. The specific calculation formula is as follows: Where R is the total risk assessment score, ranging from 0 to 100. The five evaluation dimensions are: occupant condition, cabin environment condition, external road environment, vehicle condition, and battery compartment condition, which serve as the base weighting coefficients for each evaluation dimension. Among them, the weighting coefficient of the occupant status inside the cabin Weighting coefficient of cabin environment status External road environment weighting coefficient Vehicle's own state weighting coefficient Battery compartment status weighting coefficient ; The scores for each evaluation dimension range from 0 to 100. This is the dimensional coupling correction coefficient, used to correct the weight deviation after risk coupling in different dimensions, with a value range of 0.9-1.1; The dynamic threshold for risk level at time t. The risk level threshold at time t-1 This is the threshold adjustment coefficient, with a value ranging from 0.02 to 0.

05. This refers to the amount of change in standardized feature data; Dynamic risk level assessment criteria: based on Dynamic adjustment, R < ×0.5 indicates no risk. ×0.5≤R< A score of ×0.8 indicates low risk. ×0.8≤R< ×1.2 is classified as medium risk, R≥ ×1.2 is considered high risk; The collision injury-battery leakage coupling prediction submodule, based on the correlation between collision impulse, battery compartment deformation, occupant injury, and battery leakage, adopts a planar two-degree-of-freedom rigid body collision-battery deformation coupling model to predict the change in occupant velocity, injury level, and battery leakage risk after the collision. The specific calculation formula is as follows: Where P is the collision impulse, e is the vehicle recovery coefficient, and C is a coefficient related to the collision direction. , These are the velocity components of the vehicle itself and the object it collides with in the direction of impulse, respectively. , These are the distances from the center of mass of the vehicle and the object being collided to the line of action of the impulse, respectively. , These are the yaw rates of the vehicle itself and the object it collided with, respectively. , The masses of the vehicle itself and the object it collided with are respectively. , These are the moments of inertia of the vehicle itself and the object it collides with, respectively. The value represents the change in vehicle speed after a collision; D represents the battery leakage risk value; and k represents the collision impulse correction factor. For the battery compartment protection factor, For the battery compartment protection area; according to The size of the battery is combined with the AIS damage scoring standard to determine the occupant injury level, and the battery leakage risk is determined according to the D value (D<0 means no leakage risk, 0≤D<5 means low leakage risk, and D≥5 means high leakage risk).

5. The system according to claim 1, characterized in that, The intervention operations of the active intervention execution module are divided into four levels according to the risk level and the coupling result of battery leakage risk, and are designed in conjunction with the battery protection requirements of the CTB architecture, as follows: (1) No risk (R < ×0.5 points, D<0): No active intervention is performed, only real-time monitoring of various parameters is carried out, data is synchronized to the cloud collaboration module, and the structural status of the battery compartment CTB is monitored routinely. (2) Low risk ×0.5≤R< ×0.8 points, D<0): Implement early warning interventions, including in-cabin audio and visual warnings, instrument panel text prompts, seat vibration warnings, while adjusting the in-cabin air conditioning airflow and air purification system power to optimize the in-cabin environment and simultaneously push vehicle network warning information to the driver. (3) Medium risk ×0.8≤R< ×1.2 points, D<0 or 0≤D<5): Assistance control interventions are implemented, including automatic adjustment of following distance, lane departure warning correction, and battery system pre-cooling / preventive protection; the following distance is based on millimeter-wave radar + vehicle-to-everything (V2X) data, meeting… d is the safe following distance, v is the current vehicle speed, t is the safe reaction time (t is taken as 1.5-2s), and a is the maximum braking acceleration. The road correction distance is based on vehicle network data; the battery system pre-cooling is to activate the CTB-adaptive full-bonding liquid cooling system when the battery temperature exceeds 45℃, with a cooling rate ≥2℃ / min; the battery protection is to activate the battery compartment pressure relief valve when D≥0. (4) High risk (R≥ ×1.2 points, D≥5): Execute emergency control interventions, including automatic emergency braking, emergency steering to avoid hazards, cutting off the high-voltage battery circuit, pretensioning seat belts, graded airbag deployment, emergency ventilation in the cabin, and fire and explosion protection of the battery compartment. Simultaneously, send emergency warning information to the cloud, roadside equipment, and surrounding vehicles, and coordinate with the rescue system; the braking acceleration of the automatic emergency braking is ≥8 m / s². 2 The emergency steering response time is ≤0.1s, the seatbelt pretension force is ≥2000N, the airbags are deployed in stages based on collision injury predictions, inflating airbags with varying degrees of force, and the cabin emergency ventilation rate is ≥0.5m. 3 / min, the fireproof and explosion-proof battery compartment is designed to activate the in-cabin fire extinguishing system and provide thermal insulation protection for the CTB structure, cutting off the heat conduction path between the battery compartment and the crew compartment.

6. The system according to claim 1, characterized in that, The cloud-based collaborative module adopts a four-dimensional interactive architecture encompassing the vehicle, cloud, roadside, and surrounding vehicles. It incorporates a big data analysis submodule and a CTB (Carrier-to-Body) battery safety model, and optimizes the weight coefficients of the risk assessment model using an improved gradient descent algorithm. and dynamic threshold The optimized formula is as follows: in, The weight coefficients are the weights after the (k+1)th iteration. The weight coefficients are after the k-th iteration. The dynamic threshold after the (k+1)th iteration. Let be the dynamic threshold after the k-th iteration, η be the learning rate (ranging from 0.01 to 0.05), and L be the loss function. The weight coefficients of the loss function for the k-th iteration The partial derivatives, The loss function is the dynamic threshold for the k-th iteration. The partial derivatives, To optimize the correction coefficients for the dimensions, This is the threshold adjustment coefficient; The cloud-based collaborative module stores data including raw data collected by the multi-dimensional perception module, standardized data after data fusion processing, risk level assessment results, active intervention operation logs, collision accident data, battery life cycle data, and vehicle network collaborative data. The storage period is no less than 5 years, supporting data traceability, accident review, and multi-vehicle collaborative risk warning. At the same time, through the CTB architecture battery safety model, it updates battery protection parameters in real time and optimizes intervention strategies.

7. The system according to claim 1, characterized in that, The active intervention execution module works in conjunction with the battery safety system, employing a six-fold closed-loop protection logic—prevention, guidance, structure, isolation, diversion, and connection—designed for the CTB architecture. Specifically, this includes: using high-heat-resistant and stable cells at the cell level; employing mesh-like nanoporous heat-insulating materials and a high-temperature resistant upper shell (≥1400℃) in the battery compartment, adapting to the CTB three-layer structure; equipping a fully bonded liquid cooling system for rapid cooling, improving heat dissipation efficiency by ≥30%; collecting battery parameters more than 10 times per second through the battery management system, immediately initiating cooling and high-voltage cutoff operations in case of an anomaly; adding a battery compartment pressure release channel and a fireproof and explosion-proof layer to block heat conduction; and linking with the vehicle network and rescue system to simultaneously push rescue information in case of battery leakage.

8. The system according to claim 1, characterized in that, The multi-dimensional sensing module adopts an aerospace-grade redundant design, with dual backups for the in-cabin, external, battery compartment, and vehicle-to-everything (V2X) sensing units. When any sensing unit fails, the backup unit can take over within milliseconds, ensuring uninterrupted sensing data. The data fusion processing module and risk level assessment module both adopt a dual-chip redundant architecture with a total computing power of no less than 1000 TOPS, ensuring the real-time performance of data processing and risk assessment, with a processing latency of ≤50ms. Additionally, a CTB architecture adaptability detection unit is added to monitor the compatibility status between the sensing units and the CTB structure in real time, avoiding sensing errors caused by structural compatibility issues.

9. The system according to claim 1, characterized in that, It also includes an occupant-customized battery risk collaborative protection submodule, which, based on occupant height, weight, posture, and physiological parameters collected by the in-cabin sensing unit, and combined with collision damage-battery leakage coupling prediction results, adjusts seat belt pretension, airbag inflation volume, and in-cabin protection posture using the following formula: Where F represents the seatbelt pretension force, m represents the occupant's weight, h represents the occupant's height, s represents the occupant's sitting posture offset, and D represents the battery leakage risk value. , , , V is the preload correction factor; V is the airbag inflation volume. This represents the change in vehicle speed after the collision. , , This is the inflation volume correction factor; Adjust the angle for seat protection posture. , This is the attitude adjustment coefficient.

Images