Dynamic Evaluation and Control System for Thermal Comfort of Automotive Seats Based on Visual Perception

Abstract

This invention, applicable to the field of automotive thermal management technology, provides a dynamic assessment and control system for automotive seat thermal comfort based on visual perception. The system uses a multimodal perception module to collect cabin parameters and occupant-related parameters. A physiological-psychological coupling module combines facial temperature and negative facial expressions to determine thermal discomfort and adjust the PMV value. An intelligent decision-making module calculates the thermal comfort index based on an improved nine-variable PMV model, determining the optimal heating target temperature by converging the PMV value to the thermal neutral range of [-0.5, +0.5]. A parameter tuning module uses the first-order transfer function of the heating pad to tune the PID parameters, suppressing temperature fluctuations to ±0.3°C. The execution control module regulates the power of the graphene heating pad using a PWM signal. This system transcends the static parameter limitations of traditional models, achieving multimodal dynamic thermal comfort assessment and precise control. Experimental results have demonstrated energy savings of 17.98%, significantly improving occupant thermal comfort and system energy efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of automobile thermal management, and in particular relates to a dynamic evaluation and control system for automobile seat thermal comfort based on visual perception. Background Art

[0002] Optimizing thermal comfort and improving energy efficiency in intelligent automotive cockpits have become important research areas. As a key component of cabin thermal management, the performance of automotive seat heating systems directly impacts both passenger experience and overall vehicle energy consumption.

[0003] Existing automotive seat heating systems are primarily controlled using the PMV (Thermal Comfort Index) model. This model relies on a fixed clothing thermal resistance (e.g., a preset 1.0 clo in winter) and a preset metabolic rate. The seat temperature is collected via a contact temperature sensor (with an error of ±1.2°C), and the heating power is controlled using a switching threshold or PWM (Pulse Width Modulation).

[0004] However, traditional solutions are limited by static parameters. The PMV model cannot obtain individual parameters such as occupant age, gender, and weight in real time, resulting in a thermal comfort prediction error of more than ±15% in dynamic environments. The control accuracy is insufficient. The switch threshold control causes temperature fluctuations of up to ±4.5°C. PWM control has high-frequency switching losses and does not consider the second-order thermal inertia characteristics of the seat heating pad (time constant τ = 120-180s). The perception dimension is single, relying on contact temperature sensors (error ±1.2°C), and lacks multimodal fusion analysis of the temperature field in the occupant's facial nose, cheek, and chin areas and micro-expressions such as anger, disgust, and sadness. The energy efficiency ratio is low. The total energy consumption of the switch threshold control mode increases by 21.9% compared with PID control, and the energy-saving potential has not been fully tapped. Summary of the Invention

[0005] The purpose of the present invention is to provide a dynamic evaluation and control system for thermal comfort of automobile seats based on visual perception, aiming to solve the technical problems existing in the prior art identified in the background technology.

[0006] The present invention is implemented as follows: a dynamic evaluation and control system for thermal comfort of automobile seats based on visual perception includes the following components: a multimodal perception module, a physiological-psychological coupling feedback module, an intelligent decision-making module, a parameter setting module, and an execution control module.

[0007] The multimodal sensing module is used to collect the average air temperature and relative humidity in the vehicle cabin, the current temperature of the car seat, and to obtain weight data by three-category weight classification of the occupants. It also uses a camera equipped with an infrared imager to obtain height, age, gender, emotion, and skin temperature of specific facial areas.

[0008] The physiological-psychological coupling feedback module receives information from the multimodal perception module and, based on a physiological-psychological coupling model, determines thermal discomfort when facial temperature deviates from a threshold (the thermoneutral range of 27-31°C in winter) and negative expressions (such as anger, disgust, sadness, and fear) are detected. If thermal discomfort (overheating / overcooling) is present, the PMV value is adjusted.

[0009] The intelligent decision-making module is used to receive information from the multimodal perception module and calculate the occupant thermal comfort index (PMV) in real time based on the improved PMV thermal comfort evaluation model (including a nine-variable dynamic thermal balance equation). It also sets the thermal neutral range (PMV∈[-0.5,+0.5], corresponding to PPD≤10%) in combination with the PPD formula, dynamically adjusts the seat heating target temperature through a numerical iterative algorithm, so that the PMV value converges to the optimal range, and calculates and outputs the current optimal heating target temperature.

[0010] The parameter tuning module is used to tune the PID controller K p 、 K i 、 K d Three parameters, first obtain the heating pad's heating characteristics, calculate the corresponding first-order transfer function G(s) based on this characteristic, adjust the PID parameters based on the first-order transfer function G(s), suppress temperature fluctuations to ±0.3℃, and use the adjusted PID parameters for the PID controller. The PID controller receives the heating target temperature and the current seat temperature, and outputs the PWM duty cycle.

[0011] The execution control module is used to receive the duty cycle information of the intelligent decision module and output the PWM duty cycle to the PWM signal generator. The PWM signal is then transmitted to the MOS tube to control the voltage of the heating resistor wire and adjust the seat heating power.

[0012] The beneficial effects of the present invention are:

[0013] The present invention proposes a multimodal dynamic thermal comfort assessment model, which breaks through the static parameter limitations of the traditional PMV model and realizes real-time prediction of dynamic thermal comfort status. It can meet the needs of different individuals for thermal environment comfort, and at the same time promote the development of smart cockpit technology towards personalization and intelligence.

[0014] This invention addresses the energy consumption issues of traditional threshold-controlled heating modes for seats. It uses a PID control mode to significantly reduce heating energy consumption. Experimental verification shows a 17.98% energy saving, while also reducing temperature fluctuations from ±2.5°C to ±0.3°C, with an overshoot of less than 2%. This significantly reduces discomfort caused by temperature fluctuations while also contributing to energy conservation and emission reductions.

[0015] This invention uses a YOLO v5 clothing classification model to analyze the thermal resistance of occupants' clothing in real time, with a recognition error of less than 5%. DeepFace is used to identify facial features and emotions, acquiring relevant information in real time. Dlib's 68-point feature detection is used to segment facial regions for temperature measurement. The real-time detection of these parameters works together to enable the system to accurately and quickly respond to occupant needs, significantly improving thermal comfort and energy efficiency.

[0016] Experimental verification of the present invention shows that the total energy consumption in a 30-minute period at 9°C is 12.07W·h, which is 73.7% less energy than traditional air conditioners. The annual average energy savings are expected to exceed 200 million kWh (equivalent to reducing CO2 emissions by 160,000 tons). The graphene heating system is also more energy-efficient than traditional resistance wire solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a structural block diagram of the automotive seat thermal comfort dynamic evaluation and control system based on visual perception provided by the present invention;

[0018] Figure 2 This is a workflow diagram of the multimodal perception module provided by the present invention;

[0019] Figure 3 A flowchart of the physiological-psychological coupling feedback module provided by the present invention;

[0020] Figure 4 The workflow diagram of the intelligent decision-making module provided by the present invention;

[0021] Figure 5 The working flow diagram of the parameter setting module provided by the present invention;

[0022] Figure 6 This is a workflow diagram of the execution control module provided by the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0024] Figure 1 This is a structural block diagram of the automotive seat thermal comfort dynamic evaluation and control system based on visual perception provided by the present invention, such as Figure 1 As shown, the system includes:

[0025] A multimodal sensing module collects the average cabin air temperature and relative humidity, the current seat temperature, and categorizes occupants' weight into three weight categories. It also uses a camera equipped with an infrared imager to obtain height, age, gender, emotion, and skin temperature in specific facial areas.

[0026] The multimodal sensing module includes a temperature and humidity sensor, a patch temperature sensor, a seat pressure sensor, and a camera equipped with an infrared imager;

[0027] Among them, the temperature and humidity sensor is used to collect the average air temperature and relative humidity in the cabin, the patch temperature sensor collects the current temperature of the car seat, the seat pressure sensor is used to realize the three-level weight classification, and the camera equipped with an infrared imager is used to collect facial and clothing information. Combined with the YOLO v5 algorithm, it dynamically identifies the type of clothing and calculates the thermal resistance. Combined with the DeepFace algorithm, it detects the age, gender and negative expressions (anger, disgust, sadness, fear, etc.) of the occupants in real time. Combined with the dlib algorithm, it divides the nose tip, cheek and chin areas and calculates the average skin temperature.

[0028] The workflow of the multimodal perception module includes the following steps:

[0029] S101. Collect the average air temperature in the cabin t a and relative humidity RH ,The average air temperature and relative humidity in the cabin are collected through temperature and humidity sensors;

[0030] S102: Collect the current temperature of the car seat t seat ,The current seat temperature of the car seat is collected through the patch temperature sensor;

[0031] S103. Collect passenger weight W ,Through the seat pressure matrix (16×16) and SVM classification, three weight classifications (45kg / 65kg / 85kg) are achieved;

[0032] S104. Collect passenger height H , collect occupant height data through the pinhole model to the camera;

[0033] S105: Collect the average temperature of the designated area on the passenger's face t f ,The average temperature of the designated facial area is collected by infrared thermal imaging camera and the nose tip, cheek and chin areas divided by dlib algorithm;

[0034] S106, collect thermal resistance of passenger clothing I cl, the camera in the cockpit takes pictures of the passenger's clothing, and combined with the classification model trained by YOLOv5, it dynamically identifies the clothing type and calculates the thermal resistance of the clothing;

[0035] S107. Collect the occupant's age Y, gender G, and facial expression. Use the camera in the cockpit to capture the occupant's face. Combined with the DeepFace algorithm, detect the occupant's age, gender, and negative expressions (anger, disgust, sadness, fear, etc.) in real time.

[0036] The physiological-psychological coupling feedback module receives information from the multimodal perception module and, based on the physiological-psychological coupling model, determines thermal discomfort when facial temperature deviates from the threshold (the thermoneutral range of 27-31°C in winter) and negative expressions (anger, disgust, sadness, fear, etc.) are detected. If thermal discomfort (overheating / overcooling) is present, the PMV value is adjusted.

[0037] The physiological-psychological coupled feedback module includes a camera equipped with an infrared imager. It combines the average skin temperature of specific facial areas and the occupant's facial expressions to determine thermal discomfort. If thermal discomfort (overheating / overcooling) is present, the PMV value is adjusted.

[0038] The workflow of the physiological-psychological coupling feedback module includes the following steps:

[0039] S201. Obtain the parameters required by the model, receive the dlib algorithm partition in the multimodal perception module, the average temperature of the nose tip, cheek, and chin areas detected by the infrared imager, and the occupant's facial expression (anger, disgust, sadness, fear, etc.) detected in real time by the DeepFace algorithm;

[0040] S202: Determine the occupant's current thermal comfort. Based on the physiological-psychological coupling model, when the facial temperature deviates from the threshold (the thermal neutral range of 27-31°C in winter) and a negative expression (anger, disgust, sadness, fear, etc.) is detected, a thermal discomfort (overheating / undercooling) state is determined.

[0041] S203. Adjust the PMV value according to thermal comfort. If there is overheating discomfort, adjust the PMV value by 1; if there is overcooling discomfort, adjust the PMV value by 1; and input the adjusted value into the intelligent decision module.

[0042] The intelligent decision-making module receives information from the multimodal perception module and calculates the occupant thermal comfort index (PMV) in real time based on an improved PMV thermal comfort evaluation model (including a nine-variable dynamic thermal balance equation). It also uses the PPD formula to set the thermal neutral range (PMV∈[-0.5,+0.5], corresponding to PPD≤10%). It dynamically adjusts the seat heating target temperature through a numerical iterative algorithm to converge the PMV value to the optimal range. It then calculates and outputs the current optimal heating target temperature.

[0043] The intelligent decision-making module includes a vehicle ECU, which is used to process, judge and make decisions on various information, calculate the optimal seat heating target temperature through an improved PMV model, and transmit it to the execution control module.

[0044] The workflow of the intelligent decision-making module includes the following steps:

[0045] S301, calculate the current PMV value, receive the real-time detection information of the multimodal perception module as a parameter, and calculate the PMV value by improving the PMV model formula:

[0046] ;

[0047] ;

[0048] in, M is the metabolic rate of the human body, Q cond Conductive heat dissipation for the seats. Q conv For convection cooling, Q rad For radiation heat dissipation, Q resp,let To dissipate the latent heat of breathing, Q resp,sens To dissipate sensible heat from breathing, Q evap To dissipate heat through sweating.

[0049] The formula for calculating the metabolic rate of different genders is:

[0050] ;

[0051] in, W is the occupant weight, H is the occupant's height, Y The age of the occupants.

[0052] Substituting the human metabolic rate formula and calculating the remaining terms, we can finally obtain the nine-variable PMV dynamic heat balance equation:

[0053] ;

[0054] in, t a is the air temperature, t seat is the current temperature of the seat, I cl is the thermal resistance of the occupant's clothing, f cl is the clothing area coefficient, G is the occupant's gender, RHis the relative humidity.

[0055] S302: Calculate the seat heating target temperature. Using the PMV-PDD model, determine the optimal PMV range (PMV∈[-0.5, +0.5], corresponding to PPD≤10%). Dynamically adjust the seat heating target temperature using a numerical iterative algorithm to converge the PMV value to the optimal range. Calculate the current optimal seat heating target temperature.

[0056] Table 1 PMV values corresponding to thermal sensation

[0057] ;

[0058] The table above shows the thermal sensation corresponding to the PMV value.

[0059] S303: Output the seat heating target temperature to the execution control module.

[0060] Parameter tuning module, used to tune the PID controller K p 、K i 、K d The three parameters are first obtained. The heating pad's heating characteristics are then used to calculate the corresponding first-order transfer function G(s). PID parameters are then tuned based on the first-order transfer function G(s) to suppress temperature fluctuations to ±0.3°C. The tuned PID parameters are then used in a PID controller. The PID controller receives the heating target temperature and the current seat temperature and outputs a PWM duty cycle.

[0061] The workflow of the parameter setting module includes the following steps:

[0062] S401. Determine the transfer function G(s). According to the heating characteristics of the heating pad, calculate the corresponding first-order transfer function G(s). The transfer function can reflect the power heating characteristics and the heating time characteristics.

[0063] S402, adjust the PID parameters, through the PID controller algorithm in Simulink, combined with the corresponding first-order transfer function and the built-in automatic parameter adjustment algorithm, suppress the temperature fluctuation to ±0.3℃, and at the same time require a faster heating time, obtain the PID controller K p 、K i 、K d Three parameters.

[0064] S403: Output PID parameters.

[0065] The parameter tuning module includes the PID controller algorithm in Simulink, which obtains the PID controller's K p 、K i 、K d Three parameters are input into the execution control module.

[0066] The execution control module is used to receive the duty cycle information from the intelligent decision-making module and output the PWM duty cycle to the PWM signal generator. The PWM signal is then transmitted to the MOS tube to control the voltage of the heating resistor wire and adjust the seat heating power.

[0067] The workflow of the execution control module includes the following steps:

[0068] S501, receive relevant parameters, receive the current seat temperature in the multimodal perception module, receive the seat heating target temperature in the intelligent decision module, receive the parameter setting module K p 、 K i 、 K d Three parameters;

[0069] S502, calculate the PWM duty cycle, and K p 、 K i 、 K d In the three-parameter tuning PID controller, the seat heating target temperature and the current seat temperature are input into the PID controller to calculate the corresponding PWM duty cycle;

[0070] S503, outputting the PWM duty cycle to the PWM signal generator;

[0071] S504: Adjust the seat heating power to reach the seat heating target temperature, and adjust the seat heating power by outputting a PWM signal to the MOS tube to control the voltage of the heating resistor.

[0072] The execution control module includes a heating element, which is arranged inside the driver's seat to heat the seat.

[0073] Wherein, the heating element may be a graphene heating pad.

[0074] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0075] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A dynamic evaluation and control system for automotive seat thermal comfort based on visual perception, characterized by: The system comprises: A multimodal sensing module collects information about the average cabin air temperature and relative humidity, current seat temperature, and individual occupant parameters and facial features. a physiological-psychological coupling feedback module, configured to determine the thermal discomfort state of the occupant based on the facial feature parameters and adjust the value of the thermal comfort index PMV; An intelligent decision-making module calculates the passenger thermal comfort index (PMV) in real time based on an improved PMV thermal comfort evaluation model, sets the thermal neutral range using the PPD formula, and dynamically adjusts the seat heating target temperature through a numerical iterative algorithm. The parameter tuning module is used to determine the first-order transfer function based on the heating pad's temperature rise characteristics, tune the parameters of the PID controller, suppress temperature fluctuations to ±0.3°C, and output the PWM duty cycle through the PID controller; an execution control module, configured to receive the seat heating target temperature, the current seat temperature, and parameters of a PID controller, calculate a PWM duty cycle through the PID controller, and adjust the seat heating power; The multimodal perception module includes: Temperature and humidity sensor, used to collect the average air temperature in the cabin t a and relative humidity RH ; SMD temperature sensor, used to collect the current temperature of the car seat t seat ; Seat pressure sensor and SVM classification algorithm are used to classify occupant weight into three categories; A camera equipped with an infrared imager, combined with the YOLO v5 algorithm, dynamically identifies clothing types and calculates clothing thermal resistance. I cl , combined with the DeepFace algorithm to detect the occupant's age Y, gender G and negative expressions, and combined with the dlib algorithm to divide the nose tip, cheek and chin areas and calculate the average skin temperature t f ; The physiological-psychological coupling feedback module determines that when the facial temperature deviates from the winter thermal neutral range and a negative expression is detected, it is determined to be a thermal discomfort state; if it is too hot, the thermal comfort index PMV value is increased by 1, and if it is too cold, the thermal comfort index PMV value is reduced by 1.

2. The system according to claim 1, wherein: In the intelligent decision-making module, the human metabolic rate M The calculation formula is: ； in, M f Represents the metabolic rate of female body, M m It represents the metabolic rate of male body; The metabolic rate of the human body M Substitute the nine-variable dynamic heat balance equation to calculate the PMV value: ； in, t a is the air temperature, t seat is the current temperature of the seat, I cl is the thermal resistance of the occupant's clothing, f cl is the clothing area coefficient, G is the occupant's gender, RH is the relative humidity.

3. The system according to claim 1, wherein: The parameter tuning module uses the PID controller algorithm and built-in automatic parameter tuning algorithm in Simulink to tune the PID parameters based on the first-order transfer function G(s), controlling the temperature fluctuation to ≤±0.3°C and the overshoot to <2%.

4. The system according to claim 1, wherein: The heating element of the execution control module is a graphene heating pad, which receives a PWM signal through a MOS tube to adjust the voltage of the heating resistor wire and dynamically adjust the seat heating power.

5. The system according to claim 1, wherein: The workflow of the multimodal perception module includes: Collect air temperature t a , relative humidity RH , current seat temperature t seat , occupant weight W , occupant height H , the average temperature of the designated area of the occupant's face t f , thermal resistance of passenger clothing I cl , age of crew members Y ,gender G and facial expression data; The YOLO v5 algorithm is used to identify clothing types, the DeepFace algorithm is used to detect age, gender, and expression, and the dlib algorithm is used to divide the facial temperature measurement area.

6. The system according to claim 1, wherein: The intelligent decision-making module uses a numerical iterative algorithm to converge the PMV value to the range of [-0.5, +0.5], corresponding to PPD ≤ 10%, and outputs the optimal heating target temperature to the execution control module.

Images