Warming blanket temperature closed-loop control method and system based on human body position recognition feedback

Abstract

The invention relates to the technical field of non-electrical variable control or regulation systems, and discloses a temperature rising blanket temperature closed-loop control method and system based on human body position recognition feedback, and the method comprises the steps: calculating a form compactness coefficient representing a heat dissipation boundary based on pressure data, and calling a mapping function to convert the form compactness coefficient into an equivalent thermal impedance estimated value; introducing the estimated value and dynamically constraining the forward channel gain of the temperature closed-loop controller according to a negative correlation strategy; according to the method, a real-time mapping mechanism of the contact topological characteristics and the control gain is constructed, so that the problem of model mismatch caused by body position change of the controlled object is solved; when the high-impedance contact state is detected, energy input is automatically reduced, the local heat accumulation effect is restrained, and the convergence and safety of the thermodynamic control process under the variable boundary condition are ensured.

Description

Technical Field

[0001] This invention relates to a closed-loop temperature control method and system for a heating blanket based on human body position recognition feedback, belonging to the technical field of control or regulation systems for non-electrical variables. Background Technology

[0002] Currently, for precision temperature control applications of flexible loads, conventional control architectures employ closed-loop control strategies driven by the deviation between setpoints and feedback values. These systems presuppose that the controlled object possesses time-invariant thermodynamic characteristics. The controller gain is tuned based on physical parameters such as heat capacity, conductivity, and heat dissipation surface area under standard operating conditions to maintain the system's steady-state energy balance. This design constitutes the mainstream foundation of current flexible temperature control systems. In actual operating conditions, changes in the position of the controlled object cause a step change in the contact topology between the flexible heating interface and the controlled object. Fluctuations in the effective heat transfer area directly alter the system's heat dissipation boundary conditions. Non-uniform distribution of contact pressure causes drastic fluctuations in interface contact thermal resistance. If the control algorithm uses a fixed control law tuned to the standard state, when the controlled object transitions to a curled, low-heat dissipation mode, the original forward channel gain leads to excessive energy input, causing temperature response overshoot. Conversely, when the contact changes from tight to loose, the system suffers from response hysteresis due to insufficient gain.

[0003] To address safety concerns, existing technologies attempt to mitigate risks through hardware redundancy or setting absolute thresholds, but these fail to address the fundamental issue of thermodynamic model mismatch. A utility model patent with authorization number CN210301383U discloses a temperature detection device for a heating blanket. This device uses a dual-controller architecture to monitor the inlet and outlet temperatures of the air duct. Heating is shut off when the detected value exceeds a preset fixed threshold. While this strategy utilizes hardware redundancy to reduce the risk of runaway due to single-point failures, the core control logic remains at a passive defense level. The system's understanding of the controlled object is essentially a black box, ignoring the drift in heat dissipation boundary conditions caused by patient positioning, changes in coverings, or limb folding. Within the normal operating range, where the extreme threshold is not triggered, the control algorithm still uses a fixed power output logic. This leads to increased load thermal resistance. Under non-standard operating conditions, the imbalance between energy supply and actual dissipation causes local heat accumulation, making it difficult to simultaneously meet the clinical requirements for temperature control accuracy and dynamic safety. Simply increasing the density of temperature sensors or increasing the sampling frequency cannot solve the above problems. Temperature, as a state variable of a thermodynamic system, has a large inertial characteristic and transmission delay relative to energy input. Relying solely on temperature as a single variable feedback means that the control system must wait for heat accumulation to occur and manifest as a temperature rise before initiating the adjustment mechanism. The inherent time-domain lag makes the system unable to suppress dynamic deviations caused by sudden changes in load characteristics. In addition, the folding of fabric creates localized insulating areas, which have blind spots in conventional temperature monitoring, causing actuators to continuously output power under erroneous state estimations, leading to the risk of localized overheating.

[0004] Therefore, how to establish a mechanism for real-time sensing of changes in the contact topology of the controlled object, and thereby reconstruct control parameters online to achieve an adaptive adjustment control mechanism for time-varying thermal loads, is the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A closed-loop temperature control method for a heating blanket based on human body position recognition feedback, comprising the following steps: The contact feature extraction step involves collecting real-time pressure detection data from the pressure sensing array in the functional layer of the heating blanket and discretizing it into a two-dimensional pressure distribution matrix. Based on the connected component analysis of non-zero elements, the morphological compactness coefficient characterizing the heat dissipation boundary conditions of the controlled object is calculated. The morphological compactness coefficient is used to quantify the effective contact state between the controlled object and the heating blanket. The thermal impedance parameter estimation step calls a preset thermal resistance mapping function to convert the shape compactness coefficient into the equivalent thermal impedance estimate of the controlled object at the current moment. The thermal resistance mapping function is configured to establish a positive correlation between the shape compactness coefficient and the equivalent thermal impedance estimate to characterize the attenuation characteristics of the system's heat dissipation capacity in the curled or heavily wrapped state of the controlled object. In the gain adaptive adjustment step, the estimated equivalent thermal impedance is introduced as a gain adjustment factor into the temperature closed-loop controller. According to the negative correlation gain adjustment strategy, dynamic constraints are applied to the forward channel gain of the temperature closed-loop controller. The logic of the dynamic constraint is: when the estimated equivalent thermal impedance is detected to increase, the forward channel gain is reduced to suppress the heat accumulation effect caused by the increase in local thermal resistance. The power output and response verification steps utilize a temperature closed-loop controller with forward channel gain to process the temperature deviation signal to generate a target heating power, and use this target heating power to drive the heating execution unit to achieve adaptive temperature control of the time-varying heat load object.

[0006] Preferably, the contact feature extraction step further includes a folding anomaly identification sub-step: performing gradient calculation on the two-dimensional pressure distribution matrix to generate a pressure gradient field; scanning the pressure gradient field and identifying linear continuous regions where the gradient value exceeds a preset folding threshold, and marking the linear continuous region as a folded adiabatic zone; correspondingly, the power output and response verification step further includes regional shielding control logic: constructing a logic control mask that maps to the spatial distribution of the heating execution unit, setting the mask position corresponding to the coordinates of the folded adiabatic zone to a blocking state; using the logic control mask to perform a spatial AND operation on the target heating power, blocking the power output of the heating execution unit corresponding to the folded adiabatic zone, so as to eliminate the overheating risk caused by local adiabatic while maintaining closed-loop control of the non-folded region.

[0007] Preferably, the gain adaptive adjustment step further includes internal heat disturbance feedforward compensation logic: separating high-frequency components located within a preset flutter frequency band from real-time pressure detection data, calculating the micro-motion energy index characterizing the flutter intensity of the controlled object; mapping the micro-motion energy index to an estimated internal heat generation value using a preset flutter heat generation model; and performing the following feedforward compensation calculation when generating the target heating power: ,in, The target heating power for the final output, The base power is calculated based on the temperature deviation signal. The preset feedforward compensation coefficient, The value is the estimated value of endogenous heat production. Through feedforward compensation calculation, before the temperature sensor detects the lag response of temperature rise caused by vibration, the input of external heat source is actively reduced to offset the superposition effect of endogenous metabolic heat of the controlled object.

[0008] Preferably, the thermal resistance mapping function in the thermal resistance parameter estimation step is constructed based on a preset nonlinear heat transfer model. This function defines the shape compactness coefficient as the ratio of the number of activated nodes in the two-dimensional pressure distribution matrix to the area of ​​the smallest bounding rectangle of the connected domain formed by the activated nodes. The positive correlation is specifically configured such that as the shape compactness coefficient increases, the output of the equivalent thermal resistance estimate increases exponentially to simulate the nonlinear shrinkage process of the effective heat dissipation surface area when the controlled object changes from a flat state to a curled state. The method also includes an online thermal inertia calibration step, used for... Correcting the estimated equivalent thermal impedance: When the temperature closed-loop controller is running in steady state, a step power excitation signal with a preset amplitude and duration is applied to the heating execution unit; the transient temperature rise response data of the temperature sensor to the step power excitation signal is collected, and the measured temperature rise slope is calculated; the ratio of the theoretical standard temperature rise slope based on the current compactness coefficient to the measured temperature rise slope is calculated to generate an individualized thermal capacity calibration coefficient; the individualized thermal capacity calibration coefficient is used to perform a secondary weighted correction on the forward channel gain to compensate for the model parameter drift caused by individual differences of the controlled object or the thermal resistance of clothing.

[0009] Preferably, the gain adaptive adjustment step further includes asymmetric filtering: calculating the time-domain fluctuation variance of the two-dimensional pressure distribution matrix within a preset time window; if the time-domain fluctuation variance exceeds a preset stability threshold, it is determined that the controlled object is in a transient disturbance state, and the current forward channel gain and the integral term value of the temperature closed-loop controller remain unchanged; if the time-domain fluctuation variance is lower than the stability threshold, it is allowed to update the forward channel gain, and the currently executed gain value is smoothly transitioned to the target gain value using a first-order hysteresis filtering algorithm. The first-order hysteresis filtering algorithm is configured to have an asymmetric time constant, so that the response speed of gain decrease is faster than the response speed of gain increase.

[0010] Preferably, in the folding anomaly identification sub-step, the gradient calculation uses the Laplacian operator or the Sobel operator to perform discrete convolution on the two-dimensional pressure distribution matrix; the preset folding threshold is set to be higher than the statistical upper limit of the maximum pressure gradient generated by normal limb compression of the controlled object, so as to distinguish the different pressure field morphological characteristics generated by contact with flexible organisms and folding of rigid fabrics.

[0011] Preferably, the temperature closed-loop controller is a PID controller, and the forward channel gain specifically acts on the proportional coefficient of the PID controller; the negative correlation gain adjustment strategy is specifically implemented as follows: a reference proportional coefficient corresponding to the standard tiling condition is set, and the proportional coefficient is set in real time as the product of the reference proportional coefficient and the attenuation factor. The attenuation factor is negatively correlated with the estimated value of equivalent thermal impedance and is limited to a value range of zero to one.

[0012] Preferably, the micro-motion energy index is obtained by performing a short-time Fourier transform on the high-frequency components and calculating the power spectral density integral within a specified frequency band; the preset tremor frequency band is limited to 5 Hz to 20 Hz to match the characteristic frequency range of involuntary contraction of human skeletal muscles and to filter out low-frequency interference signals caused by turning over or breathing.

[0013] Preferably, the method operates in a temperature control system containing an embedded microcontroller unit, and the data source for the two-dimensional pressure distribution matrix is ​​a resistive pressure sensor array laid in the functional layer of the heating blanket; the power output and response verification steps also include safety protection logic: real-time monitoring of the temperature response slope after the heating execution unit is powered on, if the temperature response slope exceeds the theoretical range predicted based on the current equivalent thermal impedance estimation value, then the contact data is determined to be abnormal and the output of the temperature closed-loop controller is locked to the safety basic threshold.

[0014] A closed-loop temperature control system for a heated blanket based on human body position recognition feedback includes: The contact feature extraction module is configured to collect real-time pressure detection data of the pressure sensing array in the functional layer of the heating blanket and discretize it into a two-dimensional pressure distribution matrix. Based on the connected component analysis of non-zero elements, the module calculates the morphological compactness coefficient that characterizes the heat dissipation boundary conditions of the controlled object. The morphological compactness coefficient is used to quantify the effective contact state between the controlled object and the heating blanket. The thermal impedance parameter estimation module is configured to call a preset thermal resistance mapping function to convert the shape compactness coefficient into the equivalent thermal impedance estimation value of the controlled object at the current moment. The thermal resistance mapping function is configured to establish a positive correlation between the shape compactness coefficient and the equivalent thermal impedance estimation value to characterize the attenuation characteristics of the system's heat dissipation capacity in the curled or heavily wrapped state of the controlled object. The gain adaptive adjustment module is configured to introduce the estimated equivalent thermal impedance as a gain adjustment factor into the temperature closed-loop controller, and apply dynamic constraints to the forward channel gain of the temperature closed-loop controller according to the negative correlation gain adjustment strategy. The logic of the dynamic constraint is: when the estimated equivalent thermal impedance is detected to increase, the forward channel gain is reduced to suppress the heat accumulation effect caused by the increase in local thermal resistance. The power output and response verification module is configured to use the temperature closed-loop controller including the forward channel gain to process the temperature deviation signal to generate the target heating power, and use the target heating power to drive the heating execution unit to realize the temperature adaptive control of the time-varying heat load object.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the human body position recognition feedback heating blanket, a contact topology feature and thermal impedance model parameter mapping mechanism is constructed to solve the model mismatch problem of conventional fixed parameter controllers facing objects with strong time-varying thermal loads. The real-time spatial contact state between the controlled object and the heating interface is quantified into a heat exchange scale coverage index and a heat dissipation boundary constraint compactness. Based on this, the forward channel gain or integral time constant of the closed-loop controller is dynamically scheduled. Based on the physical field distribution feedback variable gain control strategy, the control system automatically maintains the control law matching the current load characteristics when the controlled object flips or curls, causing changes in thermodynamic structure. This suppresses temperature overshoot or response hysteresis caused by sudden changes in load thermal resistance, and ensures the convergence and steady-state accuracy of the thermodynamic control process.

[0016] 2. By utilizing the high-frequency micro-motion component of the pressure sensor array to construct a feedforward suppression channel for endogenous thermal disturbances, the traditional temperature control system overcomes the limitation of response lag caused by relying solely on temperature negative feedback. The spectral characteristics representing the vibration intensity of the controlled object are separated from the mechanical domain signal and transformed into an estimated value of endogenous heat generation in the thermodynamic domain. Negative power compensation is directly superimposed at the controller output. The cross-physical domain interference observation and decoupling mechanism enables the control system to synchronously cancel the power of sudden thermal disturbance sources inside the controlled object before the temperature sensor detects the temperature rise change. This eliminates the risk of system overheating caused by the superposition of external heating and internal heat generation at the energy input source and improves the system's ability to suppress nonlinear time-varying disturbances.

[0017] 3. Based on the spatial gradient characteristics of the physical field distribution matrix, the system achieves logical identification and active constraint of local adiabatic singularities, eliminating semantic ambiguity in conventional pressure monitoring under folding conditions. By calculating the spatial gradient field of the two-dimensional distribution matrix, it distinguishes between smooth conventional load contact and mechanical folding regions with high gradient characteristics, generating a logic control mask corresponding to the heating array. Using the mask, it bypasses the conventional PID control loop operation logic and directly implements power blocking for specific heating units in the adiabatic trap state. Based on the local topological shielding mechanism of structural features, it prevents local heat accumulation from damaging system safety, maintains normal closed-loop control in other non-singular regions, and ensures the continuity of operation and safety of the control system under complex physical boundary conditions. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the flow architecture of the human body position recognition feedback-based closed-loop temperature control method for heating blankets according to the present invention. Figure 2 This is a comparison chart of temperature overshoot suppression under sudden change in heat load conditions according to the present invention; Figure 3 This is a schematic diagram of the system architecture for integrating contact sensing and gain scheduling in this invention. Detailed Implementation

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] This invention proposes a closed-loop temperature control method for a heated blanket based on human body position recognition feedback. The method comprises a closed-loop temperature control system consisting of a resistive pressure sensor array, an embedded microcontroller unit, and segmented heating execution units. Its core logic lies in constructing an observation and feedback loop for non-electrical variables. Specifically, it utilizes changes in the physical contact topology to invert the thermodynamic parameter drift of the controlled object in real time, thereby dynamically adjusting the gain of the control law for the temperature control loop. The system's main control flow includes four core processing stages: contact feature extraction, thermal impedance parameter estimation, adaptive gain adjustment, and power output and response verification. In the contact feature extraction stage, the core engineering problem addressed by the system is how to transform the irregular spatial morphology of the controlled object into computer-processable digital features to address the uncertainty of heat dissipation boundary conditions caused by changes in body position. The microcontroller unit operates at a preset sampling frequency, for example... Hertz, reads the voltage signal from the two-dimensional pressure sensor array laid in the functional layer of the heating blanket, the microcontroller quantizes the acquired analog signal into a digital signal via an analog-to-digital converter, and maps it to... Two-dimensional pressure distribution matrix .

[0021] In this matrix, each element The numerical value represents the pressure amplitude at the corresponding physical coordinate point. The microcontroller performs a connected component analysis algorithm on this matrix, filtering out background noise using a preset pressure noise threshold and extracting all non-zero active nodes. It then uses a scanline algorithm or a depth-first search algorithm to identify the largest connected region belonging to the same controlled object. Based on this connected region, the microcontroller calculates the morphological compactness coefficient representing the heat dissipation boundary constraint. This coefficient The computational logic is configured as follows: count the total number of activated nodes within a connected region. And calculate the area of ​​the minimum bounding rectangle of the connected region in the two-dimensional plane. Form compactness coefficient equal and The ratio of the controlled object to the heated blanket, a dimensionless number, quantifies the density of the contact surface between the controlled object and the heated blanket. When the controlled object changes from a flat state to a rolled-up state, the effective heat dissipation area decreases while the contact density increases, leading to... The numerical values ​​increase, providing a quantitative basis for subsequent thermal resistance estimation. To address the differences in pressure sensor array hardware specifications, a standardized mapping procedure based on physical space parameters is executed during the contact feature extraction stage. The microcontroller's non-volatile memory pre-sets the current physical row spacing of the pressure sensor array. Column spacing Processing two-dimensional pressure distribution matrix At that time, the processor discretizes the index coordinates of the matrix elements. Convert to physical plane coordinates ,in and Form compactness coefficient The computational logic operates based on the physical plane coordinate system, and the number of nodes to be activated... Corrected to the physical coverage area of ​​the activated region. area of ​​minimum circumscribed rectangle The calculation is based on the geometric area per unit of physical length, and the physical mapping procedure makes the output shape compactness coefficient... To decouple the physical quantity from the sensor pixel density, and to ensure that the reference parameters of the subsequent thermal resistance mapping function do not need to be reconstructed after replacing the sensing components with different resolutions.

[0022] In the thermal impedance parameter estimation stage, the system aims to solve the model mismatch problem of the controlled object caused by changes in contact state. Since the curled-up or heavily wrapped state of the human body will cause nonlinear decay of local heat dissipation capacity, directly using a fixed parameter model will easily lead to heat accumulation. Therefore, the microcontroller calls the thermal resistance mapping function pre-stored in memory. The morphological compactness coefficient will be calculated in real time. Converted into the estimated equivalent thermal impedance of the controlled object at the current moment. This thermal resistance mapping function is constructed based on a nonlinear heat transfer model, and a system is established. and The positive correlation between them is, in a specific embodiment, configured as an exponential function: ,in, The reference thermal resistance is under standard flat tiling conditions. This is the shape compactness coefficient under standard flat tiling conditions. Using a pre-calibrated thermistor coefficient, this calculation allows the system to track the increasing thermal resistance of the controlled object due to its curled-up position in real time, outputting a physical quantity characterizing the current heat dissipation difficulty. During the gain adaptive adjustment phase, the system dynamically adjusts the controller's response characteristics based on changes in thermal resistance to suppress temperature overshoot under varying boundary conditions. The microcontroller operates digitally... Temperature closed-loop controller, and the equivalent thermal impedance estimate obtained from the above calculation. Introduced as a gain adjustment factor into the control loop, following a negative correlation gain adjustment strategy, the microcontroller unit... The forward channel gain of the controller, especially the proportional gain. Apply dynamic constraints, specifically the constraint logic is as follows: set a baseline scale coefficient corresponding to the standard tiling condition. Real-time scaling factor Set as The product of the attenuation factor and the attenuation factor is constructed as follows: ,in Using the reference thermal resistance value, when the controlled object is detected to be in a high-resistance curled-up state, i.e. When the value increases, the system automatically decreases the scaling factor. This adjustment mechanism follows the principle of reverse heat load compensation, and actively reduces the sensitivity and output of the controller under operating conditions with a high risk of heat accumulation.

[0023] To further address the risk of localized overheating caused by accidental fabric folding commonly encountered in practical use, the contact feature extraction stage integrates folding anomaly recognition logic. The system utilizes discrete convolution kernels such as the Laplacian operator or the Sobel operator to analyze the two-dimensional pressure distribution matrix. Perform spatial gradient calculations to generate a pressure gradient field. The microcontroller scans the gradient field and identifies gradient values ​​that exceed a preset folding threshold. The region, this threshold A statistical upper limit is set above the maximum pressure gradient generated by normal human limb compression, for example, set to the normal contact gradient. This allows for the effective differentiation of different pressure field patterns generated by contact with flexible organisms and folding of rigid fabrics. If a high gradient region with a linear and continuous distribution is identified, the system marks it as a folded insulation zone. During the power output phase, regional shielding control logic is executed, and the system constructs a logic control mask that maps one-to-one with the spatial distribution of the heating execution units. For coordinates marked as folded adiabatic zones, the corresponding mask positions are set to a blocked state (logical). The remaining areas are set to the ON state (logical). When the microcontroller generates the final drive signal, it will... The controller performs a spatial AND operation between the target heating power calculated by the controller and the mask. This mechanism can block the power output of the heating unit corresponding to the folded area, maintaining normal closed-loop control in the non-folded area while eliminating the risk of burns caused by localized insulation. A preset folding threshold is also included. The specific quantitative settings adopt a boundary definition procedure based on large-sample statistical distribution. During the parameter solidification stage before leaving the factory, a pressure distribution sample library containing no less than 1000 sets of controlled objects in normal use states (flat, lying on their side, and curled up, without folding) is constructed. The processor performs Laplacian or Sobel operator convolution operation on each frame of data in the sample library, extracts the maximum spatial gradient of each frame image to generate a gradient peak statistical distribution sequence, and selects the value corresponding to the 99.9th percentile of the statistical distribution sequence as the preset folding threshold. The parameters are set based on the fact that only 0.1% of normal human contact generates a signal close to the threshold.

[0024] In response to potential endogenous thermal disturbances such as postoperative tremors or chills in the controlled object, this invention integrates feedforward compensation logic during the gain adaptive adjustment stage. The microcontroller performs frequency domain analysis on the real-time data stream of the pressure sensor array and separates frequencies located in the preset tremor frequency band using a bandpass filter, for example... Hertz The system performs a short-time Fourier transform on the high-frequency components within the Hertz range and calculates the power spectral density integral within a specified frequency band to obtain a microkinetic energy index characterizing the flutter intensity of the controlled object. Using a pre-set tremor heat production model, the system maps this index to an estimated value of endogenous heat production. When generating the target heating power, the system performs feedforward compensation calculations: ,in For the final output power, The base power is calculated based on temperature deviation. As the feedforward compensation coefficient, this calculation allows the control system to proactively reduce external heat source input based on mechanical vibration signals before the temperature sensor detects the lag response caused by vibration. This counteracts the superposition effect of endogenous metabolic heat in the controlled object, achieving proactive decoupling from complex thermal disturbances. To address the dispersion of model parameters due to individual differences and clothing thickness, this method also includes an online thermal inertia calibration process. When the temperature closed-loop controller is in steady-state operation, the microcontroller applies a preset amplitude, such as the rated power, to the heating actuator. and duration such as The system simultaneously acquires the transient temperature rise response data of the temperature sensor to the step power excitation signal, and calculates the measured temperature rise slope. The microcontroller calculates the theoretical standard temperature rise slope based on the current form factor compactness coefficient. And calculate the ratio. This coefficient serves as an individualized thermal capacity calibration factor. The forward channel gain is subject to a secondary weighted correction. If the measured temperature rise slope is greater than the theoretical value, it indicates that the actual heat capacity of the controlled object is small, and the system correspondingly reduces the gain; conversely, the gain is increased. This mechanism enhances the system's adaptive correction capability to different user body types and thermal resistance characteristics of the covering material. Regarding the stability of the control law update, the system employs an asymmetric filtering strategy to filter transient disturbances. The microcontroller calculates the two-dimensional pressure distribution matrix within a preset time window. The time-domain fluctuation variance within seconds is considered. If this variance exceeds a preset stability threshold, the controlled object is determined to be in a transient process such as overturning or agitation. In this case, the current forward channel gain and integral term values ​​are kept unchanged to prevent drastic parameter jumps. If the variance is below the threshold, gain updates are allowed, and a first-order hysteresis filtering algorithm is used to smoothly transition the current gain to the target gain. This filtering algorithm is configured with an asymmetric time constant: the time constant for gain decrease is set relatively small, for example... Seconds, to achieve a fast power reduction response; the gain rise time constant is set relatively large, such as... Seconds to achieve slow power recovery.

[0025] Example 1: In a typical postoperative anesthesia recovery monitoring scenario, the controlled subject, i.e., the patient, is in a state of confusion accompanied by involuntary thermoregulation. Initially, the patient is laid flat on the surface of the warming blanket in a standard supine position, and the system is in steady-state heating mode. At this time, the morphological compactness coefficient is [not specified]. Maintaining at a lower baseline level corresponds to a lower estimated equivalent thermal resistance. , Controller proportional coefficient Maintaining the standard gain value, the controlled object changes position due to stress response, changing from supine to lateral and with limb curling, while simultaneously inducing persistent muscle shivering. This physical process causes a step-like abrupt change in the contact topology between the controlled object and the heating blanket, reducing the effective heat exchange area. At the same time, the closed cavity formed by limb curling leads to a sharp increase in local contact thermal resistance. If the original control law is used, the system will be unable to detect the change in heat dissipation boundary conditions and will continue to output high power until the temperature sensor detects a temperature rise, at which point negative feedback adjustment will be initiated. This will inevitably lead to temperature overshoot. In the operation of this embodiment, the microcontroller unit captures the two-dimensional pressure distribution matrix in real time through the pressure sensor array. The morphological changes and connected component analysis results show that, although the number of activated nodes... The variation is small, but the area of ​​the minimum bounding rectangle of the connected region is... The morphological compactness coefficient is significantly reduced due to limb contraction. The system exhibits a rapid upward trend, based on a preset thermal resistance mapping function. The estimated value of the increased equivalent thermal resistance of the output Based on this estimate, the adaptive gain adjustment logic is triggered, and the gain is reduced in real time according to the negative correlation strategy. Controller proportional coefficient This correction of control parameters occurs before the heat accumulation effect manifests as a temperature rise, achieving model mismatch compensation for the variable structural thermal load. Simultaneously, the muscle chills of the controlled object manifest as a high-frequency micro-motion signal superimposed on the fundamental wave of the contact pressure in the pressure sensing array. The system performs parallel frequency domain analysis to separate... Hertz The signal components within the Hertz frequency band were analyzed, and the micro-motion energy index was calculated by integration. Using the tremor heat production model, this index was converted into an estimate of endogenous heat production. In generating the final target heating power At that time, the system starts from the basics Subtract from output power The proportional compensation amount ensures that this feedforward control action is synchronized with the occurrence of endogenous thermal disturbances on the time axis, preventing the superposition of external heating power and internal metabolic heat generation. Through the synergistic effect of the above-mentioned contact topology feedback and micro-motion spectrum analysis, the system makes the output power curve of the heating execution unit show a downward trend opposite to the change in thermal resistance when the controlled object undergoes drastic positional changes and accompanied by vibrations. Without the intervention of temperature sensors, the system can suppress potential temperature overshoot risks in advance and maintain the smooth convergence of the temperature of the contact surface of the controlled object.

[0026] Example 2: This example verifies the effectiveness of the gain adaptive adjustment mechanism in the method of the present invention using data, specifically targeting the typical time-varying heat load condition where the controlled object changes from a standard flat state to a high-resistance curled state. The test platform includes a high-precision medical heating blanket with segmented heating capability, and has... A flexible sensor layer of pressure sensing matrix and an embedded microcontroller system capable of recording temperature and power data were used to simulate the controlled object. The experiment employed a human body model with a surface-mounted high-precision thermocouple array, possessing static thermophysical characteristics matching the basal metabolic heat capacity of an adult. The system was set to a target temperature of [temperature value missing]. Temperature monitoring accuracy set to The control cycle is set to The experimental group (the scheme of this invention) uses a contact topology identification and gain adaptive adjustment mechanism, with a baseline scaling factor. Set as Estimated equivalent thermal impedance Based on the form compactness coefficient Real-time correction The control group (existing technology) disables topology recognition, and the controller uses a fixed proportional coefficient in the standard supine position. The experiment was conducted in two phases: the human model was placed in a standard supine position, and the system temperature was allowed to stabilize. And the power fluctuation rate is less than The steady-state conditions, in At a certain moment, the model is quickly adjusted to a tight, side-lying, curled-up posture to simulate a sudden reduction in effective heat dissipation area. Furthermore, the local contact thermal resistance increases. The system continuously records the temperature of the model contact surface during strong time-varying load abrupt changes. Maximum overshoot and steady-state recovery time The table below presents the performance comparison data of this embodiment and the control group under strong load change conditions.

[0027] Table 1: Performance Comparison Data Table

[0028] Data results show that, After a sudden change in strong time-varying heat load, a fixed proportional coefficient is used. The overshoot of the contact surface temperature in the control group model was as high as It far exceeds the safety threshold and takes a long time. Only then can it return to a steady state. This hysteresis and overshoot are inevitable consequences of traditional control strategies when faced with thermodynamic model mismatch. In contrast, the present invention, through contact topology identification and gain adaptive scheduling mechanism, allows the microcontroller to measure the morphological compactness coefficient at the instant of body position change. The increase will affect the estimated equivalent thermal resistance value. As the system increases, it reduces the scaling factor in real time. This causes the energy input to be synchronously reduced as the contact thermal resistance increases. Specifically, in the present invention, the maximum overshoot is... Controlled At an extremely low level, and the system is It immediately returns to steady state, even compared to the control group where the concentration was intentionally lowered. Compared with other solutions, the solution of this invention still has advantages in overshoot and recovery time.

[0029] Example 3: This example combines Figures 1 to 3 This section describes the closed-loop temperature control method and system for heating blankets based on human body position recognition feedback, such as... Figure 1 As shown, real-time pressure detection data is first collected through a pressure sensor array. This data is then transmitted to a contact feature extraction step to calculate the morphological compactness coefficient characterizing the heat dissipation boundary. Subsequently, the thermal impedance parameter estimation step is initiated, where the morphological compactness coefficient is converted into an equivalent thermal impedance estimate by calling a mapping function. This estimate is then sent to a gain adaptive adjustment step, where the forward channel gain is dynamically constrained based on a negative correlation strategy. Simultaneously, the temperature deviation signal from the temperature sensor is transmitted to a temperature closed-loop controller. This controller incorporates the equivalent thermal impedance estimate as a gain adjustment factor and processes the signal in conjunction with the forward channel gain. The processed signal then enters a power output and response verification step to generate a target heating power and achieve temperature adaptive control. Finally, the target heating power is used to control the heating execution unit to drive the heating component, thereby completing the closed-loop control process based on human body position recognition feedback.

[0030] like Figure 2 As shown in the figure, the horizontal axis represents time in minutes, and the vertical axis represents temperature in seconds. It includes three curves: a dashed target temperature representing the set value, a dashed control group temperature representing the existing technology, and a fixed temperature. =0.5 and the solid line representing the temperature of the present invention, when a sudden change in heat load occurs between the 10th and 15th minutes of the time axis, the control group curve shows a peak value exceeding 40.0. The curve of the present invention always closely matches 37.0. Target temperature baseline; such as Figure 3As shown, the architecture is divided into two parts: an upper-layer digital computing domain (embedded intelligent hub) and a lower-layer physical interaction domain (flexible body of the heating blanket). The physical interaction domain includes a resistive pressure sensor array as a data acquisition source and a segmented heating execution matrix as an energy output end. The controlled object, the human body, applies pressure to the physical domain and receives heat. The resulting real-time pressure matrix data stream is uploaded to the digital computing domain. After processing through a chain of contact topology perception calculation of form compactness, dynamic mapping estimation of thermal resistance, and gain adaptive scheduling constraint of output gain, an adaptive target power command is generated and sent back to the physical domain, forming a complete closed-loop control loop.

[0031] Example 4: This example describes the systematic determination procedure for the core control parameters upon which the method of the present invention relies, to ensure the reproducibility of the system and the engineering basis of the parameters before actual deployment, specifically for the thermal resistance mapping function. Reference thermal resistance in and thermal sensitivity coefficient To determine the optimal system configuration, an offline multi-condition test calibration method was adopted. A test model equipped with a high-precision thermocouple array was used to simulate three typical body positions: standard supine, lateral, and curled-up. Under each condition, the heating unit power was fixed at [value missing]. The measurement system at temperature Steady-state power consumption when reaching steady state The system's equivalent thermal impedance pass The calculation yielded, where For ambient temperature, multiple groups Corresponding form compactness Import the dataset into the least squares fitting algorithm and in exponential form. Fit it, where The fitting process ultimately determined the reference value for the morphological compactness of the standard supine position. and The parameters are stored in the microcontroller's memory as the baseline for the real-time thermal impedance estimation model.

[0032] For folding threshold To determine the optimal system parameters, the system employs a pressure gradient statistical analysis method to collect multiple sets of two-dimensional pressure distribution matrices during normal human body position changes. And collect data when the fabric is accidentally folded. Apply to each set of matrices The operator performs discrete convolution operations to calculate its pressure gradient field. Statistical analysis was performed on the gradient peak values ​​generated by all normal human contact, and their values ​​were calculated. quantiles folding threshold Set as ,in For the safety factor, the value range is: to This is used to ensure that the system responds only to mechanically folded regions with sharp linear characteristics, for the estimated endogenous heat production. The system relies on a tremor-induced heat generation model, and the feedforward compensation coefficients are determined through cross-physical domain benchmarking experiments. In a controlled environment, muscle shivering of varying intensities was induced in controlled subjects, and the endogenous heat production power during the shivering process was recorded. Simultaneously, the micro-motion energy index on the pressure array of the heating blanket was collected. Established through regression analysis and The mapping relationship between them is determined. .

[0033] And individualized calibration coefficients in the online thermal inertia calibration process. The calculation is performed when the system is in steady state. At that time, the microcontroller applies , A step excitation signal at rated power is applied, and the temperature rise response slope is measured. If after incentive Internal temperature from Rise to The measured temperature rise slope for At this point, the theoretical standard temperature rise slope predicted based on the current body position model... table lookup The individualized thermal capacity calibration coefficient calculated by the microcontroller unit for This coefficient This indicates that the actual heat capacity of the controlled object is less than the standard model value, and the system will use... For the current scaling factor A second weighted correction is performed. The time constant setting for asymmetric filtering is determined based on the system's thermodynamic large inertia characteristics and safety redundancy principles. The gain decrease time constant is... Set as The physical basis for this is the safety response requirement for thermal accumulation and the gain rise time constant. Set as Its engineering basis is the response characteristics and stability threshold of thermodynamic large inertial systems. Set as the pressure distribution matrix element in The mean square deviation within the time window is less than If the value is below this threshold, the body position is considered stable, and gain updates are allowed. The determination of these parameters ensures that the control strategy has the best dynamic adaptability while maintaining stability.

[0034] Example 5: In a laboratory setting specifically designed to verify the control stability of a time-varying heat load system, offline engineering calibration of the thermal impedance mapping function was performed. A test bench was constructed, comprising a heating blanket body, a high-precision pressure sensor array, and multiple distributed thermocouple arrays. The thermocouple arrays were used to accurately measure the actual temperature gradient between the heat source and the equivalent surface under different contact topologies. The room temperature of the test environment was set to be Relative humidity controlled at To maintain a constant state, during the experiment, a set of standardized flexible thermal models with different geometries were used. The heat capacity and surface emissivity of the materials were similar to those of human skin. These models were placed according to preset contact patterns, such as flat, semi-curled, and fully wrapped, ensuring continuous operation in each pattern until the system reached thermodynamic steady state. For each contact pattern, the processor collected real-time pressure data from the pressure sensor array and calculated the morphological compactness coefficient for that contact pattern. Simultaneously, the average temperature of the heat source measured by the thermocouple array under steady-state conditions was recorded. and ambient temperature and the steady-state input power of the heating unit. At this point, the actual equivalent thermal resistance under this contact mode Based on the principle of steady-state thermal equilibrium, the expression is: ,in and The unit is °C. The unit is , The unit is °C / W. By systematically changing the contact mode and applying simulated pressure, a total of [data / values] were collected covering [areas / areas]. Full range, for example from to ,by step size Group matching data.

[0035] Collected Group matching data The nonlinear fitting module, input to the host controller, is used to construct the morphological compactness coefficient. Estimated equivalent thermal resistance Thermal resistance mapping function between To ensure the function To ensure stability and fitting accuracy across the entire working interval, a piecewise multinomial model is used to fit the data points. Less than The low-compactness region is fitted with a first-order linear fit. lie in arrive The transition region is fitted using a quadratic polynomial. Greater than In the high-compactness region, linear fitting is reapplied, and the goal of the fitting process is to make the estimated values... Compared with actual value Mean square error between minimize, Defined as ,in Total number of data points , and The unit is °C / W. The unit is °C² / W². Finally, this verified minimum... No more than The piecewise polynomial model is embedded as a pre-set thermal resistance mapping function into the non-volatile memory of the heating blanket controller, and is used in actual operation to calculate the morphological compactness coefficient in real time. Output equivalent thermal impedance estimate This provides real-time feedforward parameter input for the adaptive gain adjustment of the temperature closed-loop controller.

[0036] Example 6: In the pre-deployment calibration of systems used for high-precision temperature control, standardized procedures must be performed to ensure the system's parameter adaptability to different individual heat capacities and environmental disturbances. The calibration procedure mainly focuses on determining the stability threshold in asymmetric filtering. and asymmetric time constant and Stability threshold To distinguish between transient agitation and steady-state positional changes in a controlled object, the system collects data in offline mode. Two-dimensional stress distribution matrix of different individuals in sleep, mild turning over, and micro-movement / restlessness states. Data, processor Time window calculation of time-domain variance of each data set Through statistical analysis, the lower limit of variance for mild rolling over movements was determined. for To ensure that gain updates are performed only when the controlled object is confirmed to be in a static or stable state, a stability threshold is set. Set as ,Right now Set as If in real time The controller then allows updating the forward channel gain. Otherwise freeze And integral terms.

[0037] An asymmetric time constant is used for smooth gain transition, ensuring a safety-first principle; the gain fall time constant... Set as The physical basis for this is the safety response requirement for thermal accumulation, namely, when a high impedance state is detected, the power reduction should be fast and the gain rise time constant should be constant. Set as Its engineering basis is the response characteristics of thermodynamic large inertial systems, namely, slow power recovery, to avoid power oscillations caused by the lag in temperature sensor detection. In actual operation, when gain needs to be updated... To target gain At this time, the processor achieves a smooth transition of parameters through a first-order hysteresis filtering algorithm, which follows the formula... , among which, when Smoothing factor ;when hour, , Sampling period By setting The asymmetry of the control strategy enables proactive measures to avoid overheating risks. During startup, the system also performs a self-check of safety protection logic, specifically monitoring the temperature response slope of the heating actuator after power-on. ,like Exceeding the current estimated equivalent thermal impedance Predicted theoretical slope of If the threshold is exceeded, the contact data is deemed abnormal, and the processor will lock the output of the temperature closed-loop controller to the safe baseline threshold. .

[0038] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0039] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A closed-loop temperature control method for a heating blanket based on human body position recognition feedback, characterized in that, Includes the following steps: The contact feature extraction step involves collecting real-time pressure detection data from the pressure sensing array in the functional layer of the heating blanket and discretizing it into a two-dimensional pressure distribution matrix. Based on the connected component analysis of non-zero elements, the morphological compactness coefficient characterizing the heat dissipation boundary conditions of the controlled object is calculated. The morphological compactness coefficient is used to quantify the effective contact state between the controlled object and the heating blanket. The thermal impedance parameter estimation step calls a preset thermal resistance mapping function to convert the shape compactness coefficient into the equivalent thermal impedance estimate of the controlled object at the current moment. The thermal resistance mapping function is configured to establish a positive correlation between the shape compactness coefficient and the equivalent thermal impedance estimate to characterize the attenuation characteristics of the system's heat dissipation capacity in the curled or heavily wrapped state of the controlled object. In the gain adaptive adjustment step, the estimated equivalent thermal impedance is introduced as a gain adjustment factor into the temperature closed-loop controller. According to the negative correlation gain adjustment strategy, dynamic constraints are applied to the forward channel gain of the temperature closed-loop controller. The logic of the dynamic constraint is: when the estimated equivalent thermal impedance is detected to increase, the forward channel gain is reduced to suppress the heat accumulation effect caused by the increase in local thermal resistance. The power output and response verification steps utilize a temperature closed-loop controller with forward channel gain to process the temperature deviation signal to generate a target heating power, and use this target heating power to drive the heating execution unit to achieve adaptive temperature control of the time-varying heat load object.

2. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 1, characterized in that, The contact feature extraction step also includes a folding anomaly identification sub-step: performing gradient calculation on the two-dimensional pressure distribution matrix to generate a pressure gradient field; scanning the pressure gradient field and identifying linear continuous regions where the gradient value exceeds a preset folding threshold, and marking the linear continuous region as a folded adiabatic zone; the power output and response verification step also includes regional shielding control logic: constructing a logic control mask that maps to the spatial distribution of the heating execution unit, setting the mask position corresponding to the coordinates of the folded adiabatic zone to the blocking state; using the logic control mask to perform a spatial AND operation on the target heating power, blocking the power output of the heating execution unit corresponding to the folded adiabatic zone, so as to eliminate the overheating risk caused by local adiabatic while maintaining closed-loop control of the non-folded region.

3. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 1, characterized in that, The gain adaptive adjustment step also includes internal heat disturbance feedforward compensation logic: high-frequency components within a preset flutter frequency band are separated from real-time pressure detection data; a micro-motion energy index characterizing the flutter intensity of the controlled object is calculated; the micro-motion energy index is mapped to an estimated internal heat generation value using a preset flutter heat generation model; and the following feedforward compensation calculation is performed when generating the target heating power: ,in, The target heating power for the final output, The base power is calculated based on the temperature deviation signal. The preset feedforward compensation coefficient, The value is the estimated value of endogenous heat production. Through feedforward compensation calculation, before the temperature sensor detects the lag response of temperature rise caused by vibration, the input of external heat source is actively reduced to offset the superposition effect of endogenous metabolic heat of the controlled object.

4. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 1, characterized in that, The thermal resistance mapping function in the thermal resistance parameter estimation step is constructed based on a preset nonlinear heat transfer model. This function defines the shape compactness coefficient as the ratio of the number of activated nodes in the two-dimensional pressure distribution matrix to the area of ​​the smallest bounding rectangle of the connected domain formed by the activated nodes. The positive correlation is specifically configured such that as the shape compactness coefficient increases, the output of the equivalent thermal resistance estimate increases exponentially to simulate the nonlinear shrinkage process of the effective heat dissipation surface area when the controlled object changes from a flat state to a curled state. The method also includes an online thermal inertia calibration step to correct... Equivalent thermal impedance estimation: When the temperature closed-loop controller is running in steady state, a step power excitation signal with a preset amplitude and duration is applied to the heating execution unit; the transient temperature rise response data of the temperature sensor to the step power excitation signal is collected, and the measured temperature rise slope is calculated; the ratio of the theoretical standard temperature rise slope based on the current compactness coefficient to the measured temperature rise slope is calculated to generate an individualized thermal capacity calibration coefficient; the individualized thermal capacity calibration coefficient is used to perform a secondary weighted correction on the forward channel gain to compensate for the model parameter drift caused by individual differences of the controlled object or the thermal resistance of clothing.

5. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 1, characterized in that, The gain adaptive adjustment step also includes asymmetric filtering: calculating the time-domain fluctuation variance of the two-dimensional pressure distribution matrix within a preset time window; if the time-domain fluctuation variance exceeds the preset stability threshold, it is determined that the controlled object is in a transient disturbance state, and the current forward channel gain and the integral term value of the temperature closed-loop controller remain unchanged. If the time-domain fluctuation variance is lower than the stability threshold, the forward channel gain is allowed to be updated, and the currently executed gain value is smoothly transitioned to the target gain value using a first-order hysteresis filtering algorithm. The first-order hysteresis filtering algorithm is configured to have an asymmetric time constant, so that the response speed of gain decrease is faster than the response speed of gain increase.

6. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 2, characterized in that, In the folding anomaly identification sub-step, gradient calculation uses the Laplacian operator or the Sobel operator to perform discrete convolution on the two-dimensional pressure distribution matrix; the preset folding threshold is set to be higher than the statistical upper limit of the maximum pressure gradient generated by normal limb compression of the controlled object, so as to distinguish the different pressure field morphological characteristics generated by contact with flexible organisms and folding of rigid fabrics.

7. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 1, characterized in that, The temperature closed-loop controller is a PID controller. The forward channel gain specifically acts on the proportional coefficient of the PID controller. The negative correlation gain adjustment strategy is specifically implemented by setting a reference proportional coefficient corresponding to the standard tiling condition, and setting the proportional coefficient in real time as the product of the reference proportional coefficient and the attenuation factor. The attenuation factor is negatively correlated with the estimated value of equivalent thermal impedance and is limited to a value range of zero to one.

8. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 3, characterized in that, The micro-motion energy index is obtained by performing a short-time Fourier transform on the high-frequency components and calculating the power spectral density integral within a specified frequency band; the preset tremor frequency band is limited to 5 Hz to 20 Hz to match the characteristic frequency range of involuntary contraction of human skeletal muscles and to filter out low-frequency interference signals caused by turning over or breathing.

9. The closed-loop temperature control method for a heating blanket based on human body position recognition feedback according to claim 1, characterized in that, The method operates in a temperature control system containing an embedded microcontroller unit. The data source for the two-dimensional pressure distribution matrix is ​​a resistive pressure sensor array laid in the functional layer of the heating blanket. The power output and response verification steps also include safety protection logic: real-time monitoring of the temperature response slope after the heating execution unit is powered on. If the temperature response slope exceeds the theoretical range predicted based on the current equivalent thermal impedance estimate, the contact data is determined to be abnormal and the output of the temperature closed-loop controller is locked to the safety basic threshold.

10. A closed-loop temperature control system for a heating blanket based on human body position recognition feedback, used to implement the method of claim 1, characterized in that, include: The contact feature extraction module is configured to collect real-time pressure detection data of the pressure sensing array in the functional layer of the heating blanket and discretize it into a two-dimensional pressure distribution matrix. Based on the connected component analysis of non-zero elements, the module calculates the morphological compactness coefficient that characterizes the heat dissipation boundary conditions of the controlled object. The morphological compactness coefficient is used to quantify the effective contact state between the controlled object and the heating blanket. The thermal impedance parameter estimation module is configured to call a preset thermal resistance mapping function to convert the shape compactness coefficient into the equivalent thermal impedance estimation value of the controlled object at the current moment. The thermal resistance mapping function is configured to establish a positive correlation between the shape compactness coefficient and the equivalent thermal impedance estimation value to characterize the attenuation characteristics of the system's heat dissipation capacity in the curled or heavily wrapped state of the controlled object. The gain adaptive adjustment module is configured to introduce the estimated equivalent thermal impedance as a gain adjustment factor into the temperature closed-loop controller, and apply dynamic constraints to the forward channel gain of the temperature closed-loop controller according to the negative correlation gain adjustment strategy. The logic of the dynamic constraint is: when the estimated equivalent thermal impedance is detected to increase, the forward channel gain is reduced to suppress the heat accumulation effect caused by the increase in local thermal resistance. The power output and response verification module is configured to use the temperature closed-loop controller including the forward channel gain to process the temperature deviation signal to generate the target heating power, and use the target heating power to drive the heating execution unit to realize the temperature adaptive control of the time-varying heat load object.

Images