Constant temperature control method, control device and equipment for baby bathing equipment

Abstract

The embodiment of the invention provides a constant-temperature control method, device and equipment for infant bathing equipment. The method comprises the steps that S1, infant limb movement data and water fluctuation characteristics in the bathing equipment are collected in real time through a multi-mode sensor group; s2, the limb movement data and the water body fluctuation characteristics are input into a thermal disturbance prediction model, a water temperature change prediction value of each area is obtained, and the thermal disturbance prediction model is constructed by fusing a space-time convolutional network and a long-short-term memory network; s3, a risk area with the water temperature decreasing rate exceeding a first threshold value is marked according to the predicted value; and S4, activating an annular heating array at the boundary of the risk area, and performing decreasing temperature compensation from outside to inside. According to the method, through multi-modal feature fusion and a special thermal disturbance prediction model, the prediction precision of local water temperature change is remarkably improved, and a bathing environment more suitable for babies is created.

Description

Technical Field

[0001] This application relates to the field of temperature control, and particularly to a constant temperature control method, control device and equipment for baby bathing equipment. Background Art

[0002] During the baby bathing process, limb movements can cause violent fluctuations in the local water body, leading to a sudden drop in water temperature. Traditional constant temperature control relies on temperature feedback and has hysteresis. Most existing baby bathing constant temperature equipment uses fixed-position temperature sensors combined with the PID control algorithm. For example, regional heating is achieved through multi-point temperature measurement methods, but it cannot effectively predict the sudden drop in water temperature caused by limb movements. There are two defects in the existing related technologies: one is that relying on temperature hysteresis feedback leads to compensation delay; the other is that no spatio-temporal correlation model of action characteristics and heat conduction is established, making it difficult to achieve precise local temperature control.

[0003] Therefore, there is an urgent need for a constant temperature control method, control device and equipment for baby bathing equipment to overcome the above deficiencies in the existing technology. Summary of the Invention

[0004] This application provides a constant temperature control method, control device and equipment for baby bathing equipment. This method significantly improves the prediction accuracy of local water temperature changes through multi-modal feature fusion and a dedicated thermal disturbance prediction model; at the same time, composite data constructs a representation ability to enhance the action-fluctuation coupling relationship, spatio-temporal convolution captures the propagation characteristics of limb movements, and LSTM models the heat conduction delay effect, effectively reducing the prediction error.

[0005] In a first aspect, a constant temperature control method for baby bathing equipment is provided, and the method includes:

[0006] S1: Real-time collect baby limb movement data and water body fluctuation characteristics in the bathing equipment through a multi-modal sensor group;

[0007] S2: Input the limb movement data and the water body fluctuation characteristics into a thermal disturbance prediction model to obtain predicted values of water temperature changes in each region, where the thermal disturbance prediction model is constructed by fusing a spatio-temporal convolutional network and a long short-term memory network;

[0008] S3: Mark risk regions where the water temperature drop rate exceeds a first threshold according to the predicted values;

[0009] S4: Activate an annular heating array at the boundary of the risk region and perform decreasing temperature compensation from the outside to the inside.

[0010] It should be understood that by capturing the activity characteristics of the baby through multi-modal sensing data, using an improved spatio-temporal thermal perturbation prediction model to prospectively identify risk areas, and combining three-layer progressive heating to form a physical isolation barrier, the traditional passive compensation is transformed into active defense, effectively eliminating the temperature fluctuations caused by the suddenness of movements, shortening the local water temperature recovery time, and creating a bathing environment more suitable for the baby.

[0011] Combined with the first aspect, in some implementations of the first aspect, the thermal perturbation prediction model includes:

[0012] A multi-scale spatio-temporal perception module that uses parallel small-scale local convolution kernels and large-scale global convolution kernels to extract the detailed features and spatial correlations of the limb movement data respectively;

[0013] A dynamic feature fusion module that encodes the water body fluctuation features into tensors of the same dimension as the limb movement data;

[0014] A prediction module that includes a bidirectional long short-term memory network and is used to output the predicted value of the water temperature change.

[0015] It should be understood that the model captures action details and global correlations through multi-scale spatio-temporal perception, realizes cross-modal interaction between the baby's limb movements and water fluctuation features through dynamic feature fusion, enhances the time series prediction ability through the bidirectional memory network, improves the recognition accuracy of actions with different amplitudes through multi-scale convolution, and establishes a feature competition mechanism with physical constraints through the dynamic fusion gate; the bidirectional LSTM enhances time series modeling while maintaining causality.

[0016] Combined with the first aspect, in some implementations of the first aspect, the method further includes an online update step:

[0017] Real-time collect the actual water temperature change data after compensation heating;

[0018] Calculate the dynamic error matrix between the predicted value and the measured value;

[0019] When the L2 norm of the dynamic error exceeds the second threshold:

[0020] Freeze the network parameters of the spatio-temporal convolution layer, adjust the weight matrix of the gated recurrent unit layer, and optimize the attenuation coefficient of the water temperature propagation path.

[0021] It should be understood that the dynamic error matrix contains the spatio-temporal distribution data of the temperature difference between the prediction and the measurement in each region; the dynamic error refers to the amount of prediction deviation accumulated over time; the L2 norm is the square root of the sum of the squares of the matrix elements and is used to quantify the overall error. This step optimizes the propagation parameters while maintaining the stability of feature extraction through a hierarchical update strategy, enabling the model to continuously adapt to actual environmental changes.

[0022] In combination with the first aspect, in some implementations of the first aspect, the method for processing the multi-modal sensor group data includes:

[0023] Perform spatio-temporal registration on the limb movement data and the water body fluctuation characteristics, achieve data synchronization through timestamp alignment, and map the limb movement data and the water body fluctuation characteristics to a unified spatial coordinate system;

[0024] Construct a composite feature vector by aligning the limb movement data and the water body fluctuation characteristics through timestamp alignment and spatial coordinate mapping.

[0025] It should be understood that the effectiveness of the data collected by the multi-modal sensor group is enhanced through specific signal processing methods, the correlation between the motion characteristics and the fluctuation characteristics is improved, and high-precision input is provided for model prediction.

[0026] In combination with the first aspect, in some implementations of the first aspect, the multi-modal sensor group includes:

[0027] A microwave motion sensor for capturing the limb movement data, and the microwave motion sensor is arranged at the edge of the bathtub;

[0028] A piezoelectric fluctuation sensor for capturing the water body fluctuation characteristics, and the piezoelectric fluctuation sensor is arranged in the side wall of the bathing device.

[0029] It should be understood that fine movements are captured non-contact by the microwave sensor, and the piezoelectric sensor accurately measures the wave propagation. The spatial layout design of the two ensures that there is no dead angle in data collection, provides high-fidelity input for the thermal disturbance prediction model, and can effectively reduce the false alarm rate. In addition, the microwave sensor is arranged at the edge of the bathtub to reduce signal occlusion, and the piezoelectric sensor is embedded in the side wall to accurately capture the wave propagation.

[0030] In combination with the first aspect, in some implementations of the first aspect, the annular heating array includes a plurality of independently temperature-controlled units spirally distributed along the inner wall of the bathtub; when the annular heating array is activated, three heating rings are formed around the risk area: the outer ring operates at the basic power to block heat diffusion, the middle ring adjusts the power according to the square root of the predicted decline rate, and the inner ring uses pulse heating to directly compensate the core area.

[0031] It should be understood that the three heating rings form a gradient protection: the outer ring blocks the heat diffusion path, the middle ring dynamically adjusts to maintain a buffer, and the inner ring quickly compensates the core area. Compared with the traditional uniform heating method, this method is 30% more energy-efficient and can reduce the temperature recovery time.

[0032] In combination with the first aspect, in certain implementations of the first aspect, during the heating compensation process, the stable flow control is started synchronously: Sound waves with a preset frequency spectrum are emitted by an ultrasonic generator arranged at the bottom of the bathtub, and the frequency parameter of the sound waves is dynamically coupled with the rotation speed of the current heating ring to suppress heat loss caused by water flow disturbance.

[0033] It should be understood that by using sound wave stable flow control to suppress water flow disturbance caused by heating, reducing unexpected heat loss, and coupling the design of sound and heat parameters to improve the stable flow efficiency, the effective utilization of the compensated heat is ensured.

[0034] In combination with the first aspect, in certain implementations of the first aspect, the accuracy of the thermal disturbance prediction model is verified in real time through a distributed temperature sensing network. When the prediction deviation exceeds the allowable range for N consecutive times, the following operations are triggered:

[0035] Pause the current thermal disturbance prediction model;

[0036] Activate the emergency control mode based on temperature gradient feedback;

[0037] At the same time, start the model reconstruction process until the prediction accuracy is restored.

[0038] It should be understood that a dual-mode control guarantee system is constructed to seamlessly switch to the emergency mode when the thermal disturbance prediction model fails, and at the same time start the self-repair process to improve the system reliability and avoid potential safety hazards caused by the failure of traditional single-mode control.

[0039] The second aspect provides a control device, which includes a processor and a memory. The processor is coupled with the memory. The memory is used to store computer programs or instructions, and the processor is used to execute the computer programs or instructions in the memory so that any implementation in the first aspect is executed.

[0040] The third aspect further provides a baby bathing device, which includes the control device described in the second aspect. Description of the Drawings

[0041] Figure 1 It is a flowchart of the implementation of a constant temperature control method for a baby bathing device provided by an embodiment of the present application. Detailed Embodiments

[0042] The terms used in the following embodiments are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and appended claims of the present application, the singular forms "a", "an", "", "the foregoing", "said", and "this" are also intended to include expressions such as "one or more", unless the context clearly indicates otherwise. It should also be understood that in the following embodiments of the present application, "at least one", "one or more" means one, two or more than two. The term "and / or" is used to describe the association relationship of associated objects and indicates that three relationships may exist; for example, A and / or B may mean: A exists alone, A and B exist simultaneously, and B exists alone, where A and B may be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0043] Reference to "one embodiment" or "some embodiments" or the like described in this specification means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments" and the like that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0044] During the baby bath process, limb movements can cause violent fluctuations in the local water body, leading to a sudden drop in water temperature. Traditional constant temperature control relies on temperature feedback and has hysteresis. Existing technologies mainly adopt fixed-position heating or uniform temperature control strategies and cannot accurately respond to local thermal disturbances.

[0045] The embodiments of the present application provide a constant temperature control method, a control device, and a device for a baby bath device. This method significantly improves the prediction accuracy of local water temperature changes through multi-modal feature fusion and a dedicated thermal disturbance prediction model; at the same time, composite data construction enhances the representation ability of the action-fluctuation coupling relationship, spatio-temporal convolution captures the propagation characteristics of limb movements, and LSTM models the heat conduction delay effect, effectively reducing the prediction error.

[0046] The technical solutions of the embodiments of the present application will be described below with reference to the drawings.

[0047] Figure 1 This is a flowchart for implementing a constant temperature control method for a baby bath device provided by an embodiment of the present application. In some examples, the method includes:

[0048] S1: Real-time collect the limb movement data of the baby and the water body fluctuation characteristics in the bathing device through a multi-modal sensor group;

[0049] S2: Input the limb movement data and the water body fluctuation characteristics into a thermal disturbance prediction model to obtain the predicted values of the water temperature changes in each area, where the thermal disturbance prediction model is constructed by fusing a spatio-temporal convolutional network and a long short-term memory network;

[0050] S3: Mark the risk areas where the water temperature drop rate exceeds the first threshold according to the predicted values;

[0051] S4: Activate the annular heating array at the boundary of the risk area and perform a decreasing temperature compensation from the outside to the inside.

[0052] In some examples, the thermal disturbance prediction model includes:

[0053] A multi-scale spatio-temporal perception module that uses parallel small-scale local convolutional kernels and large-scale global convolutional kernels to extract the detailed features and spatial correlations of the limb movement data respectively;

[0054] A dynamic feature fusion module that encodes the water body fluctuation characteristics into a tensor with the same dimension as the limb movement data;

[0055] A prediction module that includes a bidirectional long short-term memory network and is used to output the predicted values of the water temperature changes.

[0056] In a possible implementation, the spatio-temporal convolutional layer adopts 3 parallel branches. For example, a 3×3×3 convolutional kernel extracts local spatio-temporal features, and a 5×5×5 convolutional kernel captures global correlations to extract motion features at different spatio-temporal scales; a cross-attention mechanism calculates the correlation matrix of the motion and fluctuation features to generate a weight distribution; the bidirectional LSTM has a 2-layer stacked structure, with 64 hidden units in each layer, the time series window is set to 8 frames, and the output layer activates and outputs the predicted values through sigmoid.

[0057] In some examples, the method further includes an online update step:

[0058] Real-time collect the actual water temperature change data after compensation heating;

[0059] Calculate the dynamic error matrix between the predicted values and the measured values;

[0060] When the L2 norm of the dynamic error exceeds the second threshold:

[0061] Freeze the network parameters of the spatio-temporal convolutional layer, adjust the weight matrix of the gated recurrent unit layer, and optimize the attenuation coefficient of the water temperature propagation path.

[0062] In a possible implementation, a phased gradient update strategy is adopted - freezing the computational graph of the spatio-temporal convolutional layer and only allowing the fully connected weights of the gated recurrent unit layer to be updated through backpropagation; the water temperature propagation attenuation coefficient statistically predicts the mean of the deviation through a sliding window and proportionally corrects the parameters in the heat conduction equation; the update process is executed in batches using a momentum optimizer to maintain the convergence stability of the model.

[0063] In some examples, the method for processing the multi-modal sensor group data includes:

[0064] Performing spatio-temporal registration on the limb movement data and the water body fluctuation characteristics, realizing data synchronization through timestamp alignment, and mapping the limb movement data and the water body fluctuation characteristics to a unified spatial coordinate system;

[0065] Constructing a composite feature vector from the limb movement data and the water body fluctuation characteristics through timestamp alignment and spatial coordinate mapping.

[0066] In some examples, the multi-modal sensor group includes:

[0067] A microwave motion sensor for capturing the limb movement data, which is arranged at the edge of the bathtub;

[0068] A piezoelectric wave sensor for capturing the water body fluctuation characteristics, which is arranged in the side wall of the bathing device.

[0069] In some examples, the annular heating array includes a plurality of independently temperature-controlled units spirally distributed along the inner wall of the bathtub; when the annular heating array is activated, three heating rings are formed around the risk area: the outer ring operates at the basic power to block heat diffusion, the middle ring adjusts the power according to the square root of the predicted descent rate, and the inner ring uses pulsed heating to directly compensate the core area.

[0070] In some examples, a steady flow control is synchronously started during the heating compensation process: ultrasonic waves with a preset frequency spectrum are emitted by an ultrasonic generator arranged at the bottom of the bathtub, and the frequency parameter of the ultrasonic waves is dynamically coupled with the rotation speed of the current heating ring to suppress the heat loss caused by water flow disturbance.

[0071] In some examples, the accuracy of the thermal disturbance prediction model is verified in real time through a distributed temperature sensing network. When the prediction deviation exceeds the allowable range for N consecutive times, the following operations are triggered:

[0072] Pausing the current thermal disturbance prediction model;

[0073] Activating the emergency control mode based on temperature gradient feedback;

[0074] Simultaneously start the model reconstruction process until the prediction accuracy is restored.

[0075] An embodiment of the present application provides a control device, which includes a processor and a memory. The processor is coupled to the memory. The memory is used to store computer programs or instructions, and the processor is used to execute the computer programs or instructions in the memory, so that any implementation manner in the foregoing examples is executed.

[0076] An embodiment of the present application further provides a baby bathing device, which includes the control device described above.

[0077] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. Any equivalent modification or change made by those of ordinary skill in the art according to the content disclosed in the present invention shall be included in the protection scope recorded in the claims.

Claims

1. A constant temperature control method for a baby bathing device, characterized in that, The method includes: S1: Real-time collect the limb movement data of the baby and the water body fluctuation characteristics in the bathing device through a multi-modal sensor group; S2: Input the limb movement data and the water body fluctuation characteristics into a thermal disturbance prediction model to obtain the predicted values of the water temperature changes in each area, where the thermal disturbance prediction model is constructed by fusing a spatio-temporal convolutional network and a long short-term memory network; S3: Mark the risk areas where the water temperature drop rate exceeds the first threshold according to the predicted values; S4: Activate the annular heating array at the boundary of the risk area and perform a decreasing temperature compensation from the outside to the inside.

2. The method according to claim 1, wherein The thermal disturbance prediction model includes: A multi-scale spatio-temporal perception module that uses parallel small-scale local convolutional kernels and large-scale global convolutional kernels to extract the detailed features and spatial correlations of the limb movement data respectively; A dynamic feature fusion module that encodes the water body fluctuation characteristics into a tensor of the same dimension as the limb movement data; A prediction module that includes a bidirectional long short-term memory network and is used to output the predicted values of the water temperature changes.

3. The method according to claim 1, wherein The method further includes an online update step: Real-time collect the actual water temperature change data after compensatory heating; Calculate the dynamic error matrix between the predicted value and the measured value; When the L2 norm of the dynamic error exceeds the second threshold: Freeze the network parameters of the spatio-temporal convolutional layer, adjust the weight matrix of the gated recurrent unit layer, and optimize the attenuation coefficient of the water temperature propagation path.

4. The method according to claim 1, wherein The processing method of the multi-modal sensor group data includes: Perform spatio-temporal registration on the limb movement data and the water body fluctuation characteristics, realize data synchronization through timestamp alignment, and map the limb movement data and the water body fluctuation characteristics to a unified spatial coordinate system; Construct a composite feature vector by aligning the limb movement data and the water body fluctuation characteristics through timestamp alignment and spatial coordinate mapping.

5. The method according to claim 4, wherein The multi-modal sensor group includes: A microwave motion sensor that is used to capture the limb movement data and is arranged at the edge of the bathtub; A piezoelectric wave sensor that is used to capture the water body fluctuation characteristics and is arranged in the side wall of the bathing device.

6. The method according to claim 1, wherein The annular heating array includes a plurality of independently temperature-controlled units spirally distributed along the inner wall of the bathtub; when the annular heating array is activated, three heating rings are formed on the periphery of the risk area: the outer ring operates at the basic power to block heat diffusion, the middle ring adjusts the power according to the square root of the predicted drop rate, and the inner ring uses pulse heating to directly compensate the core area.

7. The method according to claim 1, wherein Start the steady flow control synchronously during the heating compensation process: emit sound waves with a preset frequency spectrum through an ultrasonic generator arranged at the bottom of the bathtub, and the frequency parameters of the sound waves are dynamically coupled with the rotation speed of the current heating ring to suppress the heat loss caused by water flow disturbance.

8. The method according to claim 1, wherein Verify the accuracy of the thermal disturbance prediction model in real time through a distributed temperature sensing network. When the prediction deviation exceeds the allowable range for N consecutive times, trigger the following operations: Pause the current thermal disturbance prediction model; Activate the emergency control mode based on the temperature gradient feedback; At the same time, start the model reconstruction process until the prediction accuracy is restored.

9. A control device, characterized in that, It includes a processor and a memory. The processor is coupled to the memory. The memory is used to store computer programs or instructions. The processor is used to execute the computer programs or instructions in the memory, so that the method according to any one of claims 1 to 8 is executed.

10. An infant bathing device, characterized in that, The baby bathing device includes the control device according to claim 9.

Images