Weight detection method, model training method and related device

By obtaining the stress information of multiple positions on the bearing device and inputting it into the pre-trained weight prediction model, the problem of low weight detection accuracy in non-fixed postures is solved, and accurate weight measurement in daily use scenarios is achieved.

CN114528980BActive Publication Date: 2024-11-08NINGBO BABY FIRST BABY PROD CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210147214.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-02-17
Publication Date
2024-11-08
Estimated Expiration
2042-02-17

AI Technical Summary

Technical Problem

When used in non-fixed postures, it is difficult to accurately measure weight, especially in daily load-bearing devices such as seats and beds, due to posture changes and pressure dispersion, the measurement accuracy is not ideal.

Method used

By obtaining the stress information of the bearing device at multiple positions, input it into the pre-trained weight prediction model, the model is used to discover the constraint relationship between the pressure information and the real weight, and accurately perform weight detection without restricting the user's posture .

Benefits of technology

When using the load-bearing device in daily life, the target user's real weight can be accurately obtained, the accuracy and convenience of weight measurement can be improved, and it is suitable for a variety of scenarios such as driving, working or child safety seats.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114528980B_ABST
    Figure CN114528980B_ABST
Patent Text Reader

Abstract

In the weight detection method, model training method and related devices provided by the present application, the weight detection equipment obtains the force information of the carrying device at multiple positions when the carrying device carries the target user; then, the force information at the multiple positions is input into a pre-trained weight prediction model to obtain the predicted weight of the target user; finally, the real weight of the target user is obtained based on the predicted weight; because the pre-trained weight prediction model discovers the constraint relationship between the force information at multiple positions and the real weight, the real weight of the target user can be obtained during the target user's daily use of the carrying device.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of weight detection, and specifically, to a weight detection method, a model training method and related devices. Background Art

[0002] Changes in human body weight can reflect the health status of the body to a certain extent. Therefore, people often use relatively portable electronic scales to measure their weight in life. However, due to factors such as busy work and trivial matters in life, it is difficult for people to develop the habit of frequently using electronic scales to measure their weight. Therefore, people are in urgent need of a way to measure their weight by using daily carrying devices.

[0003] Research has found that weight measurement devices such as electronic scales mostly measure weight based on the mapping relationship between the resistance of a strain gauge and pressure. Therefore, there are strict requirements on the user's posture when measuring weight, otherwise accurate weight cannot be measured. However, it is difficult to constrain people's posture when using a load-bearing device. Therefore, when measuring weight through a load-bearing device, it is difficult to adopt a principle similar to that of traditional weight measurement equipment. Summary of the invention

[0004] In order to overcome at least one of the deficiencies in the prior art, the present application provides a weight detection method, a model training method and related devices, including:

[0005] In a first aspect, the present application provides a weight detection method, which is applied to a weight detection device, and the method comprises:

[0006] Obtaining force information of the bearing device at multiple positions when the bearing device carries the target user;

[0007] Inputting the force information of the bearing device at multiple positions into a pre-trained weight prediction model to obtain the predicted weight of the target user;

[0008] The actual weight of the target user is obtained based on the predicted weight.

[0009] In a second aspect, the present application provides a model training method, which is applied to a model training device, and the method comprises:

[0010] Acquiring sample force information of the carrying device at multiple positions when the carrying device carries a sample object;

[0011] The actual weight of the sample object is used as reference information, and the neural network model is trained through the sample force information to obtain a trained weight prediction model.

[0012] In a third aspect, the present application provides a weight detection device, which is applied to a weight detection device, and the weight detection device comprises:

[0013] A weight factor module, used to obtain force information of the carrying device at multiple positions when the carrying device carries the target user;

[0014] A weight prediction module, used to input the force information of the bearing device at multiple positions into a pre-trained weight prediction model to obtain the predicted weight of the target user;

[0015] The weight calculation module is used to obtain the actual weight of the target user according to the predicted weight.

[0016] In a fourth aspect, the present application provides a weight detection device, which includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the weight detection method is implemented.

[0017] In a fifth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the weight detection method is implemented.

[0018] Compared with the prior art, this application has the following beneficial effects:

[0019] In the weight detection method, model training method and related devices provided by the present application, the weight detection equipment obtains the force information of the carrying device at multiple positions when the carrying device carries the target user; then, the force information at the multiple positions is input into a pre-trained weight prediction model to obtain the predicted weight of the target user; finally, the real weight of the target user is obtained based on the predicted weight; because the pre-trained weight prediction model discovers the constraint relationship between the force information at multiple positions and the real weight, the real weight of the target user can be obtained during the target user's daily use of the carrying device. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 A schematic diagram of the structure of a weight detection device provided in an embodiment of the present application;

[0022] Figure 2A flow chart of a weight detection method provided in an embodiment of the present application;

[0023] Figure 3 A schematic diagram of the interaction between a user terminal and a safety seat provided in an embodiment of the present application;

[0024] Figure 4 A schematic diagram of the structure of a neural network model provided in an embodiment of the present application;

[0025] Figure 5 This is a schematic diagram of the structure of a weight detection device provided in an embodiment of the present application.

[0026] Icons: 120 - memory; 130 - processor; 140 - communication unit; 201 - weight factor module; 202 - weight prediction module; 203 - weight calculation module. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0028] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0029] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0030] Due to busy work and daily life, it is difficult for people to develop the habit of using electronic scales to measure their weight. Therefore, people are in urgent need of a way to measure their weight using a daily carrying device, including a seat, a bed, a sofa, a crib, a stroller, a cradle, etc.

[0031] For example, people hope to measure their weight while driving a vehicle; or, people hope to measure their weight when using a seat to work for a long time; or, people hope to measure the weight of a child when placing the child in a crib, stroller, or cradle.

[0032] However, studies have found that it is difficult for people to maintain a fixed body posture when using a load-bearing device, resulting in the body weight not being concentrated in one direction, but being dispersed in various directions as people's posture changes. In addition, since weight measuring devices such as electronic scales mostly measure weight based on the mapping relationship between the resistance and pressure of strain gauges, they have strict requirements on people's body posture when measuring weight. Therefore, traditional weight measurement methods cannot measure people's accurate weight in the above scenarios.

[0033] For example, in the design process of child safety seats, comfort and safety need to be considered. Therefore, the child restraint system is specially designed for children's body shape, so that the cockpit angle of child seats is often large, sometimes up to 30 degrees, and the sitting posture of children in the seats is also different. Therefore, the collection and analysis of pressure data is affected, that is, the pressure at the bottom of the seat cannot represent the exact weight; therefore, the results measured by the strain gauge are greatly affected by the sitting posture of the child, resulting in less than ideal weight measurement precision and accuracy.

[0034] It should be noted that based on the discovery of the above technical problems, the inventors have proposed the following technical solutions to solve or improve the above problems through creative work. It should be noted that the defects in the solutions in the above prior art are the results obtained by the inventors after practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed in the embodiments of the present application for the above problems below should all be the contributions made by the inventors to the present application in the process of invention and creation, and should not be understood as technical contents known to those skilled in the art.

[0035] In view of the above problems, the inventors further studied and found that when people use the load-bearing device, multiple positions of the load-bearing device will bear pressure from the body, and there is a specific constraint relationship between the pressure information of the multiple positions and the actual weight; or, there is a specific constraint relationship between the pressure information of the multiple positions and the inclination information of the load-bearing device and the actual weight. For example, assuming that the load-bearing device is a seat, the pressure information of the multiple positions may include a first pressure on the bottom of the seat and a second pressure on the backrest of the seat, wherein, when the weight remains unchanged, the first pressure and the second pressure are affected by the sitting posture of the target user on the one hand, and are related to the inclination angle of the cabin on the other hand.

[0036] Therefore, this embodiment provides a weight detection method applied to a weight detection device. In this method, the weight detection device obtains the pressure information of the carrying device at multiple positions when the carrying device carries the target user; then, the pressure information at multiple positions is input into a pre-trained weight prediction model to obtain the predicted weight of the target user; finally, the real weight of the target user is obtained based on the predicted weight; because the pre-trained weight prediction model discovers the specific constraint relationship between the pressure information at multiple positions and the real weight, the real weight of the target user can be obtained during the target user's daily use of the carrying device.

[0037] In some embodiments, the weight detection device may be a user terminal that can be communicatively connected to the carrier. In some embodiments, the user terminal may be, but is not limited to, a mobile terminal, a tablet computer, a laptop computer, or a built-in device in a motor vehicle, or any combination thereof. In some embodiments, the mobile terminal may include a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, or any combination thereof. In some embodiments, the wearable device may include a smart bracelet, a smart shoelace, smart glass, a smart helmet, a smart watch, smart clothing, a smart backpack, a smart accessory, or any combination thereof. In some embodiments, the smart mobile device may include a smart phone, a personal digital assistant (PDA), a gaming device, a navigation device, or any combination thereof.

[0038] In other embodiments, the weight detection device may also be a carrier device including an embedded device, for example, the weight detection device may be a child safety seat, an ergonomic seat, etc. including an embedded device. The embedded device is used to deploy the above-mentioned weight prediction model, and collect the pressure on the seat in multiple directions and / or the inclination of the seat through corresponding sensors to detect the weight of the target user.

[0039] Therefore, if Figure 1 As shown, the weight detection device may include a memory 120, a processor 130, and a communication unit 140. The memory 120, the processor 130, and the communication unit 140 are electrically connected to each other directly or indirectly to achieve data transmission or interaction. For example, these components may be electrically connected to each other via one or more communication buses or signal lines.

[0040] The memory 120 may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electric erasable programmable read-only memory (EEPROM), etc. The memory 120 is used to store a program, and the processor 130 executes the program after receiving an execution instruction.

[0041] The communication unit 140 is used to establish a communication connection between the server and the user terminal through a network, and to send and receive data through the network. The network may include a wired network, a wireless network, an optical fiber network, a telecommunication network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a wide area network (WAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network, or a near field communication (NFC) network, or any combination thereof. In some embodiments, the network may include one or more network access points. For example, the network may include a wired or wireless network access point, such as a base station and / or a network switching node, through which one or more components of the service request processing system may be connected to the network to exchange data and / or information.

[0042] The processor 130 may be an integrated circuit chip having a signal processing capability, and the processor may include one or more processing cores (e.g., a single-core processor or a multi-core processor). By way of example only, the processor may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), an application-specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic device (PLD), a controller, a microcontroller unit, a reduced instruction set computer (RISC), or a microprocessor, or any combination thereof.

[0043] Based on the above introduction, Figure 2 The weight detection method in this embodiment is described in detail. It should be understood that the operations in the flowchart in this embodiment can be implemented in a non-sequential manner, and steps without logical contextual relationships can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart, or remove one or more operations from the flowchart under the guidance of the content of this application. Figure 2 As shown, the method includes:

[0044] S101, obtaining force information of a bearing device at multiple positions when the bearing device carries a target user.

[0045] The multiple positions of the support device are related to the type of the support device. For example, when the support device is a seat, the multiple positions of the seat may be the seat bottom and the back. For another example, when the support device is a crib, the multiple positions of the crib may be the positions for placing the child's head, limbs and torso in the crib. In addition, the force information of each position may include information such as the force magnitude and force direction of the position.

[0046] S102, inputting the force information of multiple positions into a pre-trained weight prediction model to obtain the predicted weight of the target user.

[0047] Exemplarily, assuming that the load-bearing device is the above-mentioned seat, and the force information of the seat at multiple positions includes the first pressure on the seat bottom and the second pressure on the seat back. Then the weight detection device uses the first pressure and the second pressure as input features of the weight prediction model and inputs them into the model; after the weight prediction model calculates the input features, the output result is the predicted weight of the target user.

[0048] Of course, the first pressure and the second pressure in the above example are merely examples provided to facilitate technical personnel in the field to understand this implementation. When actually implementing this solution, it is not limited to the first pressure and the second pressure, and the number of pressures can be appropriately increased based on this embodiment as needed.

[0049] S103, obtaining the actual weight of the target user according to the predicted weight.

[0050] Therefore, since the pre-trained weight prediction model has discovered the specific constraint relationship between the pressure information of the load-bearing device at multiple positions and the actual weight, the actual weight of the target user can be obtained by using the weight prediction model without restricting the target user's body posture when using the load-bearing device.

[0051] As introduced in the above embodiment, assuming that the supporting device is a seat, the pressure information of the above multiple positions may include a first pressure on the bottom of the seat and a second pressure on the seat back, wherein, when the body weight remains unchanged, the first pressure and the second pressure are affected by the sitting posture of the target user on the one hand, and are related to the inclination angle of the cabin on the other hand.

[0052] Therefore, for a load-bearing device with an inclination angle, the weight detection device can also obtain the force information of the load-bearing device at multiple positions and the inclination information of the load-bearing device when the load-bearing device carries the target user; then, the force information of the load-bearing device at multiple positions and the inclination information of the load-bearing device are input into the weight prediction model to obtain the predicted weight of the target user.

[0053] The inclination information package may include the magnitude and direction of the inclination. Thus, for a load-bearing device with an inclination, the prediction accuracy of the target user's weight can be improved by introducing the inclination information of the load-bearing device.

[0054] In order to obtain the actual weight of the target user through the predicted weight, in an optional implementation, the weight detection device can directly use the predicted weight as the actual weight of the target user.

[0055] Research has found that the target user's weight over a period of time will be affected by factors such as diet, clothing, and acceleration when using a seat. Therefore, in order to overcome the above-mentioned interference factors, in another optional implementation, the weight detection device obtains multiple predicted weights, wherein the multiple predicted weights correspond to different collection times; based on the multiple predicted weights, the target user's actual weight is calculated.

[0056] Exemplarily, the weight detection device obtains the predicted weight of the target user at multiple times during the day, then calculates the average weight corresponding to the predicted weights at multiple times, and finally uses the average weight as the actual weight of the target user on that day.

[0057] Alternatively, the weight detection device obtains the predicted weight of the target user at multiple times during the day, then selects the target predicted weight that appears most frequently from the predicted weights at multiple times, and finally uses the target predicted weight as the actual weight of the target user on that day.

[0058] It can be seen from the above embodiments that the load-bearing devices in this embodiment include seats, beds, sofas, cribs, strollers, cradles, etc., and the accelerations generated by some of the load-bearing devices when in use will interfere with the force information of the load-bearing devices at multiple positions. For example, when the load-bearing device is a child safety seat fixed to a motor vehicle, the corresponding target user can be a child of appropriate age. The study also found that when the seat is a child safety seat, the acceleration generated during the driving of the motor vehicle seriously affects the measurement results of the child's weight; for example, when the load-bearing device is a cart, the cart will also generate accelerations that interfere with the force information during the pushing process.

[0059] Therefore, the interference introduced by motor vehicles is reduced. Figure 2 Step S101 in may include the following implementations:

[0060] S101-1, obtaining the acceleration carried by the carrying device when the carrying device carries the target user.

[0061] S101-2: If the acceleration meets the acquisition condition, the force information of the bearing device at multiple positions is obtained.

[0062] For example, continuing to take a child safety seat as an example, the collection condition can be that the acceleration is less than the acceleration threshold, that is, the child safety seat is equipped with an acceleration sensor for collecting the acceleration generated by the child safety seat when the motor vehicle is running, and only when the acceleration is less than the acceleration threshold, the force information of the child safety seat at multiple positions is collected; thereby reducing the interference of acceleration on the child's weight. Of course, for a child safety seat with a large inclination angle, the inclination information of the child safety seat can also be collected when collecting the force information at multiple positions to further improve the prediction accuracy provided for the child.

[0063] In addition, it should be understood that a child safety seat is a child safety protection system composed of a safety belt assembly with buckles or flexible components, an adjustment mechanism, accessories, etc. It can be combined with additional devices such as a portable crib, an infant carrier, an auxiliary seat or a collision protection device, etc. In the event of a collision or sudden deceleration of a motor vehicle, the impact force on the child is reduced and the child's body movement is restricted to reduce injuries, thereby ensuring the child's safety in the vehicle.

[0064] Therefore, in order to protect the safety of children in vehicles, the relevant standards have made relevant provisions on the way children are restrained in seats, where the restraint method includes the installation method of the child seat and the restraint method of the seat belt. For example, when the child weighs 0 to 9 kg, the child seat needs to be installed in reverse; when the child weighs more than 9 kg, the child seat needs to be installed forward, and the restraint method of the seat belt needs to be adjusted to a five-point restraint. However, when the child safety seat is actually used, the user of the child seat (for example, the child's guardian) does not always pay attention to the child's weight changes and is prone to forget to adjust the child's restraint method in the child safety seat in time. In view of this, in an optional implementation, the weight detection method also includes the following implementations:

[0065] S104, comparing the actual weight with the weight threshold.

[0066] S105A: If the actual weight is greater than the weight threshold, provide instruction information to the user of the child seat.

[0067] The instruction information is used to instruct the user of the child seat to adjust the restraint method of the target user in the child safety seat.

[0068] For example, Figure 3As shown, assuming that the weight detection device is the user terminal in the above embodiment, when the user terminal is connected to the child safety seat through communication, and detects that the child's weight is greater than a preset weight threshold, an instruction message is provided to the user of the child seat. In some embodiments, the instruction message may be related text information displayed by the user terminal in the display interface to prompt the user of the child seat to adjust the restraint method. For example, a text message such as "xx's weight has exceeded 9kg, please adjust the restraint method of the child safety seat in time" is displayed. In other embodiments, the user terminal may also guide the user of the child seat to adjust the restraint method of the child seat in the form of pictures or videos while displaying the text instruction message.

[0069] In addition, in other optional embodiments, the weight detection method also includes the following embodiments:

[0070] S105B, if the actual weight is greater than the weight threshold, the electrical control system of the child safety seat adjusts the restraint mode in the child safety seat to match the actual weight.

[0071] Exemplarily, the user terminal in the above embodiment is used as an example. For some child safety seats with electric drive, after a communication connection is established between the user terminal and the child seat, if it is detected that the weight of the child is greater than a preset weight threshold, a control instruction can be sent to the child safety seat, so that the child safety seat responds to the control instruction and automatically adjusts the restraint mode of the child safety seat so that the restraint mode of the child safety seat matches the actual weight of the child, wherein the drive mode includes rotation adjustment, gear adjustment, etc.

[0072] Therefore, through the above implementation, after the child's weight exceeds the set weight threshold, the user of the child seat is promptly reminded to adjust the restraining method of the child seat, or the electrical control system of the child safety seat is automatically controlled to adjust the restraining method of the target user in the child safety seat, thereby ensuring the child's safety in the vehicle.

[0073] The study also found that when a child's weight just exceeds 9kg, the child's weight is usually difficult to stabilize at 9kg, but will fluctuate around 9kg; therefore, the weight detection equipment will repeatedly prompt the user to adjust the restraint method of the child safety seat.

[0074] In view of this, the weight threshold is the sum of the reference weight and the confidence parameter, and the confidence parameter is used to overcome the fluctuation of the actual weight. Among them, the relationship between the weight threshold, the reference weight and the confidence parameter can be expressed as:

[0075] G thr =9+σ

[0076] In the formula, Gthr represents the weight threshold, 9 represents the reference weight, and σ represents the confidence parameter. Those skilled in the art may adjust the confidence parameter appropriately as needed, and the confidence parameter in this embodiment may be 0.5.

[0077] In this embodiment, in order to obtain the weight prediction model in the above embodiment, a model training method applied to a model training device is also provided. In this method, the model training device obtains sample pressure information of the carrying device at multiple positions when the carrying device carries a sample object; then, the real weight of the sample object is used as reference information, and the neural network model is trained through the sample pressure information at multiple positions to obtain a trained weight prediction model.

[0078] The sample object changes with the type of seat. For example, if the seat is an ordinary seat, the sample object may be adults of different weights, and the sample pressure information at multiple positions may be the pressure information at multiple positions of the seat when adults of different weights use the seat.

[0079] For another example, if the seat is a child safety seat, the sample objects may be children of different weights, and the sample pressure information at multiple positions may be the pressure information at multiple positions of the child safety seat when children of different weights use the child seat.

[0080] Of course, for a load-bearing device with an inclination angle, when training and providing a prediction model, in addition to obtaining the sample pressure information of the load-bearing device at multiple positions when the load-bearing device carries a sample object, it is also necessary to obtain the sample inclination information of the load-bearing device; then, the actual weight of the sample object is used as reference information, and the neural network model is trained through the sample pressure information and sample inclination information at multiple positions to obtain a trained weight prediction model.

[0081] It should also be understood that the selection of the neural network model is directly related to the prediction effect of the final weight prediction model. Therefore, this embodiment constructs the neural network model and trains the neural network model in the following manner.

[0082] Number of neural network layers: Since the task of this embodiment is a regression task based on a multi-layer neural network, when the task is not related to the time series, a regression task of general or low complexity only requires two layers of neural networks to complete, that is, a neural network model with only one hidden layer.

[0083] Number of neurons: Considering that the number of neurons in the hidden layer should not be too large, otherwise it is likely to lead to overfitting, therefore, the empirical formula can be used Determine the number of neurons.

[0084] Where Nh Represents the number of neurons in the hidden layer, N i With N o represents the number of neurons in the input layer and the output layer respectively, and α is a regulation constant between 1 and 10, which can be adjusted by a person skilled in the art according to actual needs. For example, for a seat with an inclination angle, the number of neurons in the input layer and the hidden layer can be selected as 3, wherein the 3 neurons in the input layer are used to input the pressure information on the bottom of the seat, the pressure information on the back of the seat, and the inclination angle information of the seat respectively.

[0085] Activation function: Since the regression task can decide whether to use the activation function as needed. In this embodiment, the Relu function is used as the activation function of the hidden layer. The expression using the Relu function is:

[0086]

[0087] Loss function: The loss function is used to define the quality of the training results. This embodiment uses the mean square error (MSE) as the loss function. The expression of the mean square error (MSE) is:

[0088]

[0089] In the formula, y and Represent the actual output and predicted output of the neural network model respectively.

[0090] Learning rate and batch size: The learning rate and batch size affect the training efficiency during model training. They need to be continuously optimized according to the task to keep the loss curve sliding downward.

[0091] Through the above related introduction, the neural network model in this embodiment has the following results: Figure 4 As shown, the mathematical expression of the neural network model is:

[0092]

[0093] a j =ReLu(h j )(j=1,2,3)

[0094]

[0095] In the formula, the features of the three input units are represented as x, the three hidden neurons are h and the output unit is y. The weight of the first network layer is represented as:

[0096]

[0097] The bias of the first network layer is expressed as:

[0098]

[0099] The weights of the second network layer are expressed as:

[0100]

[0101] The bias of the second network layer is expressed as:

[0102]

[0103] Therefore, according to the above expression, the output of the neural network model is The relationship between weights and biases, so the loss function L is a multivariate function containing multiple weights and biases. The weights and biases are expressed in matrices as follows:

[0104] L(W1,W2,B1,B2)

[0105] Based on the loss function, the weights of the neural network model are adjusted through the reverse gradient propagation algorithm, where the back propagation process of the reverse gradient propagation algorithm satisfies the chain rule. As an example, the corresponding chain rule can be expressed as:

[0106]

[0107] According to the mathematical expression of the above neural network model, the predicted output can be obtained and and h 1 The expression between:

[0108]

[0109]

[0110] According to the expression of loss function L, we can get the predicted output The partial derivative of :

[0111]

[0112] as well as About h 1 The partial derivative of h 1 about The partial derivative of :

[0113]

[0114]

[0115] Based on the chain rule, the loss function L is related to the weight The partial derivative function is:

[0116]

[0117] Therefore, given the initialized weight and bias matrices, the partial derivatives can be calculated The value is used to increase or decrease This will cause a slight change in the loss L.

[0118] The partial derivatives of all weights and biases can be obtained in the above way. It is worth noting that the derivative of the Relu function here can be expressed as:

[0119]

[0120] Assuming that the learning rate of the neural network model is represented by η, the Stochastic Gradient Descent (SGD) optimization algorithm is used to train the network. SGD defines a method for changing weights and biases:

[0121]

[0122] From the above expression, we can see that when When it is a positive number, the weight After adjustment, it will become smaller; when When it is a negative number, After adjustment, it will become larger. After multiple rounds of training, the weight matrix and bias matrix of the network are adjusted so that the loss function L will gradually decrease, thereby obtaining a weight prediction model that meets the training requirements.

[0123] In some embedded devices, not only the deployment of the model is supported, but also the computing power requirements for training the model are met. In this case, the neural network model in this embodiment can be placed in the embedded device for training. For example, since the child safety seat including the embedded device is trained by itself. In the case of hardware limitations, the neural network model can also be deployed to the user terminal or server for training.

[0124] Based on the same inventive concept as the weight detection method, this embodiment also provides a weight detection device, which is applied to a weight detection device. The weight detection device includes at least one functional module that can be stored in a memory in the form of software. Figure 5 As shown, from the functional point of view, the weight detection device may include:

[0125] The weight factor module 201 is used to obtain force information of the carrying device at multiple positions when the carrying device carries the target user.

[0126] In this embodiment, the weight factor module 201 is used to implement Figure 2 For the detailed description of the weight factor module 201, please refer to the detailed description of step S101.

[0127] The weight prediction module 202 is used to input the force information of multiple positions into a pre-trained weight prediction model to obtain the predicted weight of the target user.

[0128] In this embodiment, the weight prediction module 202 is used to implement Figure 2 For the detailed description of the weight prediction module 202, please refer to the detailed description of step S102.

[0129] The weight calculation module 203 is used to obtain the actual weight of the target user according to the predicted weight.

[0130] In this embodiment, the weight calculation module 203 is used to implement Figure 3 For the detailed description of the weight calculation module 203, please refer to the detailed description of step S103.

[0131] It is worth noting that the present embodiment does not limit the number of software function modules included in the weight detection device, and those skilled in the art can divide them according to different module division standards. Therefore, in some embodiments, the weight detection device may also include other software function modules for implementing other steps or sub-steps of the weight detection method. Of course, in some other embodiments, the above-mentioned weight factor module 201, weight prediction module 202 and weight calculation module 203 can also be used to implement other steps or sub-steps of the weight detection method.

[0132] This embodiment also provides a weight detection device, which includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the weight detection method is implemented.

[0133] This embodiment also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the weight detection method is implemented.

[0134] It should be noted that in the above description of the present embodiment, the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. In addition, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "include one..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0135] It should also be understood that the device and method disclosed in this embodiment can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0136] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0137] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0138] The above are only various implementations of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A weight detection method, characterized in that: Applied to weight detection equipment, the method comprises: Acquiring force information of a child safety seat at multiple positions when the child safety seat carries a target user, the force information at the multiple positions including a first pressure borne by a base of the seat and a second pressure borne by a backrest of the seat; Inputting the force information of the child safety seat at multiple positions into a pre-trained weight prediction model to obtain the predicted weight of the target user; wherein the model training device obtains sample force information of the child safety seat at multiple positions when the child safety seat carries a sample object; using the actual weight of the sample object as reference information, training the neural network model through the sample force information at multiple positions to obtain the trained weight prediction model; Acquire a plurality of the predicted weights, wherein the plurality of the predicted weights correspond to different collection times respectively; Calculating the actual weight of the target user based on the multiple predicted weights; comparing the actual weight to a weight threshold; If the actual weight is greater than the weight threshold, providing instruction information to the user of the child seat, the instruction information being used to instruct the user of the child seat to adjust the restraint method of the target user in the child safety seat; Alternatively, if the actual weight is greater than the weight threshold, the electrical control system controlling the child safety seat adjusts the restraint mode in the child safety seat to match the actual weight.

2. The weight detection method according to claim 1, characterized in that: The obtaining of force information of the child safety seat at multiple positions when the child safety seat carries a target user includes: acquiring an acceleration of the child safety seat when the child safety seat carries a target user; If the acceleration meets the collection condition, the force information of the child safety seat at multiple positions is obtained.

3. The weight detection method according to claim 1, characterized in that: The weight threshold is the sum of the reference weight and a confidence parameter, and the confidence parameter is used to overcome the fluctuation of the actual weight.

4. The weight detection method according to claim 1, characterized in that: The obtaining of force information of the child safety seat at multiple positions when the child safety seat carries a target user includes: Acquiring force information of the child safety seat at multiple positions and inclination information of the child safety seat; The step of inputting the force information of the child safety seat at multiple positions into a pre-trained weight prediction model to obtain the predicted weight of the target user includes: The force information of the child safety seat at multiple positions and the inclination information of the child safety seat are input into the weight prediction model to obtain the predicted weight of the target user.

5. A weight detection device, characterized in that: Applied to weight detection equipment, the weight detection device comprises: A weight factor module, used to obtain force information of a child safety seat at multiple positions when the child safety seat carries a target user, wherein the force information of the multiple positions includes a first pressure borne by a base of the seat and a second pressure borne by a backrest of the seat; A weight prediction module, used for inputting the force information of the child safety seat at multiple positions into a pre-trained weight prediction model to obtain the predicted weight of the target user, wherein the model training device obtains the sample force information of the child safety seat at multiple positions when the child safety seat carries a sample object; uses the real weight of the sample object as reference information, trains the neural network model through the sample force information at multiple positions, and obtains the trained weight prediction model; A weight calculation module, used to obtain a plurality of predicted weights, wherein the plurality of predicted weights correspond to different collection times; Calculating the actual weight of the target user based on the multiple predicted weights; The weight detection device is further used to compare the actual weight with a weight threshold; If the actual weight is greater than the weight threshold, providing instruction information to the user of the child seat, the instruction information being used to instruct the user of the child seat to adjust the restraint method of the target user in the child safety seat; Alternatively, if the actual weight is greater than the weight threshold, the electrical control system controlling the child safety seat adjusts the restraint mode in the child safety seat to match the actual weight.

6. A weight detection device, characterized in that: The weight detection device includes a processor and a memory, the memory stores a computer program, and when the computer program is executed by the processor, the weight detection method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the weight detection method according to any one of claims 1 to 4 is implemented.