A multi-dimensional force sensor decoupling method based on an empirical formula assisted neural network

Abstract

The application is suitable for the technical field of data processing, and provides a multi-dimensional force sensor decoupling method based on an experience formula assisted neural network, which comprises the following steps: obtaining strain response dataset information of a target sensor, determining an experience formula based on the strain response dataset information, generating a target decoupling model based on the experience formula, and inputting output voltage signals of a sensor to be decoupled into the target decoupling model to reliably output decoupled multi-dimensional load information. The application can integrate a physical experience formula as prior knowledge into a network model, so that the model obtains physical law constraints in the training process, effectively reduces the dependence on the number of calibration samples, and alleviates the overfitting risk under the condition of a small sample. Meanwhile, the introduction of the prior knowledge can guide the network parameters to converge to the global optimal solution, avoid falling into a local extreme value, and thus improve the decoupling precision and the calculation efficiency, and finally enhance the measurement reliability and the engineering applicability of the sensor under the condition of multiple working conditions.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and more specifically, to a decoupling method for a multidimensional force sensor based on an empirical formula-assisted neural network. Background Technology

[0002] In the decoupling of multidimensional force sensors with complex structures, traditional matrix decoupling algorithms often fail to meet the requirements for high-precision decoupling. In recent years, neural network algorithms, with their powerful nonlinear mapping and generalization capabilities, have become the mainstream multidimensional force decoupling method and have achieved relatively ideal decoupling results.

[0003] However, existing neural network algorithms still have significant limitations in practical applications: on the one hand, model training heavily relies on a large amount of calibration data to ensure decoupling performance, and obtaining a large amount of data through sensor physical calibration experiments is not only time-consuming and cumbersome, but also costly; on the other hand, these algorithms mostly adopt iterative optimization strategies based on gradient descent, which are prone to getting trapped in local optima during training, resulting in limited model convergence accuracy and difficulty in ensuring the stability and reliability of decoupling results, thus exhibiting poor reliability and requiring further improvement. Summary of the Invention

[0004] Based on this, this application provides a decoupling method for a multidimensional force sensor based on an empirical formula-assisted neural network to solve the problem of poor reliability in the prior art.

[0005] In a first aspect, embodiments of this application provide a decoupling method for a multidimensional force sensor based on an empirical formula-assisted neural network, the method comprising: Based on a pre-defined sensor structure model, obtain the strain response dataset information of the target sensor; Based on the strain response dataset information, an empirical formula is determined, wherein the empirical formula is used to characterize the relationship between the strain of the target sensor and the actual force. An initial neural network model is constructed, and training dataset information is constructed. The initial neural network model includes a front-end decoupling network and a back-end empirical parameter network connected in sequence. The network parameters of the back-end empirical parameter network are initialized according to the parameters in the empirical formula. Based on the training dataset information, the initial neural network model is trained in stages to generate the target decoupling model; The output voltage signal of the sensor to be decoupled is input into the target decoupling model, and the decoupled multidimensional load information is output.

[0006] Compared with existing technologies, the beneficial effects are as follows: The multi-dimensional force sensor decoupling method based on empirical formula-assisted neural network provided in this application allows the terminal device to first obtain the strain response dataset information of the target sensor based on a preset sensor structure model. Then, based on the strain response dataset information, the empirical formula is effectively determined, an initial neural network model is constructed, and training dataset information is built. Then, based on the training dataset information, the initial neural network model is trained in stages to quickly generate the target decoupling model. Finally, the output voltage signal of the sensor to be decoupled is input to the target decoupling model to reliably output the decoupled multi-dimensional load information. This realizes the integration of physical empirical formulas as prior knowledge into the network, effectively reducing the dependence on the amount of calibration data, guiding the model to converge to the global optimal solution, improving decoupling accuracy and efficiency, significantly improving reliability, and solving the problem of poor reliability to a certain extent.

[0007] Secondly, embodiments of this application provide a multidimensional force sensor decoupling system based on an empirical formula-assisted neural network, the system comprising: Strain response dataset information acquisition module: used to acquire strain response dataset information of the target sensor based on a preset sensor structure model; Empirical formula determination module: used to determine an empirical formula based on the strain response dataset information, wherein the empirical formula is used to characterize the relationship between the strain of the target sensor and the actual force. Initial Neural Network Model Construction Module: Used to construct an initial neural network model and to construct training dataset information. The initial neural network model includes a front-end decoupling network and a back-end empirical parameter network connected in sequence. The network parameters of the back-end empirical parameter network are initialized according to the parameters in the empirical formula. Target decoupling model generation module: used to perform phased training processing on the initial neural network model based on the training dataset information to generate a target decoupling model; Multidimensional load information output module: used to input the output voltage signal of the sensor to be decoupled to the target decoupling model and output the decoupled multidimensional load information.

[0008] Thirdly, embodiments of this application provide a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described in the first aspect above.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described in the first aspect above.

[0010] It is understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0012] Figure 1 This is a flowchart illustrating a multidimensional force sensor decoupling method provided in an embodiment of this application; Figure 2 This is a schematic diagram of a sensor structure model provided in an embodiment of this application; Figure 3 This is a schematic diagram of an initial neural network model provided in an embodiment of this application; Figure 4 This is a flowchart illustrating step S400 in a multidimensional force sensor decoupling method provided in an embodiment of this application; Figure 5 This is a block diagram of a multi-dimensional force sensor decoupling system provided in an embodiment of this application; Figure 6 This is a schematic diagram of a terminal device provided in an embodiment of this application. Detailed Implementation

[0013] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0014] In the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0015] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0016] To illustrate the technical solution described in this application, specific embodiments are provided below.

[0017] Please see Figure 1 , Figure 1 This is a flowchart illustrating the multidimensional force sensor decoupling method based on an empirical formula-assisted neural network provided in this embodiment. In this embodiment, the execution subject of the multidimensional force sensor decoupling method is a terminal device. It is understood that the types of terminal devices include, but are not limited to, tablet computers, laptops, ultra-mobile personal computers (UMPCs), netbooks, personal digital assistants (PDAs), etc., and this embodiment does not impose any restrictions on the specific type of terminal device.

[0018] Please see Figure 1 The multi-dimensional force sensor decoupling method provided in this application includes, but is not limited to, the following steps: In S100, based on a preset sensor structure model, the strain response dataset information of the target sensor is obtained.

[0019] Specifically, the terminal device can effectively acquire the strain response dataset information of the target sensor based on the preset sensor structure model. The sensor structure model is used to describe the three-dimensional structural model constructed for the multi-dimensional force sensor. The multi-dimensional force sensor can be a three-beam sensor, and the target sensor is used to describe the multi-dimensional force sensor to be decoupled.

[0020] In some possible implementations, in order to obtain strain response dataset information, step S100 includes, but is not limited to, the following steps: In S110, multiple sets of force / torque loads of different directions and magnitudes are applied to the preset sensor structure model to obtain the strain response dataset information of the target sensor.

[0021] For example, please refer to Figure 2 The terminal device can apply multiple sets of force / torque loads of different directions and magnitudes to a preset sensor structure model, perform finite element simulation and analysis, and obtain strain response dataset information of the target sensor. The strain response dataset information includes multiple strain response data information, which are used to describe the strain value information corresponding to an application position in the sensor structure model. The strain value information can be finite element results.

[0022] In S200, empirical formulas are determined based on information from the strain response dataset.

[0023] Specifically, after the terminal device acquires the strain response dataset information, it can quickly determine the empirical formula based on the strain response dataset information. The empirical formula is used to characterize the relationship between the strain of the target sensor and the actual force.

[0024] In some possible implementations, the above empirical formula can be: , In the formula, This represents the force component acting on the target sensor along the X-axis. This represents the force component acting on the target sensor along the Y-axis. This represents the force component acting on the target sensor along the Z-axis. This represents the torque component acting on the target sensor about the X-axis; This represents the torque component acting on the target sensor about the Y-axis; Let X be the torque component acting on the target sensor about the Z-axis, where the X-axis, Y-axis, and Z-axis are the three coordinate axes of the target sensor's own Cartesian rectangular coordinate system. , , , , and These represent strain values ​​at different measurement locations of the target sensor, for example. These are the strain values ​​on both sides of beam No. 1; The strain value is the value on the side of beam No. 1; These are the strain values ​​on both sides of beam No. 2; The strain value is the value on the side of beam No. 2; These are the strain values ​​on both sides of beam No. 3; The strain value is the value on the side of beam No. 3; , , , , and These are the proportionality coefficients between each strain value and the applied load, for example, The value can be , The value can be , The value can be , The value can be , The value can be , The value can be .

[0025] In one possible implementation, the terminal device can transform the empirical formula into a matrix, which can be: .

[0026] Specifically, since the actual input to the decoupling algorithm model is a voltage signal, rather than the strain value upon which the empirical formula is based, the change in the dimensions of the input physical quantities means that the proportionality coefficient between the strain combination value and the applied load in the original formula is no longer directly applicable. Therefore, in the actual introduced matrix, only parameters other than the proportionality coefficient are retained, so the final matrix is: .

[0027] In S300, an initial neural network model is constructed, and training dataset information is also built.

[0028] Specifically, after the terminal device determines the empirical formula, the terminal device can construct an initial neural network model and construct training dataset information. The initial neural network model includes a front-end decoupling network and a back-end empirical parameter network connected in sequence. The network parameters of the back-end empirical parameter network are initialized according to the parameters in the empirical formula.

[0029] Without loss of generality, the number of input layer nodes of the front-end decoupling network is 6, the number of hidden layer nodes of the front-end decoupling network is 15, the number of output layer nodes of the front-end decoupling network is 6, the input of the back-end empirical parameter network is the output of the front-end decoupling network, and the output of the back-end empirical parameter network is the decoupled multidimensional load information.

[0030] For example, please refer to Figure 3 When the terminal device determines the above Following the matrix, the terminal device can add a 6-6 neural network to the existing 6-15-6 neural network to... The parameters of the matrix are used as initialization parameters for the newly added 6-6 neural network, while the preceding 6-15-6 decoupling network is randomly initialized. For example, the terminal device can apply multidimensional force / torque loads to the target sensor through a force loader and record the voltage outputs of each channel to build a training dataset. The training dataset is used for subsequent model training, where the input of the model is the voltage values ​​of each channel, and the output of the model is the force / torque values ​​acting on the sensor in each direction.

[0031] In S400, the initial neural network model is trained in stages based on the training dataset information to generate the target decoupling model.

[0032] Specifically, after the terminal device constructs the training dataset information, the terminal device can perform phased training on the initial neural network model based on the training dataset information to effectively generate the target decoupling model. The phased training process can adopt a two-stage step-by-step training strategy.

[0033] In some possible implementations, to achieve the decoupling model for generating the objective, please refer to [link / reference]. Figure 4 Step S400 includes, but is not limited to, the following steps: In S410, all parameters of the back-end empirical parameter network are frozen, and only the front-end decoupling network is trained. During the training process, a genetic algorithm is used for global optimization, and the training is iterated for the preset first iteration round.

[0034] Specifically, the terminal device can freeze all parameters of the back-end empirical parameter network and train only the front-end decoupled network. During the training process, a genetic algorithm is used for global optimization, and the training is iteratively trained for the preset first iteration round.

[0035] For example, in the first stage, the terminal device can first freeze the parameters of the back-end 6-6 network, and then use a genetic algorithm to train the front-end 6-15-6 network for 100 rounds to obtain a better initial feature mapping by utilizing global optimization capabilities.

[0036] In S420, the freeze on the back-end empirical parameter network is lifted, and the gradient descent method is used to jointly fine-tune the overall network composed of the front-end decoupling network and the back-end empirical parameter network. The preset second round is iterated to generate the target decoupling model.

[0037] Specifically, after the terminal device has completed the first iteration of training, it can unfreeze the back-end empirical parameter network and use gradient descent to jointly fine-tune the overall network composed of the front-end decoupling network and the back-end empirical parameter network, iterating through the preset second round to effectively generate the target decoupling model.

[0038] For example, in the second stage, the terminal device can unfreeze the back-end network after 100 rounds of training and use gradient descent to perform 50 rounds of joint fine-tuning training on the overall 6-15-6-6 network. This combines the global search advantage of the genetic algorithm with the local fine-tuning capability of the gradient descent method, so that the empirical parameter network is deeply matched to the individual characteristics of the single sensor, and finally outputs a high-precision dedicated decoupled model.

[0039] In the S500, the output voltage signal of the sensor to be decoupled is input to the target decoupling model, and the decoupled multidimensional load information is output.

[0040] Specifically, after the terminal device generates the target decoupling model, the terminal device can input the output voltage signal of the sensor to be decoupled into the target decoupling model and output the decoupled multidimensional load information, thereby ensuring the stability and reliability of the decoupling result and effectively improving reliability.

[0041] It should be noted that this application obtains empirical formulas for sensor strain and stress through finite element simulation. At the same time, an empirical parameter network layer is added after the original neural network, and the parameters in the empirical formula are used to initialize it. During training, a phased strategy is adopted: first, the parameters of the empirical parameter network layer are fixed, and the front-end network is randomly initialized and trained; after reaching a preset number of rounds, the empirical parameter network is unfrozen, and the entire network is jointly trained to fine-tune the parameters to adapt them to the characteristics of a single sensor. In this way, physical empirical formulas are integrated into the network as prior knowledge, which effectively reduces the dependence on the amount of calibration data and guides the model to converge to the global optimum, improving decoupling accuracy and efficiency, and greatly improving reliability.

[0042] The implementation principle of the multidimensional force sensor decoupling method based on empirical formula-assisted neural network in this application embodiment is as follows: The terminal device can first obtain the strain response dataset information of the target sensor based on the preset sensor structure model, and then effectively determine the empirical formula based on the strain response dataset information. Then, an initial neural network model is constructed, and training dataset information is constructed. Then, according to the training dataset information, the initial neural network model is trained in stages to quickly generate the target decoupling model. Finally, the output voltage signal of the sensor to be decoupled is input to the target decoupling model to reliably output the decoupled multidimensional load information, thereby significantly reducing the amount of calibration data and guiding the model to converge to the global optimal solution, improving decoupling accuracy and efficiency, and greatly improving reliability.

[0043] It should be noted that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0044] Embodiments of this application also provide a multidimensional force sensor decoupling system based on an empirical formula-assisted neural network. For ease of explanation, only the parts relevant to this application are shown, such as... Figure 5 As shown, the system 50 includes: Strain response dataset information acquisition module 51: used to acquire strain response dataset information of the target sensor based on a preset sensor structure model; Empirical formula determination module 52: used to determine empirical formulas based on strain response dataset information, wherein the empirical formulas are used to characterize the relationship between the strain of the target sensor and the actual force. Initial Neural Network Model Construction Module 53: Used to construct the initial neural network model and to construct training dataset information. The initial neural network model includes a front-end decoupling network and a back-end empirical parameter network connected in sequence. The network parameters of the back-end empirical parameter network are initialized according to the parameters in the empirical formula. Target decoupling model generation module 54: used to perform phased training processing on the initial neural network model based on the training dataset information, and generate the target decoupling model; Multidimensional load information output module 55: used to input the output voltage signal of the sensor to be decoupled to the target decoupling model and output the decoupled multidimensional load information.

[0045] Optionally, the above-mentioned target decoupling model generation module 54 includes: The first stage training submodule is used to freeze all parameters of the back-end empirical parameter network and train only the front-end decoupling network. During the training process, a genetic algorithm is used for global optimization and iterative training is performed for the preset first iteration round. The second-stage training submodule is used to unfreeze the subsequent empirical parameter network. It employs gradient descent to jointly fine-tune the overall network composed of the preceding decoupling network and the subsequent empirical parameter network, iterating through the preset second round to generate the target decoupling model.

[0046] Optionally, the number of input layer nodes of the front-end decoupling network is 6, the number of hidden layer nodes of the front-end decoupling network is 15, the number of output layer nodes of the front-end decoupling network is 6, the input of the back-end empirical parameter network is the output of the front-end decoupling network, and the output of the back-end empirical parameter network is the decoupled multidimensional load information.

[0047] Optionally, the strain response dataset information acquisition module 51 mentioned above includes: The strain response dataset information acquisition submodule is used to apply multiple sets of force / torque loads of different directions and magnitudes to a preset sensor structure model to acquire the strain response dataset information of the target sensor. The strain response dataset information includes multiple strain response data information, which are used to describe the strain value information corresponding to an application position in the sensor structure model.

[0048] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0049] This application also provides a terminal device, such as... Figure 6 As shown, the terminal device 60 in this embodiment includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When the processor 61 executes the computer program 63, it implements the steps described in the multidimensional force sensor decoupling method embodiment, for example... Figure 1 Steps S100 to S500 are shown; or, when processor 61 executes computer program 63, it implements the functions of each module in the above-described device, for example... Figure 5 The functions of the strain response dataset information acquisition module 51 to the multi-dimensional load information output module 55 are shown.

[0050] The terminal device 60 can be a desktop computer, laptop, handheld computer, or cloud server, etc., and includes, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will understand that... Figure 6 This is merely an example of terminal device 60 and does not constitute a limitation on terminal device 60. It may include more or fewer components than shown, or combine certain components, or different components. For example, terminal device 60 may also include input / output devices, network access devices, buses, etc.

[0051] The processor 61 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.; the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0052] The memory 62 can be an internal storage unit of the terminal device 60, such as the hard disk or memory of the terminal device 60. The memory 62 can also be an external storage device of the terminal device 60, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device 60. Furthermore, the memory 62 can include both internal storage units and external storage devices of the terminal device 60. The memory 62 can also store computer program 63 and other programs and data required by the terminal device 60. The memory 62 can also be used to temporarily store data that has been output or will be output.

[0053] One embodiment of this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0054] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the methods, principles and structures of this application should be covered within the scope of protection of this application.

Claims

1. A decoupling method for a multidimensional force sensor based on an empirical formula-assisted neural network, characterized in that, The method includes: Based on a pre-defined sensor structure model, obtain the strain response dataset information of the target sensor; Based on the strain response dataset information, an empirical formula is determined, wherein the empirical formula is used to characterize the relationship between the strain of the target sensor and the actual force. An initial neural network model is constructed, and training dataset information is constructed. The initial neural network model includes a front-end decoupling network and a back-end empirical parameter network connected in sequence. The network parameters of the back-end empirical parameter network are initialized according to the parameters in the empirical formula. Based on the training dataset information, the initial neural network model is trained in stages to generate the target decoupling model; The output voltage signal of the sensor to be decoupled is input into the target decoupling model, and the decoupled multidimensional load information is output.

2. The method according to claim 1, characterized in that, The empirical formula is: ， In the formula, The force component acting on the target sensor in the X-axis direction. The force component acting on the target sensor in the Y-axis direction. The force component acting on the target sensor in the Z-axis direction. The torque component acting on the target sensor about the X-axis. The torque component acting on the target sensor about the Y-axis. The torque component acting on the target sensor about the Z-axis. , , , , and These represent the strain values ​​at different measurement locations of the target sensor. , , , , and These are the proportionality coefficients between each strain value and the applied load.

3. The method according to claim 1, characterized in that, The step of performing phased training on the initial neural network model based on the training dataset information to generate the target decoupling model includes: Freeze all parameters of the back-end empirical parameter network and train only the front-end decoupling network. During the training process, a genetic algorithm is used for global optimization and iterative training is performed for a preset number of first iteration rounds. Unfreeze the latter empirical parameter network, and use gradient descent to jointly fine-tune the overall network composed of the former decoupling network and the latter empirical parameter network. Iterate through the preset second round to generate the target decoupling model.

4. The method according to claim 3, characterized in that, The number of input layer nodes of the front-end decoupling network is 6, the number of output layer nodes of the front-end decoupling network is 6, the input of the back-end empirical parameter network is the output of the front-end decoupling network, and the output of the back-end empirical parameter network is the decoupled multidimensional load information.

5. The method according to claim 1, characterized in that, The process of acquiring strain response dataset information of the target sensor based on a preset sensor structure model includes: Multiple sets of force / torque loads of different directions and magnitudes are applied to a preset sensor structure model to obtain strain response dataset information of the target sensor. The strain response dataset information includes multiple strain response data information, which are used to describe the strain value information corresponding to an application position in the sensor structure model.

6. A decoupling system for a multidimensional force sensor based on an empirical formula-assisted neural network, characterized in that, The system includes: Strain response dataset information acquisition module: used to acquire strain response dataset information of the target sensor based on a preset sensor structure model; Empirical formula determination module: used to determine an empirical formula based on the strain response dataset information, wherein the empirical formula is used to characterize the relationship between the strain of the target sensor and the actual force. Initial Neural Network Model Construction Module: Used to construct an initial neural network model and to construct training dataset information. The initial neural network model includes a front-end decoupling network and a back-end empirical parameter network connected in sequence. The network parameters of the back-end empirical parameter network are initialized according to the parameters in the empirical formula. Target decoupling model generation module: used to perform phased training processing on the initial neural network model based on the training dataset information to generate a target decoupling model; Multidimensional load information output module: used to input the output voltage signal of the sensor to be decoupled to the target decoupling model and output the decoupled multidimensional load information.

7. The system according to claim 6, characterized in that, The target decoupling model generation module includes: The first-stage training submodule is used to freeze all parameters of the back-end empirical parameter network and train only the front-end decoupling network. During the training process, a genetic algorithm is used for global optimization and iterative training is performed for a preset number of first iteration rounds. The second-stage training submodule is used to unfreeze the subsequent empirical parameter network, and to perform joint fine-tuning training on the overall network composed of the preceding decoupling network and the subsequent empirical parameter network using the gradient descent method. The training is iterated in the preset second round to generate the target decoupling model.

8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 5.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images