Door hinge subsystem determination method, device, electronic device and storage medium

Abstract

The present invention provides a method, device, electronic device and storage medium for determining a door hinge subsystem, and belongs to the field of household appliance technology. The method comprises: determining a target result based on a Jaya algorithm and a target network model; and determining a target door hinge subsystem based on the target result. The method, device, electronic device and storage medium for determining a door hinge subsystem provided by the present invention can train a network model using random samples, so that the network model can learn the correspondence between the door hinge subsystem and the shape design parameters and the maximum stress value, thereby providing a theoretical basis for the design of the door hinge subsystem. By combining the Jaya algorithm with the network model to determine the door hinge subsystem, automatic optimization of the design of the door hinge subsystem can be achieved, thereby significantly improving design efficiency and reducing costs.

Description

Technical Field

[0001] The present invention relates to the field of household appliance technology, and in particular to a method, device, electronic device and storage medium for determining a door hinge subsystem. Background Art

[0002] Under the traditional R&D model, the optimization direction of door hinges of household appliances (such as dishwashers, refrigerators, ovens, etc.) mainly relies on the experience of engineers, or through trial and error methods to find a door hinge subsystem that meets the requirements.

[0003] However, through experience or trial and error, it is usually necessary to go through multiple prototypes or multiple mold modifications before the optimal design solution can be selected. Due to the lack of theoretical guidance, invalid verification solutions often exist in trial and error. The above two methods require a relatively long cycle and high R&D costs. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the related art. To this end, the present invention proposes a door hinge subsystem determination method, which can realize automatic optimization of the door hinge subsystem design, thereby significantly improving the design efficiency and reducing the cost.

[0005] The present invention also provides a door hinge subsystem determination device.

[0006] The present invention also provides an electronic device, a non-transitory computer-readable storage medium, and a computer program product.

[0007] A door hinge subsystem determination method according to an embodiment of the first aspect of the present invention includes:

[0008] Determine the target result based on Jaya algorithm and target network model;

[0009] determining a target door hinge subsystem based on the target result;

[0010] Wherein, the target network model is obtained by training the initial network model based on multiple random samples;

[0011] Each random sample includes multiple shape design parameters and the maximum stress value of the door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by multiple shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of the door hinge subsystem; the target result includes the minimum value of the maximum stress value and the shape design parameters corresponding to the minimum value.

[0012] The door hinge subsystem determination method provided by the present invention can train the network model by using random samples, so that the network model can learn the corresponding relationship between the door hinge subsystem and the shape design parameters and the maximum stress value, thereby providing a theoretical basis for the design of the door hinge subsystem. By combining the Jaya algorithm with the network model to determine the door hinge subsystem, the automatic optimization of the door hinge subsystem design can be achieved, thereby significantly improving the design efficiency and reducing the cost.

[0013] According to an embodiment of the present invention, determining the target result based on the Jaya algorithm and the target network model includes:

[0014] Determining an objective function of the Jaya algorithm based on the maximum stress value, and determining an initial population of the Jaya algorithm based on the shape design parameters;

[0015] An iterative operation is performed based on the initial population, the objective function and the target network model until the number of iterations reaches a maximum evolutionary generation, and the target result is determined from the population determined by the last iterative operation.

[0016] According to one embodiment of the present invention, the iterative operation includes:

[0017] Determine the optimal solution and the worst solution of individuals in the current population based on the target network model and the objective function;

[0018] Based on the iterative formula of Jaya algorithm, the optimal solution and the worst solution, each individual in the current population is updated;

[0019] Determine that the solution of the target individual after updating is smaller than the solution of the target individual before updating, and retain the updated target individual to the next generation population;

[0020] Determine that the solution of the target individual after updating is greater than or equal to the solution of the target individual before updating, and retain the target individual before updating to the next generation population.

[0021] The door hinge subsystem determination method provided by the present invention determines the design scheme of the door hinge subsystem by using the Jaya algorithm and the target network model, which can quickly optimize the design scheme of the door hinge subsystem, thereby significantly improving the design efficiency. On the other hand, it can also ensure the optimization result of the design scheme of the door hinge subsystem, thereby improving the accuracy of the design of the door hinge subsystem.

[0022] According to one embodiment of the present invention, the target network model is trained in the following manner:

[0023] Repeat the training operation until the preset conditions are met, and use the initial network model that meets the preset conditions as the target network model;

[0024] The training operation includes:

[0025] Inputting a plurality of shape design parameters of the current random sample into the initial network model to obtain a first maximum stress value output by the initial network model;

[0026] Determine that an error between the first maximum stress value and the maximum stress value of the current random sample is greater than an error threshold, and update the initial network model according to the error;

[0027] The preset conditions include:

[0028] The error is less than or equal to the error threshold.

[0029] According to one embodiment of the present invention, the initial network model is a resNet network model;

[0030] The resNet network model includes an input layer, a convolutional layer, a first pooling layer, a plurality of residual modules, a second pooling layer, a dropout layer, a fully connected layer and an output layer connected in sequence;

[0031] The plurality of residual modules include at least one first residual module and at least one second residual module connected in series;

[0032] The number of convolution kernels of the convolution layer of the second residual module is twice the number of convolution kernels of the convolution layer of the previous first residual module connected thereto, and each residual module satisfies at least one of the following:

[0033] Do not downsample the input data;

[0034] Use a small convolution kernel size greater than 1.

[0035] The door hinge subsystem determination method provided by the present invention can make the resNet network model better applicable to the design of the door hinge subsystem by targetedly adjusting the structure of the resNet network model, thereby improving the design efficiency of the door hinge subsystem.

[0036] According to one embodiment of the present invention, the method further includes:

[0037] Performing stratified sampling on each shape design parameter to obtain the plurality of random samples;

[0038] The maximum stress value of the door hinge subsystem corresponding to the random sample is determined by the door hinge subsystem simulation model based on the multiple shape design parameters.

[0039] According to one embodiment of the present invention, the door hinge subsystem is used in a household appliance.

[0040] A door hinge subsystem determination device according to an embodiment of the second aspect of the present invention includes:

[0041] A training module, used to train the initial network model based on multiple random samples to obtain a target network model;

[0042] The optimization module is used to determine the target result based on the Jaya algorithm and the target network model;

[0043] a determination module, configured to determine a target door hinge subsystem based on the target result;

[0044] Among them, each random sample includes multiple shape design parameters and the maximum stress value of the door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by multiple shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of a door hinge subsystem; the target result includes the minimum value of the maximum stress value and each shape design parameter corresponding to the minimum value.

[0045] The door hinge subsystem determination device provided by the present invention can train the network model by using random samples, so that the network model can learn the corresponding relationship between the door hinge subsystem and the shape design parameters and the maximum stress value, thereby providing a theoretical basis for the design of the door hinge subsystem. By combining the Jaya algorithm with the network model to determine the door hinge subsystem, the automatic optimization of the door hinge subsystem design can be achieved, thereby significantly improving the design efficiency and reducing the cost.

[0046] According to an embodiment of the third aspect of the present invention, the electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the door hinge subsystem determination method as in the embodiment of the first aspect is implemented.

[0047] According to the non-transitory computer-readable storage medium of the fourth aspect embodiment of the present invention, a computer program is stored thereon, and when the computer program is executed by a processor, the door hinge subsystem determination method as in the first aspect embodiment is implemented.

[0048] The computer program product according to the fifth aspect of the present invention comprises a computer program, wherein when the computer program is executed by a processor, the method for determining the door hinge subsystem as in the first aspect of the present invention is implemented.

[0049] The above one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects:

[0050] The network model can learn the correspondence between the door hinge subsystem and the shape design parameters and the maximum stress value, thus providing a theoretical basis for the design of the door hinge subsystem. By combining the Jaya algorithm with the network model to determine the door hinge subsystem, the automatic optimization of the door hinge subsystem design can be achieved, thereby significantly improving the design efficiency and reducing costs.

[0051] Furthermore, by using the Jaya algorithm and the target network model to determine the design scheme of the door hinge subsystem, on the one hand, the door hinge subsystem design scheme can be quickly optimized, thereby significantly improving the design efficiency. On the other hand, the optimization result of the door hinge subsystem design scheme can also be guaranteed, thereby improving the accuracy of the door hinge subsystem design.

[0052] Furthermore, by adjusting the structure of the resNet network model in a targeted manner, the resNet network model can be better applied to the design of the door hinge subsystem, thereby improving the design efficiency of the door hinge subsystem.

[0053] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 It is a flow chart of the door hinge subsystem determination method provided by the present invention.

[0056] Figure 2 It is a schematic diagram of the shape design variables provided by the present invention.

[0057] Figure 3 It is a structural diagram of the resNet network model provided by the present invention.

[0058] Figure 4 The flowchart is an example of applying the door hinge subsystem determination method provided by the present invention.

[0059] Figure 5 It is a schematic diagram of the linear correlation coefficient of the resNet network model using the door hinge subsystem determination method provided by the present invention.

[0060] Figure 6It is a schematic diagram of the accuracy change during the training process of the resNet network model using the door hinge subsystem determination method provided by the present invention.

[0061] Figure 7 It is a schematic diagram of the iterative process of the maximum stress value of the door hinge subsystem using the door hinge subsystem determination method provided by the present invention.

[0062] Figure 8 It is a schematic diagram of error comparison of the maximum stress value of the door hinge subsystem using the door hinge subsystem determination method provided by the present invention.

[0063] Fig. 9 It is a schematic diagram for comparing the structures of door hinge subsystems to which the door hinge subsystem determination method provided by the present invention is applied.

[0064] Fig.10 It is a structural schematic diagram of the door hinge subsystem determination device provided by the present invention.

[0065] Fig.11 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0066] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0067] In the description of the embodiments of the present invention, it should be noted that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limitations on the embodiments of the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0068] In the description of the embodiments of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present invention can be understood according to specific circumstances.

[0069] In the embodiments of the present invention, unless otherwise clearly specified and limited, the first feature being "above" or "below" the second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, the first feature being "above", "above" and "above" the second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. The first feature being "below", "below" and "below" the second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0070] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0071] Figure 1 FIG. 2 is a flow chart of the door hinge subsystem determination method provided by the present invention. Figure 1 As shown, the door hinge subsystem determination method provided by the present invention may include:

[0072] Step 110: Determine the target result based on the Jaya algorithm and the target network model;

[0073] Step 120, determining a target door hinge subsystem based on the target result;

[0074] Among them, the target network model is obtained by training the initial network model based on multiple random samples;

[0075] Each random sample includes multiple shape design parameters and the maximum stress value of the door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by multiple shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of a door hinge subsystem; the target result includes the minimum value of the maximum stress value and each shape design parameter corresponding to the minimum value.

[0076] It should be noted that the door hinge subsystem determination method of the present invention can be applied to various devices and apparatuses, such as household appliances, etc. The following takes the application of the door hinge subsystem determination method of the present invention to a dishwasher as an example to explain the technical solution of the present invention in detail, but it is not intended to limit the protection scope of the present invention.

[0077] The execution subject of the door hinge subsystem determination method of the present invention may be a computer device, such as a mobile phone, a tablet computer, a notebook computer, a PDA, etc. The present invention does not make any specific limitation to this.

[0078] Optionally, a simulation model of the dishwasher door hinge subsystem can be established. For example, the simulation model can be made using MotionView software. The simulation model can include a skeleton model (door hinge subsystem simulation model), an outer door subsystem simulation model, a pull cord spring subsystem simulation model, a lock buckle subsystem simulation model, and a sealing subsystem simulation model. The door hinge subsystem simulation model can be composed of an inner liner, an inner liner hinge, and a door hinge. The outer door subsystem simulation model can be composed of an inner door, an outer door, and other outer door components. The pull cord spring subsystem simulation model can be composed of a pull cord, a fixed pulley, and a spring. The lock buckle subsystem simulation model can be composed of a door lock buckle and an inner liner lock buckle. The inner liner hinge can be composed of a lock box, a tension spring, and two buckle teeth. The sealing subsystem simulation model can be composed of 34 forces. The sealing subsystem can use a macroscopic modeling method to cut the entire sealing strip into 34 equal parts, and the mechanical behavior of each sealing strip is described using a macroscopic constitutive model, such as using a 2-segment linear segment mathematical expression to describe the sealing strip macroscopic constitutive model.

[0079] Furthermore, the Optistruct tool can be used to solve the flexible body of the hinge structure of the dishwasher door hinge subsystem simulation model, and then the hinge structure in the dishwasher door hinge subsystem simulation model is replaced by the flexible body.

[0080] Optionally, when establishing a simulation model of the dishwasher door hinge subsystem, the MotionSolve tool (such as the Transient solver) can be used to solve the model. The outer door subsystem is connected to the frame through a hinge, a hinge rotation pair is set, and a motion speed perpendicular to the door body is set at the position where the outer door is opened, the size is 0.1mm / ms, and the total analysis time is 4000ms.

[0081] It should be noted that the design of the door hinge subsystem by engineers is essentially a trial and error test by constantly changing the shape of the hinge. In the embodiment of the present invention, specific software such as HyperMorph can be used to create multiple shape design variables based on the characteristics of the hinge subsystem, such as Figure 2 As shown. Figure 2In the example, the total number of shape design variables is 14, namely Shape 1 to Shape 14.

[0082] For the door hinge subsystem, it can be regarded as the weighted sum of multiple shape design variables. A shape design variable can be a set of characteristic parameters of the door hinge subsystem. For example, the shape design variable can be specifically a vector, which includes various characteristic parameters of the door hinge subsystem corresponding to the shape design variable, such as through hole size, plate thickness, plate bending direction and angle, etc.

[0083] by Figure 2 For example, let the coefficients of shape design variables Shape 1~Shape 14 be m1~m 14 , then the door hinge subsystem can be expressed as: , i is a positive integer; where, is the i-th shape design parameter, The value range is 0.0~1.0.

[0084] After determining the design parameters of each shape, the dishwasher door hinge subsystem simulation model can simulate the maximum stress value of the door hinge subsystem based on this.

[0085] The initial network model can be trained based on multiple random samples to obtain a target network model.

[0086] Each random sample includes a plurality of shape design parameters and a maximum stress value of a door hinge subsystem corresponding to the random sample.

[0087] Specifically, multiple shape design parameters can be used as samples, and the corresponding maximum stress values ​​of the door hinge subsystem can be used as labels to conduct supervised training on the initial network model, so that the initial network model can accurately learn the corresponding relationship between different combinations of multiple shape design parameters and the maximum stress value. When the initial network model meets the preset conditions (for example, the accuracy reaches a certain threshold), it can be used as the target network model for subsequent steps.

[0088] In step 110, a target result may be determined based on the Jaya algorithm and the target network model.

[0089] It is understandable that the Jaya algorithm can optimize and solve the target network model to determine the minimum value among various possible maximum stress values ​​and various shape design parameters corresponding to the minimum value.

[0090] After the target result is determined, the dishwasher door hinge subsystem simulation model can be used in step 120 to automatically generate a design solution for the target door hinge subsystem based on the shape design parameters corresponding to the minimum value.

[0091] The door hinge subsystem determination method provided by the present invention can train the network model by using random samples, so that the network model can learn the corresponding relationship between the door hinge subsystem and the shape design parameters and the maximum stress value, thereby providing a theoretical basis for the design of the door hinge subsystem. By combining the Jaya algorithm with the network model to determine the door hinge subsystem, the automatic optimization of the door hinge subsystem design can be achieved, thereby significantly improving the design efficiency and reducing the cost.

[0092] In one embodiment, based on the Jaya algorithm and the target network model, determining the target result may include:

[0093] Determine the objective function of the Jaya algorithm based on the maximum stress value, and determine the initial population of the Jaya algorithm based on the shape design parameters;

[0094] An iterative operation is performed based on the initial population, the objective function and the target network model until the number of iterations reaches a maximum evolutionary generation, and the target result is determined from the population determined by the last iterative operation.

[0095] Specifically, when the Jaya algorithm is initialized, the objective function of the Jaya algorithm can be determined based on the maximum stress value of the door hinge subsystem, and the individuals in the population can be determined based on the shape design parameters. The length of the population individuals is determined based on the number of shape design parameters, and the number of individuals in the population can be preset.

[0096] Furthermore, after setting the population size (number of individuals) and the maximum evolutionary generations (number of iterations) of the Jaya algorithm, the initial population can be randomly generated.

[0097] Next, the initial population can be iterated based on the objective function and the Jaya algorithm until the number of iterations reaches the maximum evolutionary generation, and the target result is determined from the population determined by the last iterative operation.

[0098] Among them, the iteration operation may include:

[0099] Determine the optimal solution and the worst solution for individuals in the current population based on the target network model and the objective function;

[0100] Update each individual in the current population based on the iterative formula, optimal solution and worst solution of Jaya algorithm;

[0101] Determine that the solution of the target individual after updating is smaller than the solution of the target individual before updating, and retain the updated target individual to the next generation population;

[0102] Determine that the solution of the target individual after updating is greater than or equal to the solution of the target individual before updating, and retain the target individual before updating to the next generation population.

[0103] Specifically, the solution of each individual in the current population can be determined based on the target network model and the objective function. That is, for each individual, the corresponding shape design parameters are input into the target network model to obtain the maximum stress value output by the target network model. The maximum stress value is used as the solution of the objective function, and the solution is the solution corresponding to the individual.

[0104] After determining the solution of each individual in the current population, the optimal solution and the worst solution of the individuals in the current population can be determined; then, each individual in the current population (each shape design parameter) can be updated according to the iterative formula, the optimal solution and the worst solution of the Jaya algorithm.

[0105] Then, it can be determined whether the solution of each individual after the update is better than the solution before the update.

[0106] For example, assuming that in the current population, the updated solution of a target individual is smaller than the solution of the target individual before the update, the updated target individual will be retained in the next generation population;

[0107] If the solution of a target individual after updating is greater than or equal to the solution of the target individual before updating, the target individual before updating will be retained in the next generation population.

[0108] By repeatedly executing the above iterative operation until the number of iterations reaches the maximum evolutionary generation, the target result can be determined from the population determined by the last iterative operation.

[0109] It should be noted that the specific size of the maximum evolutionary generation and the population size can be adjusted according to actual conditions, and the present invention does not specifically limit this. Generally speaking, the larger the maximum evolutionary generation and the population size, the smaller the maximum stress value corresponding to the target result is likely to be (that is, the better the design of the door hinge subsystem is).

[0110] The door hinge subsystem determination method provided by the present invention determines the design scheme of the door hinge subsystem by using the Jaya algorithm and the target network model, which can quickly optimize the design scheme of the door hinge subsystem, thereby significantly improving the design efficiency. On the other hand, it can also ensure the optimization result of the design scheme of the door hinge subsystem, thereby improving the accuracy of the design of the door hinge subsystem.

[0111] In one embodiment, the target network model is trained as follows;

[0112] Repeat the training operation until the preset conditions are met, and use the initial network model that meets the preset conditions as the target network model;

[0113] The training operation includes:

[0114] Inputting multiple shape design parameters of the current random sample into the initial network model to obtain a first maximum stress value output by the initial network model;

[0115] Determine that the error between the first maximum stress value and the maximum stress value of the current random sample is greater than an error threshold, and update the initial network model according to the error;

[0116] The pre-conditions include:

[0117] The error is less than or equal to the error threshold.

[0118] Specifically, for each random sample, a plurality of shape design parameters corresponding to the random sample may be input into the initial network model to obtain a first maximum stress value output by the initial network model.

[0119] Next, the error between the first maximum stress value and the maximum stress value corresponding to the random sample may be calculated.

[0120] If the error is greater than the error threshold, it means that the initial network model has not reached the preset accuracy, so the error information needs to be back-propagated so that the initial network model can update parameters based on the error according to the optimization algorithm, such as updating the model weights or thresholds. The updated initial network model will be used for the training of the next random sample.

[0121] If the error is less than or equal to the error threshold, it means that the initial network model has reached the preset accuracy, so the current initial network model can be used as the target network model for subsequent processing.

[0122] The door hinge subsystem determination method provided by the present invention realizes the training of the initial network model by back propagation of the error based on the maximum stress value calculation result, which can significantly improve the convergence speed and accuracy of the initial network model, thereby further improving the efficiency of the door hinge subsystem design.

[0123] In one embodiment, the random sample can be obtained as follows:

[0124] Perform stratified sampling on each shape design parameter to obtain multiple random samples;

[0125] Among them, the maximum stress value of the door hinge subsystem corresponding to the random sample is determined by the door hinge subsystem simulation model based on multiple shape design parameters.

[0126] Specifically, the embodiment of the present invention may adopt various stratified sampling methods to perform stratified sampling on each shape design parameter to obtain multiple random samples. For example, each shape design parameter may be sampled based on Latin hypercube to obtain multiple random samples.

[0127] After determining a plurality of random samples, the door hinge subsystem simulation model may be used to perform simulation based on various shape design parameters of each random sample, thereby calculating the maximum stress value corresponding to each random sample.

[0128] The door hinge subsystem determination method provided by the present invention determines multiple random samples by adopting a stratified sampling method, and uses the hinge subsystem simulation model to determine the maximum stress value corresponding to each random sample. On the one hand, it can improve efficiency, and on the other hand, it can also ensure the objectivity of the data, thereby facilitating the model to learn the correspondence between the door hinge subsystem and the shape design parameters and the maximum stress value.

[0129] In one embodiment, the initial network model may be a resNet network model;

[0130] The resNet network model includes an input layer, a convolutional layer, a first pooling layer, multiple residual modules, a second pooling layer, a dropout layer, a fully connected layer, and an output layer connected in sequence;

[0131] The plurality of residual modules include at least one first residual module and at least one second residual module connected in series;

[0132] The number of convolution kernels of the convolution layer of the second residual module is twice the number of convolution kernels of the convolution layer of the previous first residual module connected thereto, and each residual module satisfies at least one of the following:

[0133] Do not downsample the input data;

[0134] Use a small convolution kernel size greater than 1.

[0135] like Figure 3 As shown, the resNet network model may include an input layer, a convolutional layer, a first pooling layer, a plurality of residual modules, a second pooling layer, a dropout layer, a fully connected layer, and an output layer connected in sequence.

[0136] The plurality of residual modules include two first residual modules and two second residual modules connected in series. It can be understood that Figure 3 It is only used for illustration and is not used to limit the specific structure of the resNet network model, especially the specific number of multiple residual modules and the connection relationship between the first residual module and the second residual module.

[0137] It should be noted that, considering that the data volume of the random samples of the present invention is relatively small (for example, compared with the image data), and in order to ensure the complexity of the network, the present invention proposes the above-mentioned improved resNet network model.

[0138] In the improved resNet network model, both the first residual module and the second residual module may not downsample the input data to ensure that the data is not compressed. The number of convolution kernels of the convolution layer of the second residual module can be twice the number of convolution kernels of the convolution layer of the previous first residual module connected to it. This design can double the data channel, which helps the model learn the input data.

[0139] Furthermore, the first residual module and the second residual module may include the same number of convolutional layers or different numbers of convolutional layers, but each convolutional layer may use a small convolution kernel larger than 1, for example , wait.

[0140] The door hinge subsystem determination method provided by the present invention can make the resNet network model better applicable to the design of the door hinge subsystem by targetedly adjusting the structure of the resNet network model, thereby improving the design efficiency of the door hinge subsystem.

[0141] To facilitate understanding of the technical solution of the present invention, an example of applying the door hinge subsystem determination method provided by the present invention is used for illustration below.

[0142] like Figure 4 As shown, this example includes the following steps:

[0143] Step 401, performing stratified sampling on shape design parameters to obtain shape design parameters of each random sample in a plurality of random samples;

[0144] Step 402: inputting the shape design parameters of each random sample into the door hinge subsystem simulation model to obtain the maximum stress value corresponding to each random sample;

[0145] Step 403: training the initial network model based on multiple random samples to obtain a first maximum stress value output by the initial network model;

[0146] Step 404, determining the error between the first maximum stress value and the maximum stress value of the current random sample;

[0147] Step 405: If the error is greater than the error threshold, proceed to step 406;

[0148] If the error is less than or equal to the error threshold, proceed to step 407;

[0149] Step 406: back-propagate the error information to update the initial network model according to the error;

[0150] Step 407: Complete model training and output the target network model;

[0151] Step 408, determining the initial population and objective function of the Jaya algorithm;

[0152] Step 409: Determine the solution of each individual in the current population based on the target network model and the target function;

[0153] Step 410: determine the best solution and the worst solution in the current population;

[0154] Step 411, updating each individual in the current population based on the iterative formula of Jaya algorithm, the optimal solution and the worst solution;

[0155] Step 412: determine that the solution of the target individual after the update is less than the solution of the target individual before the update, and retain the updated target individual in the next generation population; determine that the solution of the target individual after the update is greater than or equal to the solution of the target individual before the update, and retain the target individual before the update in the next generation population;

[0156] Step 413, determine whether the number of iterations reaches the maximum evolutionary generation, if yes, proceed to step 414; if no, return to step 409;

[0157] Step 414: Output the target result.

[0158] The specific implementation method of each step in this example can be found above and will not be repeated here.

[0159] In one example, the resNet network model of the present invention uses 6 residual modules, 100 groups of random samples, Relu function as activation function, ADAM optimizer, 30 miniBatch, 800 maximum iterations, 1e-8 error target, and 0.01 initial learning rate.

[0160] like Figure 5 As shown in the figure, the calculated value of the neural network is very close to the calculated value of the finite element, and the linear correlation coefficient is above 0.996, indicating that the linear correlation between the calculated value of the neural network and the calculated value of the finite element is very significant. The training process of the resNet network model is as follows Figure 6 As shown, the Loss error of the resNet network model reaches 2.0e-3 at the 122nd step, and the convergence accuracy is very high. It can be seen that the resNet network model of the present invention has a very high fitting accuracy.

[0161] Furthermore, in this example, the population size of the Jaya algorithm is 10, the maximum number of evolutionary generations is 100, and the length of each population individual is 14. Figure 7 As shown, it can be seen from the iterative process that based on the Jaya algorithm of the present invention, the maximum stress value of the door hinge subsystem after 47 generations of evolutionary algebraic iterations is 89.5N.

[0162] In this example, by applying the door hinge subsystem determination method of the present invention, the coefficients of various shape design variables can be obtained as shown in Table 1.

[0163] Table 1 Example table of coefficients of shape design variables

[0164]

[0165] Substitute the target result obtained by Jaya algorithm optimization into the simulation model of the door hinge subsystem, such as Figure 8 As shown, the maximum stress value of the door hinge subsystem is 97.8MPa, which is about 8.49% different from the maximum stress value of the door hinge subsystem obtained by the resNet network model, which is 89.5MPa. It can be seen that the door hinge subsystem determination method provided by the present invention can obtain a more ideal optimization design solution, which can meet the actual requirements of engineering design.

[0166] Through optimization analysis, we can see that the maximum stress value of the door hinge subsystem has dropped from 203.2MPa to 97.8MPa, a decrease of 51.8%, which has obvious positive benefits. The shape of the hinge structure before and after optimization is as follows: Fig. 9 shown.

[0167] The door hinge subsystem determination device provided by the present invention is described below. The door hinge subsystem determination device described below and the door hinge subsystem determination method described above can refer to each other and can achieve the same technical effects, which will not be repeated here.

[0168] Fig.10 FIG. 1 is a schematic diagram of the structure of the door hinge subsystem determination device provided by the present invention. Fig.10 As shown, the door hinge subsystem determination device provided by the present invention may include:

[0169] An optimization module 1010, for determining a target result based on a Jaya algorithm and a target network model;

[0170] A determination module 1020, configured to determine a target door hinge subsystem based on the target result;

[0171] Wherein, the target network model is obtained by training the initial network model based on multiple random samples;

[0172] Each random sample includes multiple shape design parameters and the maximum stress value of the door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by multiple shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of the door hinge subsystem; the target result includes the minimum value of the maximum stress value and the shape design parameters corresponding to the minimum value.

[0173] In one embodiment, the optimization module 1010 is specifically used to:

[0174] Determining an objective function of the Jaya algorithm based on the maximum stress value, and determining an initial population of the Jaya algorithm based on the shape design parameters;

[0175] An iterative operation is performed based on the initial population, the objective function and the target network model until the number of iterations reaches a maximum evolutionary generation, and the target result is determined from the population determined by the last iterative operation.

[0176] In one embodiment, the iterative operation includes:

[0177] Determine the optimal solution and the worst solution of individuals in the current population based on the target network model and the objective function;

[0178] Based on the iterative formula of Jaya algorithm, the optimal solution and the worst solution, each individual in the current population is updated;

[0179] Determine that the solution of the target individual after updating is smaller than the solution of the target individual before updating, and retain the updated target individual to the next generation population;

[0180] Determine that the solution of the target individual after updating is greater than or equal to the solution of the target individual before updating, and retain the target individual before updating to the next generation population.

[0181] In one embodiment, the target network model is trained in the following manner:

[0182] Repeat the training operation until the preset conditions are met, and use the initial network model that meets the preset conditions as the target network model;

[0183] The training operation includes:

[0184] Inputting a plurality of shape design parameters of the current random sample into the initial network model to obtain a first maximum stress value output by the initial network model;

[0185] Determine that an error between the first maximum stress value and the maximum stress value of the current random sample is greater than an error threshold, and update the initial network model according to the error;

[0186] The preset conditions include:

[0187] The error is less than or equal to the error threshold.

[0188] In one embodiment, the initial network model is a resNet network model;

[0189] The resNet network model includes an input layer, a convolutional layer, a first pooling layer, a plurality of residual modules, a second pooling layer, a dropout layer, a fully connected layer and an output layer connected in sequence;

[0190] The plurality of residual modules include at least one first residual module and at least one second residual module connected in series;

[0191] The number of convolution kernels of the convolution layer of the second residual module is twice the number of convolution kernels of the convolution layer of the previous first residual module connected thereto, and each residual module satisfies at least one of the following:

[0192] Do not downsample the input data;

[0193] Use a small convolution kernel size greater than 1.

[0194] In one embodiment, the door hinge subsystem determination device further includes a sampling module (not shown in the figure) for:

[0195] Performing stratified sampling on each shape design parameter to obtain the plurality of random samples;

[0196] The maximum stress value of the door hinge subsystem corresponding to the random sample is determined by the door hinge subsystem simulation model based on the multiple shape design parameters.

[0197] In one embodiment, the door hinge subsystem is used in a household appliance.

[0198] Fig.11 An example of a physical structure diagram of an electronic device is shown in FIG. Fig.11 As shown, the electronic device may include: a processor 1110, a communications interface 1120, a memory 1130 and a communication bus 1140, wherein the processor 1110, the communications interface 1120 and the memory 1130 communicate with each other via the communication bus 1140. The processor 1110 may call the logic instructions in the memory 1130 to execute the door hinge subsystem determination method provided by the above-mentioned method embodiments, for example, including:

[0199] Determine the target result based on Jaya algorithm and target network model;

[0200] determining a target door hinge subsystem based on the target result;

[0201] Wherein, the target network model is obtained by training the initial network model based on multiple random samples;

[0202] Each random sample includes multiple shape design parameters and the maximum stress value of the door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by multiple shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of the door hinge subsystem; the target result includes the minimum value of the maximum stress value and the shape design parameters corresponding to the minimum value.

[0203] In addition, the logic instructions in the above-mentioned memory 1130 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for 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 method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0204] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the door hinge subsystem determination method provided by the above-mentioned method embodiments, for example, including:

[0205] Determine the target result based on Jaya algorithm and target network model;

[0206] determining a target door hinge subsystem based on the target result;

[0207] Wherein, the target network model is obtained by training the initial network model based on multiple random samples;

[0208] Each random sample includes multiple shape design parameters and the maximum stress value of the door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by multiple shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of the door hinge subsystem; the target result includes the minimum value of the maximum stress value and the shape design parameters corresponding to the minimum value.

[0209] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to perform the door hinge subsystem determination method provided by the above methods, for example, including:

[0210] Determine the target result based on Jaya algorithm and target network model;

[0211] determining a target door hinge subsystem based on the target result;

[0212] Wherein, the target network model is obtained by training the initial network model based on multiple random samples;

[0213] Each random sample includes multiple shape design parameters and the maximum stress value of the door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by multiple shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of the door hinge subsystem; the target result includes the minimum value of the maximum stress value and the shape design parameters corresponding to the minimum value.

[0214] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0215] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0216] Finally, it should be noted that the above embodiments are only used to illustrate the present invention, rather than to limit the present invention. Although the present invention is described in detail with reference to the embodiments, it should be understood by those skilled in the art that various combinations, modifications or equivalent substitutions of the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and should be included in the scope of the claims of the present invention.

Claims

1. A method for determining a door hinge subsystem, characterized in that: include: Determine the target result based on Jaya algorithm and target network model; determining a target door hinge subsystem based on the target result; Wherein, the target network model is obtained by training the initial network model based on multiple random samples; Each random sample includes a plurality of shape design parameters and a maximum stress value of a door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by a plurality of shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of a door hinge subsystem; the target result includes a minimum value of the maximum stress value and each shape design parameter corresponding to the minimum value; The initial network model is a resNet network model; The resNet network model includes an input layer, a convolutional layer, a first pooling layer, a plurality of residual modules, a second pooling layer, a dropout layer, a fully connected layer and an output layer connected in sequence; The plurality of residual modules include at least one first residual module and at least one second residual module connected in series; The number of convolution kernels of the convolution layer of the second residual module is twice the number of convolution kernels of the convolution layer of the previous first residual module connected thereto, and each residual module satisfies at least one of the following: Do not downsample the input data; Use a small convolution kernel size greater than 1.

2. The door hinge subsystem determination method according to claim 1, characterized in that: Determining the target result based on the Jaya algorithm and the target network model includes: Determining an objective function of the Jaya algorithm based on the maximum stress value, and determining an initial population of the Jaya algorithm based on the shape design parameters; An iterative operation is performed based on the initial population, the objective function and the target network model until the number of iterations reaches a maximum evolutionary generation, and the target result is determined from the population determined by the last iterative operation.

3. The door hinge subsystem determination method according to claim 2, characterized in that: The iterative operation includes: Determine the optimal solution and the worst solution of individuals in the current population based on the target network model and the objective function; Based on the iterative formula of Jaya algorithm, the optimal solution and the worst solution, each individual in the current population is updated; Determine that the solution of the target individual after updating is smaller than the solution of the target individual before updating, and retain the updated target individual to the next generation population; Determine that the solution of the target individual after updating is greater than or equal to the solution of the target individual before updating, and retain the target individual before updating to the next generation population.

4. The door hinge subsystem determination method according to claim 1, characterized in that: The target network model is trained in the following way: Repeat the training operation until the preset conditions are met, and use the initial network model that meets the preset conditions as the target network model; The training operation includes: Inputting a plurality of shape design parameters of the current random sample into the initial network model to obtain a first maximum stress value output by the initial network model; Determine that an error between the first maximum stress value and the maximum stress value of the current random sample is greater than an error threshold, and update the initial network model according to the error; The preset conditions include: The error is less than or equal to the error threshold.

5. The door hinge subsystem determination method according to any one of claims 1 to 4, characterized in that: The method further comprises: Performing stratified sampling on each shape design parameter to obtain the plurality of random samples; The maximum stress value of the door hinge subsystem corresponding to the random sample is determined by the door hinge subsystem simulation model based on the multiple shape design parameters.

6. The door hinge subsystem determination method according to any one of claims 1 to 4, characterized in that: The door hinge subsystem is used in household appliances.

7. A door hinge subsystem determination device, characterized in that: include: The optimization module is used to determine the target result based on the Jaya algorithm and the target network model; a determination module, configured to determine a target door hinge subsystem based on the target result; Wherein, the target network model is obtained by training the initial network model based on multiple random samples; Each random sample includes a plurality of shape design parameters and a maximum stress value of a door hinge subsystem corresponding to the random sample; the door hinge subsystem is determined by a plurality of shape design parameters, each shape design parameter is determined by a shape design variable and its corresponding coefficient, and a shape design variable is determined based on a characteristic parameter of a door hinge subsystem; the target result includes a minimum value of the maximum stress value and each shape design parameter corresponding to the minimum value; The initial network model is a resNet network model; The resNet network model includes an input layer, a convolutional layer, a first pooling layer, a plurality of residual modules, a second pooling layer, a dropout layer, a fully connected layer and an output layer connected in sequence; The plurality of residual modules include at least one first residual module and at least one second residual module connected in series; The number of convolution kernels of the convolution layer of the second residual module is twice the number of convolution kernels of the convolution layer of the previous first residual module connected thereto, and each residual module satisfies at least one of the following: Do not downsample the input data; Use a small convolution kernel size greater than 1.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the door hinge subsystem determination method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the door hinge subsystem determination method according to any one of claims 1 to 6 is implemented.

Images