Door body damage detection model construction method and device, and electronic equipment

Abstract

The application provides a door body damage detection model construction method and device and electronic equipment, and belongs to the field of artificial intelligence, comprising the following steps: constructing a damage behavior simulation model based on physical benchmark data and damage coefficients of door body core components at each discrete time point; training an initial network model using a training data set to obtain the door body damage detection model; the training data in the training data set comprises an opening door angle and a damage coefficient of the damage behavior simulation model at a discrete time point, and the label of the training data is the opening door force at the discrete time point. The application benchmarks the simulation model through model test design, simultaneously establishes a 3D parameterized damage behavior simulation model considering damage, and finally reduces the order of the discrete damage behavior simulation model through deep learning to obtain a network model capable of realizing fast and accurate real-time monitoring of door body damage, thereby filling the industry gap.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, and in particular to a door body damage detection model construction method and device and electronic equipment. BACKGROUND

[0002] The door body system of an electric appliance such as a dishwasher may be damaged in core components during long-term use, affecting the performance and safety of the equipment. Therefore, real-time and accurate monitoring of the damage state of the door body system to improve the reliability and service life of the equipment is a current industry demand.

[0003] Under the traditional door body system health management mode, there are generally two ways to obtain the health state of the door body system: 1) Collect customer usage feedback. If the customer finds that the door body system is abnormal during use, they will find customer service for feedback, and the after-sales team will investigate the abnormal situation to find the cause of the door body system failure. Through long-term accumulation, relevant failure data can be collected. 2) Establish a corresponding laboratory to conduct door body system quick door closing acceleration fatigue durability life tests. This method can understand the evolution of the health state of the door body system, but due to the differences in the action timing of the door body system in actual use, the damage evolution result is inaccurate.

[0004] Therefore, there is currently no technology for real-time monitoring of the damage of the door body system in the industry, and this industry gap needs to be filled. SUMMARY

[0005] The present application provides a door body damage detection model construction method and device and electronic equipment to fill the technical gap in the current accurate, real-time, and rapid detection of the damage evolution process of the door body system of a household appliance. A door body damage detection model based on a deep learning network algorithm and considering damage evolution behavior is proposed, which can effectively solve this problem.

[0006] The present application provides a door body damage detection model construction method, comprising the following steps:

[0007] Based on physical benchmarking data and damage coefficients of each door body core component at each discrete time point on a preset timing, a damage behavior simulation model is constructed for each discrete time point. The physical benchmarking data is obtained by performing physical opening and closing experiments on each door body core component of a real physical door body.

[0008] An initial network model is trained using a training data set to obtain the door body damage detection model. Any training data in the training data set includes an opening angle of the damage behavior simulation model at its corresponding discrete time point and the damage coefficients of each door body core component. The label of the training data is the opening force at the discrete time point.

[0009] The application provides a door damage detection model construction method, which is based on physical benchmark data and damage coefficients of core components of a door at each discrete time point in a preset time sequence, and comprises the following steps of:

[0010] The physical benchmark data are used to benchmark each lossless simulation model, so as to obtain normal behavior simulation models of the core components of the door.

[0011] The normal behavior simulation models are combined to form a door system simulation model.

[0012] The door system simulation model is updated according to the damage coefficients of the core components of the door at each discrete time point in the preset time sequence, so as to obtain the damage behavior simulation models under different damage states at each discrete time point.

[0013] The application provides a door damage detection model construction method, wherein the physical benchmark data comprise force time curve samples obtained through physical opening and closing experiments under angle control of each core component of the door.

[0014] The physical benchmark data are used to benchmark each lossless simulation model, so as to obtain normal behavior simulation models of the core components of the door.

[0015] For the lossless simulation model of any core component of the door, the model parameters of the lossless simulation model are adjusted according to the curve difference between a virtual force time curve output by the lossless simulation model and the force time curve samples, until the curve difference is within a preset range, so as to obtain the adjusted lossless simulation model as the normal behavior simulation model.

[0016] The application provides a door damage detection model construction method, wherein the core components of the door comprise an outer door subsystem and all key subsystems, the key subsystems comprise one or more of a hinge subsystem, a pull rope spring subsystem, a lock buckle subsystem and a sealing subsystem, and the hinge subsystem is composed of a skeleton subsystem and the outer door subsystem.

[0017] Correspondingly, the lossless simulation model comprises an outer door subsystem simulation model, a hinge subsystem simulation model, a pull rope spring subsystem simulation model, a lock buckle subsystem simulation model and a sealing subsystem simulation model.

[0018] The application provides a door damage detection model construction method, wherein the normal behavior simulation models are combined to form a door system simulation model.

[0019] Fuse the normal behavior simulation model of the outer door subsystem, the hinge subsystem, the pull rope spring subsystem, the lock catch subsystem and the sealing subsystem corresponding to the outer door subsystem simulation model, the hinge subsystem simulation model, the pull rope spring subsystem simulation model, the lock catch subsystem simulation model and the sealing subsystem simulation model, and obtain the door body system simulation model.

[0020] According to the door body damage detection model construction method provided by the application, the door body system simulation model is updated according to the damage coefficient of each door body core component at each discrete time point in the preset time sequence, and the damage behavior simulation model at each discrete time point under different damage states is obtained, comprising:

[0021] Based on the current opening and closing times of any door body core component at any discrete time point and the fatigue opening and closing time limit of the any door body core component, the damage coefficient of the any door body core component at the any discrete time point is determined;

[0022] Based on the damage coefficient, the scaling coefficient of the target parameter related to door body damage of the any door body core component is determined;

[0023] At the any discrete time point, the door body system simulation model is updated according to the scaling coefficients of all target parameters related to door body damage of the door body core components, and the damage behavior simulation model at the any discrete time point is obtained;

[0024] All the damage behavior simulation models are obtained by traversing all the discrete time points.

[0025] According to the door body damage detection model construction method provided by the application, the target parameters of the hinge subsystem include the deformation parameters at the hinge and pull rope connection point, the target parameters of the pull rope spring subsystem include the pull rope stiffness and the extension spring stiffness, the target parameters of the lock catch subsystem include the locking spring stiffness, and the target parameters of the sealing subsystem include the sealing strip stiffness.

[0026] According to the door body damage detection model construction method provided by the application, any training data in the training data set is used to train the initial network model, and the door body damage detection model is obtained, comprising:

[0027] According to the preset time sequence, the training data corresponding to each discrete time point is sequentially input into the initial network model, and the opening door force prediction sequence output by the initial network model is obtained;

[0028] According to the network error between the door opening force prediction sequence and the door opening force label sequence, the model parameters of the initial network model are updated step by step until the model converges, and the door body damage detection model is obtained.

[0029] The door opening force label sequence is composed of door opening forces at all discrete time points in the preset time sequence.

[0030] According to the door body damage detection model construction method provided by the application, the initial network model is obtained based on a long short-term memory network model.

[0031] The application further provides a door body damage detection model construction device, mainly comprising:

[0032] The damage behavior simulation model construction unit is used for constructing a damage behavior simulation model at each discrete time point based on physical benchmark data and damage coefficients of each door body core component at each discrete time point in the preset time sequence; the physical benchmark data is obtained by performing a physical opening and closing experiment on each door body core component of a real physical door body;

[0033] The initial network model training control unit is used for training an initial network model by using a training data set to obtain the door body damage detection model; any training data in the training data set comprises an opening angle of the damage behavior simulation model at the corresponding discrete time point and the damage coefficients of each core component, and the label of the training data is the opening force at the discrete time point.

[0034] The application further provides an electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the door body damage detection model construction method of any one of the above.

[0035] The application further provides a non-transitory computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the door body damage detection model construction method of any one of the above.

[0036] The application further provides a computer program product comprising a computer program, and the computer program is executed by a processor to implement the door body damage detection model construction method of any one of the above.

[0037] The door body damage detection model construction method, device and electronic device provided by the application fill the industry gap by designing a model test to benchmark the simulation model, establishing a 3D parameterized damage behavior simulation model considering damage, and finally reducing the order of the discrete damage behavior simulation model through deep learning to obtain a network model capable of realizing fast and accurate real-time monitoring of door body damage. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0039] Figure 1 This is a flowchart illustrating the method for constructing the door damage detection model provided by the present invention.

[0040] Figure 2 This is a schematic diagram of the method for determining the damage coefficient of the hinge subsystem provided by the present invention.

[0041] Figure 3 This is a schematic diagram of the stiffness curve of the sealing strip provided by the present invention.

[0042] Figure 4 This is a schematic diagram of the training process of the door damage detection model provided by the present invention.

[0043] Figure 5 This is a schematic diagram of the structure of the device for constructing the door damage detection model provided by the present invention.

[0044] Figure 6 This is one of the schematic diagrams showing the performance of the door damage detection model provided by this invention on the training dataset.

[0045] Figure 7 This is the second schematic diagram showing the performance of the door damage detection model provided by this invention on the training dataset.

[0046] Figure 8 This is the third schematic diagram showing the performance of the door damage detection model provided by this invention on the training dataset.

[0047] Figure 9 This is the fourth schematic diagram showing the performance of the door damage detection model provided by this invention on the training dataset.

[0048] Figure 10 This is the fifth illustration of the performance of the door damage detection model provided by this invention on the training dataset.

[0049] Figure 11 This is one of the schematic diagrams showing the performance of the door damage detection model provided by this invention on a test dataset.

[0050] Figure 12 This is the second schematic diagram showing the performance of the door damage detection model provided by this invention on the test dataset.

[0051] Figure 13FIG. 3 is a schematic diagram of performance of the door damage detection model provided by the present application on a test data set.

[0052] Figure 14 FIG. 4 is a schematic diagram of performance of the door damage detection model provided by the present application on a test data set.

[0053] Figure 15 FIG. 5 is a schematic diagram of performance of the door damage detection model provided by the present application on a test data set.

[0054] Figure 16 FIG. 6 is a schematic diagram of evolution of a loss function value of the door damage detection model provided by the present application during a training process.

[0055] Figure 17 FIG. 7 is a schematic diagram of evolution of a loss function value of the door damage detection model provided by the present application during a training process.

[0056] Figure 18 FIG. 8 is a schematic diagram of a structure of an electronic device provided by the present application. DETAILED DESCRIPTION

[0057] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0058] It should be noted that in the description of the present application, the terms "comprising", "containing" or any other variant thereof are intended to cover a non-exclusive inclusion, so that a process, method, article or apparatus that comprises a list of elements does not only include those elements, but can also include other elements that are not expressly listed or inherent to such process, method, article or apparatus. Without more limitations, the element defined by the phrase "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising the element. The terms "upper", "lower" and the like indicate the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be connected inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0059] The terms "first", "second", and the like in the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally a class, and do not limit the number of objects, for example, the first object can be one or more.

[0060] The health state management scenario of electrical equipment, especially household electrical equipment, is a very broad category. Taking a dishwasher as an example, in the actual use of the user, the scenarios that need to be managed in the health state are: 1) monitoring of the health state of the dishwasher washing system: in the washing process, whether the water pressure is abnormal, whether the water temperature is abnormal, whether the booster pump is abnormal, whether the breather is abnormal, whether the drainage system is abnormal, etc. The monitoring and prediction of the health state of the entire washing system, fault diagnosis, are generally related to the related control system; 2) monitoring of the health state of the dishwasher structure system: whether the various systems of the dishwasher, including the door body system, the inner tank system, the chassis system, etc. have been damaged, analyzing the damage value, the remaining life, and judging whether the life limit has been reached, etc.

[0061] In the user's use process, the door body system of the electrical appliance, such as the pull rope spring subsystem, the sealing strip subsystem, the lock catch subsystem, the hinge subsystem and other door body core components, has a probability of failure after being used for a certain period of time, for example, the following failure conditions may occur: 1) pull rope spring subsystem: the extension spring appears to be loose or broken, and the pull rope appears to be worn or broken; 2) sealing strip subsystem: the sealing property of the sealing strip gradually decreases, and the rebound force decreases; 3) lock catch subsystem: the position of the lock catch pin appears to be worn, and the locking spring appears to be loose or broken; 4) hinge subsystem: the hinge structure appears to be deformed or broken.

[0062] In the design process of the door body system, in order to quickly obtain the damage evolution process of the entire life cycle of the door body system, a corresponding laboratory is established to perform door body system quick closing door accelerated fatigue durability life test, and failure data obtained by combining feedback in the use process of customers for failure analysis, but since the action timing of each door body system is complex and personalized, the damage evolution is fast or slow, so the existing these ways often cannot realize systematic and real-time analysis of the damage of the door body system, and how to solve this demand is a problem that needs to be solved in the industry.

[0063] In today's rapid development of digitalization, the fusion of digital twin technology and reinforcement learning algorithm and deep learning algorithm may become a new health management mode for the door body system. In order to solve the current technical demand in the industry, the present application proposes a construction method and device of a door body damage detection model and an electronic device, and the main implementation principle includes the following steps:

[0064] Firstly, a model test is designed to obtain physical benchmarking data as a benchmarking standard for a non-destructive simulation model. Secondly, a normal behavior simulation model is constructed by benchmarking the physical benchmarking data. Thirdly, based on the corrected normal behavior simulation model, a reasonable mechanical hypothesis is proposed by using fatigue damage theory, and a damage behavior simulation model considering damage evolution is established, and a training data set for deep learning is output. Then, based on the deep learning network algorithm, an AI reduced order model corresponding to the damage behavior simulation model is output as a high-fidelity digital twin model considering damage evolution behavior, i.e. a door body damage detection model. Subsequently, the door body damage detection model can be learned based on the reinforcement learning algorithm, and a corresponding door body damage recognition intelligent agent can be obtained, which can be used for real-time monitoring of the damage state of the door body system.

[0065] It should be noted that the object to which the door body damage detection model to be constructed by the present application can be any door body system, not limited to the door body system of various electrical appliances, for example: it can be a dishwasher door body system, a refrigerator door body system, a washing machine door body system, or other systems containing a door body structure. For the convenience of description, the dishwasher door body system is taken as an example for subsequent description, which is not regarded as a specific limitation on the protection scope of the present application.

[0066] The construction method of the door body damage detection model provided by the present application, the device and the electronic equipment will be described below. Figures 1-18 The construction method of the door body damage detection model provided by the present application, the device and the electronic equipment will be described below.

[0067] Figure 1 The flowchart of the construction method of the door body damage detection model provided by the present application is shown in FIG. 1, which includes but is not limited to the following steps: Figure 1

[0068] Step 101, based on the physical benchmark data and the damage coefficient of each door body core component at each discrete time point on the preset time sequence, a damage behavior simulation model of each discrete time point is constructed.

[0069] The physical benchmark data is obtained by performing a physical opening and closing experiment on each door body core component of a real physical door body, and the specific acquisition method can be:

[0070] A representative dishwasher door body system is selected to ensure that each core component (such as the pull rope spring subsystem, the sealing subsystem, the lock catch subsystem, etc.) is in an initial undamaged state. In the physical opening and closing experiment, an angle control method is adopted, that is, the angle increment of each opening is kept consistent (for example, each time by 1°), and the opening force data of the door body system under different opening angles is recorded. Specifically, a high-precision force sensor and an angle sensor can be used to collect opening force data and opening angle at each discrete time point (i.e. discrete sampling time point) on the preset time sequence, to form opening force time sequence data and opening angle time sequence data. The obtained opening force time sequence data and opening angle time sequence data can be represented by opening force time sequence curve and opening angle time sequence curve respectively, and these data will be used as the physical benchmark data for subsequent benchmarking of the undamaged simulation model.

[0071] The damage coefficient is mainly used to quantify the damage degree of each door body core component at different use stages, which can generally be determined by the ratio between the current opening and closing times of each door body core component and the fatigue opening and closing time limit.

[0072] For example, for the pull rope spring subsystem, the damage coefficient D at any discrete time point can be defined as: , wherein is the current opening and closing times at the any discrete time point, ​The fatigue opening and closing times limit of the pull rope spring subsystem.

[0073] Further, based on the physical benchmark data and the damage coefficient, a damage behavior simulation model at each discrete time point can be constructed, which can be achieved by the following steps:

[0074] Firstly, according to the initial parameters of each door body core component, an initialization simulation model of each door body core component in the door body system is constructed, which can be called a damage-free simulation model.

[0075] Then, the physical benchmark data are used to benchmark each damage-free simulation model, so that the normal behavior simulation model of each door body core component can be obtained. The normal behavior simulation model of all door body core components can be used to construct a door body system simulation model.

[0076] Finally, at each discrete time point on the preset time sequence, the parameters of the corresponding component in the door body system simulation model are updated according to the damage coefficient of each door body core component at the discrete time point, for example, at a discrete time point, the stiffness of the pull rope spring is adjusted according to the damage coefficient D of the pull rope spring subsystem. In this way, at each discrete time point, a damage behavior simulation model considering damage evolution behavior can be obtained.

[0077] Suppose that 50 discrete sampling points are set on the entire preset time sequence, then 50 damage behavior simulation models under different damage states can be obtained.

[0078] Step 102, training the initial network model with the training data set to obtain the door damage detection model.

[0079] Any training data in the training data set includes the opening angle of the damage behavior simulation model at its corresponding discrete time point and the damage coefficient of each door body core component, and the label of the training data is the opening force at the discrete time point.

[0080] Specifically, the preparation of the training data set can be performed first, and each training data in the training data set is closely related to the damage behavior simulation model obtained at each discrete sampling point. First, the opening force data of the door body system at different opening and closing angles in the preset time sequence are tested by experiment, that is, the opening angle and opening force of each discrete sampling point can be obtained. At the same time, according to the opening and closing times of each door body core component in the door body system at each discrete sampling point, the damage coefficient of each door body core component is calculated.

[0081] Further, the opening door angle obtained at each discrete time point and the damage coefficient of each door body core component form a training data, and the opening door force obtained at the discrete time point is taken as the label of the training data, so that a set of training samples corresponding to each discrete time point can be obtained,

[0082] Suppose that the door body core component has five damage coefficients D1, D2, D3, D4 and D5 at any discrete time point, then the input of the training sample at the discrete time point is (opening door angle, D1, D2, D3, D4, D5), and the corresponding label is (opening door force).

[0083] A suitable neural network model is selected as the first initial network model, for example, a long short-term memory network LSTM is selected, and then the training data set is used for training, and the specific training steps can be:

[0084] (1) input a training data in the training data set into the first initial network model.

[0085] (2) calculate the opening door force prediction value output by the first initial network model, and calculate the network error (such as mean square error) between the opening door force prediction value and the label corresponding to the training data.

[0086] (3) update the parameters of the first initial network model using the back propagation algorithm to optimize the model performance, and the goal is to reduce the network error between the output of the first initial network model and the label.

[0087] (4) repeat the above process until the model converges, and obtain the trained door body damage detection model.

[0088] The construction method of the door body damage detection model provided by the application fills the gap in the industry.

[0089] Based on the content of the above embodiment, as an optional embodiment, the door body core component includes an outer door subsystem and all key subsystems, the key subsystems include one or more of a hinge subsystem, a pull rope spring subsystem, a lock catch subsystem and a sealing subsystem, and the hinge subsystem is composed of a skeleton subsystem and the outer door subsystem.

[0090] Correspondingly, the non-destructive simulation model includes an outer door subsystem simulation model, a hinge subsystem simulation model, a pull rope spring subsystem simulation model, a lock catch subsystem simulation model and a sealing subsystem simulation model.

[0091] Take the door system of a dishwasher as an example, the key subsystems mainly include a hinge subsystem, a pull rope spring subsystem, a lock catch subsystem, and a sealing subsystem, and the above key subsystems as a whole are composed of six core components: a spring (in order to distinguish, the spring of the pull rope spring subsystem is called an extension spring, and the spring of the lock catch subsystem is called a locking spring), a rope, a pulley, a hinge wall, a door body, a door lock catch, and a sealing strip.

[0092] Correspondingly, the simulation model of the entire door system of the dishwasher is fused from the simulation models of five door core component related subsystems, i.e., an outer door subsystem simulation model, the hinge subsystem simulation model, the pull rope spring subsystem simulation model, the lock catch subsystem simulation model, and the sealing subsystem simulation model.

[0093] The traditional benchmarking method needs to obtain the key mechanical parameters of each door core component through standard mechanical test experiments. The standard mechanical test experiments generally need to manufacture each door core component into a standard test piece, and it may need to pass through multiple test experiments to complete the acquisition of the key mechanical parameters. Many colleges or research institutes cannot complete all the test experiments, and the test period is generally long, and the test cost is also relatively high.

[0094] Therefore, the application directly benchmarks the door system simulation model of the dishwasher at the system level by designing a key model test.

[0095] Next, how to construct the benchmarking model of each door core component is introduced.

[0096] The hinge subsystem mainly includes a skeleton and an outer door, and in order to be consistent with the subsequent constructed non-destructive simulation model, the state design of the benchmarking model test is to remove the lock catch, the sealing strip, and the rope, and to retain the door hinge.

[0097] The pull rope spring subsystem mainly includes a skeleton, an outer door, and a rope, a spring, etc., and in order to be consistent with the subsequent constructed non-destructive simulation model, the state design of the benchmarking model test is to remove the lock catch, the sealing strip, and to retain the rope, the outer door, the hinge, etc.

[0098] The lock catch subsystem mainly includes a skeleton, an outer door, and a lock catch, and in order to be consistent with the subsequent constructed non-destructive simulation model, the state design of the benchmarking model test is to remove the sealing strip, the rope, etc., and to retain the lock catch, the outer door, the hinge, etc.

[0099] The sealing subsystem mainly includes a skeleton, an outer door, and a sealing strip, and in order to be consistent with the subsequent constructed non-destructive simulation model, the state design of the benchmarking model test is to remove the lock catch, the rope, and to retain the outer door, the hinge, and the sealing strip, etc.

[0100] Taking the door body system of the above-described dishwasher as an example, a damage behavior simulation model is constructed for each discrete time point based on physical benchmarking data and damage coefficients of each door body core component at each discrete time point on a preset time sequence, specifically including but not limited to:

[0101] Each lossless simulation model is benchmarked using the physical benchmarking data to obtain a normal behavior simulation model for each door body core component.

[0102] The normal behavior simulation models are combined to form a door body system simulation model.

[0103] The door body system simulation model is updated according to the damage coefficients of each door body core component at each discrete time point on the preset time sequence to obtain the damage behavior simulation model under different damage states at each discrete time point.

[0104] The lossless simulation model is established based on theoretical and empirical data of each door body core component to simulate the behavior of the door body core component under a damage-free state. The general construction process includes:

[0105] 1) According to the physical properties and mechanical behavior of each door body core component, lossless simulation models are constructed. These lossless simulation models mainly include: hinge subsystem, pull rope spring subsystem, lock catch subsystem, sealing subsystem, etc.

[0106] 2) Initialize the model parameters related to each door body core component in these lossless simulation models, such as the stiffness, strength, and friction coefficient of each component.

[0107] Further, the above lossless simulation models are benchmarked using physical benchmarking data to adjust the model parameters of each lossless simulation model, so that the simulation output of the lossless simulation model is consistent with the real test data, and the normal behavior simulation model of each door body core component is obtained.

[0108] As an optional embodiment, to simplify and standardize the benchmarking process, when benchmarking the above lossless simulation models using physical benchmarking data, the corresponding benchmarking process can be set as follows:

[0109] (1) In the benchmarking test process of the lossless simulation model of each door body core component, since the door body undergoes the same time sequence action from the locked state to opening to a preset angle and returning to the original locked state, relevant data is collected, including recording the size of each door opening force at the discrete time points and the corresponding door opening angle.

[0110] (2) The test process adopts angle control, that is, the time used for opening the door by 1° is the same in each opening and closing process of each test. Through the above test, the force time sequence curve and the angle time sequence curve of the real physical prototype can be obtained as the physical benchmark data for subsequent simulation.

[0111] (3) The force sensor is used to collect the opening force, and the angle sensor is used to collect the opening angle, and the whole opening process can be automatically controlled by the mechanical arm.

[0112] Further, the normal behavior simulation models of the door body core components are combined to form a complete door body system simulation model, specifically including fusing the normal behavior simulation models of the pull rope spring subsystem, the sealing strip subsystem, the lock catch subsystem, the hinge subsystem and other door body core components to form a complete door body system simulation model. The fused door body system simulation model can also be verified at the system level to ensure that the interaction and overall behavior of the door body core components meet the actual physical system.

[0113] The damage coefficient is used to quantify the damage degree of each door body core component at different use stages. The damage coefficient D i of each door body core component can be defined according to the fatigue characteristics of the door body core component, and the value range of each damage coefficient is [0, 1], wherein 0 represents no damage and 1 represents complete failure.

[0114] Then, the damage coefficient at each discrete time point is calculated according to the current opening and closing times of each door body core component and the fatigue opening and closing time limit.

[0115] Further, the related model parameters in the door body system simulation model can be updated according to the damage coefficient of each door body core component at each discrete time point. For example, if the damage coefficient of the pull rope spring subsystem at a certain discrete time point is D2, the stiffness of the pull rope spring of the pull rope spring subsystem in the door body system simulation model can be updated according to the damage coefficient D2.

[0116] In this way, the damage behavior simulation model at each discrete time point is obtained by continuously running and obtaining the updated door body system simulation model within the preset time sequence.

[0117] The application provides a door damage detection model construction method, which comprises the following steps: obtaining normal behavior simulation models of each component by using physical benchmarking data to benchmark the non-destructive simulation models of each door core component; then, collecting the normal behavior simulation models to form a complete door system simulation model; finally, updating the door system simulation model according to the damage coefficients at each discrete time point to obtain damage behavior simulation models under different damage states, which solves the technical problem that the simulation software cannot directly establish damage evolution, provides a solid foundation for subsequent construction of a door damage detection model, and ensures the accuracy and reliability of the model.

[0118] As an optional embodiment, the physical benchmarking data comprises force time curve samples obtained through physical opening and closing experiments under angle control for each door core component.

[0119] The following will introduce in detail how to benchmark each non-destructive simulation model one by one by using the physical benchmarking data to obtain the normal behavior simulation model of each door core component, and the process comprises the following steps:

[0120] For the non-destructive simulation model of any door core component, the model parameters of the non-destructive simulation model are adjusted according to the curve difference between the virtual force time curve output by the non-destructive simulation model and the force time curve sample, until the curve difference is within a preset range, and the adjusted non-destructive simulation model is obtained as the normal behavior simulation model.

[0121] The physical benchmarking data is obtained through physical opening and closing experiments on each door core component of a real physical door, and the actual collection process can be realized by adopting the following steps taking a dishwasher door system as an object:

[0122] Firstly, a representative dishwasher door system is selected to ensure that each important component (such as a pull rope spring, a sealing strip, a lock catch, etc.) is in an initial non-damaged state.

[0123] In the experiment, an angle control method is adopted, that is, the angle increment of each opening is kept consistent (for example, 1° is increased each time), and the opening force data of the door at different opening angles is recorded.

[0124] At the same time, high-precision force sensors and angle sensors are used to collect time sequence data of the opening force and the opening angle, respectively, to form force time curve samples, and these physical benchmarking data will be used as the basis for subsequent simulation model benchmarking.

[0125] The lossless simulation model is constructed in advance based on the physical characteristics and mechanical behaviors of core components of each door body, and mainly includes a hinge subsystem simulation model, a pull rope spring subsystem simulation model, a lock catch subsystem simulation model and a sealing subsystem simulation model, etc. The hinge subsystem simulation model is mainly composed of a skeleton subsystem simulation model and an outer door subsystem simulation model.

[0126] The skeleton subsystem simulation model is mainly composed of a simulation model of a liner, a liner hinge and a door hinge. MotionView and MotionSolve are used as front and back processors respectively, unit selection is performed according to the stress characteristics of each component, and rigid body units are used to simulate all components of the liner, the liner hinge and the door hinge. All components of the liner are fixedly connected, the liner and the liner hinge are fixedly connected, the liner hinge and the ground are fixedly connected, and the door hinge and the liner hinge are set as rotary pairs.

[0127] The outer door subsystem simulation model is mainly composed of a simulation model of an inner door, an outer door and other components. MotionView and MotionSolve are used as front and back processors respectively, unit selection is performed according to the stress characteristics of each component, and rigid body units are used to simulate the inner door, the outer door and other components. The inner door, the outer door and other components are fixedly connected, and the inner door and the door hinge are fixedly connected.

[0128] The pull rope spring subsystem simulation model is mainly composed of a simulation model of a pull rope, a fixed pulley, a spring group and other components. Unit selection is performed according to the stress characteristics of each component, the fixed pulley is simulated by a rigid body unit, the pull rope is simulated by a nonlinear finite element unit, and the spring is simulated by a spring unit. One end of the pull rope is connected with the door hinge, the other end of the pull rope is connected with the spring, the spring is fixedly connected with the ground, and contact is defined between the pull rope and the fixed pulley.

[0129] The lock catch subsystem simulation model is mainly composed of a simulation model of a lock catch and a liner lock catch and other components. The liner lock catch is composed of a lock box, a tension spring and two catch teeth. MotionView and MotionSolve are used as front and back processors respectively, unit selection is performed according to the stress characteristics of each component, and rigid body units are used to simulate the door lock catch and the liner lock catch. The door lock catch is fixedly connected with the inner door, the liner lock catch is fixedly connected with the liner, the two catch teeth are respectively connected with the two ends of the tension spring, each catch tooth is defined as a rotary pair, and contact is defined among the catch tooth, the lock box and the door lock catch.

[0130] The sealing subsystem simulation model can be composed of a plurality of forces (for example, 34), adopts a macro modeling method, divides the entire sealing strip into 34 equal parts, and describes the mechanical behavior of each sealing strip by using a macro constitutive model. According to the previous test experience, a two-segment linear segment mathematical expression is used to describe the macro constitutive model of the sealing strip. MotionView and MotionSolve are used as pre- and post-processors, respectively, and the elements are selected according to the stress characteristics of each component. The force of the sealing strip is a pair of action and reaction forces, and 34 forces are used, each acting on the inner container and the inner door.

[0131] Further, the lossless simulation model of each door body core component is compared with the physical benchmark data of each door body core component, and the normal behavior simulation model of each door body core component is obtained. Taking the lossless simulation model of any door body core component as an example, the following steps can be used to achieve it:

[0132] First, run the lossless simulation model to generate a virtual force time curve. The virtual force time curve describes the opening door force change predicted by the lossless simulation model at different opening angles.

[0133] Specifically, the Transient solver in MotionSolve can be used for solving, the hinge connection between the outer door subsystem and the skeleton subsystem is set, the rotary pair and the position of the outer door opening are set, and the door body rotation motion is also set. The total analysis time is 20s.

[0134] During the simulation of the lossless simulation model, the door body experiences the same time sequence action, from the locked state of the door body, to the preset angle, and then returns to the original locked state. The simulation adopts angle control, that is, the time used for opening 1° in each opening and closing process of the lossless simulation model is the same. Through the above simulation, the virtual force time curve and the virtual angle time curve of the virtual lossless simulation model can be obtained as the simulation output data. Since the angle control method is used, the angle is consistent at the preset time sequence, so only the virtual force time curve needs to be focused on in the subsequent analysis.

[0135] Then, the virtual force time curve generated by the lossless simulation model is compared with the force time curve sample obtained by the physical opening and closing experiment of the door body core component, and the curve difference between the two is calculated.

[0136] Optionally, the curve difference can be quantified by mean square error (MSE) or other statistical indicators, which will not be described here.

[0137] Further, the model parameters of the lossless simulation model can be adjusted according to the calculated curve difference. The operation can be iterated for multiple times until the curve difference is within the preset range.

[0138] For example, if the mean square error between the virtual force time sequence curve and the force time sequence sample (i.e. the test curve) is less than a certain threshold (such as 0.01N²), it is considered that the model parameter adjustment is completed, at this time, the virtual force time sequence curve output by the lossless simulation model is basically consistent with the force time sequence sample of the real physical prototype, that is, the model is completed.

[0139] The adjusted lossless simulation model is a normal behavior simulation model, which can accurately simulate the behavior of the door body core component in the undamaged state.

[0140] In the above manner, the normal behavior simulation model corresponding to each door body core component can be obtained by respectively performing the model matching on the lossless simulation model of each door body core component.

[0141] Finally, the complete door body system simulation model can be obtained by fusing the normal behavior simulation models of all door body core components obtained by the model matching.

[0142] Taking the door body system of the dishwasher as an example, the normal behavior simulation models of the outer door subsystem, the hinge subsystem, the pull rope spring subsystem, the lock catch subsystem and the sealing subsystem can be obtained by performing the model matching on the outer door subsystem simulation model, the hinge subsystem simulation model, the pull rope spring subsystem simulation model, the lock catch subsystem simulation model and the sealing subsystem simulation model, and finally, the door body system simulation model corresponding to the door body system of the dishwasher can be obtained by fusing all the normal behavior simulation models in the simulation software.

[0143] The method for constructing the door body damage detection model provided by the application uses physical model matching data to perform model matching on the lossless simulation models of the door body core components, adjusts the model parameters to make the simulation output consistent with the experimental data, and finally obtains the normal behavior simulation model which can accurately reflect the real behavior of the door body, which lays a solid foundation for subsequent construction of the door body damage detection model and effectively improves the reliability of the model.

[0144] Based on the content of the above embodiment, as an optional embodiment, the door body system simulation model is updated according to the damage coefficient of each door body core component at each discrete time point in the preset time sequence, and the damage behavior simulation model under different damage states at each discrete time point is obtained, which comprises:

[0145] The damage coefficient of any door body core component at any discrete time point is determined based on the current opening and closing times of the door body core component at the discrete time point and the fatigue opening and closing time limit of the door body core component.

[0146] determine a scaling coefficient of a target parameter related to door damage of any door core component based on the damage coefficient;

[0147] update the door system simulation model according to the scaling coefficient of the target parameter related to door damage of all door core components at any discrete time point, and obtain the damage behavior simulation model at the discrete time point;

[0148] traverse all the discrete time points to obtain all the damage behavior simulation models.

[0149] Through the description of the above embodiment, a normal behavior simulation model of the door system is established, which is idealized as intact throughout the time sequence, but to build a high-fidelity digital twin model, the damage evolution process of the real physical prototype needs to be considered, and how to consider the change of the structural health state of the door system is a technical difficulty in the industry.

[0150] The present application proposes reasonable basic assumptions through fatigue damage mechanics theory, and establishes a theoretical relationship between the number of door openings and structural damage evolution. The following will continue to explain in detail how to update the door system simulation model according to the damage coefficient of each door core component at each discrete time point in the preset time sequence to obtain the damage behavior simulation model under different damage states.

[0151] The action time sequence experienced by the door system throughout its life cycle is opening the door and then closing the door, and repeating this until its service life. In view of this, the present application proposes several reasonable mechanical basic assumptions:

[0152] 1) If the opening angle of a certain two opening and closing doors is the same, then the damage of each door core component caused by the two opening and closing doors is the same. The present application adopts a special experimental design, so that the opening angle of each opening and closing door is the same, and the damage of each opening and closing door is the same. Assuming that the fatigue life of a certain door core component is N times, then the damage of each opening and closing door is N / 1, which is represented by D value.

[0153] 2) The damage of each door core component of the door system can be linearly accumulated. When the damage of a certain door core component accumulates to a certain extent (such as reaching a certain limit value), the door core component reaches the service life. According to Miner fatigue damage mechanics, this limit value can be assumed to be 1.

[0154] 3) In order to collect the curve of the relationship between the door opening force and the door opening angle for subsequent model order reduction sampling, it is assumed that the relationship between the door opening force and the door opening angle changes significantly only after m times of opening and closing, and that the relationship between the door opening force and the door opening angle does not change significantly within m times, which can be regarded as the same, where m can be 100 or 200 or the like.

[0155] Through the above assumptions, the number of door openings and the evolution of structural damage are linked. The fatigue life of each door core component under the test design is obtained by simulation or test method, which can perfectly solve the technical problem of establishing damage evolution for the dishwasher door system level that simulation software cannot directly realize.

[0156] In the process of constructing the damage behavior simulation model, the object is the complete door system simulation model, which can be solved by using the Transient solver. The outer door system is connected to the skeleton through a hinge, a rotary pair is set, the position of the outer door opening is set, the door body rotation is set, the total analysis time is 20s, etc. In the simulation process of the damage behavior simulation model under each damage state, the door system undergoes the same time sequence action, from the door body locking state, opening to the preset angle, and returning to the original locking state. The simulation also uses angle control synchronously, that is, the time used for opening 1° in each opening and closing simulation process is the same. Through the above simulation, the force time sequence curve of the virtual simulation machine under each damage state can be obtained (since angle control method is used, the angle time sequence curve is actually consistent, so it can be ignored), which is the output data of the damage behavior simulation model.

[0157] The step of obtaining the damage behavior simulation model under different damage states at each discrete time point can be divided into the following steps in general:

[0158] Step 1: For each door core component, at each discrete time point, calculate the damage coefficient according to the current opening and closing times and the fatigue opening and closing times limit. According to experimental data or theoretical analysis, determine the fatigue opening and closing times limit N lim of each door core component, which indicates that the door core component may fail due to fatigue when reaching the number of times. Record the actual opening and closing times N cuee of each door core component during use, then the damage coefficient can be calculated according to the formula: D i =N cuee / N lim ; where D i represents the damage coefficient of the i-th door core component, the value range is [0, 1], 0 represents no damage, and 1 represents complete failure.

[0159] Step 2: Determine the scaling coefficient of the target parameter related to door damage based on the damage coefficient of each door core component, including: determining the key parameters related to the damage of each door core component, such as the stiffness of the pull rope spring, the stiffness of the lock spring, the resilience of the sealing strip, etc. According to the damage coefficient D i of each door core component at each discrete sampling point, the scaling coefficient of the target parameter can be calculated.

[0160] For example, for the pull rope spring subsystem, the scaling coefficient of the pull rope stiffness can be expressed as: S i =1-D i , which means that as the damage coefficient increases, the target parameter (such as the pull rope stiffness) will decrease accordingly.

[0161] Step 3: At each discrete time point, update the door system simulation model based on the scaling coefficient of the target parameter of all door core components to obtain the damage behavior simulation model at that discrete time point, including: for each door core component, update the corresponding target parameter in the door system simulation model according to its scaling coefficient S i . For example, for the pull rope spring subsystem, update its initial pull rope stiffness K0 to K updated : K updated =K0*S i . Finally, run the door system simulation model using the updated target parameters to obtain the damage behavior simulation model at that discrete time point, and output the opening force time series data.

[0162] Step 4: Repeat the above steps for all preset discrete time points to obtain the damage behavior simulation model at each discrete time point. For example, starting from the first discrete time point, gradually traverse all discrete time points, and at each discrete time point, update the door system simulation model according to the current damage coefficient and the scaling coefficient of the target parameter, and run the simulation. Collect the results of the damage behavior simulation model at each discrete time point to form a complete set of damage behavior simulation models.

[0163] The present application determines the damage coefficient based on the current opening and closing times of each door core component and the fatigue opening and closing time limit, and updates the target parameters in the door system simulation model according to the damage coefficient, thereby obtaining the damage behavior simulation model under different damage states. This process provides detailed damage state data for subsequent construction of a door damage detection model, ensuring that the model can accurately reflect the dynamic damage behavior of the door system at different stages of use.

[0164] Based on the above embodiments, as an optional embodiment, the target parameters of the hinge subsystem include the deformation parameters at the connection point between the hinge and the pull cord; the target parameters of the pull cord spring subsystem include the pull cord stiffness and the extension spring stiffness; the target parameters of the latch subsystem include the locking spring stiffness; and the target parameters of the sealing subsystem include the sealing strip stiffness.

[0165] The following section uses the dishwasher door system as an example to describe the damage to its four subsystems using corresponding damage coefficients. The damage coefficients for the entire door system simulation model include:

[0166] 1) Regarding the hinge damage coefficient D1 corresponding to the hinge subsystem:

[0167] Based on the stress analysis of the hinge, the damage is described by the deformation at the connection point between the hinge and the rope. Since the hinge is a rigid body in the model and will not deform, the damage is characterized by the positional change of the connection point, that is, the deformation parameter at the connection point between the hinge and the rope.

[0168] Figure 2 This is a schematic diagram of the method for determining the damage coefficient of a hinge subsystem provided by the present invention, as shown below. Figure 2 As shown, the vertical line represents the original position of the hinge and the rope connection point, and the slanted line represents the state position of the hinge and the rope connection point after deformation. The deformation parameter of the connection point can be expressed as: R*θ. Then the damage coefficient D1=θ1 / θ0, 0≤θ1≤θ0, 0≤D1≤1. R represents the radial distance from the hinge center to the rope connection point, which is used to convert the angle change θ into the actual displacement.

[0169] 2) Regarding the rope damage coefficient D2 corresponding to the rope spring subsystem:

[0170] Considering that the pull rope in the pull rope spring subsystem is composed of nylon-wrapped steel wire rope, and its force is mainly borne by the steel wire rope, which is a pure tensile state, the change in the stiffness of the pull rope will be used in this invention to characterize a target parameter for judging damage to the pull rope spring subsystem.

[0171] Specifically, if we define the initial stiffness of the pull rope as K0, and its stiffness after undergoing multiple switching operations at any discrete time point as K1, then we can define the damage coefficient of the pull rope spring subsystem at that discrete time point as D2 = (K0 - K1) / K0. Where 0 ≤ K1 ≤ K0, 0 ≤ D2 ≤ 1.

[0172] 3) Regarding the rope damage coefficient D3 corresponding to the rope spring subsystem:

[0173] Meanwhile, the present application also takes into account that the change of the stiffness of the retractable spring in the pull rope spring subsystem is also an important target parameter reflecting the damage degree of the subsystem, and the corresponding damage formula can be expressed as: D3=(K2-K3) / K2, wherein 0≤K3≤K2, and 0≤D3≤1. Wherein, the initial stiffness of the retractable spring is K2, and the stiffness after experiencing multiple switches at any discrete time point is K3.

[0174] 4) Regarding the sealing subsystem corresponding sealing strip damage coefficient D4:

[0175] The sealing strip only has a certain rebound force due to the recovery of the sealing strip at the moment when the door body is opened, and the rebound force of the sealing strip disappears after the door body is separated from the framework subsystem. Therefore, for the determination of the damage coefficient of the sealing subsystem, the damage behavior simulation model is related to the specific damage behavior simulation model, and the most direct target parameter is the change of the stiffness of the sealing strip.

[0176] Figure 3 is the sealing strip stiffness curve provided by the present application, wherein the horizontal axis represents the compression deformation of the sealing strip under the action of external force, which is usually expressed in length units (such as millimeters or inches), and the vertical axis represents the force applied on the sealing strip, which is usually expressed in force units (such as Newton or pound force). As shown in Figure 3 , the sealing strip stiffness curve describes the force response of the sealing strip at different compression deformations, reflecting the stiffness characteristics of the sealing strip. The slope of the stiffness curve (i.e. the ratio of force to compression deformation) reflects the stiffness of the sealing strip, and the greater the slope of the stiffness curve, the harder the sealing strip; the smaller the slope of the stiffness curve, the softer the sealing strip.

[0177] Therefore, when the scaling coefficient of the target parameter related to the damage of the door body at any discrete time point of the sealing subsystem is performed, the damage degree of the sealing strip is quantified, and the corresponding damage coefficient D4 is determined, D4=0 represents no damage of the sealing strip, and D4=1 represents complete failure of the sealing strip.

[0178] Further, the scaling coefficient of the sealing strip stiffness curve can be calculated according to the damage coefficient D4: S4=1-D4, for example: if D4=0.2, the scaling coefficient of the sealing strip stiffness curve=0.8.

[0179] Then, at each discrete time point, the sealing strip stiffness curve can be updated according to the scaling coefficient S4: K5=K4× S 4, wherein K4 is the initial stiffness curve slope of the sealing strip, and K5 is the updated stiffness curve slope.

[0180] Using the updated stiffness curve to run the door body system simulation model, the damage behavior simulation model at the discrete time point can be obtained.

[0181] 5) the spring damage coefficient D5 corresponding to the lock catch subsystem:

[0182] At the initial moment of opening the door, the opening force is mainly the lock catch force generated by the lock catch, which is mainly the tensile force caused by the deformation of the locking spring. Therefore, the damage of the entire lock catch subsystem can be understood as being mainly caused by the locking spring, and the change in the stiffness of the locking spring in the door body system simulation model can be defined to correlate the corresponding damage coefficient.

[0183] Specifically, the initial stiffness of the locking spring can be defined as K6, and the stiffness after multiple opening and closing can be defined as K7, and the damage coefficient D5 = (K6-K7) / K6, wherein 0≤K7≤K6, and 0≤D5≤1.

[0184] Based on the above description, the construction of the 3D parameterized damage behavior simulation model considering damage evolution can be completed. A total of five target parameters D1-D5 associated with each door body core component are set. At each discrete time point, the corresponding damage coefficients D1-D5 are calculated according to the number of door openings. Then, according to the definition of the above damage coefficients, the door body system simulation model is automatically updated to obtain a corresponding damage behavior simulation model.

[0185] The construction method of the door body damage detection model provided by the application determines the damage coefficient based on the current opening and closing times of each door body core component and the fatigue opening and closing time limit, and updates the target parameters in the door body system simulation model according to the damage coefficient, so as to obtain the damage behavior simulation model under different damage states. This process provides detailed damage state data for subsequent construction of the door body damage detection model, ensuring that the model can accurately reflect the dynamic behavior of the door body system at different use stages.

[0186] Based on the content of the above embodiment, as an optional embodiment, any training data in the training data set is used to train the initial network model to obtain the door body damage detection model, including:

[0187] According to the preset time sequence, each training data corresponding to the discrete time point is sequentially input into the initial network model to obtain an opening force prediction sequence output by the initial network model;

[0188] According to the network error between the opening force prediction sequence and the opening force label sequence, the model parameters of the initial network model are gradually updated until the model converges, and the door body damage detection model is obtained;

[0189] The opening force label sequence is composed of the opening forces of all discrete time points on the preset time sequence.

[0190] Since the detection calculation of the door body damage is performed directly according to all the damage behavior simulation models obtained by the method mentioned in the above embodiment, there are limitations such as low calculation efficiency, limited generalization ability and complex data processing, and especially these door body damage detection models are usually based on physical equations and complex mechanical calculations, a large amount of calculation resources and time are required for each running of the simulation model, and the generated data volume is usually large, especially when multiple discrete time points and multiple damage states are considered, complex algorithms and a large amount of calculation resources are required for processing and analyzing these data, which cannot become a digital twin model suitable for actual application.

[0191] The application creatively adopts a deep learning network structure algorithm to train the data generated by the 3D parameterized damage behavior simulation model, and outputs a reduced-order damage behavior simulation model considering damage behavior, as a digital twin model of the door body system considering damage evolution.

[0192] As an optional embodiment, the application selects a long short-term memory (LSTM) network as the basic architecture of the initial network model, because the LSTM can effectively process time series data and capture the time dependence relationship in the input data. However, in actual application process, other network models can also be considered as basic model architecture for training to obtain the door body damage detection model, for example, a recurrent neural network (RNN), a gated recurrent unit (GRU), a transformer network model, or in a scene where higher detection accuracy is required, a hybrid model composed of multiple network models can be used as the basic model architecture.

[0193] Figure 4 is a training process schematic diagram of the door body damage detection model provided by the application, and the following will be described in detail Figure 4 how to realize the training of the door body damage detection model by taking the LSTM as the basic model architecture as an example, as shown in FIG. 8. Mainly including but not limited to the following steps:

[0194] Step 1, based on the method provided in the foregoing embodiment, the damage behavior simulation model corresponding to each discrete time point is obtained, so as to generate a training data set by using the damage behavior simulation models, and each data sample in the training data set includes input data and label data. The input data is the opening angle data of each discrete time point and the damage coefficient of each door body core component, and the corresponding label data is the corresponding opening force data.

[0195] Specifically, taking the door system of a general dishwasher as an example, the entire life cycle of the door system will run about 50,000 times, and every 1,000 times (according to the actual selection of different sampling frequencies), a door system simulation model is taken, and according to the assumptions provided in the above embodiments, that is, the method, the damage value coefficient of each subsystem (door core component) of the door system simulation model can be obtained each time, and according to the damage coefficient, a damage behavior simulation model corresponding to the door system simulation model can be obtained, so that 50 door system simulation models can be obtained. All door system simulation models are input into MotionSolve for solving to obtain the opening door force time series data set of the dishwasher, and according to the opening door force time series data set, a training data set for the improved LSTM deep learning neural network algorithm can be constructed.

[0196] Step 2, select an LSTM network as the initial network model architecture, which can receive the opening door angle and the damage coefficient of each door core component, the LSTM layer processes the time series data and extracts the time-dependent features, and the output layer outputs the predicted opening door force.

[0197] Step 3, according to the preset time sequence, the training data corresponding to each discrete time point is input into the initial network model in turn, including the input data and label data recorded in the training data set for each discrete time point, and the input data (opening door angle and damage coefficient) is input into the initial network model.

[0198] Run the initial network model to obtain the opening door force prediction sequence output by it, that is, use the initial network model to infer the input data to generate the opening door force prediction sequence (collect the prediction results of each discrete time point to form a complete opening door force prediction sequence).

[0199] Step 4, calculate the network error according to the difference between the prediction sequence and the label sequence. Obtain the true opening door force data of each discrete time point from the training data set to form the opening door force label sequence. Use mean square error (MSE) or other statistical indicators to calculate the network error between the prediction sequence and the label sequence, for example:

[0200] ;

[0201] Where N is the total number of discrete time points, and represent the opening door force error of the i-th discrete time point.

[0202] Step 5, the parameters of the initial network model can be updated step by step according to the calculated network error until the model converges, and the specific steps include:

[0203] 1) Use the backpropagation algorithm to calculate the gradient of the error with respect to the model parameters.

[0204] 2) Update model parameters according to gradient using optimization algorithm (such as Adam).

[0205] 3) Repeat the above process until the network error no longer decreases significantly, and the model converges.

[0206] Step 6, after the above training process, the parameters of the initial network model are gradually optimized, and a high-precision door damage detection model is finally obtained. The door damage detection model can accurately predict the corresponding opening force according to the input opening angle and damage coefficient, thereby realizing real-time detection of the door damage state.

[0207] The construction method of the door damage detection model provided by the application trains the initial network model using the training data set, gradually optimizes the model parameters, and finally obtains a high-precision door damage detection model, which can significantly improve the calculation efficiency, enhance the generalization ability, simplify the data processing, and meet the needs of real-time monitoring and complex working conditions. It is a key step to realize efficient and accurate door damage detection.

[0208] Figure 5 is a structural schematic diagram of the construction device of the door damage detection model provided by the application, as Figure 5 shown, mainly includes but is not limited to:

[0209] The damage behavior simulation model construction unit 51 is used to construct the damage behavior simulation model of each discrete time point based on the physical benchmark data and the damage coefficients of each door core component at each discrete time point on the preset time sequence.

[0210] Among them, the physical benchmark data is obtained by performing physical opening and closing experiments on each door core component of the real physical door;

[0211] The initial network model training control unit 52 is used to train the initial network model using the training data set, and obtain the door damage detection model.

[0212] Among them, any training data in the training data set includes an opening angle of the damage behavior simulation model at its corresponding discrete time point and damage coefficients of each core component, and the label of the training data is the opening force at the discrete time point.

[0213] It should be noted that the construction device of the door damage detection model provided by the application can realize the construction method of the door damage detection model provided by any of the above embodiments when it is actually run, and will not be described here.

[0214] The application provides a door body damage detection model construction device, a simulation model is calibrated through model test design, a 3D parameterized damage behavior simulation model considering damage is established, and finally, a network model capable of realizing rapid and accurate real-time monitoring of door body damage is obtained through deep learning on the discrete damage behavior simulation model, thereby filling the industry gap.

[0215] In order to fully illustrate the advantages of the door body damage detection model construction method provided by the application in actual detection, the following will be described in detail in combination with related test data.

[0216] Figure 6 is one of the performance schematic diagrams of the door body damage detection model provided by the application in the training data set, Figure 7 is another performance schematic diagram of the door body damage detection model provided by the application in the training data set, Figure 8 is a third performance schematic diagram of the door body damage detection model provided by the application in the training data set, Figure 9 is a fourth performance schematic diagram of the door body damage detection model provided by the application in the training data set, Figure 10 is a fifth performance schematic diagram of the door body damage detection model provided by the application in the training data set. Wherein, the horizontal coordinates all represent time steps, each unit corresponds to 0.025 seconds, for example, 400 on the horizontal coordinate represents 10 seconds; the vertical coordinates all represent the numerical values of the predicted values and the actual values, for example, the opening door force, the angle or other related parameters in the representation, the unit is Newton (N), one of the two force time curves represents the predicted value, and the other represents the actual value (the dashed line represents the predicted value, and the solid line represents the actual value).

[0217] In the initial network model training process shown in Figures 6-10 , the total opening door times (the data size in the training data set) are 20900, 27100, 34800, 4200 and 9700, respectively, and the comparison relationship between the predicted value and the actual value is mainly compared, wherein the horizontal coordinate also represents the time step, and the vertical coordinate represents the opening door force. The force time curve fully demonstrates the convergence of the door body damage detection model in the training process, and can effectively prove that the door body damage detection model obtained after a limited number of training has high prediction accuracy.

[0218] Figure 11 is one of the performance schematic diagrams of the door body damage detection model provided by the application in the test data set, Figure 12 is another performance schematic diagram of the door body damage detection model provided by the application in the test data set, Figure 13 is a third performance schematic diagram of the door body damage detection model provided by the application in the test data set, Figure 14 is a fourth performance schematic diagram of the door body damage detection model provided by the application in the test data set,Figure 15 Figure 5 is a schematic view of performance of the door damage detection model provided by the present application on a test data set.

[0219] wherein, in the initial network model training process shown in Figures 11-15 In the initial network model training process shown in the figure, the total number of door opening times (the data size in the test data set) is 10400, 11400, 12200, 12900 and 13600, respectively. By comparing the predicted value with the actual value, the convergence of the door damage detection model in the test process is shown, which can help to evaluate the generalization ability and prediction accuracy of the model on unseen data. Through the verification of the test data set, it can be determined that the door damage detection model obtained by the present application has good generalization and high prediction accuracy.

[0220] Figure 16 Figure 1 is a schematic view of the evolution of the loss function value of the door damage detection model provided by the present application in the training process, Figure 17 Figure 2 is a schematic view of the evolution of the loss function value of the door damage detection model provided by the present application in the training process. Wherein, the horizontal coordinates all represent the number of iterations, and the vertical coordinates all represent the loss function value. The curve is mainly used to represent the change of the loss function value with the increase of the number of training iterations. If the loss function value decreases rapidly and tends to be stable, it means that the model converges quickly. If the loss function value tends to a small value eventually, it means that the model training effect is good.

[0221] As shown in the figure, Figures 16-17 As shown in the figure, the loss function value Loss of the door damage detection model at the 171st step reaches 2.53, with high convergence accuracy. Thus, a deep learning network model with high fitting accuracy is obtained as the damage behavior digital twin model of the dishwasher door system.

[0222] The above experiment fully proves the construction method of the door damage detection model provided by the present application, and a LSTM deep learning network model with high fitting accuracy is obtained, which has the ability to effectively utilize the time sequence characteristics and solve the problem of long time sequence prediction.

[0223] Figure 18 Figure 1 is a schematic view of the structure of an electronic device provided by the present application, such as Figure 18As shown, the electronic device can include a processor 1810, a communications interface 1820, a memory 1830, and a communications bus 1840, wherein the processor 1810, the communications interface 1820, and the memory 1830 complete mutual communication through the communications bus 1840. The processor 1810 can invoke a logical instruction in the memory 1830 to execute a construction method of a door body damage detection model, the method comprising: constructing a damage behavior simulation model of each discrete time point based on physical benchmarking data and damage coefficients of each door body core component at the preset time sequence on each discrete time point; the physical benchmarking data is obtained by performing a physical opening and closing experiment on each door body core component of a real physical door; training an initial network model using a training data set to obtain the door body damage detection model; any training data in the training data set includes an opening angle of the damage behavior simulation model at its corresponding discrete time point and the damage coefficients of each door body core component, and the label of the training data is the opening force at the discrete time point.

[0224] In addition, the logical instructions in the memory 1830 described above can be implemented in the form of a software functional unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0225] In another aspect, the present application also provides a computer program product, which comprises a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions that, when executed by a computer, enable the computer to perform the method for constructing a door damage detection model provided by any of the above embodiments, the method comprising: constructing a damage behavior simulation model of each discrete time point based on physical benchmarking data and damage coefficients of each door core component at each discrete time point on a preset time sequence; the physical benchmarking data is obtained by performing a physical opening and closing experiment on each door core component of a real physical door; training an initial network model using a training data set to obtain the door damage detection model; any training data in the training data set comprises an opening angle of the damage behavior simulation model at its corresponding discrete time point and the damage coefficients of each door core component, and a label of the training data is an opening force at the discrete time point.

[0226] In yet another aspect, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for constructing a door damage detection model provided by any of the above embodiments, the method comprising: constructing a damage behavior simulation model of each discrete time point based on physical benchmarking data and damage coefficients of each door core component at each discrete time point on a preset time sequence; the physical benchmarking data is obtained by performing a physical opening and closing experiment on each door core component of a real physical door; training an initial network model using a training data set to obtain the door damage detection model; any training data in the training data set comprises an opening angle of the damage behavior simulation model at its corresponding discrete time point and the damage coefficients of each door core component, and a label of the training data is an opening force at the discrete time point.

[0227] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the present embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0228] Those skilled in the art can clearly understand the technical solutions of the various embodiments from the above description of the embodiments, and the various embodiments can be implemented by means of software with the necessary general hardware platforms, and of course, can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that makes a contribution, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0229] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for constructing a door damage detection model, characterized in that, The method comprises the following steps: Based on the physical benchmark data and the damage coefficient of each door core component at each discrete time point on the preset time sequence, a damage behavior simulation model at each discrete time point is constructed; the physical benchmark data is obtained by physical opening and closing experiment on each door core component of the real physical door; An initial network model is trained using a training data set to obtain a door damage detection model; any training data in the training data set includes an opening angle of the damage behavior simulation model at its corresponding discrete time point and the damage coefficient of each door core component, and the label of the training data is the opening force at the discrete time point; The damage behavior simulation model at each discrete time point is constructed based on the physical benchmark data and the damage coefficient of each door core component at each discrete time point on the preset time sequence, which comprises: The normal behavior simulation model of each door core component is obtained by benchmarking each lossless simulation model using the physical benchmark data; The normal behavior simulation models are combined to form a door system simulation model; The door system simulation model is updated according to the damage coefficient of each door core component at each discrete time point on the preset time sequence, and the damage behavior simulation model under different damage states at each discrete time point is obtained, which comprises: Based on the current opening and closing times of any door core component at any discrete time point and the fatigue opening and closing time limit of the any door core component, the damage coefficient of the any door core component at the any discrete time point is determined; Based on the damage coefficient, the scaling coefficient of the target parameter related to door damage of the any door core component is determined; At the any discrete time point, the door system simulation model is updated according to the scaling coefficients of all target parameters related to door damage of the door core components, and the damage behavior simulation model at the any discrete time point is obtained; All the damage behavior simulation models are obtained by traversing all the discrete time points.

2. The method of claim 1, wherein the method further comprises: The physical benchmark data includes force time sequence curve samples obtained by physical opening and closing experiment under angle control for each door core component; The normal behavior simulation model of each door core component is obtained by benchmarking each lossless simulation model using the physical benchmark data, which comprises: For the lossless simulation model of any door core component, the model parameters of the lossless simulation model are adjusted according to the curve difference between the virtual force time sequence curve output by the lossless simulation model and the force time sequence curve sample, until the curve difference is within a preset range, and the adjusted lossless simulation model is obtained as the normal behavior simulation model.

3. The method of claim 1 or 2, wherein the method further comprises: The door core component includes an outer door subsystem and all key subsystems, the key subsystems include one or more of a hinge subsystem, a pull rope spring subsystem, a lock catch subsystem and a sealing subsystem; the hinge subsystem is composed of a skeleton subsystem and an outer door subsystem. Correspondingly, the non-damage simulation model includes an outer door subsystem simulation model, a hinge subsystem simulation model, a pull rope spring subsystem simulation model, a lock catch subsystem simulation model, and a sealing subsystem simulation model.

4. The method of claim 3, wherein the method further comprises: The normal behavior simulation model is used to simulate the normal behavior of each door body core component at each discrete time point in the preset time sequence. The normal behavior simulation model is used to simulate the normal behavior of each door body core component at each discrete time point in the preset time sequence.

5. The method of claim 3, wherein the method further comprises: The target parameters of the hinge subsystem include a deformation parameter at a hinge and pull rope connection point, the target parameters of the pull rope spring subsystem include pull rope stiffness and extension spring stiffness, the target parameters of the lock catch subsystem include lock spring stiffness, and the target parameters of the sealing subsystem include sealing strip stiffness.

6. The method of claim 1, wherein the method further comprises: The initial network model is trained by using any training data in the training data set, and the door body damage detection model is obtained. According to the network error between the opening door force prediction sequence and the opening door force label sequence, the model parameters of the initial network model are gradually updated until the model converges, and the door body damage detection model is obtained. The opening door force label sequence is composed of the opening door forces of all discrete time points in the preset time sequence. The initial network model is obtained based on a long short-term memory network model.

7. The method of claim 1, wherein the method further comprises: The initial network model is obtained based on a long short-term memory network model. 8.A device for constructing a door damage detection model, characterized in that, The damage behavior simulation model construction unit is used to construct a damage behavior simulation model at each discrete time point based on physical benchmark data and damage coefficients of each door body core component at each discrete time point in the preset time sequence. The normal behavior simulation model is used to simulate the normal behavior of each door body core component at each discrete time point in the preset time sequence. According to the damage coefficients of each door body core component at each discrete time point in the preset time sequence, the door body system simulation model is updated to obtain the damage behavior simulation model under different damage states at each discrete time point. Based on the current opening and closing times of any door body core component at any discrete time point and the fatigue opening and closing time limit of the any door body core component, the damage coefficient of the any door body core component at the any discrete time point is determined. Based on the damage coefficient, the scaling coefficient of the target parameter related to door body damage of the any door body core component is determined. ​ At any discrete time point, the door system simulation model is updated according to the scaling coefficients of all door body core components and target parameters related to door body damage, to obtain the damage behavior simulation model at the discrete time point; All the damage behavior simulation models are obtained by traversing all the discrete time points; An initial network model training control unit is configured to train an initial network model using a training data set to obtain the door body damage detection model; any training data in the training data set includes an opening angle of a door body at a corresponding discrete time point and damage coefficients of each core component of the damage behavior simulation model, and a label of the training data is an opening force at the discrete time point.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the method for constructing the door body damage detection model according to any one of claims 1 to 7. 10.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the method for constructing the door body damage detection model according to any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the method for constructing the door body damage detection model according to any one of claims 1 to 7.

Images