Washing machine AI control model training method and device and washing machine

Abstract

The invention discloses a training method and device of a washing machine AI control model and a washing machine, and relates to the technical field of artificial intelligence application. Setting a model training module, wherein a peripheral parameter fault threshold is preset; each washing machine comprises a first control module, a dewatering motor, a load obtaining module and an eccentricity obtaining module, wherein the dewatering motor, the load obtaining module and the eccentricity obtaining module are connected with the first control module. The first control module is in communication connection with the model training module; a plurality of peripheral detection modules are arranged, are respectively connected with each washing machine complete machine, are used for acquiring peripheral parameters, and are in communication connection with the model training module; circularly obtaining the dehydration load, the dehydration eccentricity and the peripheral parameters in the accelerated dehydration control stage; judging whether the peripheral parameters reach or exceed a peripheral parameter fault threshold value; and if yes, collecting the peripheral parameters, the corresponding dehydration load and the dehydration eccentricity as first training data, and training an accelerated dehydration AI control model applied to the accelerated dehydration control stage of the washing machine. The training efficiency and accuracy of the AI control model are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence application, and specifically, relates to a training method and device for an AI control model of a washing machine, and a washing machine. Background Art

[0002] At present, the training data used to develop intelligent control of washing machine systems mainly comes from the operating data of the washing machine itself. The lack of data representing the operating status of the machine makes the trained model for controlling the washing machine less accurate and less intelligent. In addition, the data on the operating status of the machine needs to be manually recorded and imported, which results in low training efficiency.

[0003] The above information disclosed in the background technology is only used to increase the understanding of the background technology of the present application, and therefore, it may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0004] In view of the problems in the prior art of only using the washing machine's own operating data as training data, which leads to low training efficiency, low accuracy of the intelligent control model and insufficient intelligence, the present invention proposes a training method, device and washing machine for an AI control model of a washing machine, and automatically acquires external peripheral data representing the operating status of the machine to participate in the training of the AI ​​control model, thereby improving the training efficiency and the accuracy and intelligence of the AI ​​control model.

[0005] In order to achieve the above invention / design purpose, the present invention adopts the following technical solution to implement it.

[0006] A training method for a washing machine AI control model, comprising: S1. Set up the model training module and preset the peripheral parameter fault threshold; S2, setting a plurality of experimental washing machines, which include a first control module and a dehydration motor connected to the first control module, a load acquisition module, and an eccentricity acquisition module, for acquiring the load and the eccentricity; the first control module is communicatively connected to the model training module; S3, setting a plurality of peripheral detection modules, respectively connected to each of the washing machines, for obtaining peripheral parameters, communicating with the model training module, and outputting the peripheral parameters to the model training module; S4, placing clothes with an initial load in the washing machine for dehydration; cyclically obtaining the dehydration load, dehydration eccentricity, and the peripheral parameters in the accelerated dehydration control stage; S5, determining whether the peripheral parameter reaches or exceeds the peripheral parameter failure threshold; if yes, executing S6; if no, executing S4; S6: Collect the peripheral parameters, the corresponding dehydration load, and the dehydration eccentricity as first training data for training an accelerated dehydration AI control model.

[0007] In some specific embodiments, the peripheral detection module includes a second control module and a vibration sensor and a force sensor connected to the second control module; the vibration sensor and the force sensor are used to detect the vibration value and the force value of the washing machine and transmit them to the second control module; the second control module is connected to the model training module for communication, and transmits the obtained vibration value and the force value to the model training module; the model training module recognizes them as the peripheral parameters; The peripheral parameter fault threshold includes a vibration threshold and a force threshold; judging whether the peripheral parameter reaches or exceeds the peripheral parameter fault threshold is that the vibration value reaches or exceeds the vibration threshold and / or the force value reaches or exceeds the force threshold.

[0008] In some specific embodiments, the peripheral detection module includes a second control module and a noise sensor and a force sensor connected to the second control module; the noise sensor and the force sensor are used to detect the noise value and the force value of the washing machine and transmit them to the second control module; the second control module is connected to the model training module for communication, and transmits the obtained noise value and the force value to the model training module; the model training module recognizes them as the peripheral parameters; The peripheral parameter fault threshold includes a noise threshold and a force threshold; judging whether the peripheral parameter reaches or exceeds the peripheral parameter fault threshold is that the noise value reaches or exceeds the noise threshold and / or the force value reaches or exceeds the force threshold.

[0009] In some specific embodiments, it also includes: S1', presetting a plurality of different distribution parameters in the model training module, which include a first rotation speed and a first acceleration; presetting a plurality of distribution eccentricity thresholds corresponding to different distribution loads; S2', the model training module sends the distribution parameters to the first control module to control the distribution control stage of the operation of each washing machine; when the distribution control stage is completed, the distribution load and distribution eccentricity of the washing machine are obtained; S3', determine whether the distribution eccentricity reaches or exceeds the eccentricity threshold corresponding to the distributed load; if yes, execute S4'; if no, execute S5'; S4', termination of dehydration control; S5', obtaining the initial load, the corresponding distribution parameters, the distributed load, and the distributed eccentricity and collecting them as second training data for training the distributed AI control model.

[0010] A training device for a washing machine AI control model, comprising a plurality of washing machines, a plurality of peripheral detection modules, and a model training module; Each washing machine is used to load clothes with an initial load, and includes a first control module, a dehydration motor connected thereto, a load acquisition module, and an eccentricity acquisition module; Each of the peripheral detection modules is connected to each of the washing machines respectively, and is used to detect the peripheral parameters of the washing machine when the washing machine is running; The model training module is a hardware entity with a processor and a memory, which is respectively communicated with each of the first control modules and each of the peripheral detection modules, and is configured with a data training tool, with a preset peripheral parameter fault threshold, and is configured to obtain the peripheral parameters, dehydration load, and dehydration eccentricity in real time during the accelerated dehydration control stage of each washing machine; determine whether the peripheral parameters reach the peripheral parameter fault threshold; if so, collect the dehydration load and the dehydration eccentricity and collect them as the first training data for training the accelerated dehydration AI control model.

[0011] In some specific embodiments, the peripheral detection module includes a second control module and a vibration sensor and four force sensors connected to the second control module; the vibration sensor and each force sensor are used to detect the vibration value of the whole washing machine and the force vector of each foot and transmit them to the second control module; the second control module is communicatively connected with the model training module, and transmits the vibration value and each force vector to the model training module; the model training module obtains the force value of the whole washing machine from each force vector, and identifies the vibration value and the force value as the peripheral parameters; The peripheral parameter fault threshold includes a vibration threshold and a force threshold; determine whether the vibration value reaches or exceeds the vibration threshold and / or whether the force value reaches the force threshold; if so, collect the dehydration load and the dehydration eccentricity as the first training data for training the accelerated dehydration AI control model.

[0012] In some specific embodiments, the peripheral detection module includes a second control module and a noise sensor and four force sensors connected to the second control module; the noise sensor and each of the force sensors are used to detect the noise value and the force vector of each foot and transmit them to the second control module; the second control module is communicatively connected to the model training module, and transmits the noise value and each of the force vectors to the model training module; the model training module obtains the force value of the whole washing machine from each of the force vectors, and identifies the noise value and the force value as the peripheral parameters; The peripheral parameter fault threshold includes a noise threshold and a force threshold; determine whether the noise value reaches or exceeds the noise threshold and / or whether the force value reaches the force threshold; if so, collect the dehydration load and the dehydration eccentricity as the first training data for training the accelerated dehydration AI control model.

[0013] In some specific embodiments, the model training module also presets multiple groups of distribution parameters and eccentricity thresholds corresponding to different distribution loads, and is configured as follows: Sending the distribution parameters to each of the first control modules, controlling each of the washing machines to operate according to the distribution parameters during the distribution control phase, and returning the distribution load and distribution eccentricity measured after the distribution control phase ends; Determine whether the distribution eccentricity reaches or exceeds the eccentricity threshold corresponding to the distribution load; if so, stop dehydration control; if not, control to enter the accelerated dehydration control stage, and collect the initial load, the distribution parameters, the distributed load, and the distribution eccentricity as the second training data for training the distribution AI control model.

[0014] A washing machine comprises a first control module, a dehydration motor, a load acquisition module and an eccentricity acquisition module respectively connected thereto; the first control module is configured with an accelerated dehydration AI control model obtained by the training device of the above-mentioned washing machine AI control model, which detects or predicts whether a dehydration failure occurs based on the dehydration load and dehydration eccentricity obtained in real time; and if so, controls the dehydration to be interrupted.

[0015] In some specific embodiments, the first control module is also configured with a distributed AI control model, which uses corresponding distribution parameters to control the operation of the distribution control stage according to the initial load, so that the distribution eccentricity after the distribution control stage runs meets the requirements for entering the accelerated dehydration stage.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: The training method, device and washing machine of the washing machine AI control model of the present invention train the accelerated dehydration AI control model by combining the load and eccentricity data obtained by the load and eccentricity acquisition module of the washing machine itself and the peripheral parameter data obtained by the peripheral detection module arranged outside the washing machine to form the first training data for accelerating the control of the dehydration control stage; the data obtained and output by the washing machine itself and the external data obtained by the peripheral detection module of the washing machine are combined to form the first training data, thereby expanding the dimension of the data used to train the AI ​​control model and improving the accuracy and precision of the training; the acquisition of training data and the training of the AI ​​control model are automatically realized through communication with the washing machine and the peripheral detection module, thereby improving the degree of automation and further improving the efficiency of training data collection and the efficiency of AI control model training.

[0017] After reading the specific embodiments of the present invention in conjunction with the accompanying drawings, other features and advantages of the present invention will become more clear. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 2 is a schematic diagram of the structure of a training device for a washing machine AI control model according to an embodiment; Figure 2 is a schematic diagram of a dehydration process of a washing machine according to an embodiment; Figure 3 is a flow chart of a method for training a washing machine AI control model according to an embodiment; Figure 4 is a schematic diagram of a first training data acquisition process according to an embodiment; Figure 5 is a schematic diagram of a first training data acquisition process according to an embodiment; Figure 6 is a schematic diagram of a second training data acquisition process according to an embodiment; Figure 7 2 is a schematic diagram of the structure of a training device for a washing machine AI control model according to an embodiment; Figure 8 It is a schematic diagram of the composition structure of a training device for a washing machine AI control model according to an embodiment.

[0020] In the figure, 1. Model training module; 2. Washing machine; 3. Peripheral detection module; 4. First communication module; 21. First control module; 22. Dehydration motor; 23. Load acquisition module; 24. Eccentricity acquisition module; 31. Second communication module; 32. Second control module; 33. Force sensor; 34. Vibration sensor; 35. Noise sensor. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In the description of the present invention, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore, should not be understood as a limitation on the present invention.

[0023] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. In the description of the implementation method, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0024] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features.

[0025] In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0026] Reference Figure 2 The present invention discloses a training method for an AI control model of a washing machine, and the AI ​​control model obtained by training is applied to the control of the dehydration process of the washing machine.

[0027] The dehydration process of the washing machine includes a distribution control stage, corresponding to Figure 1 The AC section in the washing machine is used to control the rotation speed of the inner drum of the washing machine so that the clothes to be dehydrated are distributed in the inner drum and the eccentricity is reduced; generally, the washing machine is accelerated to a first rotation speed and maintained at the first rotation speed for a first time; and then the washing machine is accelerated from the first rotation speed to the second rotation speed at a first acceleration for a second time to evenly distribute the clothes in the inner drum.

[0028] The dehydration process of the washing machine also includes an accelerated dehydration judgment stage and an accelerated dehydration control stage; the accelerated dehydration judgment stage is to maintain the second speed for a third time, corresponding to Figure 1 The CD segment in the accelerated dehydration judgment stage is used, and the distribution load and distribution eccentricity are detected or obtained, and it is determined whether to enter the accelerated dehydration control stage according to the obtained distribution load and distribution eccentricity; the accelerated dehydration control stage is to deepen the dehydration of the clothes in the inner drum at the second acceleration for the fourth time length, corresponding to Figure 1 The washing machine that has passed this stage will not have any dehydration failure in the subsequent dehydration stage. Dehydration failures include vibration failures, movement failures, etc.

[0029] A vibration fault is when the vibration value or noise value of the washing machine 2 reaches or exceeds a specified threshold value; a movement fault is when the force value of the washing machine 2 reaches or exceeds a specified threshold value.

[0030] The distributed load is the load in the inner drum obtained in the accelerated dehydration judgment stage; the distributed eccentricity is the eccentricity of the clothes in the inner drum obtained in the accelerated dehydration judgment stage.

[0031] Reference Figure 1 , Figure 2 , Figure 3 The training method of the washing machine AI control model includes obtaining the parameters of the operation process in the accelerated dehydration control stage as the first training data for training the accelerated dehydration AI control model, and the accelerated dehydration AI control model is used for the intelligent control of the accelerated dehydration control stage of the washing machine, specifically including: S1, setting a model training module 1, which presets a peripheral parameter fault threshold; S2, setting a plurality of washing machine complete machines 2, which are respectively connected to the model training module 1 in communication; the washing machine complete machine 2 includes a first control module 21 and a dehydration motor 22, a load acquisition module 23, and an eccentricity acquisition module 24 connected to the first control module 21; the dehydration load and dehydration eccentricity of the washing machine complete machine 2 in the accelerated dehydration control stage are obtained by the load acquisition module 23 and the eccentricity acquisition module 24 and transmitted to the model training module 1; the dehydration load is the real-time load in the accelerated dehydration control stage; the dehydration eccentricity is the real-time eccentricity in the accelerated dehydration control stage; S3, corresponding to each washing machine 2, a peripheral detection module 3 is respectively set; the peripheral parameters of the washing machine 2 in the accelerated dehydration control stage are obtained through the peripheral detection module 3 and transmitted to the model training module 1; S4, the model training module 1 cyclically obtains the dehydration load, dehydration eccentricity, and peripheral parameters in the accelerated dehydration control stage; S5, determining in real time whether the peripheral parameter reaches or exceeds the peripheral parameter fault threshold; if yes, executing S6; if no, executing S4; S6. Obtain the dehydration eccentricity and dehydration load of the corresponding peripheral parameters, and collect them as the first training data for training the accelerated dehydration AI control model.

[0032] The training method of the washing machine AI control model of the present invention combines the load and eccentricity data obtained by the load and eccentricity acquisition module 24 of the washing machine 2 itself and the peripheral parameter data obtained by the peripheral detection module 3 arranged outside the washing machine 2 to form the first training data to train the accelerated dehydration AI control model for the control of the accelerated dehydration control stage; the data obtained and output by the washing machine 2 itself and the external data obtained by the peripheral detection module 3 of the washing machine 2 are combined to form the first training data, so as to expand the dimension of the data used to train the AI ​​control model and improve the accuracy and precision of the training; the acquisition of training data and the training of the AI ​​control model are automatically realized through communication with the washing machine 2 and the peripheral detection module 3, so as to improve the degree of automation and further improve the efficiency of training data collection and the efficiency of AI control model training.

[0033] The specific process and principle of the training method of the washing machine AI control model of the present invention are described in detail below through specific embodiments.

[0034] In some specific embodiments, reference Figure 1 , Figure 2 , Figure 3 , Figure 4 The training method of the accelerated dehydration AI control model specifically includes: S7, the model training module 1 presets the peripheral parameter fault threshold including the vibration threshold and the force threshold, and is connected to the multiple washing machine complete machines 2 for communication; S8. A peripheral detection module 3 is respectively provided for each washing machine 2; the peripheral detection module 3 includes a second control module 32 and a vibration sensor 34 and a force sensor 33 connected to the second control module 32; the second control module 32 is in communication connection with the model training module 1; the model training module 1 cyclically obtains the vibration value and force value of the washing machine 2 during the accelerated dehydration control stage of the washing machine 2 through the peripheral detection module 3, and cyclically obtains the dehydration load and dehydration eccentricity during the accelerated dehydration control stage through the load acquisition module 23 and the eccentricity acquisition module 24 of the washing machine 2 itself; the dehydration load is the real-time load during the accelerated dehydration control stage; the dehydration eccentricity is the real-time eccentricity during the accelerated dehydration control stage; S9, the model training module 1 determines in real time whether the vibration value reaches or exceeds the vibration threshold and / or whether the force value reaches or exceeds the force threshold; if yes, execute S10; if no, execute S8; S10, obtaining the dehydration eccentricity and dehydration load corresponding to the vibration value and / or the force value, and collecting them as the first training data for training the accelerated dehydration AI control model.

[0035] When the vibration value reaches or exceeds the vibration threshold, a vibration fault occurs; when the force value reaches or exceeds the force threshold, a movement fault occurs; and a movement fault means that the force value of the washing machine includes the lateral force that overcomes the friction force to move the washing machine horizontally.

[0036] In some specific embodiments, reference Figure 1 , Figure 2 , Figure 3 , Figure 5 The training method of the accelerated dehydration AI control model specifically includes: S7', the model training module 1 presets a noise threshold and a stress threshold, and communicates with a plurality of washing machines 2; that is, the peripheral parameter fault threshold includes a noise threshold and a stress threshold; S8', respectively set up peripheral detection modules 3 for each washing machine 2; the peripheral detection module 3 includes a second control module 32 and a noise sensor 35 and a force sensor 33 connected to the second control module 32; the second control module 32 is connected to the model training module 1 in communication; the model training module 1 obtains the noise value and force value of the washing machine 2 during the accelerated dehydration control stage of the washing machine 2 through the peripheral detection module 3, and obtains the dehydration load and dehydration eccentricity of the accelerated dehydration control stage in real time through the load acquisition module 23 and eccentricity acquisition module 24 of the washing machine 2 itself; the dehydration load is the real-time load during the accelerated dehydration control stage; the dehydration eccentricity is the real-time eccentricity during the accelerated dehydration control stage; S9', the model training module 1 determines in real time whether the noise value reaches or exceeds the noise threshold and / or whether the force value reaches or exceeds the force threshold; if yes, execute S10'; if no, execute S8'; S10', obtaining the dehydration eccentricity and dehydration load corresponding to the noise value and / or the force value, and collecting them as the first training data for training the accelerated dehydration AI control model.

[0037] When the vibration value exceeds the vibration threshold, a vibration fault occurs; when the force value exceeds the force threshold, a movement fault occurs; and when a movement fault occurs, the force value of the washing machine includes a lateral force that overcomes the friction force to move the washing machine horizontally.

[0038] In some specific embodiments, reference Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 The training method of the washing machine AI control model also includes obtaining parameters of distributed loads and distributed eccentricities corresponding to different initial loads, first speeds, and first accelerations as training data for training the distributed AI control model in the distributed control stage, specifically including: S1', presetting a plurality of different distribution parameters in the model training module 1, which include a first rotation speed and a first acceleration; presetting a plurality of distribution eccentricity thresholds corresponding to different distribution loads; S2', the model training module 1 sends the distribution parameters to the first control module 21 to control the operation of each washing machine 2 in the distribution control stage; S3', when the distribution control stage is completed, that is, the dehydration judgment stage is accelerated, the distribution load and distribution eccentricity of the washing machine 2 are obtained; S4', determine whether the distribution eccentricity reaches or exceeds the eccentricity threshold corresponding to the distributed load; if yes, execute S5'; if no, execute S6'; S5', terminate dehydration control; S6', obtaining the initial load, the corresponding distribution parameters, the distributed load, and the distributed eccentricity as the second training data for training the distributed AI control model; and entering the accelerated dehydration control stage.

[0039] The training method of the washing machine AI control model in this embodiment obtains the distributed AI control model by obtaining the initial load and distribution parameter training that meet the distribution eccentricity threshold, which is used to control the distributed control stage of the dehydration process to avoid the situation where the distribution eccentricity exceeds the eccentricity threshold, thereby improving the intelligence of the dehydration process control, the accuracy of the control and the efficiency of the dehydration.

[0040] In some specific embodiments, the eccentricity can be obtained by installing an acceleration sensor at the bearing of the inner drum of the washing machine or the outer drum support structure to collect the acceleration signal of the inner drum of the washing machine in real time when it rotates; Fourier transform the acceleration signal, analyze the amplitude changes of the fundamental frequency and harmonic components, and obtain the eccentricity value. That is, the eccentricity value is measured by the amplitude change value of the fundamental frequency and harmonic components of the acceleration.

[0041] In some specific embodiments, the acceleration is obtained by calculating the speed change rate by differentially calculating the speed data of the dehydration motor 22 in the uniform speed stage, and the difference between the maximum acceleration and the minimum acceleration in the continuous interval is used to measure the eccentricity in the accelerated dehydration judgment stage.

[0042] In some specific embodiments, the real-time torque current of the dehydration motor 22 is obtained and the real-time torque current signal is subjected to low-pass digital filtering to eliminate high-frequency noise interference; the maximum torque current and the minimum torque current in a plurality of consecutive dehydration cycles are extracted, and the average values ​​of the maximum torque current and the minimum torque current are respectively obtained, and the difference between them is calculated to represent the eccentricity value. That is, e=K(Imax_avg-Imin_avg ) in, e is the eccentricity; K is the model calibration factor to avoid model errors; Imax_avg is the average value of each maximum torque current; Imin_avg is the average value of each minimum torque current.

[0043] In some specific embodiments, in the training method of the washing machine AI control model, multiple washing machine machines 2 are respectively placed with different initial loads, or multiple washing machine machines 2 are grouped to place clothes with different initial loads, and each washing machine 2 or each group of washing machine 2 takes turns to place different initial loads, and at the same time receive the distribution parameters sent by the model training module 1, and control the distribution control stage of each washing machine 2 with the received distribution parameters, so as to obtain the first training data and the second training data.

[0044] The training method of the washing machine AI control model of the present invention obtains first training data and second training data by placing clothes with different initial loads in multiple washing machines 2 or placing clothes with different initial loads in groups and controlling them with the same distribution parameters at the same time, thereby improving the efficiency of obtaining the first training data and the second training data and the breadth and dimension of the data, and improving the efficiency and accuracy of model training.

[0045] In some specific embodiments, the accelerated dehydration AI control model is a first fitting function of the dehydration load and the dehydration eccentricity obtained by training with the first training data, and the washing machine intelligently determines whether a peripheral parameter failure occurs based on whether the self-tested dehydration load and dehydration eccentricity satisfy the first fitting function, and interrupts dehydration when a peripheral parameter failure occurs; otherwise, continues to accelerate the dehydration control.

[0046] The training method of the washing machine AI control model of this embodiment obtains a first fitting function by fitting the first training data, so that the washing machine can intelligently determine whether to interrupt dehydration according to whether the dehydration load and dehydration eccentricity meet the first fitting function, thereby protecting the washing machine and extending the life of the washing machine.

[0047] In some specific embodiments, the distribution control model is a second fitting function of the initial load and distribution parameters obtained through the second training data, so that the washing machine intelligently configures different distribution parameters according to the different initial loads and the second fitting function measured by self-test to control the operation of the distribution control stage of the washing machine, so that the distribution eccentricity after operation in the distribution control stage meets the requirements for entering the accelerated dehydration control stage, thereby improving the dehydration rate.

[0048] The training method of the washing machine AI control model of the present embodiment obtains the second fitting function by fitting the second training data, so that the washing machine can intelligently configure the distribution parameters according to the initial load, so that the distribution eccentricity meets the requirements for accelerated dehydration control, and reduces the probability of peripheral parameter failure when entering the accelerated dehydration control from the distribution control stage, thereby improving the dehydration efficiency and protecting the washing machine.

[0049] In some specific embodiments, when the distributed control stage of the washing machine 2 is controlled to operate with distributed parameters, the actual speed and acceleration values ​​of the dehydration motor 22 of the washing machine 2 are measured and output by the speed detection device of the dehydration motor 22 itself and the acceleration sensor provided in the washing machine 2 itself, which are used to correct the distribution parameters used for the second training data.

[0050] The training method of the washing machine AI control model of this embodiment corrects the distribution parameters corresponding to the initial load, distributed load, and distributed eccentricity in the second training data through the rotation speed and acceleration measured by the washing machine 2 itself during the distribution control stage, thereby improving the accuracy of the second training data.

[0051] In some specific embodiments, the eccentricity thresholds corresponding to different distributed loads are obtained by setting multiple standard load blocks to construct multiple eccentricities of each initial load for distributed control; the eccentricity value at which a certain probability of peripheral parameter failure occurs is the eccentricity threshold corresponding to the distributed load.

[0052] Reference Figure 1 , Figure 2 , Figure 3The present invention also discloses a training device for an AI control model of a washing machine, and the AI ​​control model obtained through training is applied to the dehydration process of the washing machine.

[0053] The dehydration process of the washing machine includes a distribution control stage, which is used to control the rotation speed of the inner drum of the washing machine so that the clothes to be dehydrated are distributed in the inner drum to reduce eccentricity; generally, it is accelerated to a first rotation speed and maintained at the first rotation speed for a first time; and then accelerated from the first rotation speed to the second rotation speed with a first acceleration for a second time to evenly distribute the clothes in the inner drum.

[0054] The dehydration process of the washing machine also includes an accelerated dehydration judgment stage and an accelerated dehydration control stage; the accelerated dehydration judgment stage is to maintain the second speed for a third time, and to detect the distributed load and the distributed eccentricity in the accelerated dehydration judgment stage, and to determine whether to enter the accelerated dehydration control stage based on the obtained distributed load and the distributed eccentricity; the accelerated dehydration control stage is to deepen the dehydration of the clothes in the inner drum at the second acceleration for a fourth time.

[0055] The training device of the washing machine AI control model includes multiple washing machine complete machines 2, multiple peripheral detection modules 3, and a model training module 1.

[0056] The washing machine 2 includes a first control module 21, a dehydration motor 22, a load acquisition module 23, and an eccentricity acquisition module 24 respectively connected to the first control module 21; the load acquisition module 23 is used to obtain the initial load, distributed load, and dehydration load and transmit them to the first control module 21; the eccentricity acquisition module 24 is used to obtain the distributed eccentricity and dehydration eccentricity and transmit them to the first control module 21; the washing machine 2 is loaded with clothes with an initial load for testing.

[0057] Each peripheral detection module 3 is respectively provided corresponding to each washing machine 2 and is respectively connected to each washing machine 2 in a one-to-one correspondence, and is used to detect and output peripheral parameters of the washing machine 2 when the washing machine 2 is in operation.

[0058] The model training module 1 is a hardware body with a processor and a memory, and is configured with a data training tool, and is respectively connected to each first control module 21 and each peripheral detection module 3 in communication. The model training module 1 is preset with a peripheral parameter fault threshold, and is configured as follows: The peripheral parameters, dehydration load, and dehydration eccentricity of the accelerated dehydration control stage are obtained cyclically, and it is determined in real time whether the peripheral parameters reach or exceed the peripheral parameter fault threshold; if so, the dehydration load and dehydration eccentricity are collected as the first training data for training the accelerated dehydration AI control model; if not, the accelerated dehydration control is completed and the corresponding data is discarded.

[0059] The training device of the washing machine AI control model of the present invention combines the peripheral parameters obtained by the peripheral detection module 3 when a peripheral parameter failure occurs and the dehydration load and dehydration eccentricity obtained by the washing machine 2 itself to form a first training data to train an accelerated dehydration AI control model, which is used to accelerate the control of the dehydration control stage; the washing machine 2's own data and external detection data are combined to train the AI ​​control model to increase the dimension of the model training data, improve the accuracy and intelligence of the AI ​​control model, protect the washing machine, and extend the life of the washing machine; the first training data is automatically acquired through communication between the washing machine 2 itself and the peripheral detection module 3 and the model training module 1, thereby improving the efficiency of training data collection and model training efficiency.

[0060] The specific structure and principle of the training device for the washing machine AI control model of the present invention are described in detail below through specific embodiments.

[0061] In some specific embodiments, reference Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 7 ,The fault threshold of peripheral parameters includes vibration threshold and force threshold.

[0062] The peripheral detection module 3 includes a second control module 32 and a vibration sensor 34 and four force sensors 33 connected to the second control module 32, which obtain the vibration value of the accelerated dehydration control stage and the force vector of the four feet in real time and transmit them to the second control module 32; the second control module 32 is communicated with the model training module 1 through the second communication module 31, and transmits the obtained vibration value and four force vectors to the model training module 1.

[0063] The model training module 1 obtains the dehydration load and dehydration eccentricity in the accelerated dehydration control stage in real time, and obtains the force value of the washing machine 2 from the four force vectors; it judges in real time whether the vibration value reaches or exceeds the vibration threshold and / or whether the force value reaches or exceeds the force threshold; if so, the dehydration load and dehydration eccentricity are collected as the first training data for training the accelerated dehydration AI control model; if not, the accelerated dehydration control is completed and the corresponding data is discarded.

[0064] The training device of the washing machine AI control model of the present invention accelerates the dehydration AI control model by training the first training data composed of the vibration value and force value of the external detection data obtained by the peripheral detection module 3 and the dehydration load and dehydration eccentricity obtained by itself, which is used to accelerate the control of the dehydration control stage. The self-data and external detection data are combined with the training control model to increase the dimension of the model training data, improve the accuracy and intelligence of the control model, improve the dehydration success rate and efficiency, protect the washing machine, and extend the life of the washing machine.

[0065] When the vibration value exceeds the vibration threshold, a vibration fault occurs; when the force value exceeds the force threshold, a movement fault occurs; and when a movement fault occurs, the force value of the washing machine includes a lateral force that overcomes the friction force to move the washing machine horizontally.

[0066] In some specific embodiments, reference Figure 1 , Figure 2 , Figure 3 , Figure 5 , Figure 8 ,The fault threshold of peripheral parameters includes noise threshold and stress threshold.

[0067] The peripheral detection module 3 includes a second control module 32 and a noise sensor 35 and four force sensors 33 connected to the second control module 32, which obtain the noise value of the accelerated dehydration control stage and the force vector of the four feet in real time and transmit them to the second control module 32; the second control module 32 is communicated with the model training module 1 through the second communication module 31, and transmits the obtained noise value and four force vectors to the model training module 1.

[0068] The model training module 1 cyclically obtains the dehydration load and dehydration eccentricity in the accelerated dehydration control stage, and obtains the force value of the washing machine 2 from the four force vectors; it judges in real time whether the noise value reaches or exceeds the noise threshold and / or whether the force value reaches or exceeds the force threshold; if so, the dehydration load and dehydration eccentricity are collected as the first training data for training the accelerated dehydration AI control model; if not, the accelerated dehydration control is completed and the corresponding data is discarded.

[0069] The training device of the washing machine AI control model of the present invention accelerates the dehydration AI control model by training the first training data composed of the external detection data noise value, force value obtained by the peripheral detection module 3 and the dehydration load and dehydration eccentricity obtained by itself, which is used to accelerate the control of the dehydration control stage. The self-data and external detection data are combined with the training control model to increase the dimension of the model training data, improve the accuracy and intelligence of the control model, improve the dehydration success rate and efficiency, protect the washing machine, and extend the life of the washing machine.

[0070] When the vibration value exceeds the vibration threshold, a vibration fault occurs; when the force value exceeds the force threshold, a movement fault occurs; and when a movement fault occurs, the force value of the washing machine includes a lateral force that overcomes the friction force to move the washing machine horizontally.

[0071] In some specific embodiments, the eccentricity acquisition module 24 is an acceleration sensor installed at the bearing of the inner drum of the washing machine or the outer drum support structure, which collects the acceleration signal of the inner drum of the washing machine when it rotates in real time and transmits it to the first control module 21; the first control module 21 performs Fourier transform on the acceleration signal, analyzes the amplitude change of the fundamental frequency and harmonic components, and obtains the eccentricity value. That is, the eccentricity value is measured by the amplitude change value of the fundamental frequency and harmonic components of the acceleration.

[0072] In some specific embodiments, the eccentricity acquisition module 24 is a speed sensor, which obtains the speed data of the dehydration motor 22 in the uniform speed stage and transmits it to the first control module 21 for differential calculation of the speed change rate to obtain the acceleration, and calculates the difference between the maximum acceleration and the minimum acceleration in the continuous interval to accelerate the acquisition of the eccentricity value in the dehydration judgment stage.

[0073] In some specific embodiments, the eccentricity acquisition module 24 is a current detection module, which obtains the real-time torque current of the dehydration motor 22, performs low-pass digital filtering on the real-time torque current signal, and transmits it to the first control module 21 to eliminate high-frequency noise interference; the first control module 21 extracts the maximum torque current and the minimum filtered torque current in a plurality of consecutive dehydration cycles, and respectively obtains the average value of each maximum torque current and each minimum torque current, and calculates the difference between them to represent the eccentricity value. That is, e=K(Imax_avg-Imin_avg ) in, e is the eccentricity; K is the model calibration factor to avoid model errors; Imax_avg is the average value of each maximum torque current; Imin_avg is the average value of each minimum torque current.

[0074] In some specific embodiments, the peripheral detection module 3 also includes a test bench; the force sensor 33 is a multi-dimensional force sensor; the four force sensors 33 are respectively fixedly connected to the four feet of the washing machine 2 and the test bench; the force vectors of the four feet measured by the four force sensors 33 are transmitted to the model training module 1; the model training module 1 obtains the force value of the washing machine 2 according to each force vector.

[0075] The test bench has four locking joints with adjustable width and length, which are fixedly connected to four force sensors 33 .

[0076] In some specific embodiments, the training device of the washing machine AI control model also includes a first communication module 4; the first control module 21 is a driving module of the dehydration motor 22, which is communicatively connected with the model training module 1 through the first communication module 4.

[0077] In some specific embodiments, reference Figure 1 , Figure 2 , Figure 3 , Figure 6 The model training module 1 presets multiple sets of distribution parameters and multiple eccentricity thresholds corresponding to different distribution loads, and is configured as follows: Sending distributed parameters to each first control module 21 to control each washing machine 2 to operate according to the distributed parameters in the distributed control stage, and obtaining the distributed load and distributed eccentricity of the washing machine 2 when entering the accelerated dehydration judgment stage after the distributed control stage ends; Compare the distribution eccentricity with the eccentricity threshold corresponding to its distribution load; when the distribution eccentricity does not exceed the eccentricity threshold, the control enters the accelerated dehydration control stage, and the corresponding initial load and distribution parameters are collected as the first training data for training the distribution AI control model; when the distribution eccentricity exceeds the eccentricity threshold, the experiment fails and the corresponding data is discarded.

[0078] The training device of the washing machine AI control model of the present invention obtains the initial load and the second training data composed of the distribution parameters through the washing machine itself to train the distributed AI control model for the operation control of the dehydration motor 22 in the distributed control stage; improves the accuracy and intelligence of the control model, improves the dehydration success rate and efficiency, and enhances the user experience.

[0079] In some specific embodiments, the accelerated dehydration control model is a first fitting function of the dehydration load and the dehydration eccentricity obtained by training with the first training data, and whether a peripheral parameter failure occurs is intelligently determined by judging whether the dehydration load and dehydration eccentricity measured by the washing machine meet the first fitting function based on the self-test, and the dehydration is interrupted when it is determined that there is a peripheral parameter failure; otherwise, the accelerated dehydration control continues.

[0080] The training device of the washing machine AI control model of this embodiment obtains a first fitting function by fitting the first training data, so that the washing machine can intelligently judge whether a peripheral parameter failure has occurred or is about to occur based on the dehydration load and the dehydration eccentricity, thereby improving the intelligence and accuracy of the control, protecting the washing machine, and extending the life of the washing machine.

[0081] In some specific embodiments, the distribution control model is a second fitting function of the initial load and distribution parameters obtained from the second training data, so that the washing machine intelligently configures different distribution parameters to control the operation of the distribution control stage of the washing machine according to different initial loads and the second fitting function measured by self-test, so that the distribution eccentricity after operation in the distribution control stage meets the requirements for entering the accelerated dehydration control stage, thereby improving the dehydration rate.

[0082] The training device of the washing machine AI control model of the present embodiment obtains a second fitting function by fitting the second training data, so that the washing machine can intelligently configure the distribution parameters according to the initial load. The intelligent configuration of the distribution parameters according to the initial load not only makes the distribution eccentricity meet the dehydration requirements, but also reduces the probability of vibration failure and movement failure when the distributed load and distribution eccentricity are obtained in the distribution control stage to accelerate the dehydration control, thereby improving the dehydration rate and enhancing the user experience.

[0083] The present invention also discloses a washing machine, which includes a first control module 21, a dehydration motor 22, a load acquisition module 23, and an eccentricity acquisition module 24 respectively connected to the first control module 21; the first control module 21 is configured with an accelerated dehydration AI control model obtained in the above embodiment, which is used to judge whether a peripheral parameter failure has occurred or is about to occur in the washing machine according to the dehydration load cyclically acquired by the load acquisition module 23 and the dehydration eccentricity cyclically acquired by the eccentricity acquisition module 24, and if so, control the dehydration motor 22 to stop running and interrupt dehydration; if not, control the dehydration motor 22 to continue running.

[0084] The washing machine of the present invention controls the operation of the accelerated dehydration control stage by configuring the accelerated dehydration AI control model obtained by the above-mentioned AI training method and AI training device, thereby improving the intelligence of the operation of the washing machine, protecting the washing machine, and extending the life of the washing machine.

[0085] The specific structure and control principle of the washing machine of the present invention are described in detail below through specific embodiments.

[0086] In some specific embodiments, the first control module 21 is configured with a distributed AI control model obtained in the embodiments of the above-mentioned training method or training device, which is used to control the operation of the distributed control stage according to the initial load configuration distribution parameters obtained by the load acquisition module 23, and in the accelerated dehydration judgment stage, determine whether to enter the accelerated dehydration control stage according to the distributed load obtained by the load acquisition module 23 and the distributed eccentricity obtained by the eccentricity acquisition module 24.

[0087] This embodiment improves the success rate of dehydration by intelligently configuring distribution parameters according to the initial load.

[0088] In some specific embodiments, the load acquisition module 23 is a pressure sensor installed at the bottom of the washing machine or at a supporting point of the suspension system, and the load is calculated by measuring the pressure value.

[0089] In some specific embodiments, the load acquisition module 23 uses a load sensor placed at the connection between the inner cylinder and the frame to measure the load value in real time. In some specific embodiments, the eccentricity acquisition module 24 is an acceleration sensor arranged at the bearing of the inner drum of the washing machine or the outer drum support structure; the acceleration signal of the inner drum of the washing machine is collected in real time when it rotates and transmitted to the first control module 21; the first control module 21 performs Fourier transform on the acceleration signal, analyzes the amplitude changes of the fundamental frequency and harmonic components, and obtains the eccentricity value.

[0090] In some specific embodiments, the eccentricity acquisition module 24 is a speed sensor, which obtains the speed data of the dehydration motor 22 in the uniform speed stage and transmits it to the first control module 21 for differential calculation of the speed change rate to obtain the acceleration, and calculates the difference between the maximum acceleration and the minimum acceleration in the continuous interval to accelerate the acquisition of the eccentricity value in the dehydration judgment stage.

[0091] In some specific embodiments, The real-time torque current signal is processed by low-pass digital filtering and then transmitted to the first control module 21 to eliminate high-frequency noise interference; the first control module 21 extracts the maximum torque current and the minimum filtered torque current in a plurality of consecutive dehydration cycles, and respectively obtains the average value of each maximum torque current and each minimum torque current, and calculates the difference between them to represent the eccentricity value. That is, e=K(Imax_avg-Imin_avg ) in, e is the eccentricity; K is the model calibration factor to avoid model errors; Imax_avg is the average value of each maximum torque current; Imin_avg is the average value of each minimum torque current.

[0092] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for a person skilled in the art to modify the technical solutions described in the aforementioned embodiments or to replace some of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions claimed to be protected by the present invention.

Claims

1. A training method for a washing machine AI control model, characterized in that: include: S1. Set up the model training module and preset the peripheral parameter fault threshold; S2, setting a plurality of experimental washing machines, which include a first control module and a dehydration motor connected to the first control module, a load acquisition module, and an eccentricity acquisition module, for acquiring the load and the eccentricity; the first control module is communicatively connected to the model training module; S3, setting a plurality of peripheral detection modules, respectively connected to each of the washing machines, for obtaining peripheral parameters, communicating with the model training module, and outputting the peripheral parameters to the model training module; S4, placing clothes with an initial load in the washing machine for dehydration; cyclically obtaining the dehydration load, dehydration eccentricity, and the peripheral parameters in the accelerated dehydration control stage; S5, determining whether the peripheral parameter reaches or exceeds the peripheral parameter fault threshold; If yes, execute S6; if no, execute S4; S6: Collect the peripheral parameters, the corresponding dehydration load, and the dehydration eccentricity as first training data for training an accelerated dehydration AI control model.

2. The training method of the washing machine AI control model according to claim 1, characterized in that: The peripheral detection module includes a second control module and a vibration sensor and a force sensor connected to the second control module; the vibration sensor and the force sensor are used to detect the vibration value and the force value of the washing machine and transmit them to the second control module; the second control module is connected to the model training module for communication, and transmits the obtained vibration value and the force value to the model training module; the model training module recognizes them as the peripheral parameters; The peripheral parameter fault threshold includes a vibration threshold and a force threshold; judging whether the peripheral parameter reaches or exceeds the peripheral parameter fault threshold is that the vibration value reaches or exceeds the vibration threshold and / or the force value reaches or exceeds the force threshold.

3. The training method of the washing machine AI control model according to claim 1, characterized in that: The peripheral detection module includes a second control module and a noise sensor and a force sensor connected to the second control module; the noise sensor and the force sensor are used to detect the noise value and the force value of the washing machine and transmit them to the second control module; The second control module is in communication connection with the model training module, and transmits the obtained noise value and the force value to the model training module; The model training module identifies it as the peripheral parameter; The peripheral parameter fault threshold includes a noise threshold and a force threshold; judging whether the peripheral parameter reaches or exceeds the peripheral parameter fault threshold is that the noise value reaches or exceeds the noise threshold and / or the force value reaches or exceeds the force threshold.

4. The training method of the washing machine AI control model according to any one of claims 1 to 3, characterized in that: Also includes: S1', presetting a plurality of different distribution parameters in the model training module, which include a first rotation speed and a first acceleration; presetting a plurality of distribution eccentricity thresholds corresponding to different distribution loads; S2', the model training module sends the distribution parameters to the first control module to control the distribution control stage of the operation of each washing machine; when the distribution control stage is completed, the distribution load and distribution eccentricity of the washing machine are obtained; S3', determine whether the distribution eccentricity reaches or exceeds the eccentricity threshold corresponding to the distributed load; if yes, execute S4'; if no, execute S5'; S4', terminate dehydration control; S5', obtaining the initial load, the corresponding distribution parameters, the distributed load, and the distributed eccentricity and collecting them as second training data for training the distributed AI control model.

5. A training device for a washing machine AI control model, characterized in that: include: A plurality of washing machines, which are used to load clothes with an initial load, include a first control module, a dehydration motor respectively connected thereto, a load acquisition module, and an eccentricity acquisition module; A plurality of peripheral detection modules, which are respectively connected to the washing machines and are used to detect the peripheral parameters of the washing machines when they are in operation; The model training module is a hardware entity with a processor and a memory, which is respectively communicated with each of the first control modules and each of the peripheral detection modules, is configured with a data training tool, has a preset peripheral parameter fault threshold, and is configured to obtain the peripheral parameters, dehydration load, and dehydration eccentricity in real time during the accelerated dehydration control stage of each washing machine; determine whether the peripheral parameters reach the peripheral parameter fault threshold; if so, collect the dehydration load and the dehydration eccentricity and collect them as the first training data for training the accelerated dehydration AI control model.

6. The training device for the washing machine AI control model according to claim 5, characterized in that: The peripheral detection module includes a second control module and a vibration sensor and four force sensors connected to the second control module; the vibration sensor and each force sensor are used to detect the vibration value of the washing machine and the force vector of each foot respectively and transmit them to the second control module; the second control module is connected to the model training module for communication, and transmits the vibration value and each force vector to the model training module; the model training module obtains the force value of the washing machine from each force vector, and identifies the vibration value and the force value as the peripheral parameters; The peripheral parameter fault threshold includes a vibration threshold and a force threshold; determine whether the vibration value reaches or exceeds the vibration threshold and / or whether the force value reaches the force threshold; if so, collect the dehydration load and the dehydration eccentricity as the first training data for training the accelerated dehydration AI control model.

7. The training device for the washing machine AI control model according to claim 5, characterized in that: The peripheral detection module includes a second control module and a noise sensor and four force sensors connected to the second control module; the noise sensor and each of the force sensors are used to detect the noise value and the force vector of each foot and transmit them to the second control module; The second control module is in communication connection with the model training module, and transmits the noise value and each of the force vectors to the model training module; the model training module obtains the force value of the whole washing machine from each of the force vectors, and identifies the noise value and the force value as the peripheral parameters; The peripheral parameter fault threshold includes a noise threshold and a stress threshold; Determine whether the noise value reaches or exceeds the noise threshold and / or whether the force value reaches the force threshold; If yes, the dehydration load and the dehydration eccentricity are collected as the first training data for training the accelerated dehydration AI control model.

8. The training device for the washing machine AI control model according to any one of claims 5 to 7, characterized in that: The model training module also presets multiple sets of distribution parameters and eccentricity thresholds corresponding to different distribution loads, and is configured as follows: Sending the distribution parameters to each of the first control modules, controlling each of the washing machines to operate according to the distribution parameters during the distribution control phase, and returning the distribution load and distribution eccentricity measured after the distribution control phase ends; Determining whether the distribution eccentricity reaches or exceeds the eccentricity threshold corresponding to the distribution load; If yes, stop the dehydration control; if no, control enters the accelerated dehydration control stage, and collects the initial load, the distribution parameters, the distributed load, and the distributed eccentricity as the second training data for training the distribution AI control model.

9. A washing machine, characterized in that: It includes a first control module, a dehydration motor, a load acquisition module, and an eccentricity acquisition module respectively connected thereto; the first control module is configured with an accelerated dehydration AI control model obtained by a training device for a washing machine AI control model according to any one of claims 5 to 8, which detects or predicts whether a dehydration failure occurs based on the dehydration load and dehydration eccentricity obtained in real time; and if so, controls the dehydration to be interrupted.

10. The washing machine according to claim 9, characterized in that: The first control module is also configured with a distributed AI control model, which uses corresponding distribution parameters to control the operation of the distribution control stage according to the initial load, so that the distribution eccentricity after the distribution control stage runs meets the requirements for entering the accelerated dehydration stage.

Images