Dishwasher control method, apparatus, and electronic device

Abstract

The application provides a kind of dish washer control method, device and electronic equipment, it is related to smart home technical field.The method comprises: obtaining the original image of the utensil to be washed, and the compliance verification and enhancement pretreatment are carried out on the original image, and the image to be identified is obtained;The utensil material and oil stain distribution data are output by inputting the image to be identified into the preset double-branch deep learning model;Based on utensil material and oil stain distribution data, combined with preset user preference data, the cleaning strategy is output through the preset double-objective function optimization model;The cleaning strategy is executed, and the cleaning strategy is dynamically modified based on the real-time data of the dishwasher at intervals of a preset time;Fundamentally avoid improper washing caused by material misjudgment, and can intelligently balance energy consumption, water consumption and time efficiency while ensuring excellent cleaning effect according to the unique situation of each washing, significantly improve the intelligent level, resource utilization efficiency and user satisfaction of the dishwasher.

Description

Technical Field

[0001] This invention relates to the field of smart home technology, and in particular to a dishwasher control method, device, and electronic device. Background Technology

[0002] Currently, dishwashers with intelligent recognition functions are becoming increasingly common. These machines typically integrate image acquisition modules to photograph the loaded tableware and match basic oil stain conditions based on image features. However, such technologies often only analyze the macroscopic color and area of ​​oil stains, failing to deeply analyze image details. The complex textures of the tableware materials themselves (such as the luster of ceramic glaze or the natural grain of bamboo and wood) can easily be confused with the optical characteristics of oil residue under certain lighting conditions. This causes the system to mistakenly identify material textures as stains requiring cleaning, or ignore actual oil stains adhering to the recesses in the texture. This fundamental recognition error renders subsequent washing program recommendations inaccurate.

[0003] For the washing process, existing dishwashers mainly rely on several preset fixed programs (such as heavy-duty wash, normal wash, and energy-saving wash). The parameters of these programs (such as water temperature, duration, and water pressure) are preset fixed values, designed for an assumed "average" level of grease and dish combination. In actual use, the material of the user's tableware, the type and stubbornness of the grease, the density of the load, and the user's own energy-saving or time-sensitive preferences vary greatly. Fixed programs cannot respond to these specific variables, thus generally producing two contradictory results: first, providing excessive washing intensity for light grease or delicate materials, resulting in waste of water and energy and potential damage to tableware; second, insufficient cleaning power for heavy or special grease (such as solidified animal fat or burnt substances), requiring a second wash and affecting the user experience. Summary of the Invention

[0004] The purpose of this invention is to provide a dishwasher control method, device, and electronic device. First, through image processing and deep learning models, it achieves high-precision differentiation and quantitative analysis of tableware materials and oil stain properties. Then, it constructs a dual-objective optimization model that integrates multi-source real-time data and user preferences, dynamically generates and can adjust the washing strategy in real time during execution, fundamentally avoiding improper washing caused by material misjudgment. Furthermore, based on the unique context of each wash, it can intelligently balance energy consumption, water consumption, and time efficiency while ensuring excellent cleaning results, significantly improving the dishwasher's intelligence level, resource utilization efficiency, and user satisfaction.

[0005] In a first aspect, the present invention provides a dishwasher control method, comprising: The original image of the tableware to be cleaned is obtained, and the original image is subjected to compliance verification and enhanced preprocessing to obtain the image to be identified. The image to be identified is input into a preset dual-branch deep learning model, which outputs data on tableware material and oil stain distribution. Based on data on tableware material and oil stain distribution, combined with preset user preference data, a cleaning strategy is output through a preset dual-objective function optimization model; the dual-objective function optimization model is based on the dual optimization objectives of washing effect and washing energy efficiency. The system executes a cleaning strategy and dynamically adjusts it based on real-time dishwasher data at preset intervals.

[0006] In some preferred embodiments of the present invention, the steps of acquiring the original image of the tableware to be cleaned, and performing compliance verification and enhancement preprocessing on the original image to obtain the image to be identified include: Acquire the original image using external hardware devices; Compliance verification is performed based on preset verification rules to filter out interfering images; wherein the verification rules include at least one of the following: scene, sharpness, and pixels; The original image that has passed the compliance verification will undergo local illumination correction based on brightness differentiation and nonlinear stretching principles. The edge detection threshold will be dynamically adjusted using the Laplacian operator and edge threshold adaptive adjustment technology for different types of oil contamination to obtain the image to be identified. The types of oil contamination include liquid oil contamination and solid oil contamination.

[0007] In some preferred embodiments of the present invention, the dual-branch learning model includes: a first branch constructed based on a computer vision object detection architecture and a second branch constructed based on image semantic segmentation; The first branch is trained using transfer learning and data augmentation based on pre-defined labeled data to achieve simultaneous detection of tableware materials and oil stains; the labeled data includes: various tableware materials and various types of oil stains; The second branch identifies the load density distribution of stacked tableware based on image semantic segmentation, and obtains three-dimensional data of material, oil stains and load.

[0008] In some preferred embodiments of the present invention, the method further includes: Cross-validation is performed based on the output structure of the first branch and the output result of the second branch; If the test fails, the system will perform a reconfirmation by magnifying the local image and guiding the user to confirm the result through visual interaction once the confidence level meets the preset confidence level verification threshold.

[0009] In some preferred embodiments of the present invention, the dual-branch deep learning model is a lightweight model based on structured pruning to remove redundant convolutional kernels; wherein the weights of the dual-branch deep learning model are 8-bit integers.

[0010] In some preferred embodiments of the present invention, the washing effect includes: a constraint that the oil removal rate is greater than a preset first threshold and a constraint that the material damage rate is less than a preset second threshold; the method further includes: constraining the washing effect optimization target by the following formula: R = f(T, P, t1, t2); D = g(T, P); Where R is the oil removal rate, f() is the mapping model based on the historical washing database, T is the water temperature, P is the water pressure, t1 is the soaking time, t2 is the spraying time, D is the material damage rate, and g() is the material damage model.

[0011] In some preferred embodiments of the present invention, the method further includes: constraining the washing energy efficiency optimization target by the following formula: C = α × W + β × E + γ × T1; Where C is the total resource consumption, α is the first preference weight, W is the water consumption, β is the second preference weight, E is the power consumption, γ is the third preference weight, and T1 is the total washing time.

[0012] In some preferred embodiments of the present invention, the step of executing a cleaning strategy and adjusting the cleaning strategy based on real-time dishwasher data at preset time intervals includes: Real-time status data of the dishwasher is collected at a preset target time and compared with the expected status data in the cleaning strategy. If the deviation exceeds the preset deviation value, the cleaning strategy is adjusted based on real-time status data.

[0013] In a second aspect, the present invention provides a dishwasher control device, comprising: The image preprocessing module is used to acquire the original image of the tableware to be cleaned, and to perform compliance verification and enhanced preprocessing on the original image to obtain the image to be identified. The material and oil stain prediction module is used to input the image to be identified into a preset dual-branch deep learning model and output tableware material and oil stain distribution data. The cleaning strategy generation module is used to output a cleaning strategy based on tableware material and oil stain distribution data, combined with preset user preference data, through a preset dual-objective function optimization model; wherein, the dual-objective function optimization model is based on the dual optimization objective constraints of washing effect and washing energy efficiency. The execution module is used to execute the cleaning strategy and dynamically adjust the cleaning strategy based on real-time data from the dishwasher at preset intervals.

[0014] Thirdly, the present invention provides an electronic device including a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the dishwasher control method provided in the first aspect above.

[0015] This invention brings the following beneficial effects: This invention provides a dishwasher control method, device, and electronic device, comprising: acquiring an original image of the tableware to be cleaned, performing compliance verification and enhanced preprocessing on the original image to obtain an image to be identified; inputting the image to be identified into a preset dual-branch deep learning model to output tableware material and oil stain distribution data; based on the tableware material and oil stain distribution data, combined with preset user preference data, outputting a cleaning strategy through a preset dual-objective function optimization model; wherein, the dual-objective function optimization model is based on dual optimization objective constraints of washing effect and washing energy efficiency; executing the cleaning strategy, and dynamically correcting the cleaning strategy based on real-time dishwasher data at preset intervals; firstly, through image processing and deep learning models, high-precision differentiation and quantitative analysis of tableware material and oil stain properties are achieved, and then a dual-objective optimization model integrating multi-source real-time data and user preferences is constructed, dynamically generating and adjusting the washing strategy in real time during execution, fundamentally avoiding improper washing due to material misjudgment, and intelligently balancing energy consumption, water consumption, and time efficiency according to the unique context of each wash, while ensuring excellent cleaning effect, significantly improving the dishwasher's intelligence level, resource utilization efficiency, and user satisfaction. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 A flowchart of a dishwasher control method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a dishwasher control device provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0018] Icons: 310 - Image preprocessing module; 320 - Material and oil stain prediction module; 330 - Cleaning strategy generation module; 340 - Execution module; 400 - Memory; 401 - Processor; 402 - Bus; 403 - Communication interface. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," "third," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] Furthermore, terms such as "horizontal," "vertical," and "sag" do not imply that components must be absolutely horizontal or suspended, but rather that they can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal relative to "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0024] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0025] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0026] Example 1 This invention provides a dishwasher control method, see [link to relevant documentation]. Figure 1 The flowchart shown in this embodiment of the invention provides a dishwasher control method, which includes: Step S102: Obtain the original image of the tableware to be cleaned, and perform compliance verification and enhanced preprocessing on the original image to obtain the image to be identified.

[0027] Specifically, after receiving the uploaded image, the server uses a three-layer technical architecture of "image enhancement preprocessing + dual-branch deep learning model + cross-scene data adaptation" to achieve synchronous and accurate detection of the type / thickness and material type of oil stains on tableware, providing underlying data support for the intelligent washing parameter recommendation of the dishwasher.

[0028] Furthermore, in some preferred embodiments of the present invention, the steps of acquiring the original image of the tableware to be cleaned, and performing compliance verification and enhanced preprocessing on the original image to obtain the image to be identified include: acquiring the original image through an external hardware device; performing compliance verification based on preset verification rules to filter out interfering images; wherein, the verification rules include at least one of the following: scene, sharpness, and pixels; performing local illumination correction on the original image that has passed the compliance verification based on brightness differentiation and nonlinear stretching principles, and dynamically adjusting the edge detection threshold using the Laplacian operator and edge threshold adaptive adjustment technology for different types of oil stains identified to obtain the image to be identified; wherein, the types of oil stains include: liquid oil stains and solid oil stains.

[0029] Specifically, users take or upload images of dishes via a mobile app. The system first performs compliance checks, such as determining if the image is a kitchen scene (e.g., including a sink, countertop, or dishwasher basket), whether the image is sharp enough (no severe motion blur), and whether the pixel count meets minimum requirements (e.g., above 2 megapixels). This filters out completely irrelevant (e.g., selfies) or low-quality, unprocessable images. This step is fundamental to ensuring accurate recognition and effectively prevents subsequent algorithm misjudgments or failures due to input image quality issues, thus improving system reliability from the outset.

[0030] For images that pass verification, an improved Retinex-NMS (image enhancement) algorithm is used for preprocessing enhancement. This algorithm first divides the image into highlight, normal, and shadow areas based on brightness, and then processes them separately: logarithmic compression is applied to highlight areas to suppress overexposure, histogram equalization is performed on shadow areas to enhance details, and normal areas retain their original features. This local illumination correction method avoids the loss of oil stain details or material texture distortion caused by global enhancement. Next, the Laplacian operator is used for edge detection, and the edge detection threshold is adaptively adjusted according to the different characteristics of liquid oil stains (blurred edges) and solid oil stains (clear edges). This technique enhances the boundary difference between oil stains and the surface of the tableware, providing crucial data for accurately calculating the area and thickness of the oil stains, thereby improving the robustness and accuracy of the entire recognition process.

[0031] Step S104: Input the image to be recognized into the preset dual-branch deep learning model, and output the tableware material and oil stain distribution data.

[0032] Specifically, the pre-processed images are fed into a dual-branch deep learning model deployed on a server or edge for core recognition. This model is specifically designed to simultaneously decouple tableware material information from oil stain information, fundamentally solving the problem of traditional methods misidentifying material texture as oil stains. Through this step, the system can extract refined data from a single image, including tableware type, oil stain category, distribution, area, and thickness. This data is the fundamental basis for generating personalized washing strategies and the cornerstone for achieving a balance between energy efficiency and effectiveness.

[0033] Furthermore, in some preferred embodiments of the present invention, the dual-branch learning model includes: a first branch constructed based on a computer vision object detection architecture and a second branch constructed based on image semantic segmentation; the first branch is trained by transfer learning and data augmentation based on preset labeled data to achieve simultaneous detection of tableware materials and oil stains; wherein, the labeled data includes: multiple tableware materials and multiple types of oil stains; the second branch identifies the load density distribution of tableware stacks based on image semantic segmentation to obtain three-dimensional data of material-oil stain-load.

[0034] Specifically, the first branch employs an improved YOLOv8 object detection architecture. It is trained on a large-scale, finely labeled dataset containing 10 common tableware materials (e.g., ceramic, tempered glass, crystal glass, stainless steel, solid wood, bamboo, melamine resin (imitation porcelain), silicone, high-temperature resistant plastics (e.g., PP), and coated cast iron) and 8 types of oil stains (e.g., liquid vegetable oil, liquid animal oil, solidified animal fat, protein residue (e.g., egg liquid), starch residue (e.g., porridge stains), mixed oil stains, burnt carbon deposits, and spicy seasoning pigment residues). This is achieved through transfer learning and targeted data augmentation (simulating different kitchen lighting, oil stain diffusion effects, and water stain reflection). This allows the branch to simultaneously output the material type of each tableware instance in the image, as well as the type, area, and estimated thickness of the oil stains attached to it.

[0035] The second branch employs a semantic segmentation network (such as a variant of U-Net) to specifically identify the stacking of tableware, their mutual occlusion relationships, and space occupancy, outputting a load density distribution map of the entire cavity. Finally, the system merges the results from both branches to generate a three-dimensional data volume containing a "material matrix," an "oil heat map," and a "load distribution map," thereby comprehensively and three-dimensionally describing the static load and contamination status of this washing task.

[0036] Furthermore, in some preferred embodiments of the present invention, the method further includes: performing cross-validation based on the output structure of the first branch and the output result of the second branch; if the cross-validation fails, reconfirming the cross-validation is performed by local image magnification detection, and guiding the user to confirm the cross-validation through visual interaction when the confidence level meets the preset confidence level verification threshold.

[0037] Specifically, to ensure the reliability of the recognition results, the system is designed with a multi-dimensional cross-validation mechanism. The validation logic is based on the consistency between prior knowledge and data, for example: Material-oil stain logic verification: If the first branch determines that a certain area is made of "silicone" material (usually not resistant to high temperature), but at the same time detects "burnt carbon" (usually produced by high temperature), then the contradiction flag is triggered.

[0038] Spatial consistency check: The "load distribution map" output by the second branch shows a certain area as a densely stacked occlusion zone, but the first branch gives high-confidence detection results for "crystal glass" and "large area of ​​oil stains" in the same area. Since it is difficult to obtain a clear judgment under severe occlusion, this result has low reliability and triggers a re-verification.

[0039] Optical characteristics verification: The detected "oil stain" area exhibits strong specular reflection characteristics, which is inconsistent with the diffuse reflection or greasy luster characteristics of common oil stains. It is possible that the light spot of stainless steel material was misjudged.

[0040] When the verification mechanism is triggered, the system marks the result as "suspected contradiction" or "low confidence area" and automatically calls up a high-resolution local image (if available) or performs super-resolution magnification and re-detection on the suspicious area of ​​the original image. Furthermore, if the overall confidence level of the model output is within a preset fuzzy range (e.g., 80%-90%), the system will not arbitrarily accept the result. Instead, it guides the user to make a final confirmation through visual interactive methods on the APP interface, such as highlighted prompts, magnified local images, and text options (e.g., "Is this oil stain? [Yes / No]"). This "human-machine collaborative verification" mechanism effectively reduces the risk of misjudgment that a single model may produce in complex scenarios. By introducing the strong prior of user judgment, it further improves the accuracy of the data input into the decision-making system.

[0041] Furthermore, in some preferred embodiments of the present invention, the dual-branch deep learning model is a lightweight model based on structured pruning to remove redundant convolutional kernels; wherein the weights of the dual-branch deep learning model are 8-bit integers.

[0042] Specifically, considering the real-time requirements of the recognition service and the potential need for partial deployment on a mobile app, the original large model was lightweighted. Structured pruning techniques were employed to systematically evaluate and remove redundant convolutional kernels and channels with low contribution. Subsequently, the model's weight parameters were converted from 32-bit floating-point numbers to 8-bit integers for storage and computation (i.e., INT8 quantization). After these optimizations, the model size was significantly compressed from 200MB to less than 50MB, inference speed was improved several times, and high recognition accuracy (e.g., material recognition accuracy ≥98%) was maintained through fine-tuning. This allows complex deep learning models to run efficiently on cloud servers and has the potential for deployment to edge devices (such as the built-in computing units of high-performance dishwashers or mobile phones), ensuring the feasibility, responsiveness, and smooth user experience of the entire system application.

[0043] Step S106: Based on the tableware material and oil stain distribution data, combined with preset user preference data, a cleaning strategy is output through a preset dual-objective function optimization model; wherein, the dual-objective function optimization model is based on dual optimization objective constraints of washing effect and washing energy efficiency.

[0044] Specifically, the system integrates three-dimensional data of the dishes and grease, real-time status data (such as initial water temperature, current water level, and detergent balance) obtained from the dishwasher's sensors via Wi-Fi, and user-defined preference tags set in the app. This multi-dimensional data serves as input to drive a mathematical model with "washing effect" and "washing energy efficiency" as dual optimization objectives. Under all hard constraints, it solves for a set of optimal, executable washing parameter instructions. This replaces the drawbacks of the traditional fixed-mode "one-size-fits-all" approach, achieving a fundamental shift from "humans adapting to machines" to "machines adapting to both humans and objects," thus completing truly personalized and adaptive washing planning.

[0045] Furthermore, in some preferred embodiments of the present invention, the washing effect includes: a constraint that the oil removal rate is greater than a preset first threshold and a constraint that the material damage rate is less than a preset second threshold; the method further includes: constraining the washing effect optimization target by the following formula: R=f(T, P, t1, t2); D = g(T, P); where R is the oil removal rate, f() is a mapping model established based on the historical washing database, T is the water temperature, P is the water pressure, t1 is the soaking time, t2 is the spraying time, D is the material damage rate, and g() is the material damage model.

[0046] Specifically, the washing effect target is an uncompromising hard constraint that must be met first. The system sets the oil removal rate R to be no less than 99% (first threshold) and the material damage rate D to be no more than 0.1% (second threshold). To achieve this target, the system constructs an oil removal rate mapping model R=f(T, P, t1, t2) based on a large amount of accumulated historical washing experiment data. This model reflects the complex nonlinear relationship between water temperature, water pressure, soaking and spraying time, and the cleaning effect. Simultaneously, a material damage model D=g(T, P) is constructed for the 10 listed material types to assess the potential damage risk of different water temperature and pressure combinations to various materials (e.g., high temperature causing solid wood to crack, strong water pressure causing ceramic glaze abrasion). During optimization, the algorithm must find a solution within the intersection of the "effective cleaning zone" and the "safe material zone" defined by the model, thus fundamentally guaranteeing the cleaning effect and preventing damage to tableware due to improper programming, achieving a dual guarantee of "safety" and "effectiveness" in cleaning.

[0047] Furthermore, in some preferred embodiments of the present invention, the method further includes constraining the washing energy efficiency optimization target by the following formula: C=α×W+β×E+γ×T1; where C is the total resource consumption, α is the first preference weight, W is the water consumption, β is the second preference weight, E is the power consumption, γ is the third preference weight, and T1 is the total washing time.

[0048] Specifically, under the hard constraint of achieving optimal washing performance, the system uses "minimizing total resource consumption C" as its energy efficiency optimization objective. Total resource consumption C is the weighted sum of water consumption W, power consumption E, and total washing time T1. The weighting coefficients α, β, and γ are not fixed values ​​but are dynamically adjusted based on the user's real-time preferences set in the app, aligning system behavior with user intent. For example: When the user selects the "Prioritize Water Conservation" mode, α=0.5, β=0.3, and γ=0.2 are set. The optimization algorithm will try its best to find water-saving solutions, which may appropriately extend the soaking time (t1) to utilize chemical action, or reduce the water pressure in non-core stages.

[0049] When the user selects the "Prioritize Energy Saving" mode, with α=0.2, β=0.6, and γ=0.2, the algorithm will tend to lower the overall water temperature, optimize the heating strategy, and reduce the running time of high-power water pumps.

[0050] When the user selects the "Quick Wash" mode, α=0.1, β=0.3, γ=0.6 are set. The algorithm will maximize water temperature and pressure within permissible limits, while minimizing the duration of each stage, prioritizing speed.

[0051] When the user selects the "Standard Wash" mode, set α=0.33, β=0.33, γ=0.34 for balanced optimization.

[0052] This design directly quantifies users' personalized needs and integrates them into the optimization process, thereby achieving a true "energy efficiency-effect" balance by dynamically balancing the relationship between water, electricity, and time while ensuring cleaning effectiveness, thus meeting diverse user needs.

[0053] Step S108: Execute the cleaning strategy and dynamically adjust the cleaning strategy based on real-time data from the dishwasher at preset intervals.

[0054] Specifically, the optimal cleaning strategy generated above (including specific parameter sequences such as water temperature, water pressure, and duration of each stage) is sent to the dishwasher controller for execution. However, the actual washing environment is subject to uncertainties and disturbances (such as inlet water temperature being much lower than the preset temperature in winter, slow water temperature rise due to heat absorption by the load, and fluctuations in detergent dissolution). Therefore, the system does not simply "set a function and leave it at that," but actively monitors and dynamically adjusts throughout the washing process, forming a closed-loop control of "planning-execution-monitoring-replanning." This ensures that the actual washing trajectory always approaches the theoretically optimal path, thereby guaranteeing the stable achievement of the final washing effect and energy efficiency targets, and improving the system's robustness in real-world home environments.

[0055] Furthermore, in some preferred embodiments of the present invention, the step of executing a cleaning strategy and adjusting the cleaning strategy based on real-time data of the dishwasher at preset time intervals includes: collecting real-time status data of the dishwasher at a preset target time and comparing it with the expected status data in the cleaning strategy; if the deviation is greater than a preset deviation value, adjusting the cleaning strategy based on the real-time status data.

[0056] Specifically, the system is set to collect key status data such as water temperature, actual water level, and turbidity in the cavity in real time at critical points approximately 1 / 3 and 2 / 3 of the total washing time (e.g., at the 20th and 40th minutes of a 60-minute program) via the dishwasher's built-in sensor array and Wi-Fi communication. These measured values ​​are then precisely compared with the expected theoretical values ​​at those points in the current washing strategy. If the deviation of a key parameter (such as water temperature) exceeds a set threshold (e.g., 10%), it is determined that the actual washing process has significantly deviated from the preset optimal trajectory. At this point, the system immediately restarts the dual-objective function optimization module, but this time the optimization is based on the current real-time state (not the initial state) as a new starting point. It integrates the latest sensor data to quickly calculate and generate a set of adjusted parameters for the remaining washing stage (e.g., extending the heating time of the main wash stage due to insufficient temperature rise, appropriately increasing the spray water pressure to compensate for water temperature, or adjusting the number of subsequent rinses based on the current turbidity to determine the dissolution of oil stains), and immediately issues the adjustment for execution. This mid-process dynamic correction mechanism enables the system to effectively cope with various disturbances in the real environment, significantly enhancing the adaptability of the washing process and the consistency of the final result, ensuring a stable output of high-quality washing effects under different initial conditions.

[0057] This invention provides a dishwasher control method, comprising: acquiring an original image of the tableware to be cleaned, performing compliance verification and enhanced preprocessing on the original image to obtain an image to be identified; inputting the image to be identified into a preset dual-branch deep learning model to output tableware material and oil stain distribution data; based on the tableware material and oil stain distribution data, combined with preset user preference data, outputting a cleaning strategy through a preset dual-objective function optimization model; wherein, the dual-objective function optimization model is based on dual optimization objective constraints of washing effect and washing energy efficiency; executing the cleaning strategy, and dynamically correcting the cleaning strategy based on real-time dishwasher data at preset intervals; firstly, through image processing and deep learning models, high-precision differentiation and quantitative analysis of tableware material and oil stain properties are achieved, and then a dual-objective optimization model integrating multi-source real-time data and user preferences is constructed, dynamically generating and adjusting the washing strategy in real time during execution, fundamentally avoiding improper washing due to material misjudgment, and intelligently balancing energy consumption, water consumption and time efficiency according to the unique context of each wash, while ensuring excellent cleaning effect, significantly improving the intelligence level, resource utilization efficiency and user satisfaction of the dishwasher.

[0058] Example 2 Based on the above embodiments, this invention provides a dishwasher control device, see [link to previous embodiment]. Figure 2 The schematic diagram shown in this embodiment of the invention provides a dishwasher control device, which includes: The image preprocessing module 310 is used to acquire the original image of the tableware to be cleaned, and to perform compliance verification and enhanced preprocessing on the original image to obtain the image to be identified.

[0059] The material and oil stain prediction module 320 is used to input the image to be identified into a preset dual-branch deep learning model and output tableware material and oil stain distribution data.

[0060] The cleaning strategy generation module 330 is used to output a cleaning strategy based on the tableware material and oil stain distribution data, combined with preset user preference data, through a preset dual-objective function optimization model; wherein, the dual-objective function optimization model is based on the dual optimization objective constraints of washing effect and washing energy efficiency.

[0061] The execution module 340 is used to execute the cleaning strategy and dynamically adjust the cleaning strategy based on real-time data from the dishwasher at preset intervals.

[0062] Furthermore, in some preferred embodiments of the present invention, the image preprocessing module 310 is used to acquire the original image through an external hardware device; perform compliance verification based on preset verification rules to filter out interfering images; wherein, the verification rules include at least one of the following: scene, sharpness, and pixels; perform local illumination correction on the original image that has passed the compliance verification based on brightness differentiation and nonlinear stretching principles, and dynamically adjust the edge detection threshold using the Laplacian operator and edge threshold adaptive adjustment technology for different types of oil stains identified to obtain the image to be identified; wherein, the types of oil stains include: liquid oil stains and solid oil stains.

[0063] Furthermore, in some preferred embodiments of the present invention, the dual-branch learning model includes: a first branch constructed based on a computer vision object detection architecture and a second branch constructed based on image semantic segmentation; the first branch is trained by transfer learning and data augmentation based on preset labeled data to achieve simultaneous detection of tableware materials and oil stains; wherein, the labeled data includes: multiple tableware materials and multiple types of oil stains; the second branch identifies the load density distribution of tableware stacks based on image semantic segmentation to obtain three-dimensional data of material-oil stain-load.

[0064] Furthermore, in some preferred embodiments of the present invention, the device further includes: a cross-validation module, used to perform cross-validation based on the output structure of the first branch and the output result of the second branch; if it fails, it is reconfirmed by local image magnification detection, and when the confidence level meets the preset confidence level verification threshold, it guides the user to confirm through visual interaction.

[0065] Furthermore, in some preferred embodiments of the present invention, the dual-branch deep learning model is a lightweight model based on structured pruning to remove redundant convolutional kernels; wherein the weights of the dual-branch deep learning model are 8-bit integers.

[0066] Furthermore, in some preferred embodiments of the present invention, the washing effect includes: a constraint condition that the oil removal rate is greater than a preset first threshold and a constraint condition that the material damage rate is less than a preset second threshold; the cleaning strategy generation module 330 is used to constrain the washing effect optimization target through the following formula: R=f(T, P, t1, t2); D = g(T, P); where R is the oil removal rate, f() is a mapping model established based on the historical washing database, T is the water temperature, P is the water pressure, t1 is the soaking time, t2 is the spraying time, D is the material damage rate, and g() is the material damage model.

[0067] Furthermore, in some preferred embodiments of the present invention, the cleaning strategy generation module 330 is used to constrain the washing energy efficiency optimization target by the following formula: C=α×W+β×E+γ×T1; where C is the total resource consumption, α is the first preference weight, W is the water consumption, β is the second preference weight, E is the power consumption, γ is the third preference weight, and T1 is the total washing time.

[0068] Furthermore, in some preferred embodiments of the present invention, the execution module 340 is used to collect real-time status data of the dishwasher at a preset target time and compare it with the expected status data in the cleaning strategy; if the deviation is greater than a preset deviation value, the cleaning strategy is adjusted based on the real-time status data.

[0069] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the dishwasher control device described above can be referred to the corresponding process in the embodiments of the aforementioned dishwasher control method, and will not be repeated here.

[0070] Example 3 This invention also provides an electronic device for operating a dishwasher control method; see [link to related documentation]. Figure 3The schematic diagram of an electronic device provided by the embodiment of the present invention shown includes a memory 400 and a processor 401. The memory 400 is used to store one or more computer instructions, which are executed by the processor 401 to implement the dishwasher control method described above.

[0071] Furthermore, Figure 3 The electronic device shown also includes a bus 402 and a communication interface 403. The processor 401, the communication interface 403 and the memory 400 are connected via the bus 402.

[0072] The memory 400 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 403 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 402 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0073] Processor 401 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 401 or by instructions in software form. Processor 401 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a readily available storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 400, and processor 401 reads information from memory 400 and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiments.

[0074] This invention also provides a computer-readable storage medium storing computer-executable instructions. When these computer-executable instructions are called and executed by a processor, they cause the processor to implement the aforementioned dishwasher control method. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0075] The computer program products of the dishwasher control method, apparatus and electronic device provided in the embodiments of the present invention include a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods in the preceding method embodiments. For specific implementation, please refer to the method embodiments, which will not be repeated here.

[0076] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and / or device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0077] Furthermore, in the description of the embodiments of the present invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on the specific circumstances.

[0078] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dishwasher control method, characterized in that, include: The original image of the tableware to be cleaned is acquired, and the original image is subjected to compliance verification and enhanced preprocessing to obtain the image to be identified. The image to be identified is input into a preset dual-branch deep learning model, which outputs data on tableware material and oil stain distribution. Based on the tableware material and oil stain distribution data, combined with preset user preference data, a cleaning strategy is output through a preset dual-objective function optimization model; wherein, the dual-objective function optimization model is based on dual optimization objective constraints of washing effect and washing energy efficiency; The cleaning strategy is executed, and the cleaning strategy is dynamically modified based on real-time data from the dishwasher at preset intervals.

2. The dishwasher control method according to claim 1, characterized in that, The steps of acquiring the original image of the tableware to be cleaned, and performing compliance verification and enhanced preprocessing on the original image to obtain the image to be identified include: The original image is acquired using an external hardware device; The compliance verification is performed based on preset verification rules to filter out interfering images; wherein, the verification rules include at least one of the following: scene, sharpness, and pixels; The original image that has passed the compliance verification is locally illuminated based on brightness differentiation and nonlinear stretching principles. The edge detection threshold is dynamically adjusted using the Laplacian operator and edge threshold adaptive adjustment technology for different types of oil stains to obtain the image to be identified. The types of oil stains include liquid oil stains and solid oil stains.

3. The dishwasher control method according to claim 1, characterized in that, The dual-branch learning model includes: a first branch based on a computer vision object detection architecture and a second branch based on image semantic segmentation; The first branch is trained through transfer learning and data augmentation based on preset labeled data to achieve simultaneous detection of tableware materials and oil stains; wherein, the labeled data includes: multiple tableware materials and multiple types of oil stains; The second branch identifies the load density distribution of the stacked tableware based on the image semantic segmentation, and obtains three-dimensional data of material-oil stain-load.

4. The dishwasher control method according to claim 3, characterized in that, The method further includes: Cross-validation is performed based on the output structure of the first branch and the output result of the second branch; If the test fails, the system will perform a reconfirmation by magnifying the local image and guiding the user to confirm the result through visual interaction once the confidence level meets the preset confidence level verification threshold.

5. The dishwasher control method according to any one of claims 1 to 4, characterized in that, The dual-branch deep learning model is a lightweight model based on structured pruning to remove redundant convolutional kernels; wherein the weights of the dual-branch deep learning model are 8-bit integers.

6. The dishwasher control method according to claim 1, characterized in that, The washing effect includes the constraint that the oil removal rate is greater than a preset first threshold and the constraint that the material damage rate is less than a preset second threshold; the method further includes: constraining the washing effect optimization target by the following formula: R = f(T, P, t1, t2); D = g(T, P); Where R is the oil removal rate, f() is the mapping model established based on the historical washing database, T is the water temperature, P is the water pressure, t1 is the soaking time, t2 is the spraying time, D is the material damage rate, and g() is the material damage model.

7. The dishwasher control method according to claim 1, characterized in that, The method further includes constraining the washing energy efficiency optimization target using the following formula: C = α × W + β × E + γ × T1; Where C is the total resource consumption, α is the first preference weight, W is the water consumption, β is the second preference weight, E is the power consumption, γ is the third preference weight, and T1 is the total washing time.

8. The dishwasher control method according to claim 1, characterized in that, The steps of executing the cleaning strategy and adjusting the cleaning strategy based on real-time dishwasher data at preset time intervals include: Real-time status data of the dishwasher is collected at a preset target time and compared with the expected status data in the cleaning strategy. If the deviation is greater than the preset deviation value, the cleaning strategy is adjusted based on the real-time status data.

9. A dishwasher control device, characterized in that, include: The image preprocessing module is used to acquire the original image of the tableware to be cleaned, and to perform compliance verification and enhanced preprocessing on the original image to obtain the image to be identified. The material and oil stain prediction module is used to input the image to be identified into a preset dual-branch deep learning model and output tableware material and oil stain distribution data. The cleaning strategy generation module is used to output a cleaning strategy based on the tableware material and oil stain distribution data, combined with preset user preference data, through a preset dual-objective function optimization model; wherein, the dual-objective function optimization model is based on the dual optimization objective constraints of washing effect and washing energy efficiency; The execution module is used to execute the cleaning strategy and dynamically modify the cleaning strategy based on real-time data from the dishwasher at preset intervals.

10. An electronic device, characterized in that, The device includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the dishwasher control method according to any one of claims 1 to 8.

Images