Household appliance control method and electronic equipment

Abstract

The invention discloses a household appliance control method and electronic equipment, and relates to the technical field of household appliance control, and the method comprises the steps: obtaining first time sequence operation data of a household appliance; inputting the first time sequence operation data and preset operation data of the household appliance into a system model of the household appliance to obtain a plurality of expected operation data of the household appliance at the next moment; updating a fuzzy set boundary corresponding to a preset membership function based on the first time sequence operation data to obtain a target membership function; the target membership function is used for determining the corresponding energy consumption when the household appliances are operated according to the expected operation data; for any expected operation data, inputting the energy consumption, the preset operation data and the expected operation data into a target function, and determining a target value corresponding to the expected operation data; and generating a control instruction at the current moment based on the expected operation data corresponding to the minimum target value, and sending the control instruction to the household appliance. By adopting the method, the energy optimization of the household appliance can be realized.

Description

Technical Field

[0001] The present application relates to the field of home appliance control technology, and in particular to a control method and electronic device for home appliance equipment. Background Art

[0002] With the current trend of smart homes and energy conservation and emission reduction, home appliance energy management is crucial. Existing home appliance energy management systems are mostly based on fixed rules or simple data analysis.

[0003] For example, traditional home appliance energy management systems typically pre-set a set of fixed operating logic. For example, air conditioners are set to run at a fixed temperature and fan speed during specific hours, and refrigerators maintain a single cooling intensity. This model ignores fluctuations in the real-time operating status of home appliances. For example, the actual cooling efficiency of air conditioner compressors may change due to factors such as age and indoor heat load. In such cases, continued operation based on fixed rules will lead to energy waste.

[0004] Based on this, the existing home appliance energy management system lacks the ability to dynamically respond to the real-time operating status and environmental changes of home appliances, and cannot achieve precise energy optimization based on actual conditions. Summary of the Invention

[0005] The present application provides a control method and electronic device for household appliances, which can solve the problem that existing household appliance energy management systems are unable to achieve energy optimization of household appliances based on the real-time operating status and environmental changes of household appliances.

[0006] In a first aspect, a method for controlling a household appliance is provided, comprising:

[0007] Acquire first time-series operation data of the home appliance; the first time-series operation data includes state data of the home appliance at a current moment and environmental parameters of the environment in which the home appliance is located;

[0008] Inputting the first time-series operating data and the preset operating data of the home appliance into a system model of the home appliance to obtain a plurality of expected operating data of the home appliance at the next moment; the system model is used to describe the law of change of the operating state of the home appliance over time; and the expected operating data is used to represent the operating state of the home appliance at the next moment;

[0009] updating a fuzzy set boundary corresponding to a preset membership function based on the first time-series operating data to obtain a target membership function; the target membership function is used to quantify the fuzzy level corresponding to each expected operating data based on the updated fuzzy set boundary to determine the energy consumption of the household appliance when operating with the expected operating data;

[0010] For any expected operating data, energy consumption, preset operating data, and expected operating data are input into an objective function to determine a target value corresponding to the expected operating data; the objective function is used to measure the performance of the home appliance in multiple preset indicators during operation; the preset indicators include energy consumption indicators; the numerical value of the target value is negatively correlated with the performance of the home appliance in the preset indicators;

[0011] A control instruction at the current moment is generated based on the expected operation data corresponding to the minimum target value, and the control instruction is sent to the home appliance.

[0012] In the above technical solution, by acquiring first-time sequential operating data of a home appliance in real time, and this first-time sequential operating data includes current state data and environmental parameters, the current operating state and environmental conditions of the home appliance can be accurately analyzed. Subsequently, the real-time first-time sequential operating data and preset operating data are input into a system model used to describe how the device's operating state changes over time. This can yield multiple expected operating data for the next moment, enabling the home appliance to subsequently make adaptive adjustments based on its actual state and dynamic changes in the environment, thereby enhancing its dynamic response capabilities to state changes and environmental changes. The expected operating data is used to characterize the operating state of the home appliance at the next moment. Subsequently, the electronic device can adaptively update the fuzzy set boundaries corresponding to the preset membership function based on the real-time first-time sequential operating data to obtain a target membership function. This allows the equal fuzzy level division to better fit the real-time data distribution, resolving the problem that traditional static boundaries cannot adapt to changes in device state or environment. Based on this, the fuzzy level corresponding to each expected operating data point is quantified based on the updated fuzzy set boundary in the membership function to determine the energy consumption of the home appliance when operating according to the expected operating data. This allows for refined quantification of device energy consumption and improves the accuracy of energy consumption calculations based on fuzzy logic. Finally, after obtaining multiple expected operating data points, the corresponding energy consumption of the home appliance when operating according to each expected operating data point can be determined. Furthermore, the energy consumption, preset operating data, and expected operating data are input into an objective function that measures the performance of the home appliance on multiple preset indicators during operation to obtain corresponding target values. Because the preset indicators in the objective function at least include the energy consumption indicator, and the target value is negatively correlated with the home appliance's performance on the preset indicators, this method can generate control instructions based on the expected operating data corresponding to the minimum target value, ensuring that the home appliance operates in an energy-optimized manner, effectively avoiding energy waste caused by the limitations of traditional operating modes and achieving energy optimization.

[0013] In one example, obtaining first time-series operation data of a household appliance includes:

[0014] Acquire second time-series operation data of the home appliance; the second time-series operation data includes operation data of the home appliance at the current moment, and the operation data includes status data and environmental parameters;

[0015] For the operation data at any moment, with the operation data as the center, determine a set of continuous operation data sets with a preset number in the second time series operation data;

[0016] The corresponding running data are replaced by the average values ​​of the running data sets to obtain the first time series running data.

[0017] In this technical solution, for each piece of operating data, a preset number of continuous operating data sets are determined within the second time-series operating data, centered around that data point. This fully exploits the local correlation of the data to capture the changing trends of the appliance's operating status over a short period of time, avoiding the isolation of individual data points. Subsequently, the average value of the operating data set is used to replace the corresponding operating data point to obtain the first time-series operating data. This effectively smooths random fluctuations in the data, removing noise interference and reducing the impact of abnormal fluctuations on subsequent analysis.

[0018] In one embodiment, obtaining first time-series operation data of a household appliance includes:

[0019] Calculate the mean and standard deviation of the first time series running data;

[0020] Calculating a first difference between the average value and a preset multiple of the standard deviation, and a sum of the average value and the preset multiple of the standard deviation;

[0021] The operating data that is not between the first difference value and the sum value in the first time series operating data is deleted.

[0022] In the above technical solution, the standard deviation reflects the degree of data dispersion, and the preset multiple allows for flexible adjustment of the range based on actual needs. When operating data falls outside this range, it can be considered an outlier due to noise or device anomalies. Therefore, by removing these outliers, the first time-series operating data can more accurately reflect the actual operating status of the home appliance, reducing the potential for misleading analysis and decision-making due to abnormal data.

[0023] In one embodiment, updating the fuzzy set boundary corresponding to the preset membership function based on the first time series running data to obtain the target membership function includes:

[0024] Determine the distribution value of each operation data in the first time series operation data at different preset data values;

[0025] Determine the preset data value corresponding to the maximum value among the multiple distribution values ​​as the target data value;

[0026] The target membership function is obtained by updating the fuzzy set boundary based on the target data value.

[0027] In the above technical solution, by analyzing the distribution characteristics of the first time series operating data at different preset data values, the preset data value corresponding to the maximum distribution value is determined as the target data value, which can accurately describe the core distribution trend of the data and avoid interference from edge data. In this way, the fuzzy set boundary of the membership function is dynamically updated based on the target value, which can make the equal fuzzy level division more consistent with the real-time data distribution and solve the problem that traditional static boundaries cannot adapt to changes in device status or environment. Furthermore, by determining the energy consumption of the expected operating data based on the updated membership function, it is possible to achieve refined quantification of device energy consumption and improve the accuracy of fuzzy logic decision-making.

[0028] In one embodiment, determining the distribution value of each operating data in the first time series operating data at different preset data values ​​includes:

[0029] For any preset data value, respectively calculate a second difference between the preset data value and each running data;

[0030] Each second difference value is input into a preset kernel density estimation function to obtain a distribution value.

[0031] In the above technical solution, calculating the difference between the preset data values ​​and each operating data point can intuitively quantify the degree of deviation between the preset values ​​and the actual data, providing a basic metric for analyzing data distribution. Furthermore, using kernel density estimation to process the difference directly and adaptively captures complex data patterns through the smoothing properties of the kernel function, effectively avoiding the information loss of discretization methods. At the same time, bandwidth parameter adjustment achieves a balance between bias and variance. Consequently, it can accurately describe the distribution characteristics of operating data at different preset values.

[0032] In one embodiment, the preset indicators further include an operation stability indicator and a performance indicator; inputting energy consumption, preset operation data, and expected operation data into an objective function to determine a target value corresponding to the expected operation data includes:

[0033] Calculating a third difference between the expected operating data and the preset operating data; the third difference is used to measure the operating stability of the home appliance between the current moment and the next moment;

[0034] Determine the performance wear parameters of the home appliance when it is running at the expected operating data; the performance wear parameters are used to measure the performance of the home appliance;

[0035] A target value is determined based on the energy consumption, the third difference, and the performance wear parameter.

[0036] In the above technical solution, by calculating the third difference corresponding to the equipment's operational stability, the performance wear parameter representing the performance wear index, and combining it with energy consumption to comprehensively determine the target value, it is possible to quantify the equipment status in multiple dimensions. The third difference can evaluate operational stability from the perspective of temporal continuity, the performance wear parameter can describe the long-term performance loss of the equipment, and energy consumption can directly reflect the energy efficiency level at the next moment. Generating target values ​​based on these three indicators can simultaneously evaluate the smoothness, reliability, and energy consumption of the equipment's operation.

[0037] In one embodiment, determining the target value based on the energy consumption, the third difference, and the performance wear parameter includes:

[0038] The energy consumption, the third difference, and the performance wear parameter are weighted with the corresponding weights to obtain a target value; the weight corresponding to the energy consumption is greater than the weights corresponding to the third difference and the performance wear parameter.

[0039] In the above technical solution, energy consumption is directly linked to long-term user costs and energy efficiency policy requirements, reflecting the core competitiveness of home appliances. Therefore, energy consumption can be weighted higher than the weights of other pre-set indicators. Furthermore, the third difference parameter focuses on short-term operational stability, which falls under the category of ensuring a basic user experience. The performance wear parameter focuses on the long-term durability of the device, which is a hidden cost consideration. By using these weighted differences, the optimization goal can be prioritized towards energy efficiency, while also taking into account stability and reliability.

[0040] In one embodiment, the method further includes:

[0041] After the home appliance executes the control command, the actual energy consumption change between the current moment and the next moment, the third difference corresponding to the operation stability index, the user satisfaction feedback, and the fault handling action are obtained. The satisfaction is used to measure the degree to which the control command meets the user's expectations.

[0042] Determine the reward value corresponding to the control instruction based on the actual energy consumption change, the third difference, satisfaction, and fault handling action;

[0043] The first time series running data and the reward value are used as a set of training samples;

[0044] The intelligent agent is trained based on multiple groups of training samples to obtain a target intelligent agent; the target intelligent agent is used to implement the method of the first aspect as described above to generate control instructions.

[0045] In this technical solution, the actual energy consumption change after the home appliance executes the control command, the third difference corresponding to the operational stability index, user satisfaction, and the fault handling action are obtained. A reward value is determined based on this information. The first time series operational data and the reward value are used as training samples. The target agent is trained on multiple sets of samples. This effectively compensates for the shortcomings of the original control command generation step in responding to device aging, sudden environmental changes, user demand switching, device performance degradation, temporary user needs, and sensor failures. This achieves global intelligent optimization of multiple preset indicators such as home appliance energy consumption, user experience, and lifespan.

[0046] In one embodiment, an edge node device is deployed in a preset area of ​​the home appliance device, and the edge node device is provided with a target agent.

[0047] In the above technical solution, edge node devices are deployed in the preset area of ​​home appliances and target intelligent entities are set, which can effectively overcome the shortcomings of traditional cloud-based centralized home appliance energy management systems. In the process of generating control instructions, the first time-series operation data can be uploaded to the local network without passing through the home appliance. From the local network, it can be uploaded to the cloud server and then returned to the home appliance. It only needs to be transmitted to the edge computing node in the preset area for processing. In other words, the edge node device can process data directly locally, thereby avoiding the transmission of data back and forth to the cloud. When the network is congested or the signal is poor, the delay in generating and executing control instructions can be reduced, and real-time processing and decision-making can be achieved.

[0048] In a second aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the methods of the first aspect when executing the computer program.

[0049] In a third aspect, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by an electronic device, the electronic device executes the control method of the home appliance in the second aspect.

[0050] In a fourth aspect, a computer program product is provided, which includes: a computer program, which, when executed by an electronic device, enables the electronic device to execute the method for controlling the household appliance in the second aspect.

[0051] It can be understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A timing interaction diagram showing a method for controlling a household appliance in an electronic device in some embodiments of the present application is shown;

[0053] Figure 2 A schematic diagram of an application scenario of controlling a home appliance in a method for controlling a home appliance provided in an embodiment of the present application is shown;

[0054] Figure 3 A schematic diagram illustrating an implementation method for generating first time-series operation data in a method for controlling a household appliance provided in an embodiment of the present application is shown;

[0055] Figure 4 A schematic diagram illustrating an implementation method for updating a membership function in a method for controlling a household appliance provided in an embodiment of the present application is shown;

[0056] Figure 5 A schematic diagram illustrating an implementation method for determining a target value in a household appliance provided in an embodiment of the present application is shown;

[0057] Figure 6 A flowchart for implementing training an intelligent agent in a method for controlling a household appliance provided in one embodiment of the present application is shown;

[0058] Figure 7 A schematic diagram of an application scenario of a method for controlling a household appliance provided in another embodiment of the present application is shown. DETAILED DESCRIPTION

[0059] With the current trend of smart homes and energy conservation and emission reduction, home appliance energy management is crucial. Existing home appliance energy management systems are mostly based on fixed rules or simple data analysis.

[0060] For example, traditional home appliance energy management systems often pre-set a set of fixed operating logic. For example, air conditioners are set to run at a fixed temperature and fan speed during specific hours, and refrigerators maintain a single cooling intensity. This model ignores fluctuations in the real-time operating status of home appliances. For example, the actual cooling efficiency of air conditioner compressors may change due to factors such as age and indoor heat load. In this case, continuing to operate based on fixed rules will lead to energy waste.

[0061] Based on this, in order to improve the dynamic response capability to the real-time operating status and environmental changes of household appliances, so that energy optimization can be achieved according to actual conditions, an embodiment of the present application provides an electronic device and a control method for household appliances. By acquiring first time-series operating data of the household appliance in real time, where the first time-series operating data includes current state data and environmental parameters, the control method can accurately analyze the current operating state and environmental conditions of the household appliance at the current moment. Subsequently, the real-time first time-series operating data and preset operating data are input into a system model used to describe the temporal changes in the operating state of the appliance. Multiple expected operating data for the next moment can be obtained, enabling the household appliance to make adaptive adjustments based on its actual state and dynamic changes in the environment, thereby enhancing its dynamic response capability to state changes and environmental changes. The expected operating data is used to represent the operating state of the household appliance at the next moment. Subsequently, the electronic device can adaptively update the fuzzy set boundary corresponding to the preset membership function based on the real-time first time-series operating data to obtain a target membership function. Furthermore, the equal fuzzy level division can be made to better fit the real-time data distribution, solving the problem that traditional static boundaries cannot adapt to changes in device state or environment. Based on this, the fuzzy level corresponding to each expected operating data point is quantified based on the updated fuzzy set boundary in the membership function to determine the energy consumption of the home appliance when operating according to the expected operating data. This allows for refined quantification of device energy consumption and improves the accuracy of energy consumption calculations based on fuzzy logic. Finally, after obtaining multiple expected operating data points, the corresponding energy consumption of the home appliance when operating according to each expected operating data point can be determined. Furthermore, the energy consumption, preset operating data, and expected operating data are input into an objective function that measures the performance of the home appliance on multiple preset indicators during operation to obtain corresponding target values. Since the preset indicators in the objective function at least include the energy consumption indicator, and the target value is negatively correlated with the home appliance's performance on the preset indicators, this method can generate control instructions based on the expected operating data corresponding to the minimum target value, ensuring that the home appliance operates in an energy-optimized manner, effectively avoiding energy waste caused by the limitations of traditional operating modes, and achieving energy optimization.

[0062] To further describe the control methods of home appliances, please refer to Figure 1 , Figure 1 A timing interaction diagram of a method for controlling a household appliance in an electronic device in some embodiments of the present application is shown. Figure 1 As shown, the electronic device includes a controller, which is configured to perform the following steps:

[0063] S101: Acquire first time-series operation data of a household appliance.

[0064] In one embodiment, the electronic device may be a cloud server, an edge node device, or the like, without limitation. In this embodiment, the electronic device is explained by taking the edge node device as an example.

[0065] For example, edge node devices can be smart gateways or edge computing boxes, which can serve as a connection hub between various home appliances (such as air conditioners, refrigerators, and washing machines) and sensor networks (temperature, humidity, and occupancy sensors, etc.). These edge node devices can receive various data collected by sensors, such as ambient temperature and humidity, and operating data of home appliances, and adjust the operating status of home appliances in real time to make them more energy-efficient and efficient. Furthermore, they can receive app commands and convert them into signals that the corresponding home appliances can recognize.

[0066] As an example, see Figure 2 , Figure 2 A schematic diagram of an application scenario of controlling a home appliance in a method for controlling a home appliance provided in an embodiment of the present application is shown.

[0067] Edge node devices can use the Rockchip RK3588 high-performance embedded processor, which boasts 6Tops of computing power and runs a customized Linux operating system. This system integrates an efficient database management system. Edge node devices can utilize a heterogeneous computing architecture, with the CPU handling general tasks such as data scheduling and system management, while the NPU accelerates AI inference tasks and improves computing power utilization. Edge node devices can also be configured with 1GB of DDR4 memory and 32GB of eMMC storage, with built-in cache and optimized data storage structures for rapid data storage and retrieval. Furthermore, edge node devices can interact with home appliances and sensor networks with low latency through communication technologies such as Wi-Fi 6 and Bluetooth 5.2, ensuring timely and stable data transmission.

[0068] In another embodiment, depending on the usage scenario and purpose, other processors with computing power comparable to RK3588 can also be selected to build edge nodes, such as MediaTek Dimensity 700, Allwinner T537, Amlogic T311D, etc.

[0069] Home appliances are equipped with customized data acquisition modules and efficient control interfaces. These modules utilize high-precision sensors, such as air conditioner outlet temperature sensors with an accuracy of ±0.2°C, refrigerator internal humidity sensors with an accuracy of ±2%, and washing machine motor speed sensors with an accuracy of ±5 rpm. These sensors accurately capture the operating parameters of home appliances. To adapt to complex electromagnetic environments, data acquisition modules utilize electromagnetic shielding and anti-interference designs to ensure accurate data collection. The control interfaces rapidly receive control commands from edge node devices, enabling adjustments to the operating status of home appliances. Furthermore, home appliances can utilize modular designs for ease of installation, maintenance, and upgrades.

[0070] The sensor network consists of a variety of sensors distributed throughout the home environment. These include temperature sensors with an accuracy of ±0.1°C, humidity sensors with an accuracy of ±2%, and infrared sensors with a detection range of up to 5 meters. Using self-organizing network technology, these sensors form an efficient data collection network, enabling real-time and comprehensive collection of environmental parameters. To reduce the power consumption of the sensor network, low-power sensors and intelligent sleep technology are employed. Sensors automatically enter a sleep state when no data collection tasks are being performed. Furthermore, routing algorithms ensure rapid data transmission to edge node devices, providing data support for energy management decisions.

[0071] The user terminal can be equipped with a dedicated app developed based on a cross-platform development framework. Users can use the app to view the operating status and energy consumption data of home appliances in real time. The energy consumption data can be updated as often as five times per second, ensuring that users can keep up to date with the operating status of home appliances. The app has a remote control function, allowing users to remotely control the power on and off of home appliances, adjust parameters, and so on, with millisecond-level response speeds through smart devices such as mobile phones and tablets. In addition, the app also supports voice control and scene mode settings. Users can control home appliances through voice commands or set different scene modes according to their needs, such as home mode, away mode, and sleep mode.

[0072] Wireless access points act as network bridges, providing wireless network access services for home appliances (such as air conditioners, refrigerators, and washing machines) and sensor networks (temperature sensors, humidity sensors, and occupancy sensors). Home appliances and sensor networks can connect to the network based on wireless access points, enabling data transmission and interaction. For example, temperature data collected by a temperature sensor can be uploaded to an edge node device or app via a wireless access point. Control commands sent by users through an app can also be transmitted to home appliances via wireless access points. Edge node devices can generate control commands based on the various data uploaded by the sensor network and home appliances and transmit them to the home appliances.

[0073] In one embodiment, the first time-series operating data is a chronologically ordered set of data related to the operation of the home appliance and a set of environmental parameters of the environment in which the home appliance is located. The data may record the operating status information of the home appliance at different times over a period of time. The first time-series operating data includes the current state data and the environmental parameters of the environment in which the home appliance is located.

[0074] In one embodiment, the above-mentioned environmental parameters can be considered as quantitative indicators of external environmental factors that can affect the operating status or performance of the home appliance. For different home appliances or application scenarios, their environmental parameters are usually different.

[0075] For example, environmental parameters include, but are not limited to, temperature, humidity, light intensity, and air pressure, etc., which are not limited thereto.

[0076] It should be noted that environmental parameters can be obtained through built-in sensors (for example, the room temperature probe of the air conditioner), or external access (for example, through the linkage of the smart home system, obtaining real-time weather data from the smart weather station), or user input, and there is no limitation on this.

[0077] For example, for an air conditioner, the electronic device may collect data every five minutes over the past hour and arrange the data, such as the compressor speed, condenser and evaporator operating temperatures, outlet air temperature, ambient temperature, and humidity, in chronological order based on the collection time, to form the aforementioned first time-series operating data. The data collected at the current moment can be considered the aforementioned status data and environmental parameters.

[0078] It is understandable that by analyzing the first time series operating data, it is possible to understand the operating patterns, performance changes, and environmental changes of household appliances over a period of time. For example, by observing the first time series operating data of an air conditioner, it can be found that as the usage time increases, the compressor speed fluctuates in certain periods, which may indicate that the compressor performance has declined.

[0079] In one embodiment, different home appliances typically correspond to different status data. For example, for an air conditioner, the status data may include the compressor speed, the operating temperatures of the condenser and evaporator, and the air outlet temperature; for a refrigerator, the status data may include the refrigerator temperature, humidity, and refrigerant pressure, etc., without limitation.

[0080] It should be noted that due to equipment aging, environmental interference and other reasons, the raw data collected by sensors often contain outliers and noise.

[0081] It is understandable that processing based on the first time series operation data containing abnormal values ​​or noise will affect the accuracy of subsequent control instruction generation.

[0082] Based on this, in order to ensure the data quality of the first time series operation data, the electronic device can remove outliers and noise based on a distance measurement method and a filtering algorithm.

[0083] As an example, the electronic device may Figure 3 The steps S301-S303 shown process the acquired raw data to obtain the first time series operation data. The details are as follows:

[0084] S301: Acquire second time-series operation data of a household appliance.

[0085] In one embodiment, the second time-series operating data includes the operating data of the household appliance at the current moment. The operating data may include the status data of the household appliance at the current moment, as well as data such as environmental parameters of the environment in which the appliance is located. When the second time-series operating data includes operating data at historical moments, the operating data corresponds to the status data of the household appliance at historical moments, as well as environmental parameters of the environment in which the appliance is located. The second time-series operating data is similar to the first time-series operating data and will not be described in detail. It should be noted that the second time-series operating data is the raw data collected by the sensor.

[0086] S302 : For the operation data at any moment, with the operation data as the center, determine a set of continuous operation data sets with a preset number in the second time series operation data.

[0087] In one embodiment, since the operating data includes multiple types of data, such as status data (air outlet temperature and cooling power of the air conditioner) and environmental parameters (temperature and humidity of the environment), at this time, for each type of operating data, it is necessary to use the operating data as the center and determine the operating data set in the second time-series operating data of the corresponding type.

[0088] For example, taking the preset value of 3 as an example, for the operating data corresponding to the i-th outlet air temperature, the i-1th outlet air temperature, the i-th outlet air temperature and the i+1th outlet air temperature can be obtained as the operating data set corresponding to the i-th outlet air temperature.

[0089] The above preset values ​​can be set based on actual conditions and are not limited to this. A larger preset value results in a smoother curve corresponding to the filtered operating data, but may lose some data details. A smaller preset value results in less noise suppression but better preservation of data variation characteristics. Therefore, the optimal preset value can be determined based on experimental results.

[0090] It should be noted that, since the operation data set needs to be generated with the operation data as the center, the number of data included in the operation data set should be an odd number.

[0091] It should be noted that, based on the above steps, a corresponding operational data set will be obtained for each operational data point. Furthermore, since the operational data set needs to be generated by acquiring multiple operational data points before and after the operational data point, for operational data points (e.g., the first or last operational data point) that do not have sufficient operational data points to satisfy the window length (the window length is a preset value), it is possible to choose not to generate the operational data set corresponding to the operational data point, or to retain the original data point as the filtered value.

[0092] Alternatively, the operating data corresponding to the phase window is used as the operating data set. For example, taking the preset value as 3, for the first operating data, the operating data set may include the first operating data and the second operating data.

[0093] S303 : Replace the corresponding operating data with the average value of the operating data set to obtain first time series operating data.

[0094] In one embodiment, the average value may be calculated by, but is not limited to, an arithmetic average, a weighted average, or the like, and is not limited thereto. After obtaining the average value of each operating data set, the average value may be substituted for the corresponding operating data. That is, the data in the first time series operating data are all the replaced average values.

[0095] As an example, after obtaining the second time series operation data, the electronic device can perform noise removal according to the following formula. The formula is as follows:

[0096]

[0097] Among them, y k represents the average value corresponding to the kth running data, m represents the preset value, x i represents the i-th running data in the running data set, and n represents the total number of second time series running data.

[0098] It should be noted that, taking the window with a preset value of m (m is an odd number) as an example, the center position of the window is the position corresponding to the kth running data. When the window slides from left to right, the starting position must ensure that the window can be completely included in the running data set. Therefore, k must be at least That is, when At the beginning, the leftmost part of the window just reaches the first position of the second time series running data (at this time, there are m data in the window). Similarly, when the window slides to the right, in order to ensure the window is complete, k is the largest to At this time, the rightmost side of the window just reaches the last position of the second time series running data.

[0099] It should be noted that for each piece of operating data, determining a preset number of continuous operating data sets centered around that data point in the second time-series operating data can fully leverage the local correlation of the data to capture the changing trends of the home appliance's operating status over a short period of time, avoiding the isolation of individual data points. Subsequently, replacing the corresponding operating data with the average value of the operating data set to obtain the first time-series operating data effectively smooths random fluctuations in the data, removing noise interference and reducing the impact of abnormal fluctuations on subsequent analysis.

[0100] In another embodiment, to further remove outliers from the first time-series operating data, the electronic device may further calculate the mean and standard deviation of the first time-series operating data. Then, a first difference between the mean and a preset multiple of the standard deviation is calculated, as well as a sum of the mean and the preset multiple of the standard deviation. Finally, the operating data in the first time-series operating data that is not between the first difference and the sum is deleted. In this case, the set of operating data that remains undeleted is the aforementioned first time-series operating data.

[0101] It can be understood that since the first time series operating data includes multiple types of operating data, for different types of operating data (for example, ambient temperature, humidity, air outlet temperature of the air conditioner, cooling power, etc.), the mean and standard deviation of each type of operating data need to be calculated separately to remove outliers in each type of operating data.

[0102] For example, the electronic device may divide the first time-series operating data into multiple independent subsets based on type, with each subset corresponding to a specific type of operating data. For example, a subset corresponding to the ambient temperature type, a subset corresponding to the ambient humidity type, and so on. Then, for each subset, the mean and standard deviation corresponding to the subset are calculated, and a first difference between the mean and a preset multiple of the standard deviation, as well as a sum of the mean and the preset multiple of the standard deviation, are calculated. Finally, the operating data in the subset that is not between the first difference and the sum is deleted.

[0103] It should be noted that if the current state data and environmental parameters in the first time series operation data are deleted, the data collected this time can be considered as abnormal data. Therefore, the electronic device does not need to perform the subsequent processing steps.

[0104] The preset multiple can be set according to actual circumstances and is not limited thereto. For example, the preset multiple can be 3. Furthermore, based on the above explanation of the method for calculating the average value, the electronic device can also calculate the standard deviation based on the arithmetic mean or weighted mean, and this is not limited thereto. The method for calculating the standard deviation based on the average value is an existing method and is not described in detail.

[0105] It's important to note that the standard deviation reflects the degree of data dispersion, and the preset multiple allows for flexible adjustment of the range based on actual needs. When operating data falls outside this range, it can be considered an outlier due to noise or device anomalies. Therefore, removing these outliers allows the first time-series operating data to more accurately reflect the actual operating status of the home appliance, reducing the potential for misleading analysis and decision-making from abnormal data.

[0106] It should be added that when executing the above steps S301-303, the first time series operation data can also be divided into multiple independent subsets based on type, and then the above steps S301-303 are executed for the operation data in each subset respectively. This will not be explained in detail.

[0107] S102: Input the first time-series operation data and the preset operation data of the home appliance into a system model of the home appliance to obtain a plurality of expected operation data of the home appliance at the next moment.

[0108] In one embodiment, the preset operating data may refer to standard operating state parameters pre-set based on the design objectives, performance parameters, usage scenarios, or user needs of the home appliance. Typically, the preset operating data serves as a reference for system control, model prediction, or operation optimization, measuring whether the actual operating state of the device meets expectations, or as input conditions to drive the system model to generate prediction results (e.g., control instructions).

[0109] For example, an air conditioner's preset operating data in "energy-saving mode" and "fast cooling mode" are different; and a refrigerator's preset temperature in "quick freezing mode" is lower than its preset temperature in "normal mode," but its cooling power is higher. In other words, the electronic device can determine the preset operating data based on the operating mode of the home appliance, or obtain the preset operating data in response to user input.

[0110] In one embodiment, the system model is used to describe how the operating state of a household appliance changes over time. Specifically, the system model can be considered a mathematical framework constructed based on the physical characteristics and operating patterns of the household appliance. It is used to describe the dynamic relationship between the input and output of the appliance (i.e., how the operating state of the household appliance changes over time). This allows for accurate modeling of system behavior through a state-space model.

[0111] Exemplarily, the system model may include a state equation that describes how the system's internal state changes over time, and an output equation that converts internal state variables into observable and measurable information.

[0112] Among them, the state equation can be as follows:

[0113]

[0114] in, represents the first-order derivative of the state variable x(t), A is the f×f system matrix, describing the relationship between the internal states of the system, and B is the f×h input matrix, describing the influence of the input variables on the state variables; f is the number of state variables, and h is the number of input variables.

[0115] The output equation can be as follows:

[0116] y(t)=Cx(t)+Du(t);

[0117] Where C is the p×f output matrix, which describes the influence of state variables on output; D is the p×h transmission matrix, which describes the direct influence of input variables on output variables; and p is the number of output variables.

[0118] Taking air conditioning as an example, x(t) may include state data such as the speed of the compressor, the temperature of the condenser and evaporator, etc.; the input variable u(t) may include data such as environmental parameters and the set temperature in the preset operating data. In other words, it can be considered that represents the state data and environmental parameters in the first time-series operating data, as well as the preset operating data. Furthermore, the output variable y(t) may include the desired operating temperature, cooling / heating power, etc. In other words, y(t) can be considered to represent the expected operating data.

[0119] In addition, the specific values ​​of matrices A, B, C, and D can be determined by analyzing key components such as the compressor, condenser, and evaporator (for example, constructing thermodynamic equations using Fourier's law and the law of conservation of energy) and combining them with experimental data (for example, temperature change curves at different speeds).

[0120] Similar to the air conditioner mentioned above, for refrigerators, it is necessary to consider factors such as their refrigeration cycle characteristics and the impact of door opening frequency on internal temperature to build a system model based on thermodynamic principles; for washing machines, corresponding system models can be established based on characteristics such as motor load changes and water level control to achieve accurate simulation of the operating status of different household appliances.

[0121] In one embodiment, the expected operating data refers to the operating state parameters of the household appliance at the next moment predicted by the system model, representing the "prediction result" or "theoretical output" of the system model for the future state of the appliance.

[0122] It should be noted that there are multiple expected operating data, each of which can be used to represent the operating state of the home appliance at the next moment. For example, the operating state of the home appliance at the next moment may be: the air outlet temperature of the air conditioner is 14°C (lowering the air outlet temperature to speed up the indoor cooling rate) and the cooling power is 1500W (increasing the compressor power to enhance the cooling capacity); or the air outlet temperature of the air conditioner is 18°C ​​(higher air outlet temperature reduces the compressor load) and the cooling power is 1000W (lower power operation to reduce energy consumption).

[0123] S103 : updating the fuzzy set boundary corresponding to the preset membership function based on the first time series operation data to obtain a target membership function.

[0124] In one embodiment, an electronic device can obtain historical energy consumption data of household appliances and then classify the energy consumption into several fuzzy levels based on demand. For example, there are three levels: low energy consumption, medium energy consumption, and high energy consumption. Subsequently, the energy consumption quantiles are calculated based on data statistics. For example, low energy consumption corresponds to the first 20% of all historical energy consumption data, medium energy consumption corresponds to the middle 60% of all historical energy consumption data, and high energy consumption corresponds to the last 20% of all historical energy consumption data. Finally, a membership function corresponding to each fuzzy level is established based on the quantiles. For example, a triangular function, a trapezoidal function, a Gaussian function, etc.

[0125] It can be understood that by quantifying energy consumption through membership functions, continuous physical quantities can be converted into fuzzy levels that are consistent with human cognition, providing a flexible decision-making basis for energy-saving control, energy efficiency evaluation and user interaction of home appliances.

[0126] However, it should be noted that in the fuzzy processing of environmental parameters, it is usually impossible to dynamically respond to the real-time changing characteristics of the data, and it is difficult to adaptively adjust the fuzzy set division, which causes the fuzzification results to deviate from the actual data distribution, affecting the accuracy.

[0127] For example, in everyday life, descriptions like "high temperature," "low temperature," and "humidity" often lack absolute standards. For example, one user might consider 28°C to be high, while another might consider 32°C to be high. Therefore, using a fixed temperature line to distinguish between "high temperature" and "low temperature" introduces conceptual ambiguity, making it difficult to adaptively adjust the fuzzy set partitioning, which in turn affects the accuracy of subsequent energy consumption calculations.

[0128] Based on this, in order to improve the accuracy of energy consumption calculation, the electronic device can execute the above S103 step to update the fuzzy set boundary corresponding to the membership function to obtain the target membership function, so that the equal fuzzy level division is more in line with the real-time data distribution, and solve the problem that the traditional static boundary cannot adapt to the device status or environmental changes.

[0129] The target membership function is used to quantify the fuzzy level corresponding to each expected operating data based on the updated fuzzy set boundary, so as to determine the energy consumption of the household appliances when they are operated with the expected operating data.

[0130] For example, the electronic device may be configured as follows: Figure 4 The steps S401-S403 shown update the membership function to calculate the above energy consumption. The details are as follows:

[0131] S401: Determine the distribution value of each operation data in the first time series operation data at different preset data values.

[0132] In one embodiment, the above-mentioned operation data is data in the first time-series operation data, represents the operation status information and environmental parameters of the household appliance at a specific moment, and is a basic unit that constitutes the first time-series operation data.

[0133] The preset data values ​​are a series of pre-set data values ​​used to analyze the distribution of operating data under different values. The values ​​can be determined based on experience, equipment characteristics, or relevant standards.

[0134] For example, the user may set the preset data value based on the actual situation of the home appliance and the analysis requirements. For example, for the first time series operation data of the temperature type, the preset data value may be set according to the operating temperature range and common temperature values ​​of the device.

[0135] The above-mentioned distribution value refers to a quantitative value of the distribution of each operating data in the first time series operating data on different preset data values, indicating the degree of association or occurrence probability of the operating data with a certain preset data value.

[0136] In one embodiment, the electronic device may pre-divide the interval range corresponding to each preset data value, and then count the amount of operating data within the interval range, and finally determine the data amount as the above distribution value.

[0137] In another embodiment, for any preset data value, the electronic device can further calculate a second difference between the preset data value and each running data, and then input each second difference into a preset kernel density estimation function to obtain a distribution value.

[0138] The kernel density estimation function is a non-parametric estimation method used to estimate the probability density function of a random variable. Specifically, the kernel density estimation function can be as follows:

[0139]

[0140] Where f(o) represents the distribution of the first time series running data at the preset value o, n represents the number of running data in the first time series running data, i represents the i-th running data, K represents the Gaussian kernel function, h is the preset bandwidth parameter, and o represents the o-th preset value. In other words, oi represents the second difference.

[0141] It's important to note that calculating the difference between the preset data values ​​and each set of operating data can intuitively quantify the degree of deviation between the preset values ​​and the actual data, providing a fundamental metric for analyzing data distribution. Furthermore, using kernel density estimation to process these differences allows for adaptive capture of complex data patterns directly through the smoothing properties of the kernel function, effectively avoiding the information loss inherent in discretization methods. Furthermore, bandwidth parameter adjustment allows for a balanced bias and variance. Consequently, this allows for an accurate description of the distribution characteristics of operating data at different preset values.

[0142] S402: Determine the preset data value corresponding to the maximum value among the multiple distribution values ​​as the target data value.

[0143] S403: Update the fuzzy set boundary based on the target data value to obtain the target membership function.

[0144] In one embodiment, the target data value is a representative data value selected from a plurality of distribution values, corresponding to the maximum value in the distribution values. Generally, the target data value can be considered as a preset data value that best represents the distribution characteristics of the first time series running data.

[0145] It should be noted that the above target data value can be one or more, and there is no limitation on this.

[0146] For example, for the first time series operating data of the temperature type, temperature can be divided into two concepts: high temperature and low temperature. In this case, the target data value can be 1. When the operating data is greater than the target data value, it can be considered that the operating data indicates a high temperature. When the operating data is less than or equal to the target data value, it can be considered that the operating data indicates a low temperature.

[0147] Alternatively, temperature can be categorized into three concepts: high temperature, medium temperature, and low temperature. In this case, there can be two target data values. When the operating data is greater than the first target data value, the operating data can be considered to indicate a high temperature; when the operating data is less than or equal to the second target data value, the operating data can be considered to indicate a low temperature; and when the operating data is less than or equal to the first target data value and greater than the second target data value, the operating data can be considered to indicate a medium temperature.

[0148] In one embodiment, the membership function is a mathematical tool used to describe the fuzzy correspondence between "operating data" and "level" (such as high, medium, and low), and the range of different levels can be defined by the fuzzy set boundaries (such as interval endpoints and membership curve shapes).

[0149] For example, three intervals can be pre-set in the electronic device, each corresponding to a membership function. In this case, after obtaining the target data value, the fuzzy set boundaries in the membership function can be replaced with the target data value to update the membership function. In other words, the interval endpoints in the intervals can be replaced to dynamically adjust the fuzzy set boundaries and membership function, so that the updated membership function better matches the actual distribution of the first time series running data, thereby improving the accuracy of the fuzzification processing.

[0150] It should be noted that using the maximum value of the distribution corresponding to the first time series operating data as a benchmark can bring the center of the fuzzy set closer to the dense area of ​​the actual operating data. This can reduce membership misjudgments caused by boundary shifts when subsequently calculating the energy consumption corresponding to the expected operating data, thereby improving the accuracy and reliability of energy consumption assessments.

[0151] In one embodiment, the electronic device can determine the energy consumption corresponding to each expected operating data item based on the updated membership function. For example, when calculating the energy consumption corresponding to the expected operating data, the fuzzy set boundary corresponding to the expected operating data can be first determined. The expected operating data can then be input into the membership function corresponding to the fuzzy set boundary to obtain a classification result. Finally, the energy consumption is determined based on the classification result and the fuzzy logic rules.

[0152] As an example, taking the first time series operating data as the ambient temperature variable, if the target data values ​​determined based on the above steps S401-S402 are 26°C and 35°C respectively, the fuzzy set boundaries and membership functions can be re-updated based on the target data values.

[0153] Specifically, taking the membership function corresponding to the high temperature fuzzy set as an example, its membership function can be as follows:

[0154]

[0155] Among them, T can be considered as the expected running data, T low represents one of the target data values ​​(for example, 26°C above), T high Indicates another target data value (for example, 35°C as mentioned above). That is, T low and T high is the fuzzy set boundary dynamically adjusted according to the above steps S401-S403.

[0156] It should be noted that when the membership function contains letters corresponding to the fuzzy set boundaries, the membership function also needs to be updated accordingly. low <T<T high The corresponding function, which contains the fuzzy set boundary T low and T high .

[0157] In another embodiment, the membership functions corresponding to the medium-temperature fuzzy set and the low-temperature fuzzy set can also be dynamically adjusted according to the above steps S401-S403, which will not be further described. For example, the fuzzy set boundaries of the membership function corresponding to the medium-temperature fuzzy set can be 24°C and 26°C, respectively, and the fuzzy set boundaries of the membership function corresponding to the low-temperature fuzzy set can be 22°C and 24°C, respectively.

[0158] As an example, when T is 25°C, since T is less than or equal to T low ,u high (T) = 0, indicating that the ambient temperature is not "high temperature". When T is 28 ° C, since T is less than or equal to T high , greater than T low , we can calculate u by the above formula high (T) = 0.44, indicating that the degree of the ambient temperature belonging to "high temperature" is about 44%. And when T is 35 ° C, since T is greater than or equal to T high , we can calculate u by the above formula high (T)=1, indicating that the ambient temperature is "high temperature".

[0159] It should be noted that the above is only the processing of the environmental parameters in the first time series operation data. The status data can also be processed as described in the above example, which will not be explained again.

[0160] It is understood that the higher or lower the ambient temperature, or the higher the heating or cooling power of the air conditioner, the higher the energy consumption required when the air conditioner is in operation. Therefore, the above energy consumption can be determined based on the classification result corresponding to the expected operating data. For example, the electronic device can perform reasoning based on preset fuzzy logic rules to obtain the above energy consumption.

[0161] As an example, the fuzzy logic rule can be IFEis A, AND Xis B, THEN Yis C, where E represents the environmental parameter, X represents the expected operating data, Y represents the energy consumption, and A, B, and C are the fuzzy sets (updated membership functions) of the corresponding first time series operating data. That is, the above fuzzy logic rule indicates that when the household appliance is in the result corresponding to A (for example, high temperature) and the expected operating data is in the result corresponding to B (for example, high cooling power), its corresponding energy consumption is the value calculated by the C membership function. In this embodiment, the specific setting of the fuzzy logic rule is not limited.

[0162] It should be noted that in this embodiment, by analyzing the distribution characteristics of the first time series operating data at different preset data values, the preset data value corresponding to the maximum distribution value is determined as the target data value, which can accurately describe the core distribution trend of the data and avoid edge data interference. In this way, dynamically updating the fuzzy set boundary of the membership function based on the target value can make the equal fuzzy level division more consistent with the real-time data distribution, solving the problem that traditional static boundaries cannot adapt to changes in device status or environment. Furthermore, determining the energy consumption of the expected operating data based on the updated membership function can achieve refined quantification of device energy consumption and improve the accuracy of fuzzy logic decision-making.

[0163] In another embodiment, the electronic device can also decompose energy consumption into the sum of the power consumption of each component based on the operating principle of the appliance to establish a mathematical expression for the device's energy consumption. The expected operating data can then be input into the mathematical expression to obtain the aforementioned energy consumption. Alternatively, the aforementioned energy consumption can be obtained by performing feature prediction on the expected operating data based on a pre-trained energy consumption prediction neural network model. In this embodiment, the method for determining energy consumption is not limited.

[0164] S104 . For any expected operating data, input the energy consumption, the preset operating data, and the expected operating data into an objective function to determine a target value corresponding to the expected operating data.

[0165] In one embodiment, the objective function is used to measure the performance of a home appliance across multiple preset indicators during operation; the preset indicators include energy consumption. Furthermore, the target value is negatively correlated with the home appliance's performance on the preset indicators. That is, the larger the target value, the worse the overall performance of the home appliance on the preset indicators; the smaller the target value, the better the overall performance of the home appliance on the preset indicators.

[0166] As an example, the electronic device may determine a correction coefficient for correcting energy consumption based on preset operating data and expected operating data, and then determine the target value by multiplying the correction coefficient by the energy consumption. The correction coefficient may be determined by pre-establishing a correlation between the preset operating data, the expected operating data, and the correction coefficient, and then determining the correction coefficient based on the correlation.

[0167] In another embodiment, the above-mentioned preset indicators may also include one or more of an operation stability indicator, a performance indicator, a comfort indicator, etc., which is not limited to this. For ease of explanation, this embodiment is described by taking the example that the preset indicators also include an operation stability indicator and a performance indicator. Figure 5 As shown in steps S501-S503, Figure 5 A schematic diagram of an implementation method for determining a target value in a household appliance provided in an embodiment of the present application is shown. Detailed description is as follows:

[0168] S501: Calculate a third difference between expected operating data and preset operating data.

[0169] In one embodiment, the third difference is used to measure the operational stability of the home appliance between the current moment and the next moment. It should be noted that the "operational stability" of a home appliance generally refers to the continuity, predictability, and low volatility of its operating state (e.g., parameters such as temperature, pressure, frequency, and power) at adjacent time points.

[0170] The preset operating data can be considered the "ideal state" of the home appliance (such as the air conditioner set temperature, refrigerator set humidity, etc.). The smaller the third difference between the expected operating data (actual or predicted data at the next moment) and the preset operating data, the closer the actual operating state of the home appliance is to the target, with smaller fluctuations and higher device operational stability. Conversely, the larger the third difference, the further the actual operating state deviates from the ideal state, with more severe operational fluctuations and lower device operational stability.

[0171] In addition, the third difference is calculated for the "current moment" and the "next moment," essentially evaluating the state changes between adjacent moments in a time series. If the third difference remains low for multiple consecutive moments, it can be considered that the home appliance can continuously maintain a state close to the preset operating data in the time dimension, and the device operation is stable. If the third difference fluctuates frequently or suddenly increases, it can be considered that the home appliance operation is unstable and may be at risk of failure or affected by external interference (such as voltage fluctuations, environmental changes, etc.).

[0172] For example, taking the type of preset operating data as humidity, if the preset humidity of the refrigerator's refrigeration area is 50% RH, the humidity at the current moment is 49% RH, and the expected humidity at the next moment is 51% RH, then the third difference is ±1% RH, indicating that the humidity fluctuation is small and the refrigeration environment is stable; if the expected humidity at the next moment suddenly changes to 65% RH, and the third difference reaches +15% RH, it can be considered that the humidity deviates significantly from the preset humidity, and there may be unstable operation due to frequent door opening or equipment failure, which requires early warning or adjustment.

[0173] S502: Determine performance wear parameters when the household appliance operates according to expected operating data.

[0174] In one embodiment, the performance wear parameters are used to measure the performance of household appliances.

[0175] Electronic devices can pre-set performance wear parameters corresponding to various types of operating data, allowing direct determination of the performance wear parameters corresponding to the desired operating data. Alternatively, relying on a data-driven model or physical model of the home appliance, a mapping relationship between quantifiable operating data and performance wear parameters can be established through correlation analysis between historical operating data and actual wear status, and the performance wear parameters corresponding to the desired operating data can be determined based on the mapping relationship.

[0176] In another embodiment, the electronic device may also calculate the performance wear parameter using the following formula. Detailed description is as follows:

[0177]

[0178] Where wear(k) represents the performance wear parameter at the kth moment, u(k) represents the expected operating data at the kth moment, and unom represents the rated operating data of the home appliance. For example, in the case of an air conditioner, unom can represent the rated operating current of the compressor.

[0179] S503 : Determine a target value based on the energy consumption, the third difference, and the performance wear parameter.

[0180] In one embodiment, the electronic device may add the energy consumption, the third difference, and the performance wear parameter to obtain the target value.

[0181] In another embodiment, since the various preset indicators have different degrees of influence on the operation of household appliances, in order to reasonably evaluate the expected operating data that meets the requirements, when calculating the target value, the energy consumption, the third difference and the performance wear parameters can be weighted with the corresponding weights to obtain the target value.

[0182] The weight corresponding to the energy consumption is greater than the weights corresponding to the third difference and the performance wear parameter. For example, the weight corresponding to the energy consumption may be 0.8, and the weights corresponding to the third difference and the performance wear parameter may be 0.1.

[0183] It's important to note that energy consumption is directly linked to long-term user costs and energy efficiency policy requirements, reflecting the core competitiveness of home appliances. Therefore, energy consumption can be weighted higher than other pre-set indicators. Furthermore, the third difference parameter focuses on short-term operational stability, a fundamental aspect of user experience, while the performance wear parameter focuses on long-term device durability, a hidden cost consideration. By using these weighted differences, optimization can prioritize energy efficiency while also taking into account stability and reliability. This approach not only meets users' core desire for low energy consumption but also ensures a balanced overall device operating status.

[0184] It is understandable that when other preset indicators (for example, operation stability indicators) need to be determined as the main indicators, the weight corresponding to the main indicator can be increased accordingly, and the weights corresponding to the remaining indicators can be reduced to calculate the above target value. This is not explained in detail.

[0185] In another embodiment, the electronic device may further calculate the energy consumption corresponding to the operating data at each moment in the first time-series operating data, and input the energy consumption at each moment, the preset operating data, and the expected operating data at the corresponding moment into the objective function and add them together to obtain a total target value. Subsequently, subsequent steps are processed based on the total target value.

[0186] It can be understood that by calculating the total target value based on the operating data at each moment in the first time series operating data, the stability and anti-interference ability of the electronic device in subsequently generating control instructions can be improved through the "historical-current" correlation relationship in each operating data.

[0187] The calculation method of the energy consumption at each moment and the corresponding expected operating data is similar to the calculation method of the energy consumption at the current moment and the corresponding expected operating data, which will not be explained in detail. The total target value can be calculated according to the following formula, which is detailed as follows:

[0188]

[0189] Among them, J represents the total target value, N represents the total number of first time series operation data, a1 represents the weight corresponding to energy consumption, E(k) represents the energy consumption corresponding to the k-th moment, a2 represents the weight corresponding to the operation stability index, u(k) represents the expected operation data at the k-th moment, uref(k) represents the preset operation data at the k-th moment, a3 represents the weight corresponding to the performance wear index, and Wear(k) represents the performance wear parameter at the k-th moment.

[0190] In this embodiment, by calculating the third difference corresponding to the device's operational stability, the performance wear parameter representing the performance wear indicator, and combining this with energy consumption to determine the target value, a multi-dimensional quantification of the device's status can be achieved. The third difference can assess operational stability from the perspective of temporal continuity, the performance wear parameter can describe the device's long-term performance loss, and energy consumption can directly reflect the energy efficiency level at the next moment. Generating target values ​​based on these three indicators can simultaneously assess the smoothness, reliability, and energy consumption of the device's operation.

[0191] S105 : Generate a control instruction at the current moment based on the expected operation data corresponding to the minimum target value, and send the control instruction to the home appliance.

[0192] In one embodiment, after obtaining multiple target values, it can be considered that the minimum target value corresponds to the minimum required energy consumption; or, the objective function is constructed by integrating energy consumption and other preset indicators (for example, operating stability, performance, etc.). In this case, the minimum target value is to achieve a better balance between multiple indicators, not simply pursuing the extreme of a certain indicator, but performing better in multiple important indicators. Then, a control instruction is generated based on the expected operating data corresponding to the minimum target value and sent to the home appliance. The control instruction is used to control the home appliance to operate with the expected operating data corresponding to the minimum target value, thereby ensuring that the home appliance can save energy and reduce consumption while meeting the requirements of other preset indicators (for example, stable operation and reduced wear) during operation, thereby achieving overall performance optimization.

[0193] In one embodiment, generating a control instruction at a current moment and sending it to a home appliance can cause the home appliance to operate with desired operating data at a next moment. Specifically, the control instruction is generated at the current moment to control the home appliance to ensure that the state data of the home appliance is adjusted to the state set by the desired operating data from the current moment to the next moment.

[0194] In one embodiment, the desired operating data is typically presented in the form of parameters (e.g., air outlet temperature 26°C, humidity 50% RH, cooling power 1500W). Based on this, the electronic device can reversely map the parameters corresponding to the desired operating data into control inputs based on a preset device mechanism model or data-driven model to generate control instructions.

[0195] It should be noted that the control instructions generated based on the above steps S101-S105 are generated based on theoretical models (e.g., system models and objective functions). However, in actual scenarios, factors such as environmental parameters (e.g., room temperature fluctuations, voltage instability), device status (aging of components, filter clogging), and user behavior (frequent opening and closing of doors) often change the actual operating conditions of home appliances in real time. In this case, if the operation of home appliances is controlled only based on the control instructions generated in steps S101-S105, the status data of the home appliances may not actually reach the preset operating data.

[0196] Based on this, to enable the home appliance to operate in a state corresponding to the preset operating data, after the home appliance responds to the control command, the electronic device can also obtain the actual operating data of the home appliance after responding to the control command. Then, based on the actual operating data and the expected operating data corresponding to the control command, the electronic device generates an optimized control command and sends it to the home appliance again. In other words, the electronic device can execute the home appliance control method and the optimized control command generation process every preset period.

[0197] It can be understood that through the above method, electronic devices can not only utilize the goal orientation of open-loop control (generating control instructions corresponding to the expected operating data of the minimum value) to ensure the accuracy of the control instruction generation, but also use the dynamic adjustment capability of closed-loop control (optimizing the generation of control instructions) to deal with uncertain factors such as model errors, environmental disturbances and equipment aging, and ultimately achieve high-precision operation, energy efficiency optimization and fault adaptability of household appliances under complex working conditions, significantly improving user experience and equipment service life.

[0198] In one embodiment, the electronic device may generate an optimized control instruction based on the actual operation data and the expected operation data corresponding to the control instruction through a PID closed-loop control method, which will not be described in detail.

[0199] In another embodiment, when generating control instructions corresponding to target values, the electronic device can employ parallel computing to simultaneously solve for the target value corresponding to each desired operating data point, thereby generating the control instructions. For example, within each sampling period (e.g., 100 milliseconds), the electronic device can employ a fast optimization algorithm that combines parallel computing with heuristic search to find the optimal solution to the objective function based on the current state data and environmental parameters, and generate control instructions in real time.

[0200] Specifically, electronic devices can deconstruct complex optimization problems into multiple independent sub-problems, and use parallel computing architecture to achieve simultaneous solution of multiple sub-problems to reduce computing time; at the same time, through heuristic search technology, prior guidance rules are constructed based on problem characteristics to quickly converge to an approximate optimal solution, ensuring that control instructions are output within milliseconds.

[0201] It is understandable that through the aforementioned parallel mechanism, the computational processes of each sub-problem are simultaneously advanced on a hardware multi-core architecture (e.g., CPU multi-threading or GPU acceleration), avoiding the timing-dependent bottlenecks of traditional serial algorithms. For example, environmental data analysis and air conditioning performance calculations can be performed simultaneously without waiting for the preceding modules to complete. Ultimately, a heuristic search algorithm is used to globally integrate the results of multiple sub-problems, rapidly generating outlet temperature control instructions that balance energy efficiency and comfort. This enables the system to make decisions within the sampling period, meeting the latency requirements for real-time control of home appliances.

[0202] In another embodiment, the electronic device described above can be an edge node device. In this scenario, the edge node device can be deployed in a preset area of ​​the home appliance device, and the edge node device is provided with a target intelligent agent, and then the above steps are implemented through the target intelligent agent.

[0203] In one embodiment, the preset area refers to a specific spatial range determined in advance. In the home appliance control scenario, the preset area can be a physical space range. For example, a room (such as a living room) in a family home, an entire house, or an office or a floor in a commercial place, etc., without limitation. In addition, the preset area can also be a logical area, such as a group of devices in a smart home system divided according to the functions or management requirements of home appliances. For example, all refrigeration equipment (refrigerators, air conditioners) are grouped into one area. Edge node devices and target intelligent bodies are deployed within the scope of the above-mentioned preset area to enable efficient and intelligent management and control of home appliances in the preset area.

[0204] The target agent can be considered an intelligent software module deployed in edge node devices, capable of making autonomous decisions. Specifically, the target agent can generate control instructions based on artificial intelligence technologies (such as machine learning and reinforcement learning) or traditional control theories (for example, PID and model predictive control).

[0205] For example, the target intelligent agent can automatically generate control instructions for household appliances based on preset indicators (such as energy consumption, operating stability and performance) and real-time data (first time series operating data), and has certain environmental perception, logical reasoning and behavioral decision-making capabilities.

[0206] In this embodiment, edge node devices are deployed in a preset area of ​​home appliances and target intelligent entities are set, which can effectively overcome the shortcomings of traditional cloud-based centralized home appliance energy management systems. In the process of generating control instructions, the first time-series operation data can be uploaded to the local network without passing through the home appliance, and then uploaded to the cloud server from the local network, and then returned to the home appliance by the original route. It only needs to be transmitted to the edge computing node in the preset area for processing. Furthermore, the transmission of data to and from the cloud can be avoided, so that when the network is congested or the signal is poor, the delay in generating and executing control instructions can be reduced from 5-10 seconds to 10-30 milliseconds, realizing real-time processing and decision-making. For example, when a user adjusts the air conditioner temperature, the response can be within 30 milliseconds, which improves the user experience and meets the application requirements of smart home appliances in high real-time scenarios.

[0207] It should be noted that in actual scenarios, the control instructions generated based on the above steps S101-S105 may still have problems such as the system model being difficult to adapt to dynamic changes such as equipment aging and environmental mutations in a timely manner, the static setting of the objective function weight being unable to match user demand switching and equipment performance degradation, and the lack of effective response to scenarios such as temporary user needs and sensor failures. Therefore, it is possible to train an intelligent agent (edge ​​computing node) to dynamically correct the system model deviation through continuous interaction with the environment, adaptively adjust the weights corresponding to each preset indicator in the objective function, and learn the optimal strategy for unknown scenarios based on the exploration mechanism, thereby making up for the inherent defects of steps S101-S105 in dynamic adaptability, multi-objective collaboration, and rare scenario processing, and achieving global intelligent optimization of energy consumption, experience, and life of home appliances.

[0208] As an example, the electronic device may Figure 6 The steps S601-S604 shown above train the agent to obtain the target agent. The details are as follows:

[0209] S601. After the home appliance executes the control instruction, obtain the actual energy consumption change of the home appliance between the current moment and the next moment, the third difference corresponding to the operation stability index, the satisfaction feedback from the user, and the fault handling action.

[0210] In one embodiment, the actual energy consumption change refers to the difference between the actual energy consumption (e.g., electricity or gas) consumed by the household appliance from the moment it executes the control instruction to the next moment and the previous moment, and is used to measure the actual impact of the control instruction on energy consumption. Electronic devices can collect real-time energy consumption data from the current moment to the next moment using energy consumption sensors (e.g., electricity meters or gas meters) built into the household appliance to calculate the actual energy consumption change.

[0211] The satisfaction level is used to measure the degree to which the control instructions meet the user's expectations. For example, the satisfaction level can be input by the user. For example, a satisfaction evaluation request can be pushed to the user through the device operation interface (e.g., touch screen, APP) or a third-party platform (e.g., questionnaire), and the user can express his or her opinion by scoring (e.g., 1-5 points) or providing text feedback.

[0212] Alternatively, the electronic device can analyze user usage behavior data (e.g., device mode switching frequency, operation duration, etc.) to indirectly infer satisfaction (requires machine learning model training). In this embodiment, the method for the electronic device to obtain satisfaction is not limited.

[0213] These fault handling actions refer to electronic devices that monitor home appliance operating data in real time, combine it with pre-set algorithms or models, identify potential device failure risks in advance, and proactively trigger a series of preventative response measures. In other words, when a functional device failure occurs, early warning and intervention can be used to reduce the probability of downtime, extend the life of the device, or avoid safety hazards.

[0214] In one embodiment, the electronic device may be pre-set with a neural network model for generating fault handling actions, which may be trained based on equipment operating parameters (e.g., current, temperature), environmental data, and historical fault cases to generate a model for identifying abnormal operating conditions and fault handling.

[0215] S602: Determine a reward value corresponding to the control instruction based on the actual energy consumption change, the third difference, the satisfaction level, and the fault handling action.

[0216] In one embodiment, the above-mentioned reward value may be the sum of the reward values ​​corresponding to the actual energy consumption change, the third difference, the satisfaction, and the fault handling action, or a weighted sum, which is not limited to this.

[0217] In one embodiment, the electronic device may be pre-set with reward rules corresponding to the actual energy consumption change, the third difference, the satisfaction level, and the fault handling action, so as to calculate the above-mentioned reward value based on the reward rules.

[0218] For example, for energy consumption indicators, the reward rules may include: energy consumption reduction rule, for every kilowatt-hour reduction, the reward is +10 points; energy consumption increase rule, for every kilowatt-hour increase, the reward is -5 points (or no reward, the penalty mechanism needs to be designed according to the specific scenario).

[0219] For the operation stability index, the reward rules may include: stability improvement rule, for every 10% improvement (stability = 1-the ratio between the third difference at the current moment and the third difference at the previous moment), a reward of +5 points; stability decline rule, for every 10% decline, a reward of -3 points (or no reward).

[0220] Satisfaction indicators can be quantified using a score (1-10 points) or a satisfaction percentage (0%-100%). The corresponding reward rules could be: For every 10% increase in satisfaction, reward +8 points; for every 10% decrease in satisfaction, reward -4 points (or no reward).

[0221] The corresponding reward rules for the fault handling action indicator can be: successfully avoiding a fault through preventive measures (such as parameters returning to normal or hidden dangers eliminated) will receive a reward of +20 points per time; failing to avoid a fault through preventive measures will receive a reward of 0 points. Electronic devices can determine this based on whether the operating data of the home appliance returns to a normal threshold (such as the motor current falling within the rated value) after executing the fault handling action, or based on user feedback confirming that the hidden danger has been resolved.

[0222] Among them, the reward rules corresponding to different home appliances can be the same or set differently, and there is no limitation on this.

[0223] S603: Use the first time series running data and the reward value as a set of training samples.

[0224] S604: Train an intelligent agent based on multiple sets of training samples to obtain a target intelligent agent.

[0225] In one embodiment, during the long-term operation of the home appliance, the electronic device may generate a set of training samples through the above steps S601-S603 each time after executing the above steps S101-S105.

[0226] In another embodiment, the training samples may also include control instructions and actual operation data. The actual operation data may include actual state data of the home appliance after executing the control instructions, as well as actual environmental parameters of the environment in which the appliance resides. The electronic device may obtain the actual operation data based on a sensor network, which will not be described in detail.

[0227] Based on the above description, the electronic device can store the current operating data st (state data and environmental parameters), action at (control instruction), reward value rt after executing the control instruction, and actual operating data st+1 (actual state data and actual environmental parameters) at the next moment in the experience database as an experience quadruple (st, at, rt, st+1). Based on this, it can be assumed that the electronic device can obtain a certain amount of training samples from the experience database to train the above-mentioned intelligent agent.

[0228] The above-mentioned intelligent agent is similar to the target intelligent agent, except that the target intelligent agent is a trained intelligent agent.

[0229] In one embodiment, when training an intelligent agent, the electronic device may pre-set the state space and action space corresponding to the home appliance to train the intelligent agent based on the state space and action space.

[0230] The state space includes information such as the state data (operating status) of home appliances and environmental parameters, such as the set temperature, actual temperature, and compressor speed of an air conditioner; the internal temperature and door opening status of a refrigerator; and other environmental parameters such as indoor temperature, humidity, and light intensity. The action space encompasses various control commands for home appliances, such as turning the air conditioner on and off, adjusting the temperature and fan speed, adjusting the refrigerator's cooling intensity, and selecting a washing machine's wash mode.

[0231] It should be noted that the state space can collect status data and environmental parameters of home appliances, allowing the intelligent agent to perceive the current operating status of home appliances in real time, providing a basis for subsequent decision-making (generation of control instructions). The action space clarifies the operational range of home appliances, avoiding the issuance of control instructions that the home appliances cannot execute (for example, temperature settings outside the adjustment range). The combination of the two, with the state space providing a basis for decision-making and the action space providing a means of execution, enables the intelligent agent to select appropriate control actions based on the current state in a complex and changing home environment, thereby achieving efficient and precise control of home appliances.

[0232] As an example, an electronic device can train an agent through the following steps. The details are as follows:

[0233] Extract multiple sets of training samples from the experience database;

[0234] For any training sample, the first time series operation data in the training sample is input into the to-be-trained intelligent agent, and a predicted reward value is output; the intelligent agent is used to output a predicted control instruction to be executed by the home appliance based on the features corresponding to the input first time series operation data; the predicted reward value is used to reflect the reward value corresponding to the execution of the predicted control instruction by the home appliance;

[0235] Calculate the training loss between the predicted reward value and the reward value in the training sample;

[0236] The parameters in the agent are updated based on the training loss value until the training conditions are met and the target agent is obtained.

[0237] In one embodiment, when extracting training samples, experience replay and prioritized experience replay techniques can be used. Experience replay technology involves storing the experience of the agent interacting with the environment (i.e., training samples) and randomly extracting experience for training to break data correlation and improve training stability.

[0238] And, the priority experience replay technology is to sample according to the importance of experience (the weight of the training sample), give priority to important experience for training, and significantly speed up the training speed. Exemplarily, the electronic device can set the weight of the training sample based on the absolute value of the reward value included in the training sample. For example, the larger the absolute value of the reward value, the more the training sample meets the optimization of the preset indicator, or the more it deviates from the optimization of the preset indicator. Furthermore, the training samples obtained by sampling according to the importance of experience (the weight of the training sample) can enable the intelligent agent to quickly correct errors or strengthen efficient strategies (strengthen control instructions) during the training process, thereby speeding up the training speed.

[0239] In one embodiment, the trained agent can output the predicted control instructions and the corresponding predicted reward values ​​for each training sample according to the steps S101-S105. Furthermore, the parameters of the various models (e.g., system model, objective function, kernel density estimation function, and membership function) in S101-S105 are updated using the calculated training loss values, allowing the agent to continuously learn the optimal strategy (optimal control instructions).

[0240] For example, the agent can enhance the denoising strength of the data cleaning algorithm (e.g., by adjusting a preset value), adjust the fuzzy set boundaries corresponding to the membership function, and adjust the weight distribution of various preset indicators (e.g., by increasing the weight of the energy consumption indicator). Through the above-mentioned update method, the generation of predictive control instructions can be adjusted to optimize and learn the strategy.

[0241] The electronic device may calculate the training loss value using a regularization loss function, a mean square error loss function, or the like. Furthermore, the model parameters may be updated based on a stochastic gradient descent method, an adaptive learning method, or the like, which will not be described in detail.

[0242] In one embodiment, the electronic device can ensure that the target agent can continuously adapt to environmental changes and user needs by regularly evaluating, updating, and adjusting its policies. For example, during hot summer months, the weight of the air conditioning cooling effect (i.e., the cooling index) in the reward function can be appropriately increased; during winter months, the balance between energy consumption and comfort can be improved to learn the optimal policy (optimal control instructions).

[0243] In this embodiment, the actual energy consumption change after the home appliance executes the control command, the third difference corresponding to the operational stability index, user satisfaction, and the fault handling action are obtained. A reward value is determined based on this information. The first time series operational data and the reward value are used as training samples. The target agent is trained on multiple sets of samples. This effectively compensates for the shortcomings of the original control command generation step in addressing device aging, sudden environmental changes, user demand switching, device performance degradation, temporary user needs, and sensor failures. This achieves global intelligent optimization of multiple preset indicators such as home appliance energy consumption, user experience, and lifespan.

[0244] In order to more clearly illustrate the solution in this application, the solution in this application is described below using specific examples. Figure 7 , Figure 7 A schematic diagram of an application scenario of a method for controlling a household appliance provided in another embodiment of the present application is shown.

[0245] Data cleaning phase:

[0246] Taking the air conditioner as an example, the sensor network can collect real-time second-order operational data such as indoor temperature, humidity, and air conditioner operating power, and send it to the edge node device. For example, it can record the indoor temperature (such as 26°C), humidity (50%), and the current air conditioner operating power (1000 watts) every 5 minutes.

[0247] Among them, since the sensor may have noise caused by small fluctuations, the edge node device can remove the noise signal through the above steps S301--S303 (noise removal algorithm) to obtain the first time series operation data that can truly reflect the environment and device status.

[0248] In addition, to further remove outliers from the first time-series operating data, the edge node device may also perform statistical analysis on the first time-series operating data to remove outliers. For example, the electronic device may calculate the mean and standard deviation of the first time-series operating data, and calculate a first difference between the mean and a preset multiple of the standard deviation, as well as a sum of the mean and the preset multiple of the standard deviation. Finally, the operating data that is not between the first difference and the sum is deleted. That is, outlier detection is performed to remove outliers.

[0249] For example, if the mean temperature in the first time series operation data is calculated to be 25.5°C, the standard deviation is 0.5°C, and the temperature value in the status data at the current moment is 50°C, then it can be considered that the temperature does not conform to the actual situation, and it is determined to be an abnormal value and processed.

[0250] Adaptive fuzzy algorithm stage:

[0251] The electronic device may perform a data distribution analysis on the first time series operation data to obtain a target data value for the first time series operation data. For example, after obtaining multiple distribution values ​​based on the kernel density estimation function in the above example, the electronic device may determine a preset data value corresponding to the maximum value among the multiple distribution values ​​as the target data value.

[0252] The electronic device can then update the fuzzy set boundaries and parameters in the preset membership function based on the target data values. For example, if the target data values ​​obtained after data distribution analysis of the first time series operating data are 26°C and 28°C, the fuzzy set boundary corresponding to the "comfort temperature" can be adjusted from the original 24°C-29°C to 26°C-28°C to accommodate the actual summer conditions.

[0253] Real-time control algorithm stage:

[0254] The electronic device can establish a system model of the home appliance, for example, establish a system model of the air conditioner, and obtain multiple expected operating data of the home appliance at the next moment based on the system model and the first time series operating data.

[0255] Furthermore, electronic devices can set a goal to minimize air conditioning energy consumption while ensuring user comfort (maintaining the temperature at 26°C-28°C). In other words, an objective function can be established that can measure preset indicators such as energy consumption and comfort.

[0256] Finally, through the optimization algorithm, the optimal expected operating data of the air conditioner under the current state (for example, temperature 26°C, medium wind speed) is solved, and a control instruction is generated and sent to the home appliance.

[0257] The electronic device can input the expected operating data into the updated membership function and determine the energy consumption corresponding to the expected operating data based on fuzzy logic rules. The generated energy can then be used in the optimization algorithm in the real-time control algorithm stage to solve for the optimal expected operating data.

[0258] Adaptive learning algorithm stage:

[0259] Electronic devices can define state spaces, action spaces, and reward functions. Specifically, the state space can include state data such as indoor temperature, humidity, wind speed, air outlet temperature, and cooling power; the action space can include control instructions such as adjusting temperature, wind speed, and operating mode; and the reward function can include positive rewards for energy consumption reductions and negative rewards for temperature deviations from the comfortable range. For example, a 10% energy consumption reduction is rewarded with 10 points, while a 1°C temperature deviation from the comfortable range is penalized with 5 points, among other reward rules.

[0260] The electronic device can train the intelligent agent. During the training process, the intelligent agent can interact with the environment (air conditioning and indoor environment) based on the above algorithm and optimize the strategy based on reward feedback. Specifically, the intelligent agent can be updated according to the corresponding embodiment of steps S601-S604.

[0261] Specifically, during training, the agent can search for optimal air conditioning control strategies (optimal control instructions), such as appropriately lowering the outlet air temperature during high daytime temperatures and raising it at night, to achieve a balance between energy conservation and comfort. Specifically, during optimization, the parameters of various models (e.g., system model, objective function, kernel density estimation function, and membership function) can be updated to continuously learn the optimal strategy.

[0262] In another embodiment, if Figure 1 As shown, the electronic device can be used to implement the control method of the household appliance described in the above method embodiment.

[0263] The electronic device may include one or more memories on which programs are stored. The programs can be run by the controller to generate instructions, so that the controller executes the control method of the household appliance described in the above method embodiment according to the instructions.

[0264] Optionally, data may be stored in the memory. Optionally, the controller may read data stored in the memory, which may be stored at the same storage address as the program or at a different storage address than the program.

[0265] The above are only specific embodiments of the present application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for controlling a household appliance, characterized in that: include: Obtaining first-order sequential operation data of household appliances; The first time-series operation data includes the state data of the household appliance at the current moment and the environmental parameters of the environment in which the household appliance is located; Inputting the first time-series operating data and the preset operating data of the home appliance into a system model of the home appliance to obtain a plurality of expected operating data of the home appliance at a next moment; the system model is used to describe the law of change of the operating state of the home appliance over time; The expected operation data is used to represent the operation state of the household appliance at the next moment; Updating the fuzzy set boundary corresponding to the preset membership function based on the first time series operation data to obtain a target membership function; The target membership function is used to quantify the fuzzy level corresponding to each of the expected operating data based on the updated fuzzy set boundary, so as to determine the energy consumption of the household appliance when operating with the expected operating data respectively; For any of the expected operating data, the energy consumption, the preset operating data, and the expected operating data are input into an objective function to determine a target value corresponding to the expected operating data; the objective function is used to measure the performance of the household appliance in multiple preset indicators during operation; the preset indicators include an energy consumption indicator; and the numerical value of the target value is negatively correlated with the performance of the household appliance in the preset indicators; A control instruction at the current moment is generated based on the expected operation data corresponding to the minimum target value, and the control instruction is sent to the home appliance.

2. The method according to claim 1, characterized in that The obtaining of first time-series operation data of the household appliance includes: Acquire second time-series operation data of the household appliance; the second time-series operation data includes operation data of the household appliance at the current moment, and the operation data includes the state data and the environmental parameters; For the operation data at any moment, taking the operation data as the center, determine a set of continuous operation data sets with a preset number in the second time-series operation data; The corresponding operating data are replaced by the average values ​​of the operating data sets to obtain the first time series operating data.

3. The method according to claim 1, characterized in that The obtaining of first time-series operation data of the household appliance includes: Calculating the mean and standard deviation of the first time series running data; Calculating a first difference between the average value and the standard deviation of a preset multiple, and a sum of the average value and the standard deviation of the preset multiple; Delete the operating data that is not between the first difference value and the sum value in the first time-series operating data.

4. The method according to claim 1, wherein The updating of the fuzzy set boundary corresponding to the preset membership function based on the first time series operation data to obtain the target membership function includes: Determine a distribution value of each operation data in the first time-series operation data at different preset data values; Determine a preset data value corresponding to a maximum value among the plurality of distribution values ​​as a target data value; The target membership function is obtained by updating the fuzzy set boundary based on the target data value.

5. The method according to claim 4, characterized in that The determining of the distribution value of each operation data in the first time series operation data at different preset data values ​​includes: For any of the preset data values, respectively calculating a second difference between the preset data value and each of the operating data; Each second difference value is input into a preset kernel density estimation function to obtain the distribution value.

6. The method according to claim 1, characterized in that The preset indicators further include an operation stability indicator and a performance indicator; and inputting the energy consumption, the preset operation data, and the expected operation data into an objective function to determine a target value corresponding to the expected operation data includes: Calculating a third difference between the expected operating data and the preset operating data; the third difference is used to measure the operating stability of the household appliance between the current moment and the next moment; determining a performance wear parameter of the household appliance when the household appliance operates according to the expected operating data; the performance wear parameter is used to measure the performance of the household appliance; The target value is determined based on the energy consumption, the third difference, and the performance wear parameter.

7. The method according to claim 6, characterized in that The determining the target value based on the energy consumption, the third difference, and the performance wear parameter includes: The energy consumption, the third difference, and the performance wear parameter are weighted with corresponding weights to obtain the target value; the weight corresponding to the energy consumption is greater than the weights corresponding to the third difference and the performance wear parameter.

8. The method according to any one of claims 1 to 7, characterized in that The method further comprises: After the home appliance executes the control instruction, obtaining an actual energy consumption change of the home appliance between the current moment and the next moment, a third difference corresponding to an operation stability indicator, user satisfaction feedback, and a fault handling action; the satisfaction rating is used to measure the extent to which the control instruction meets the user's expectations; determining a reward value corresponding to the control instruction based on the actual energy consumption change, the third difference, the satisfaction level, and the fault handling action; Taking the first time series operation data and the reward value as a set of training samples; An intelligent agent is trained based on multiple groups of training samples to obtain a target intelligent agent; the target intelligent agent is used to implement the method of generating control instructions according to any one of claims 1 to 7.

9. The method according to claim 8, characterized in that An edge node device is deployed in a preset area of ​​the home appliance, and the edge node device is provided with the target agent.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 9 is implemented.

Images