A low-carbon electricity control method and system for user-side smart meter linkage

Abstract

This invention provides a low-carbon electricity control method and system linked to a smart meter at the user end, comprising: dividing a dataset into at least one mode data and multiple power mode benchmarks corresponding to the mode data; obtaining the corresponding mode data based on status data and recording it as the current mode; calculating a real-time carbon efficiency index based on the power mode benchmarks and the current total active power data; generating decision rules based on the real-time carbon efficiency index and the current mode; the control module performing multi-device collaborative optimization according to the decision rules and sending control commands to the corresponding smart modules; updating the device status data to the dataset when the user adjusts the device, and optimizing the parameters of multi-device collaborative optimization. Using the above method, a synergistic reduction of carbon emissions throughout the house is achieved, solving the problem of energy saving by a single device but poor overall energy efficiency. While achieving energy-saving goals, it effectively ensures a good user experience. It has high versatility.

Description

Technical Field

[0001] This invention relates to low-carbon electricity control, specifically a low-carbon electricity control method and system linked to a smart meter at the user end. Background Technology

[0002] With the popularization of smart home technology, the number of smart devices in homes has increased significantly. Various appliances and furniture may be equipped with independent smart modules, supporting remote switching or energy consumption adjustment. However, most of these devices operate in an "island-like" state, lacking the ability to centrally sense and collaboratively analyze whole-house electricity consumption data. Against the backdrop of the continuous advancement of "dual carbon" goals, the demand for refined management of household energy consumption and carbon emission reduction is becoming increasingly prominent. However, existing systems struggle to integrate the scattered device status with total active power data, failing to form a unified low-carbon control strategy, and even less able to dynamically generate carbon efficiency indicators based on real-time total active power, resulting in limited overall home energy consumption optimization effects.

[0003] Another major challenge in whole-house energy management lies in balancing energy efficiency with user experience and personalized lifestyle habits. Other low-carbon electricity centralized control solutions often employ fixed strategies, such as automatically shutting down or reducing the operating power of certain devices when total power consumption is high. However, this approach easily ignores actual user needs. For example, devices actively turned on by the user may be forcibly controlled by the system, causing inconvenience. The system also struggles to effectively distinguish between normal user use and forgotten shutdowns due to prolonged device operation, lacking accurate judgment of user intent. The root cause is that the control logic fails to deeply integrate device status with user habits, leading to a conflict between energy-saving behavior and user experience.

[0004] Furthermore, current methods for assessing household carbon efficiency remain rather rudimentary. Many schemes attempt to simply determine a household's carbon index based on total electricity consumption, ignoring differences in family structure, lifestyle, and equipment types. For example, the needs of elderly and younger families differ, as do their energy consumption baselines in summer and winter. A one-size-fits-all carbon index assessment is not only inaccurate, but the resulting regulatory directives also lack rationality and acceptability. Therefore, there is an urgent need for a system capable of identifying diverse household patterns and establishing a dynamic, personalized carbon efficiency assessment and collaborative regulation mechanism based on these patterns. Summary of the Invention

[0005] To overcome existing technical problems, this invention provides a low-carbon electricity control method and system that links smart meters at the user end.

[0006] The present invention adopts the following technical solution.

[0007] A low-carbon electricity control method linked to a user-side smart meter includes the following steps:

[0008] S1. The control module acquires the total active power data of the user's electricity meter and integrates it into the dataset. It acquires the status data of multiple devices through multiple intelligent modules that can connect to the signals of the devices and the control module and transmits and integrates them into the dataset. Based on the dataset, it divides at least one mode data and multiple power mode references corresponding to the mode data.

[0009] S2. Obtain the corresponding mode data based on the status data and record it as the current mode. Calculate the real-time carbon efficiency index based on the power mode benchmark and the current total active power data. Generate decision rules based on the real-time carbon efficiency index and the current mode.

[0010] S3. The control module performs multi-device collaborative optimization according to the decision rules and sends control commands to the corresponding intelligent modules;

[0011] S4. When the user adjusts the device, the status data of the updated device is transmitted to the data center, and the parameters for multi-device collaborative optimization are optimized.

[0012] As a further improvement of the present invention, the mode data includes an outgoing mode, a home silent mode, and a home active mode, and the status data includes a power-on status and a power-off status.

[0013] The specific steps for dividing the dataset into at least one pattern data and multiple power pattern benchmarks corresponding to the pattern data include: if a device is powered on for more than a long-term threshold every day, then the device is recorded as an essential device and the remaining devices are recorded as home devices.

[0014] Obtain the power-on timestamp of each home device when it switches from a power-off state to a power-on state. Calculate the difference between the power-on timestamps of different home devices in pairs. If the difference is less than the activity threshold, the corresponding home device is recorded as an active device and integrated into the same active set. The remaining home devices are recorded as silent devices.

[0015] When all home appliances are powered off, it is recorded as the "away mode";

[0016] When all active devices in the same active cluster are powered on, or when the number of powered-on home devices exceeds the preset active power threshold, it is recorded as home active mode; otherwise, it is recorded as home silent mode when not in away mode.

[0017] Multiple power mode benchmarks corresponding to different mode data are divided based on the total active power data of the corresponding mode data.

[0018] As a further improvement of the present invention, the specific steps of dividing the total active power data corresponding to different mode data into multiple power mode references include:

[0019] The total active power data in the outgoing mode is averaged according to the fuzzy time period to obtain the first power mode benchmark for the corresponding outgoing mode and different fuzzy time periods;

[0020] Select the total active power data in home active mode and home silent mode and divide it into multiple fuzzy time periods. If the sum of the number of fuzzy time periods in home active mode and home silent mode on a given day is greater than the preset rest day threshold, then that day will be regarded as a rest day, and the remaining days will be recorded as work days.

[0021] The total active power data of the home active mode on rest days is averaged according to the fuzzy time period to obtain the second power mode benchmark corresponding to the home active mode and different fuzzy time periods. The total active power data of the home active mode on weekdays is averaged according to the fuzzy time period to obtain the third power mode benchmark corresponding to the home active mode and different fuzzy time periods.

[0022] The total active power data during rest days in home silent mode is averaged according to the fuzzy time period to obtain the fourth power mode benchmark corresponding to the home silent mode and different fuzzy time periods. The total active power data during weekdays in home silent mode is averaged according to the fuzzy time period to obtain the fifth power mode benchmark corresponding to the home silent mode and different fuzzy time periods.

[0023] As a further improvement of the present invention, the status data also includes a timeout-not-closed status;

[0024] The specific steps for obtaining the corresponding pattern data based on the status data and recording it as the current pattern include: when the status data is transmitted and integrated into the dataset, it is determined whether there is a change in the power-on or power-off status of home devices. If so, step S11 is executed.

[0025] S11. When all active devices in the same active set are powered on, or when the number of home devices powered on is greater than the preset power-on active threshold, a home active mode is obtained and the home active mode is recorded as the current mode; otherwise, step S12 is executed.

[0026] S12. When all home appliances are powered off, obtain the away mode and record the away mode as the current mode; otherwise, proceed to step S13.

[0027] S13. Obtain the latest power-on timestamp and historical power-on timestamp when the home device in the power-on state switches from the power-off state to the power-on state, and the historical power-off timestamp when switching from the power-on state to the power-off state. Determine whether the home device is in the timeout-out state by using the latest power-on timestamp, historical power-on timestamp, and historical power-off timestamp. If so, mark the home device as being in the timeout-out state. If the number of home devices in the power-on state is less than the preset power fluctuation threshold, and all home devices are in the timeout-out state, obtain the away mode and record the away mode as the current mode. Otherwise, proceed to step S14.

[0028] S14. Obtain the home silent mode and record the home silent mode as the current mode.

[0029] As a further improvement of the present invention, the specific steps for calculating the real-time carbon efficiency index based on the power mode benchmark and the current total active power data include: obtaining the benchmark power consumption by multiplying the power mode benchmark by the dynamic acquisition window duration, obtaining the current power consumption by multiplying the current total active power data by the dynamic acquisition window duration, subtracting the benchmark power consumption from the current power consumption to obtain the increased power consumption, and multiplying the increased power consumption by the dynamic carbon intensity factor to obtain the real-time carbon efficiency index.

[0030] As a further improvement of the present invention, the current modes include an out-of-home mode, a home silent mode, and a home active mode;

[0031] The specific steps for generating decision rules based on the real-time carbon efficiency index and the current mode include: if the real-time carbon efficiency index is less than or equal to 0, then generate decision rules that do not interfere with the equipment.

[0032] If the real-time carbon efficiency index is greater than 0, it is determined whether the current mode is the home active mode, the home silent mode, or the out mode. If it is the home active mode, the first strong control threshold is generated. If the real-time carbon efficiency index is greater than the first strong control threshold, the decision rules for strong control devices are generated. If the real-time carbon efficiency index is less than the first strong control threshold, the decision rules for weak control devices are generated.

[0033] If it is the home silent mode or the outgoing mode, a second strong control threshold is generated with a value less than the first strong control threshold. If the real-time carbon efficiency index is greater than the second strong control threshold, a decision rule for strong control equipment is generated. If the real-time carbon efficiency index is less than the second strong control threshold, a decision rule for weak control equipment is generated.

[0034] As a further improvement of the present invention, the decision rules include non-intervention devices, weakly controlled devices, and strongly controlled devices;

[0035] The specific steps of the control module in performing multi-device collaborative optimization according to decision rules include: calculating the carbon efficiency benefit value for each device at each adjustment step based on the adjustment step size that the intelligent module can control for different devices and the corresponding carbon efficiency change value for that adjustment step size; calculating the comfort cost value for each device at each adjustment step based on the adjustment step size of different devices and the corresponding comfort change value for that adjustment step size; multiplying the carbon efficiency benefit value by the dynamic benefit weighting coefficient to obtain the benefit value; multiplying the comfort cost value by the dynamic cost weighting coefficient to obtain the cost value; subtracting the cost value of the corresponding device from the benefit value of each device and summing the results over all devices to construct a multi-device objective function.

[0036] If the decision rule is to not intervene in the equipment, then blank adjustment instructions are obtained for each equipment;

[0037] If the decision rule is a weak control device, then determine whether the dynamic cost weight coefficient is greater than the first preference threshold. If so, then obtain the blank control instruction corresponding to the device.

[0038] If the decision rule is to emphasize the control instruction, then determine whether the dynamic cost weight coefficient is greater than the second preference threshold. If the second preference threshold is greater than the first preference threshold, then obtain the blank control instruction corresponding to the device.

[0039] For devices that do not have blank control instructions, perform iterative calculation of the multi-device objective function to maximize the control instructions for each device.

[0040] As a further improvement of the present invention, the specific steps for the device to perform the iterative calculation of maximizing the multi-device objective function are as follows: each device performs the iterative calculation of maximizing the multi-device objective function with an adjustment step size;

[0041] After sending the control command to the corresponding intelligent module, the process includes the following steps: re-execute steps S2 to S4 until the user reverses the adjustment of the non-blank control command device, or maximizes the value of the objective function to be less than or equal to 0, or all the control commands sent are blank control commands.

[0042] As a further improvement of the present invention, the specific steps of updating the status data of the device to the data center when the user adjusts the device and optimizing the parameters of multi-device collaborative optimization include: obtaining the total operation step size generated when the user adjusts the device, determining whether the total operation step size is in the same direction as the control command, if in the same direction, calculating a negative reward value, and if in the opposite direction, calculating a positive reward value.

[0043] Multiply the reward value by the preset learning rate to obtain the weight change value. Add the weight change value to the dynamic cost weight coefficient to obtain the new dynamic cost weight coefficient and overwrite the original dynamic cost weight coefficient.

[0044] This invention also proposes a low-carbon electricity control system linked to user-end smart meters, which employs a low-carbon electricity control method linked to user-end smart meters as described above, including:

[0045] Multiple intelligent modules can connect to the signals of the equipment and control module, acquire the status data of multiple devices, transmit it to the control module, and integrate it into the data set of the control module for receiving control commands and controlling the corresponding devices;

[0046] The control module is used to acquire the total active power data of user meters and integrate it into a dataset. Based on the dataset, it divides at least one mode data and multiple power mode benchmarks corresponding to the mode data. Based on the status data, it obtains the corresponding mode data and records it as the current mode. Based on the power mode benchmark and the current total active power data, it calculates the real-time carbon efficiency index. Based on the real-time carbon efficiency index and the current mode, it generates decision rules. Based on the decision rules, it performs multi-device collaborative optimization and sends control commands to the corresponding smart modules. Based on the dataset, it optimizes the parameters of multi-device collaborative optimization.

[0047] The beneficial effects of this invention are as follows:

[0048] 1. Achieving Whole-House Low-Carbon Collaborative Optimization and Energy Saving. In traditional solutions, each smart module and its corresponding device operates independently, with their control focusing only on their own energy consumption, failing to coordinate from a global perspective. This invention uses a control module to uniformly collect total active power data from the electricity meter and status data from each device, integrating them into a single dataset that covers the status data of all smart devices in the house. Based on this, the system divides at least one mode of data according to the dataset. When the control module identifies the current mode and calculates the corresponding real-time carbon efficiency index, it can generate collaborative decision rules across all devices in the house. This achieves a collaborative reduction in carbon emissions throughout the house, solving the problem of individual device energy saving resulting in suboptimal overall energy efficiency.

[0049] 2. Achieving energy-saving goals while effectively ensuring user experience. It can determine a real-time carbon efficiency index with relative values ​​for different household behaviors and usage habits, making carbon emission calculations highly versatile and providing an important foundation for the popularization of whole-house low-carbon electricity control methods. Furthermore, it can evolve with user habits, achieving long-term energy savings. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating the present invention. Detailed Implementation

[0052] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent. To better illustrate this embodiment, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions of the product.

[0053] It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings. The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0054] Reference Figure 1 This invention proposes a low-carbon electricity control method linked to a user-end smart meter, comprising the following steps:

[0055] S1. The control module acquires the total active power data of the user's electricity meter and integrates it into the dataset. It acquires the status data of multiple devices through multiple intelligent modules that can connect to the signals of the devices and the control module and transmits and integrates them into the dataset. Based on the dataset, it divides at least one mode data and multiple power mode references corresponding to the mode data.

[0056] In traditional solutions, each smart module and its corresponding device operates independently, with their control focusing only on their own energy consumption, failing to coordinate from a global perspective. This invention uses a control module to uniformly collect total active power data from the electricity meter and status data from each device, integrating them into a single dataset that covers the status data of all smart devices in the house. Based on this, the system identifies at least one mode based on the dataset. Thus, after the subsequent control module identifies the "current mode" and calculates the corresponding real-time carbon efficiency index, it can generate collaborative decision-making rules across all devices in the house. This achieves a coordinated reduction in carbon emissions throughout the house, solving the problem of individual device energy saving resulting in suboptimal overall energy efficiency. Since various smart modules can now control and read the status data of most devices, such as Mijia, Huawei, Haier Smart Home, Midea Smart Home, and Aqara, this invention will not be elaborated upon further.

[0057] As a further improvement of the present invention, the mode data includes an outgoing mode, a home silent mode, and a home active mode, and the status data includes a power-on status and a power-off status.

[0058] The specific steps for dividing the dataset into at least one pattern data and multiple power pattern benchmarks corresponding to the pattern data include: if a device is powered on for more than a long-term threshold every day, then the device is recorded as an essential device and the remaining devices are recorded as home devices.

[0059] Obtain the power-on timestamp of each home device when it switches from a power-off state to a power-on state. Calculate the difference between the power-on timestamps of different home devices in pairs. If the difference is less than the activity threshold, the corresponding home device is recorded as an active device and integrated into the same active set. The remaining home devices are recorded as silent devices.

[0060] When all home appliances are powered off, it is recorded as the "away mode";

[0061] When all active devices in the same active cluster are powered on, or when the number of powered-on home devices exceeds the preset active power threshold, it is recorded as home active mode; otherwise, it is recorded as home silent mode when not in away mode.

[0062] Multiple power mode benchmarks corresponding to different mode data are divided based on the total active power data of the corresponding mode data.

[0063] Dividing data into different modes facilitates subsequent adjustments at varying degrees. For example, in the home activity mode, the devices users turn on and the parameters of those devices are generally what users need. Therefore, the adjustment level should be appropriately reduced to avoid significantly impacting users' sense of security. This provides a favorable foundation for the popularization of low-carbon electricity control methods. Low-carbon control should be achieved with minimal impact on user experience, as it essentially reduces users' total electricity consumption, thereby lowering their electricity costs and achieving a win-win situation for both users and the environment. It's important to note that if a device consistently exceeds a long-term threshold for power consumption on a few days in the historical data, but not on other days, that device is considered a home device, not a necessary one. This avoids misleading users based on occasional behavior. Similarly, if a few days show a concentrated activity with an extra device, that extra device is considered a silent device. However, it is important to note that if home appliances are not powered off for most days, but only a few days are powered off, then the number of days should be counted as the few days. This is because users may forget to switch the appliances to the power-off state due to occasional behavior.

[0064] The long-term threshold can be set to 20 hours or 23 hours, etc., and the corresponding devices are marked as essential devices. Essential devices do not participate in subsequent multi-device collaborative optimization, which can prevent incorrect regulation of essential devices during subsequent adjustments. Essential devices generally include refrigerators, routers, aquarium pumps for fish ponds, etc. Of course, if an essential device is not used for one day, it is considered a home device. If it exceeds the long-term threshold for several consecutive days thereafter, it will be marked as an essential device again.

[0065] The activity threshold can be set to five minutes and adjusted based on the frequency of dataset updates by the intelligent module. By constructing an active set, the system can perceive and learn user intent. By capturing the difference in power-on timestamps, it can detect the sequence in which each device switches from a power-off state to a power-on state, achieving intelligent recognition of user behavior. For example, after returning home, a user might sequentially turn on the smart fingerprint lock, living room lights, air conditioner, and television, or turn on the rice cooker, induction cooker, etc. In short, when the user performs this series of operations, all active devices in the same active set will be powered on. The control module determines that there is strong human intervention in the house and switches from the away mode to the home active mode. Of course, if the number of intelligent modules is small or the user's behavior is highly random, making it impossible to learn user behavior, the threshold can be determined simply based on the number of home devices that are powered on. The active power threshold is determined based on the total number of home devices. If there are only four or five home devices, the active power threshold can be set to two or three. If there are more than twenty home devices, the active power threshold can be determined based on 50% of the total number of devices, after removing those that have not been switched on for a long time.

[0066] It's important to note that this invention doesn't use camera modules or door sensor modules to determine whether a user is out and about. Firstly, this is to avoid adding unnecessary usage costs for customers, thus laying the technological groundwork for future promotion. Secondly, users today highly value their personal privacy. If camera modules and door sensor modules could directly determine whether someone is out and about, it could easily lead to catastrophic privacy breaches. Or, if a user simply wants to go downstairs to buy something or lose something, having their absence trigger stricter controls would create strong resistance from users.

[0067] As a further improvement of the present invention, the specific steps of dividing the total active power data corresponding to different mode data into multiple power mode references include:

[0068] The total active power data in the outgoing mode is averaged according to the fuzzy time period to obtain the first power mode benchmark for the corresponding outgoing mode and different fuzzy time periods;

[0069] Select the total active power data in home active mode and home silent mode and divide it into multiple fuzzy time periods. If the sum of the number of fuzzy time periods in home active mode and home silent mode on a given day is greater than the preset rest day threshold, then that day will be regarded as a rest day, and the remaining days will be recorded as work days.

[0070] The total active power data of the home active mode on rest days is averaged according to the fuzzy time period to obtain the second power mode benchmark corresponding to the home active mode and different fuzzy time periods. The total active power data of the home active mode on weekdays is averaged according to the fuzzy time period to obtain the third power mode benchmark corresponding to the home active mode and different fuzzy time periods.

[0071] The total active power data during rest days in home silent mode is averaged according to the fuzzy time period to obtain the fourth power mode benchmark corresponding to the home silent mode and different fuzzy time periods. The total active power data during weekdays in home silent mode is averaged according to the fuzzy time period to obtain the fifth power mode benchmark corresponding to the home silent mode and different fuzzy time periods.

[0072] Fuzzy time periods, specifically taking the "outing" mode as an example, refer to the time stamp from when the user switches to "outing" mode until the next switch to any mode other than "outing." If the time exceeds 30 minutes, it can be divided into multiple 30-minute segments to obtain multiple fuzzy time periods. The last fuzzy time period can be less than 30 minutes. When calculating the average, the data is summed according to the duration with weights. Of course, if the user's lifestyle is relatively regular, and each mode's data has a long duration, then a two-hour fuzzy time period can be used. Similarly, the subsequent dynamic data collection window duration follows the same principle as determining this fuzzy time period.

[0073] Because people's behavior varies significantly across different days—for example, on a weekday, they might eat out quickly after work and only briefly check their phone upon returning home, while on a weekend, they might cook, clean, and so on—this leads to a substantial difference in the baseline energy patterns for weekdays and weekends. Therefore, roughly dividing each day into weekdays and weekends allows for a simple and quick distinction between these two distinct types of daytime behavior patterns. It's important to note that when calculating the real-time carbon efficiency index, calendar data can be used to determine whether a workday is required, such as calendar adjustments, holidays, and weekends. Users can also set special dates like single days off or alternating weeks. The current pattern can be changed when a relatively consistent behavioral pattern is detected, such as when a user turns off numerous home devices and a smart lock is opened.

[0074] The power mode benchmark update calculation can be set to be updated once a day, and the update is performed according to a certain learning rate, so that the power mode benchmark can adapt to the latest learning and avoid rapid increase or decrease of the power mode benchmark.

[0075] It should be noted that, in order to accommodate users’ random behavior, the power pattern benchmarks for different ambiguous time periods can be summed again by weight, such as the average total active power of a day.

[0076] S2. Obtain the corresponding mode data based on the status data and record it as the current mode. Calculate the real-time carbon efficiency index based on the power mode benchmark and the current total active power data. Generate decision rules based on the real-time carbon efficiency index and the current mode.

[0077] While achieving energy-saving goals, it effectively ensures a good user experience. It can determine a real-time carbon efficiency index with relative values ​​for different household behaviors and usage habits, making carbon emission calculations highly versatile and providing an important foundation for the popularization of whole-house low-carbon electricity control methods. Furthermore, it can evolve with user habits to achieve long-term energy savings.

[0078] As a further improvement of the present invention, the status data also includes a timeout-not-closed status;

[0079] The specific steps for obtaining the corresponding pattern data based on the status data and recording it as the current pattern include: when the status data is transmitted and integrated into the dataset, it is determined whether there is a change in the power-on or power-off status of home devices. If so, step S11 is executed.

[0080] S11. When all active devices in the same active set are powered on, or when the number of home devices powered on is greater than the preset power-on active threshold, a home active mode is obtained and the home active mode is recorded as the current mode; otherwise, step S12 is executed.

[0081] S12. When all home appliances are powered off, obtain the away mode and record the away mode as the current mode; otherwise, proceed to step S13.

[0082] S13. Obtain the latest power-on timestamp and historical power-on timestamp when the home device in the power-on state switches from the power-off state to the power-on state, and the historical power-off timestamp when switching from the power-on state to the power-off state. Determine whether the home device is in the timeout-out state by using the latest power-on timestamp, historical power-on timestamp, and historical power-off timestamp. If so, mark the home device as being in the timeout-out state. If the number of home devices in the power-on state is less than the preset power fluctuation threshold, and all home devices are in the timeout-out state, obtain the away mode and record the away mode as the current mode. Otherwise, proceed to step S14.

[0083] S14. Obtain the home silent mode and record the home silent mode as the current mode.

[0084] The steps for determining the current pattern differ from those for dividing the dataset into at least one pattern because the randomness of user behavior must be considered, as users may forget to turn off home devices. The determination of the power-on activity threshold has been mentioned earlier and will not be elaborated upon further in this invention. By determining whether there are devices that have exceeded the timeout period, subsequent decision rules can be easily generated to prioritize the control of those devices, and the continuous waste of resources caused by user forgetfulness can also be avoided.

[0085] The system determines whether a home appliance is in an overdue power-off state by using the latest power-on timestamp, historical power-on timestamps, and historical power-off timestamps. Specifically, it checks whether the latest power-on timestamp and the historical power-on timestamp are within a preset power-on deviation threshold, which can be three to four hours. It also checks whether the current time exceeds the historical power-off timestamp. If so, the home appliance is in an overdue power-off state.

[0086] The logic behind the scenario where the number of powered-on home devices is less than the preset power fluctuation threshold, and all devices are in an overdue "not turned off" state, is that the user turned off a small portion of the home devices, resulting in the number falling below the preset power fluctuation threshold. However, the remaining devices are still in an overdue "not turned off" state, effectively indicating that the user forgot to turn them off. It's important to note that the historical power-on timestamp and historical power-off timestamp must both be either rest days or both work days. The method for determining whether a day is a rest day or a work day can first be based on calendar attribute data, such as calendar adjustments for workdays, holidays, and weekends. Users can also set special dates such as single-day weekends or alternating weeks based on their needs. The current pattern can be changed when a relatively definite behavioral pattern is detected, such as when a large number of home devices are detected being turned off and the smart lock is detected being opened.

[0087] As a further improvement of the present invention, the specific steps for calculating the real-time carbon efficiency index based on the power mode benchmark and the current total active power data include: obtaining the benchmark power consumption by multiplying the power mode benchmark by the dynamic acquisition window duration, obtaining the current power consumption by multiplying the current total active power data by the dynamic acquisition window duration, subtracting the benchmark power consumption from the current power consumption to obtain the increased power consumption, and multiplying the increased power consumption by the dynamic carbon intensity factor to obtain the real-time carbon efficiency index.

[0088] The dynamic acquisition window duration is determined using a principle similar to that of the fuzzy time period. However, to improve the flexibility of subsequent control command generation, the dynamic acquisition window duration can be set to five minutes, ten minutes, etc. Based on the calculation of this real-time carbon efficiency index, the historical position of the current carbon emission level can be determined. The value of the dynamic carbon intensity factor can be updated periodically from the power grid.

[0089] Traditional methods assess carbon indexes solely based on absolute electricity consumption. However, directly using this approach in this invention ignores the significant differences in household structure, lifestyle, and appliance types. The same electricity consumption might be just right in house A, but far from sufficient in house B. Therefore, a one-size-fits-all approach is inappropriate. Calculating a baseline electricity consumption effectively addresses this issue. Furthermore, the power mode baseline is updated through a learning rate, allowing it to adapt to the latest user behavior and preventing rapid increases or decreases in the baseline.

[0090] As a further improvement of the present invention, the current modes include an out-of-home mode, a home silent mode, and a home active mode;

[0091] The specific steps for generating decision rules based on the real-time carbon efficiency index and the current mode include: if the real-time carbon efficiency index is less than or equal to 0, then generate decision rules that do not interfere with the equipment.

[0092] If the real-time carbon efficiency index is greater than 0, it is determined whether the current mode is the home active mode, the home silent mode, or the out mode. If it is the home active mode, the first strong control threshold is generated. If the real-time carbon efficiency index is greater than the first strong control threshold, the decision rules for strong control devices are generated. If the real-time carbon efficiency index is less than the first strong control threshold, the decision rules for weak control devices are generated.

[0093] If it is the home silent mode or the outgoing mode, a second strong control threshold is generated with a value less than the first strong control threshold. If the real-time carbon efficiency index is greater than the second strong control threshold, a decision rule for strong control equipment is generated. If the real-time carbon efficiency index is less than the second strong control threshold, a decision rule for weak control equipment is generated.

[0094] Different decision rules are generated based on the current mode and the real-time carbon efficiency index, effectively achieving a leap from "one-size-fits-all" energy saving to "scenario-adaptive" intelligent control. For example, in the active home mode where user activity is frequent, the control module adopts a relatively high first strong control threshold, such as 70. This means that only when the real-time carbon efficiency index exceeds 70, indicating significantly unreasonable energy consumption, will the control module generate decision rules for strongly controlled devices, such as repeatedly raising the air conditioner temperature, turning off the water heater that has been running for a certain period of time, or putting devices that have not been used for a certain period of time into sleep mode. If the index is between 0 and 70, only decision rules for weakly controlled devices are generated, such as appropriately raising the air conditioner temperature, turning off the water heater that has been running for a long time, or putting devices that have not been used for a long time into sleep mode. Of course, the devices that are turned off will record the historical power-on timestamp, and will be powered on or woken up again when approaching the historical power-on timestamp. This design ensures that the control module intervenes very restrainedly when the user is active at home, prioritizing comfort. Once switched to the out-of-home mode, the control module immediately adopts a stricter second strong control threshold, such as 50. At this time, even a slight energy consumption anomaly will trigger strong control, thereby achieving more thorough energy saving in unmanned scenarios.

[0095] When the real-time carbon efficiency index is less than or equal to 0, a decision rule that does not interfere with the equipment is generated, which has two effects. First, it respects and protects the user's proactive energy-saving achievements. For example, after the user manually turns off all unnecessary appliances, the control module will not perform any further operations unnecessarily. Second, it avoids unnecessary calculations and command issuance in a state that is already optimal.

[0096] Of course, to further reduce the impact of low-carbon control on users, a decision rule that does not interfere with the device can still be generated when the real-time carbon efficiency index is less than or equal to a preset dynamic control threshold. This dynamic control threshold can be 20 for the corresponding home silent mode and away mode, and 40 for the corresponding home active mode. However, to further enhance the effect of low-carbon control, it is preferentially set to 0.

[0097] S3. The control module performs multi-device collaborative optimization according to the decision rules and sends control commands to the corresponding intelligent modules;

[0098] As a further improvement of the present invention, the decision rules include non-intervention devices, weakly controlled devices, and strongly controlled devices;

[0099] The specific steps of the control module in performing multi-device collaborative optimization according to decision rules include: calculating the carbon efficiency benefit value for each device at each adjustment step based on the adjustment step size that the intelligent module can control for different devices and the corresponding carbon efficiency change value for that adjustment step size; calculating the comfort cost value for each device at each adjustment step based on the adjustment step size of different devices and the corresponding comfort change value for that adjustment step size; multiplying the carbon efficiency benefit value by the dynamic benefit weighting coefficient to obtain the benefit value; multiplying the comfort cost value by the dynamic cost weighting coefficient to obtain the cost value; subtracting the cost value of the corresponding device from the benefit value of each device and summing the results over all devices to construct a multi-device objective function.

[0100] If the decision rule is to not intervene in the equipment, then blank adjustment instructions are obtained for each equipment;

[0101] If the decision rule is a weak control device, then determine whether the dynamic cost weight coefficient is greater than the first preference threshold. If so, then obtain the blank control instruction corresponding to the device.

[0102] If the decision rule is to emphasize the control instruction, then determine whether the dynamic cost weight coefficient is greater than the second preference threshold. If the second preference threshold is greater than the first preference threshold, then obtain the blank control instruction corresponding to the device.

[0103] For devices that do not have blank control instructions, perform iterative calculation of the multi-device objective function to maximize the control instructions for each device.

[0104] The expression for the objective function for multiple devices is as follows:

[0105] ,

[0106] ,

[0107] in, It is the revenue value of the nth device. This is the cost value of the nth device. It is a dynamic benefit weighting coefficient. It is capable of controlling the adjustment step size of the nth device. It is a carbon efficiency change value that adjusts the step size. It is a dynamic cost weighting coefficient. It controls the value after adjusting a step size. and the standard values ​​set by the user The difference in comfort level.

[0108] Since user comfort must always be a primary consideration in low-carbon electricity control, when the dynamic cost weight coefficient of a device exceeds either the first or second preference threshold, that device's control is not considered. If the dynamic cost weight coefficients of all devices exceed either the first or second preference threshold, resulting in all devices receiving blank control commands, then the control round is interrupted, and the dynamic cost weight coefficients of each device are slightly reduced. This invention classifies devices according to decision rules and introduces a dynamic trade-off mechanism. This dual-protection mechanism prioritizes the method with the least impact on users when energy saving is needed, and proactively suspends unnecessary controls when users prioritize comfort (users overwrite devices, resulting in high dynamic cost weight coefficients). This achieves a dynamic balance between energy saving and comfort, avoiding the one-size-fits-all approach or single-device self-adjustment schemes common in traditional industrial energy control. Furthermore, it can flexibly adjust optimization objectives according to different scenarios, achieving globally collaborative optimization decisions.

[0109] In addition, it should be noted that the adjustment step size mentioned here... The default value is 1. However, if a device is connected to two smart modules, such as an air conditioner, one smart module can control the air conditioner with an adjustment step of 1 degree Celsius, while the other smart module can control it with an adjustment step of 0.5 degrees Celsius, then the corresponding carbon efficiency change value for the air conditioner can be calculated based on either the 0.5 degree Celsius change or the 1 degree Celsius change. If calculated based on 0.5 degrees Celsius, the smart module controlling 1 degree Celsius would have two adjustment steps, but in reality, it's only one adjustment step. Of course, in actual operation, if a device corresponds to two smart modules with different adjustment step sizes, the calculation is based on the smallest adjustment step size. However, if that smart module malfunctions, the calculation can be based on the larger adjustment step size.

[0110] For ease of understanding, as a specific example of the present invention and with simplified numerical calculations, it is assumed that the living room air conditioner user has recently maintained a temperature of 24 degrees Celsius. Therefore, the standard value is 24, the adjustment step is 1 degree Celsius, the carbon efficiency change value is 0.5, and the absolute value of the comfort change is... If the dynamic benefit weighting coefficient is 1 and the dynamic cost weighting coefficient is 0.5, the net benefit is -0.5, so the air conditioner should not be adjusted. Similarly, for adjustable brightness lamps, the dynamic cost weighting coefficient can be relatively low.

[0111] As a further improvement of the present invention, the specific steps for the device to perform the iterative calculation of maximizing the multi-device objective function are as follows: each device performs the iterative calculation of maximizing the multi-device objective function with an adjustment step size;

[0112] After sending the control command to the corresponding intelligent module, the process includes the following steps: re-execute steps S2 to S4 until the user reverses the adjustment of the non-blank control command device, or maximizes the value of the objective function to be less than or equal to 0, or all the control commands sent are blank control commands.

[0113] This solution breaks down the overall energy-saving target into a series of small, gradual adjustments by iteratively calculating and executing the adjustments in a single step. This effectively avoids the drastic impact on user experience caused by excessively large single adjustments, achieving a smooth and imperceptible energy-saving process. Users can gradually transition to a more energy-efficient operating state with almost no awareness, greatly improving the concealment of the control strategy and user acceptance, thus laying an important foundation for the widespread adoption of low-carbon electricity control.

[0114] S4. When the user adjusts the device, the status data of the updated device is transmitted to the data center, and the parameters for multi-device collaborative optimization are optimized.

[0115] As a further improvement of the present invention, the specific steps of updating the status data of the device to the data center when the user adjusts the device and optimizing the parameters of multi-device collaborative optimization include: obtaining the total operation step size generated when the user adjusts the device, determining whether the total operation step size is in the same direction as the control command, if in the same direction, calculating a negative reward value, and if in the opposite direction, calculating a positive reward value.

[0116] Multiply the reward value by the preset learning rate to obtain the weight change value. Add the weight change value to the dynamic cost weight coefficient to obtain the new dynamic cost weight coefficient and overwrite the original dynamic cost weight coefficient.

[0117] After a control command is sent, the user's subsequent actions are interpreted by the control module as feedback on their satisfaction with the control. Specifically, the control module calculates the total step size of the user's actions; for example, the user lowers the air conditioner temperature by 2°C and compares this to the direction of the original control command. If the user's action is in the same direction as the original command—for example, the smart module lowers the temperature by 1°C, but the user actually lowers it by 2°C—the control module determines that the control is still insufficient and assigns a negative reward value, such as -0.5. This negative reward, after being processed by a learning rate, such as 0.2, results in a weight change value of -0.1, slightly reducing the dynamic cost weight coefficient. This means that the control module will slightly increase its tolerance for comfort costs in subsequent decisions and may adopt more proactive energy-saving controls in the future. Conversely, if the user's operation is reversed—for example, the intelligent module lowers the temperature by 1°C, but the user raises it by 2°C—the control module determines this adjustment is excessive and exacerbates the user's discomfort. It then awards a positive reward value, such as +0.8, resulting in a weight change of +0.16. This causes the dynamic cost weight coefficient to increase rapidly, meaning the control module will place greater emphasis on comfort costs in future decisions, leading to more conservative adjustment behavior. Through this real-time feedback and parameter fine-tuning based on specific values, the system can continuously perceive and adapt to the user's actual feelings, making the global collaborative optimization strategy increasingly aligned with individual habits.

[0118] Introducing a learning rate ensures the stability and convergence of the parameter optimization process, avoiding sudden changes in system strategy due to a single drastic feedback, and guaranteeing the consistency of user experience and the predictability of system behavior. For user A, who repeatedly dims the lights in the control module and then manually brightens them, through continuous positive reward feedback, the control module eventually learns its dynamic cost weight coefficient from the default 0.5 to a higher value, such as 0.8. This means that when serving user A, the control module calculates the cost of lighting comfort very high, even exceeding the first or second preference threshold, resulting in the generation of blank control commands. Therefore, it rarely dims the lights proactively unless the carbon efficiency index is severely exceeded or the subsequent dynamic cost weight coefficient value slowly decreases. For user B, if they frequently ignore the control module's control of the humidifier, generating a negative reward, their weight coefficient may gradually decrease to 0.3. When serving user B, the control module is much less hesitant to perform operations such as turning off or dimming the humidifier. This data-driven personalized parameter optimization allows the system to deeply integrate into the lifestyles of different families.

[0119] This invention also proposes a low-carbon electricity control system linked to user-end smart meters, which employs a low-carbon electricity control method linked to user-end smart meters as described above, including:

[0120] Multiple intelligent modules can connect to the signals of the equipment and control module, acquire the status data of multiple devices, transmit it to the control module, and integrate it into the data set of the control module for receiving control commands and controlling the corresponding devices;

[0121] The control module is used to acquire the total active power data of user meters and integrate it into a dataset. Based on the dataset, it divides at least one mode data and multiple power mode benchmarks corresponding to the mode data. Based on the status data, it obtains the corresponding mode data and records it as the current mode. Based on the power mode benchmark and the current total active power data, it calculates the real-time carbon efficiency index. Based on the real-time carbon efficiency index and the current mode, it generates decision rules. Based on the decision rules, it performs multi-device collaborative optimization and sends control commands to the corresponding smart modules. Based on the dataset, it optimizes the parameters of multi-device collaborative optimization.

[0122] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A low-carbon electricity control method linked to a user-end smart meter, characterized in that, Includes the following steps: S1. The control module acquires the total active power data of the user's electricity meter and integrates it into the dataset. It acquires the status data of multiple devices through multiple intelligent modules that can connect to the signals of the devices and the control module and transmits and integrates them into the dataset. Based on the dataset, it divides at least one mode data and multiple power mode references corresponding to the mode data. S2. Obtain the corresponding mode data based on the status data and record it as the current mode. Calculate the real-time carbon efficiency index based on the power mode benchmark and the current total active power data. Generate decision rules based on the real-time carbon efficiency index and the current mode. S3. The control module performs multi-device collaborative optimization according to the decision rules and sends control commands to the corresponding intelligent modules; S4. When the user adjusts the device, the device status data is updated and transmitted to the data center, and the parameters for multi-device collaborative optimization are optimized. Current modes include Out-of-home mode, Home Silent mode, and Home Active mode; The specific steps for generating decision rules based on the real-time carbon efficiency index and the current mode include: if the real-time carbon efficiency index is less than or equal to 0, then generate decision rules that do not interfere with the equipment. If the real-time carbon efficiency index is greater than 0, it is determined whether the current mode is the home active mode, the home silent mode, or the out mode. If it is the home active mode, the first strong control threshold is generated. If the real-time carbon efficiency index is greater than the first strong control threshold, the decision rules for strong control devices are generated. If the real-time carbon efficiency index is less than the first strong control threshold, the decision rules for weak control devices are generated. If it is the home silent mode or the outgoing mode, a second strong control threshold is generated with a value less than the first strong control threshold. If the real-time carbon efficiency index is greater than the second strong control threshold, a decision rule for strong control equipment is generated. If the real-time carbon efficiency index is less than the second strong control threshold, a decision rule for weak control equipment is generated.

2. The low-carbon electricity control method for user-end smart meter linkage according to claim 1, characterized in that, The mode data includes out-of-home mode, home silent mode, and home active mode; the status data includes power-on status and power-off status. The specific steps for dividing the dataset into at least one pattern data and multiple power pattern benchmarks corresponding to the pattern data include: if a device is powered on for more than a long-term threshold every day, then the device is recorded as an essential device and the remaining devices are recorded as home devices. Obtain the power-on timestamp of each home device when it switches from a power-off state to a power-on state. Calculate the difference between the power-on timestamps of different home devices in pairs. If the difference is less than the activity threshold, the corresponding home device is recorded as an active device and integrated into the same active set. The remaining home devices are recorded as silent devices. When all home appliances are powered off, it is recorded as the "away mode"; When all active devices in the same active cluster are powered on, or when the number of powered-on home devices exceeds the preset active power threshold, it is recorded as home active mode; otherwise, it is recorded as home silent mode when not in away mode. Multiple power mode benchmarks corresponding to different mode data are divided based on the total active power data of the corresponding mode data.

3. The low-carbon electricity control method for user-end smart meter linkage according to claim 2, characterized in that, The specific steps for dividing the total active power data corresponding to different modes into multiple power mode benchmarks include: The total active power data in the outgoing mode is averaged according to the fuzzy time period to obtain the first power mode benchmark for the corresponding outgoing mode and different fuzzy time periods; Select the total active power data in home active mode and home silent mode and divide it into multiple fuzzy time periods. If the sum of the number of fuzzy time periods in home active mode and home silent mode on a given day is greater than the preset rest day threshold, then that day will be regarded as a rest day, and the remaining days will be recorded as work days. The total active power data of the home active mode on rest days is averaged according to the fuzzy time period to obtain the second power mode benchmark corresponding to the home active mode and different fuzzy time periods. The total active power data of the home active mode on weekdays is averaged according to the fuzzy time period to obtain the third power mode benchmark corresponding to the home active mode and different fuzzy time periods. The total active power data during rest days in home silent mode is averaged according to the fuzzy time period to obtain the fourth power mode benchmark corresponding to the home silent mode and different fuzzy time periods. The total active power data during weekdays in home silent mode is averaged according to the fuzzy time period to obtain the fifth power mode benchmark corresponding to the home silent mode and different fuzzy time periods.

4. A low-carbon electricity control method for user-end smart meter linkage according to claim 2, characterized in that, Status data also includes the status of not being turned off after timeout; The specific steps for obtaining the corresponding pattern data based on the status data and recording it as the current pattern include: when the status data is transmitted and integrated into the dataset, it is determined whether there is a change in the power-on or power-off status of home devices. If so, step S11 is executed. S11. When all active devices in the same active set are powered on, or when the number of home devices powered on is greater than the preset power-on active threshold, a home active mode is obtained and the home active mode is recorded as the current mode; otherwise, step S12 is executed. S12. When all home appliances are powered off, obtain the away mode and record the away mode as the current mode; otherwise, proceed to step S13. S13. Obtain the latest power-on timestamp and historical power-on timestamp when the home device in the power-on state switches from the power-off state to the power-on state, and the historical power-off timestamp when switching from the power-on state to the power-off state. Determine whether the home device is in the timeout-out state by using the latest power-on timestamp, historical power-on timestamp, and historical power-off timestamp. If so, mark the home device as being in the timeout-out state. If the number of home devices in the power-on state is less than the preset power fluctuation threshold, and all home devices are in the timeout-out state, obtain the away mode and record the away mode as the current mode. Otherwise, proceed to step S14. S14. Obtain the home silent mode and record the home silent mode as the current mode.

5. A low-carbon electricity control method for user-end smart meter linkage according to claim 1, characterized in that, The specific steps for calculating the real-time carbon efficiency index based on the power mode benchmark and the current total active power data include: multiplying the power mode benchmark by the dynamic acquisition window duration to obtain the benchmark electricity consumption; multiplying the current total active power data by the dynamic acquisition window duration to obtain the current electricity consumption; subtracting the benchmark electricity consumption from the current electricity consumption to obtain the increased electricity consumption; and multiplying the increased electricity consumption by the dynamic carbon intensity factor to obtain the real-time carbon efficiency index.

6. A low-carbon electricity control method for user-side smart meter linkage according to claim 1, characterized in that, Decision-making rules include non-intervention equipment, weakly regulated equipment, and heavily regulated equipment; The specific steps of the control module in performing multi-device collaborative optimization according to decision rules include: calculating the carbon efficiency benefit value for each device at each adjustment step based on the adjustment step size that the intelligent module can control for different devices and the corresponding carbon efficiency change value for that adjustment step size; calculating the comfort cost value for each device at each adjustment step based on the adjustment step size of different devices and the corresponding comfort change value for that adjustment step size; multiplying the carbon efficiency benefit value by the dynamic benefit weighting coefficient to obtain the benefit value; multiplying the comfort cost value by the dynamic cost weighting coefficient to obtain the cost value; subtracting the cost value of the corresponding device from the benefit value of each device and summing the results over all devices to construct a multi-device objective function. If the decision rule is to not intervene in the equipment, then blank adjustment instructions are obtained for each equipment; If the decision rule is a weak control device, then determine whether the dynamic cost weight coefficient is greater than the first preference threshold. If so, then obtain the blank control instruction corresponding to the device. If the decision rule is to emphasize the control instruction, then determine whether the dynamic cost weight coefficient is greater than the second preference threshold. If the second preference threshold is greater than the first preference threshold, then obtain the blank control instruction corresponding to the device. For devices that do not have blank control instructions, perform iterative calculation of the multi-device objective function to maximize the control instructions for each device.

7. A low-carbon electricity control method for user-end smart meter linkage according to claim 6, characterized in that, The specific steps for the device to perform the iterative calculation of maximizing the multi-device objective function are as follows: each device performs the iterative calculation of maximizing the multi-device objective function with an adjustment step size; After sending the control command to the corresponding intelligent module, the process includes the following steps: re-execute steps S2 to S4 until the user reverses the adjustment of the non-blank control command device, or maximizes the value of the objective function to be less than or equal to 0, or all the control commands sent are blank control commands.

8. A low-carbon electricity control method for user-end smart meter linkage according to claim 1, characterized in that, The specific steps for updating the device status data to the dataset when the user adjusts the device and optimizing the parameters of multi-device collaborative optimization include: obtaining the total operation step size generated when the user adjusts the device, determining whether the total operation step size is in the same direction as the control command, if in the same direction, calculating a negative reward value, and if in the opposite direction, calculating a positive reward value. Multiply the reward value by the preset learning rate to obtain the weight change value. Add the weight change value to the dynamic cost weight coefficient to obtain the new dynamic cost weight coefficient and overwrite the original dynamic cost weight coefficient.

9. A low-carbon electricity control system linked to a user-end smart meter, characterized in that, The low-carbon electricity control method using a user-end smart meter linkage as described in any one of claims 1-8 includes: Multiple intelligent modules can connect to the signals of the equipment and control module, acquire the status data of multiple devices, transmit it to the control module, and integrate it into the data set of the control module for receiving control commands and controlling the corresponding devices; The control module is used to acquire the total active power data of user meters and integrate it into a dataset. Based on the dataset, it divides at least one mode data and multiple power mode benchmarks corresponding to the mode data. Based on the status data, it obtains the corresponding mode data and records it as the current mode. Based on the power mode benchmark and the current total active power data, it calculates the real-time carbon efficiency index. Based on the real-time carbon efficiency index and the current mode, it generates decision rules. Based on the decision rules, it performs multi-device collaborative optimization and sends control commands to the corresponding smart modules. Based on the dataset, it optimizes the parameters of multi-device collaborative optimization.

Images