Smart home energy optimization management system and method

Abstract

The invention relates to the field of Internet of Things, in particular to a smart home energy optimization management system and method. The method comprises the following steps: collecting home environment data and equipment operation data in real time through smart home equipment deployed in a home environment, and constructing a dynamic environment model; based on the dynamic environment model, identifying a user behavior mode through historical data, calculating an indoor thermal load, and predicting the renewable energy supply quantity of the energy supply equipment within a specified duration in the future according to weather data; generating an equipment collaborative optimization scheme and a parameter adjustment instruction which are matched with the user behavior pattern, the indoor heat load and the renewable energy source supply quantity, and dynamically adjusting an equipment operation strategy through a depth deterministic strategy gradient (DDPG) algorithm; identifying a current scene according to the current behavior data of the user, and readjusting the equipment collaborative optimization scheme, the parameter adjustment instruction and the equipment operation strategy according to the current scene; and a control instruction for each smart home device is issued through a Zigbee / Wi-Fi protocol, and the execution effect of the device is monitored in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of Internet of Things, and in particular to an intelligent home energy optimization management system and method. BACKGROUND

[0002] With the rapid development of Internet of Things technology, the intelligent home system has gradually evolved from single device control to multi-device collaboration and intelligent decision-making. In practical applications, a large number of intelligent terminal devices with networking functions are deployed in the home environment, including air conditioners, lighting, curtains, security systems, and energy management devices, etc. Efficient collaboration is needed among these devices to improve living comfort, ensure safety, and reduce energy consumption. In related technologies, devices often operate based on preset rules or user manual instructions, lacking dynamic perception ability of the home environment state and user behavior patterns, resulting in lagging device operation strategies and inability to adjust flexibly according to actual needs. With the increasing proportion of renewable energy in the home energy structure, such as the popularization of distributed energy supply methods like photovoltaic and wind energy, how to reasonably schedule the energy flow between energy storage devices and the power grid while meeting user daily use needs has become a pressing problem. Therefore, the present application provides an intelligent home energy optimization management method and system to solve one of the above technical problems. SUMMARY

[0003] The present application aims to provide an intelligent home energy optimization management system and method that can solve at least one of the above technical problems. The specific scheme is as follows:

[0004] According to the specific embodiment of the present application, in the first aspect, the present application provides an intelligent home energy optimization management method, comprising:

[0005] By deploying various intelligent home devices in the home environment, real-time collection of home environment data and device operation data is performed, and a dynamic environment model is constructed. Based on the dynamic environment model, user behavior patterns are identified through historical data, indoor heat load is calculated, and renewable energy supply of energy supply devices for a specified future time period is predicted based on weather data. A device collaboration optimization scheme and parameter adjustment instruction matching the user behavior patterns, indoor heat load, and renewable energy supply are generated, and a deep deterministic policy gradient (DDPG) algorithm is used to dynamically adjust the device operation strategy. According to the current behavior data of the user, the current scene is identified, and the device collaboration optimization scheme, parameter adjustment instruction, and device operation strategy are adjusted again according to the current scene. Control instructions for each intelligent home device are issued through Zigbee / Wi-Fi protocol, and the device execution effect is monitored in real time.

[0006] In one embodiment, predicting the renewable energy supply of an energy supply device within a specified future time period based on weather data includes: inputting the weather data into a machine learning algorithm to enable the machine learning algorithm to determine and output the weather type and sunshine duration within the specified future time period; inputting the weather type, sunshine duration, and historical power generation data of the energy supply device as input parameters into a linear regression model to obtain the renewable energy supply of the energy supply device within the specified future time period predicted by the linear regression model.

[0007] In one embodiment, the device collaboration optimization scheme is used to define the collaboration mode between smart home devices; the parameter adjustment instruction is determined based on the current operating status of each smart home device and the collaboration mode between smart home devices defined by the device collaboration optimization scheme, and is used to define the parameter adjustment mode that each smart home device needs to perform to satisfy the device collaboration optimization scheme; the device operation strategy is used to define the optimal device operation sequence that conforms to the device collaboration optimization scheme, and the power allocation for each smart home device.

[0008] In one embodiment, the state variables of the DDPG algorithm's state space include the current energy storage capacity, the current grid electricity price, the outdoor temperature, the user behavior pattern, the current operating status of each smart home device, and the device energy consumption requirements; the action space of the DDPG algorithm includes predefined device startup priorities, predefined power allocation schemes, and energy storage scheduling strategies; the reward function of the DDPG algorithm is configured according to reward objectives, which include at least one of the following: minimizing energy costs within a specified time period, prioritizing the use of renewable energy, avoiding the current energy storage capacity reaching its maximum value, and avoiding device conflicts.

[0009] In one implementation, when the reward objective is to minimize energy costs within a specified time period, the penalty value of the reward function is calculated using the following formula: E penalty =E cost ×W energy Among them, E penalty E is the penalty value of the reward function. cost For energy usage costs, W energy Energy cost is the weighting factor; whereby energy usage cost is calculated using the following formula: E cost =P grid ×U grid +D cost ×D power Among them, E cost For energy usage costs, P grid U is the current grid electricity price. grid D represents the electrical energy purchased from the grid during the current period. costD is the energy cost coefficient of the smart home device power P is the total power of the smart home device.

[0010] In one embodiment, when the reward target is to preferentially use renewable energy, the reward value of the reward function is calculated using the following formula: R bonus = R renewable × W renewable × 100; wherein R bonus is the reward value of the reward function, R renewable is the utilization rate of renewable energy, and W renewable is the weight of renewable energy; wherein the utilization rate of renewable energy is calculated using the following formula: wherein R renewable is the utilization rate of renewable energy, E renewable is the amount of renewable energy used in the current period, and E total is the total energy consumption in the current period.

[0011] In one embodiment, when the reward target is to avoid the energy storage reaching the maximum value, the penalty value of the reward function is calculated using the following formula: SOC penalty = SOC constraint × W SOC ; wherein SOC is the energy storage, SOC penalty is the penalty value of the reward function, SOC constraint is the energy storage constraint value, and W SOC is the energy storage constraint value; wherein the energy storage constraint value satisfies the following constraint relationship:

[0012] According to the specific embodiments of the present application, in a second aspect, the present application provides a smart home energy optimization management system, comprising:

[0013] The data acquisition unit is used for collecting home environment data and device operation data in real time through the smart home devices deployed in the home environment, and constructing a dynamic environment model; the prediction unit is used for identifying user behavior patterns through historical data, calculating indoor heat load, and predicting renewable energy supply of energy supply devices in a specified future time period based on the dynamic environment model; the generation unit is used for generating a device coordination optimization scheme and parameter adjustment instruction matching the user behavior patterns, the indoor heat load, and the renewable energy supply; the processing unit is used for dynamically adjusting a device operation strategy through a deep deterministic policy gradient (DDPG) algorithm; a current scene is identified according to current behavior data of a user, and the device coordination optimization scheme, the parameter adjustment instruction, and the device operation strategy are adjusted again according to the current scene; control instructions for the smart home devices are issued through Zigbee / Wi-Fi protocols, and the execution effect of the devices is monitored in real time.

[0014] Compared with the prior art, the above scheme of the embodiment of the present application has at least the following beneficial effects: the present application provides a smart home energy optimization management method, which constructs a dynamic model by collecting home environment and device operation data, and identifies user habits in combination with historical behavior, so as to more accurately predict indoor heat load changes and future renewable energy supply conditions, and provide a basis support for subsequent optimization control. On this basis, the device operation strategy is dynamically adjusted by using a deep reinforcement learning algorithm, so that the system can continuously optimize the control logic according to the real-time state, and improve the response flexibility and energy saving efficiency. At the same time, the coordination optimization scheme, the parameter instruction, and the operation strategy are further adjusted through the identification of the current behavior of the user, so that the system has stronger scene adaptability, thereby realizing the optimal configuration of energy while meeting the comfort demand. Finally, the control instructions are issued through Zigbee / Wi-Fi protocols and the execution effect is monitored, so as to ensure the stability and controllability of the whole system, which helps to improve the overall energy efficiency and user experience of the smart home. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 A flowchart of a smart home energy optimization management method is shown;

[0016] Figure 2 A flowchart of a method for predicting renewable energy supply is shown;

[0017] Figure 3 A unit block diagram of a smart home energy optimization management system according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0018] In order to make the objects, technical solutions and advantages of the present application clearer, the following further describes the present application in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0019] The terms used in the embodiments of the present application are only for the purpose of describing particular embodiments and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. "Plural" generally includes at least two.

[0020] It should be understood that the term "and / or" used herein is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0021] It should be understood that although the terms first, second, third, etc. can be used in the embodiments of the present application to describe, these descriptions should not be limited to these terms. These terms are only used to distinguish the description. For example, without departing from the scope of the embodiments of the present application, the first can also be called the second, and similarly, the second can also be called the first.

[0022] Depending on the context, the word "if" as used herein can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "if it is determined" or "if (a stated condition or event) is detected" can be interpreted as "when it is determined" or "in response to determining" or "when (a stated condition or event) is detected" or "in response to detecting (a stated condition or event)".

[0023] It should also be noted that the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that a product or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such product or system. Without more limitations, the element defined by the sentence "including a" does not exclude the presence of other identical elements in the product or system including the element.

[0024] It should be particularly noted that the symbols and / or numbers present in the specification, if not marked in the description, are not the figure marks.

[0025] The optional embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0026] The embodiments of the present application, i.e., an embodiment of an intelligent home energy optimization management method.

[0027] The embodiments of the present application will be described in detail below with reference to the accompanying drawings. Figure 1 The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0028] Figure 1 A flowchart of an intelligent home energy optimization management method is shown, as shown in the figure, comprising the following steps. Figure 1

[0029] Step S101, through the deployment of each smart home device in the home environment, real-time collection of home environment data and device running data, and construction of a dynamic environment model.

[0030] Step S102, based on the dynamic environment model, identifying user behavior patterns through historical data, calculating indoor heat load, and predicting renewable energy supply of energy supply devices in the future specified time according to weather data.

[0031] Among them, the energy supply device refers to a device that can obtain renewable energy, such as a solar power generation device, a wind power generation device, etc.

[0032] Step S103, generating a device cooperative optimization scheme and parameter adjustment instruction matching the user behavior pattern, indoor heat load and renewable energy supply, and dynamically adjusting the device running strategy through the deep deterministic policy gradient DDPG algorithm.

[0033] Step S104, according to the current behavior data of the user, identifying the current scene, and adjusting the device cooperative optimization scheme, the parameter adjustment instruction and the device running strategy according to the current scene.

[0034] Step S105, issuing control instructions to each smart home device through Zigbee / Wi-Fi protocol, and real-time monitoring of device execution effect.

[0035] The method provided by the present application constructs a dynamic model by collecting home environment and device running data, and identifies user habits in combination with historical behavior, so as to more accurately predict indoor heat load change and future renewable energy supply, and provide a basis for subsequent optimization control. On this basis, the device running strategy is dynamically adjusted by using the deep reinforcement learning algorithm, which can continuously optimize the control logic according to the real-time state, improve the response flexibility and energy saving efficiency.

[0036] ​In the embodiments of the present application, the device coordination optimization scheme means generating a coordination working strategy between devices based on the analysis result, which is used to define the coordination mode between smart home devices to improve the overall energy efficiency. For example, if it is detected that the user leaves the living room, the TV and the light are automatically turned off, and the air conditioning supply frequency is reduced. This coordination optimization also includes how different types of devices cooperate to achieve energy saving effect, such as when the outdoor temperature is lower than the indoor temperature, the curtain is automatically opened to introduce cold air to reduce the running power of the air conditioner.

[0037] In the embodiments of the present application, the parameter adjustment instruction means fine-tuning the working parameters of a single device according to the environmental conditions and user behavior patterns, which is determined according to the current running state of each smart home device and the coordination mode between smart home devices defined by the device coordination optimization scheme, and is used to define the parameter adjustment mode required to be performed by each smart home device to meet the device coordination optimization scheme.

[0038] In an example, the parameter adjustment instruction can be used to adjust the air conditioner set value according to the outdoor temperature. For example, set to 26℃ when the outdoor temperature is 30℃, and set to 28℃ when the outdoor temperature is 25℃.

[0039] In another example, the parameter adjustment instruction can be used to reduce lighting energy consumption by dimming LED. For example, reduce the light brightness to 30% when the window is open during the day.

[0040] In the embodiments of the present application, the device running strategy is used to define the optimal device running sequence and power distribution for each smart home device in accordance with the device coordination optimization scheme. This strategy mainly targets high-energy-consumption devices, and intelligently arranges the best running time for these devices according to the current renewable energy supply situation and the state of the energy storage system. For example, during the peak period of solar power generation, high-energy-consumption devices such as washing machines and dishwashers are preferentially started.

[0041] The method provided by the present application further adjusts the coordination optimization scheme, parameter instruction and running strategy according to the current behavior of the user, so as to have stronger scene adaptability, thereby realizing the optimal configuration of energy while meeting the comfort demand. Finally, the control instruction is issued under the Zigbee / Wi-Fi protocol and the execution effect is monitored, so as to ensure the stability and controllability of the whole system, which helps to improve the overall energy efficiency and user experience of the smart home.

[0042] In some embodiments, the device coordination optimization scheme, parameter adjustment instruction and device running strategy can be adjusted again according to the current scene.

[0043] For example, for the user leaving home scenario, through real-time recognition of behavior data by the edge computing framework, unnecessary devices such as lights, televisions, and stereos are automatically turned off, and the air conditioner is switched to a low-power standby state, for example, the temperature is set to 28°C, and the fan speed is the lowest. At the same time, the energy storage system preferentially supplies power to reduce dependence on the power grid, and security monitoring is started. In the device operation strategy, high-power devices such as televisions and washing machines are preferentially turned off, and energy storage power supply is immediately started after the lights are turned off. Security devices are running throughout and real-time data is uploaded to the cloud to ensure home safety and energy saving.

[0044] For example, for the user coming home scenario, multi-device linkage is performed. For example, the entrance light is turned on (100% brightness), the living room light is adjusted to 60% brightness, the air conditioner is set to 26°C medium wind speed, the sound is played with a preset playlist (volume 50%), and the coffee machine is started and completes brewing after 15 minutes. In the device operation strategy, the light is turned on within 5 seconds after the door lock is opened, the air conditioner is started 3 seconds after the light is turned on, the sound and coffee machine are started in parallel, and the sensor permission is released immediately after the security system is disarmed, creating a convenient and warm home experience.

[0045] For example, for the user sleeping scenario, the user's behavior is detected through mattress pressure sensors and human infrared, and the bedroom environment is adjusted. For example, the light device turns off the main light and keeps the bedside light in a dimming mode, the air conditioner is set to low-frequency operation, the white noise device plays ocean sound, and the mattress sensor enables heart rate and respiratory rate monitoring. In the device operation strategy, the air conditioner is automatically started after the light is turned off, the white noise device is turned on simultaneously 10 minutes after the air conditioner is started, and the infrared detector scans for abnormal activity every 5 minutes, while unnecessary devices are turned off, ensuring comfortable sleep and health monitoring.

[0046] For example, for the user getting up at night scenario, low-brightness lighting can be provided for the user. For example, when the user gets up at night, the behavior is detected through mattress sensors and human infrared, the bedside light strip is turned on, the bathroom light is adjusted to 5% brightness, the air conditioner is maintained in silent mode, and the ventilation system is started in low-frequency circulation. In the device operation strategy, the light is turned on immediately after the human moves, the bathroom light is turned on 3 seconds after entering, the ventilation system is started 5 seconds after the light is turned on, and all devices are automatically turned off 10 minutes after leaving, ensuring safety and energy saving.

[0047] For example, for the user getting up scenario, a natural sunrise can be simulated for the user. For example, the curtains are gradually opened to 80% opening, the bedside light is gradually increased from 10% brightness to 100%, the sound plays bird chirping or piano music, the fresh air system is started to maintain 45% humidity, and the coffee machine and electric kettle are started simultaneously 1 minute after the music is played. In the device operation strategy, the curtains and lights are started simultaneously, the music is played when the light is increased to 50%, and the fresh air system and beverage devices are delayed.

[0048] For example, for a user viewing scene, the viewing environment of the user can be improved. For example, the main light in the living room is turned off during viewing, and the 5% brightness atmosphere light is kept on, the curtains are closed, the projector / TV is turned on, the sound is switched to Dolby surround mode, and the air conditioner is set to 24°C silent mode. In the device operation strategy, the curtains are started after the lights are turned off, the projector is turned on 5 seconds after the curtains are closed, the sound is switched to the mode 10 seconds after the projector is started, and the air conditioner is adjusted in temperature synchronously, so as to create an immersive home theater experience.

[0049] For example, for a guest scene, the living room environment can be adjusted. For example, the main light is adjusted to 80% brightness and set to cold white 6500K, the sound plays jazz music, the tea bar machine is started and completes brewing after 10 minutes, and the curtains are opened to 50% opening to introduce natural light. In the device operation strategy, the lights and music are started synchronously, the tea bar machine is started after 1 minute of music playing, the curtains are opened 3 seconds after the beverage device is started, and all devices are operated for no more than 30 minutes, so as to create a warm and friendly atmosphere.

[0050] It should be noted that the present application is based on determining the device cooperative optimization scheme, parameter adjustment instruction and device operation strategy that best meet the current needs of the home environment, and then adapting and optimizing the device cooperative optimization scheme, parameter adjustment instruction and device operation strategy to the current scene. The advantages are that on the one hand, when the current scene does not change (for example, during the period when the user is away), dynamic intervention can be made in combination with the actual situation of the home environment to minimize the global energy consumption, and on the other hand, when the current scene changes, a positive response can be quickly made to make the final control instruction of each smart home device meet the user's demand and achieve global optimization at all times.

[0051] In some embodiments, the renewable energy supply amount can be predicted by the following method.

[0052] Figure 2 A method flowchart for predicting a renewable energy supply amount is shown as follows. Figure 2 As shown, the method comprises the following steps.

[0053] In step S201, weather data is input into a machine learning algorithm, so that the machine learning algorithm determines and outputs the weather type and light duration in a future specified duration.

[0054] In step S202, the weather type, light duration and historical power generation data of the energy supply device are input into a linear regression model as input parameters, and the renewable energy supply amount of the energy supply device in the future specified duration predicted by the linear regression model is obtained.

[0055] In the embodiment of the present application, the weather data is analyzed by means of a machine learning algorithm to accurately determine the weather type and the light duration in the future period, and then historical power generation data is input into a linear regression model for prediction, which can effectively improve the prediction accuracy of the renewable energy output of the energy supply equipment.

[0056] The prediction mechanism proposed in the present application can prioritize the use of clean energy in advance planning, reduce dependence on the power grid, and thus reduce electricity costs and improve environmental performance. Compared with traditional experience estimation or fixed mode prediction, this method significantly improves the accuracy and stability of prediction and enhances the intelligent control ability of the system.

[0057] In some embodiments, the machine learning algorithm can use Random Forest as the core algorithm, and the input features include temperature, humidity, wind speed, cloud cover, and solar intensity in historical weather data. Through feature engineering, the original data is standardized, and combined with principal component analysis to extract key dimensions, which can effectively reduce noise interference. For example, in a renewable energy prediction scenario, the model analyzes the weather data of the past 30 days to identify the correlation threshold between cloud cover and light intensity, and uses it as the basis for outputting the weather type. At the same time, for light duration prediction, the model extracts the sunrise and sunset time and cloud cover period through time series analysis, and corrects the light attenuation coefficient based on geographic latitude and longitude, to generate a 24-hour segmented light duration prediction result.

[0058] In the construction phase of the linear regression model in the embodiment of the present application, the selection of input parameters and the allocation of feature weights directly affect the prediction accuracy.

[0059] For example, assuming that the historical power generation data of the energy supply equipment contains power output records under different weather conditions, the model needs to normalize the data to eliminate dimensional differences. For example, the weather type is encoded as a one-hot vector, the light duration is mapped as a continuous numerical value, and the historical power generation data is split into time series segments by hour granularity. In the model training process, L2 regularization is introduced to avoid overfitting, and cross-validation is used to dynamically adjust hyperparameters. For example, if the model finds that there is a strong positive correlation between light duration and photovoltaic power output, and the correlation between wind speed and wind power output is greatly affected by the season, the model can enhance its adaptability to complex scenarios by introducing interaction terms. The final output prediction result can be accurate to the hourly level of renewable energy supply, with an error range of within ±5%. In this way, high-precision energy prediction data can be obtained to provide decision support for dynamic scheduling of smart home systems.

[0060] In the present application, for the energy optimization management of smart home, it is assumed that the prediction shows that the illumination time is 6 hours a day, and the linear regression model combines the historical data to calculate that the average power generation of the photovoltaic panel in this period is 1.2 kW. According to this, the energy storage scheduling strategy can be generated, and the electricity in the first 3 hours of the high-power period is stored in the battery, and the low-power period in the last 3 hours starts to supplement the power from the power grid. This active management method based on prediction can improve the utilization rate of the energy storage system by 20%, and reduce the power purchase cost by about 15%. Compared with the traditional experience estimation, this method significantly improves the stability and controllability of the prediction through data-driven modeling.

[0061] In the above embodiment, for the actual deployment of the machine learning algorithm and the linear regression model, the synergistic mechanism between the two further enhances the robustness. For example, assuming that an extreme weather occurs in a certain area, the random forest model quickly corrects the weather type judgment by updating the weather data in real time, and feeds back the abnormal illumination time to the linear regression model. At this time, the model automatically calls the power generation data under the extreme weather as the training sample, and dynamically adjusts the regression coefficient, so as to generate a prediction result that is more consistent with the reality. This online learning mechanism can maintain the prediction accuracy in a non-steady state environment, and avoid the failure of the energy scheduling strategy due to sudden weather changes.

[0062] In the above embodiment of the present application, through multi-stage modeling and optimization, a closed loop is formed from weather data collection to energy prediction, which provides a scientific basis for energy scheduling of smart home systems. Among them, the machine learning algorithm captures the deep law of weather changes through feature extraction and pattern recognition, and the linear regression model quantifies the relationship between historical data and prediction targets by using statistical methods. The combination of the two not only improves the accuracy of the prediction, but also adapts to complex and variable environmental conditions through dynamic adjustment of parameters. This data-driven prediction mechanism provides an extensible solution for energy management in the Internet of Things environment, and has significant technological progress.

[0063] In the embodiment of the present application, the device operation strategy is dynamically adjusted by the deep deterministic policy gradient DDPG algorithm.

[0064] In some embodiments, for the state space of the DDPG algorithm, the state variables include the current energy storage capacity, the current power grid price, the outdoor temperature, the user behavior pattern, the current device operation state of each smart home device, and the device energy consumption demand.

[0065] In some embodiments, the action space of the DDPG algorithm includes the pre-defined device start priority, the pre-defined power allocation scheme, and the energy storage scheduling strategy.

[0066] In some embodiments, the reward function of the DDPG algorithm is configured according to a reward target, the reward target including at least one of minimizing energy cost in a specified time period, preferentially using renewable energy, avoiding current energy storage reaching a maximum value, and avoiding device conflicts.

[0067] As a specific embodiment, the reward target is a minimum energy cost in a specified time period, and the penalty value of the reward function is calculated using the following formula:

[0068] E penalty = E cost × W energy .

[0069] Wherein, E penalty is the penalty value of the reward function, E cost is the energy use cost, and W energy is the energy cost weight.

[0070] Wherein, the energy use cost is calculated using the following formula:

[0071] E cost = P grid × U grid + D cost × D power .

[0072] Wherein, E cost is the energy use cost, P grid is the current grid electricity price, U grid is the electricity purchased from the grid in the current period, D cost is the energy consumption cost coefficient of the smart home device, and D power is the total power of the smart home device.

[0073] In the reward function design for minimizing energy cost in the embodiment of the application, the grid electricity ratio and energy storage scheduling strategy are dynamically adjusted to achieve optimal economy. For example, assuming that the current period is in the peak of the grid electricity price, and the energy storage capacity is sufficient, the energy storage power supply will be preferentially called to reduce the grid electricity purchase in the high electricity price period. Combined with the user scenario, such as in the away-from-home scenario, the air conditioner is set to use only energy storage power supply, rather than directly taking electricity from the grid. At this time, the P grid × U grid term in the energy use cost calculation formula is significantly reduced, and the D cost × D power term has a limited increase due to the high efficiency of energy storage power supply. In this way, a lower E cost value can be obtained, so that E penaltyThe reduction is obtained in the DDPG algorithm, thereby obtaining a higher reward, and promoting continuous optimization of the scheduling strategy. For example, in a certain home scenario, if the energy storage power supply proportion is increased from 40% to 70%, the grid power purchase amount is reduced by 20 kWh, and the Ecost can be reduced by about 30 yuan / day, thereby significantly improving the economy.

[0074] As a specific embodiment, the reward target is to preferentially use renewable energy, and the reward value of the reward function is calculated by using the following formula:

[0075] R bonus =R renewable ×W renewable ×100.

[0076] Wherein, R bonus is the reward value of the reward function, R renewable is the utilization rate of renewable energy, and W renewable is the weight of renewable energy.

[0077] Wherein, the utilization rate of renewable energy is calculated by using the following formula:

[0078]

[0079] Wherein, R renewable is the utilization rate of renewable energy, E renewable is the amount of renewable energy used in the current period, and E total is the total energy consumption in the current period.

[0080] In the reward function design for preferentially using renewable energy in the embodiment of the application, the utilization rate of clean energy such as photovoltaic and wind power is maximized, carbon emissions are reduced, and dependence on fossil energy is reduced. For example, assuming that the weather is sunny on a certain day, the photovoltaic panel generates 8 kWh / hour, and the household load is 5 kWh / hour. At this time, the photovoltaic power generation is preferentially used for high-energy-consuming equipment, and the energy storage is started to store excess energy. In this way, a higher R renewable is obtained, thereby increasing R bonus . In combination with a user scenario, for example, in a movie watching scenario, unnecessary lights and air conditioners are forced to be turned off, all loads are transferred to photovoltaic power supply, and the energy storage capacity is maintained above 50% for night use. Under this strategy, R renewable may be increased from 60% to 90%, R bonus is increased by more than 30%, thereby guiding the DDPG algorithm to preferentially select a clean energy scheduling scheme.

[0081] As a specific embodiment, the reward target is to avoid the energy storage capacity reaching the maximum value, and the penalty value of the reward function is calculated by using the following formula:

[0082] SOCpenalty = SOC constraint x W SOC .

[0083] wherein SOC is the energy storage capacity, SOC penalty is the penalty value of the reward function, SOC constraint is the energy storage capacity constraint value, and W SOC is the energy storage capacity constraint value.

[0084] wherein the energy storage capacity constraint value satisfies the following constraint relationship:

[0085]

[0086] In the embodiment of the present application, in the design of the penalty function for avoiding overfilling of the energy storage capacity, the battery life attenuation and safety hazards are prevented by dynamically adjusting the energy storage charging and discharging strategy. For example, assuming that the maximum capacity of the energy storage system is 100 kWh, the current energy capacity is 95 kWh, and SOC constraint is set to 90%. At this time, SOC penalty will be significantly increased due to the proximity to the threshold value, and measures need to be taken to reduce the energy storage capacity. In combination with user scenarios, such as actively starting high-energy-consuming devices such as coffee machines and rice cookers, or selling electricity to the power grid in the homecoming scenario, to release excess energy. In this way, a more reasonable SOC distribution can be obtained, and the battery cycle life can be extended. For example, in a certain actual operation, by releasing the energy storage capacity 1 hour in advance, the SOC is reduced from 95% to 75%, avoiding the risk of battery aging caused by overcharging, and reducing the negative impact of the penalty value on the total reward.

[0087] In the present application, in the joint implementation of multiple reward functions, economic efficiency, environmental protection, and device safety can be balanced through multi-objective optimization. For example, assuming that the current period power grid price is at peak, the photovoltaic power generation capacity is 6 kWh / hour, and the energy storage capacity is 85%. At this time, the DDPG algorithm needs to meet three goals: economic efficiency, environmental protection, and safety. At this time, E cost needs to be reduced, R renewable needs to be increased, and SOC penalty needs to be reduced. For example, photovoltaic power supply is used preferentially, and the energy storage capacity is reduced from 85% to 70%, and purchase of electricity from the power grid during the peak price period is avoided. This joint strategy dynamically adjusts the priority through the weight parameter, and realizes the Pareto optimality in the total reward function. For example, if the user pays more attention to economic efficiency, the weight of W energy can be appropriately increased, and the cost of purchasing electricity from the power grid can be further reduced under the premise of ensuring the safety of the energy storage.

[0088] In the present application, through the synergistic effect of the above multi-dimensional reward function, global optimization of smart home energy management can be achieved. For example, in a complex scenario, when the user is in both away mode (economic priority) and guest mode (comfort priority) at the same time, the DDPG algorithm dynamically adjusts the weight parameters. Assuming that the current grid price is at the valley, but the energy storage capacity has exceeded 90%. At this time, the algorithm may reduce the W energy weight, increase the W SOC weight, and preferentially release the energy storage capacity to avoid overcharging, while allowing moderate use of low-cost grid energy to maintain basic lighting and air conditioning. This flexible multi-objective trade-off mechanism enables the system to adaptively optimize under different user demands and environmental constraints, ultimately achieving a balanced improvement in energy costs, environmental benefits, and device safety.

[0089] The present application also provides a system embodiment connected with the above-mentioned embodiment, for realizing the method steps of the above-mentioned embodiment, based on the same name meaning explanation and the same as the above-mentioned embodiment, having the same technical effect as the above-mentioned embodiment, which will not be repeated here.

[0090] As Figure 3 shown, the present application provides a smart home energy optimization management system 300, comprising:

[0091] A data acquisition unit 301 is configured to acquire real-time home environment data and device operation data by deploying various smart home devices in the home environment, and to construct a dynamic environment model.

[0092] A prediction unit 302 is configured to identify user behavior patterns based on the dynamic environment model, calculate indoor heat load, and predict the renewable energy supply of energy supply devices within a specified time period in the future based on historical data.

[0093] A generation unit 303 is configured to generate a device coordination optimization scheme and parameter adjustment instruction that matches the user behavior pattern, indoor heat load, and renewable energy supply.

[0094] A processing unit 304 is configured to dynamically adjust the device operation strategy through a deep deterministic policy gradient (DDPG) algorithm. According to the current behavior data of the user, the current scene is identified, and the device coordination optimization scheme, parameter adjustment instruction, and device operation strategy are adjusted again according to the current scene. Control instructions for each smart home device are issued through Zigbee / Wi-Fi protocol, and the execution effect of the device is monitored in real time.

[0095] Regarding the system in the above-mentioned embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment related to the method, and will not be described in detail here.

[0096] Although the operations in the diagrams have been described with reference to particular sequences of operations, these sequences of operations should not be construed as requiring a particular order of operations, unless stated otherwise, and except that described in the context of a particular figure. In some cases, multitasking and parallel processing can be advantageous.

[0097] The methods, systems, devices, and storage media of the present application can be implemented using standard programming techniques, with rules-based logic or other logic that can be implemented in software, hardware, or both. Also, the term "system" as used herein and in the claims can encompass both a hardware implementation and / or a device that is implemented using one or more lines of software code.

[0098] Any of the steps, operations, or procedures described herein can be performed or implemented using one or more hardware or software modules, alone or in combination with other devices. In one embodiment, a software module is implemented using a computer program product comprising a computer-readable medium containing computer program code, which can be executed by a computer processor for performing any or all of the steps, operations, or procedures described.

[0099] The foregoing description of implementations of the application has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to be limited to the precise form disclosed, and various modifications and variations are possible in light of the above teachings or can be acquired from practice of the application. One skilled in the art will recognize the equivalent of what has been described in the foregoing description although not explicitly described. Many of the details of the application can be implemented without departuring from the spirit and scope of the application.

[0100] With respect to the systems in the above-described embodiments, in which the specific manner in which the various modules perform operations has been described in detail in the embodiments relating to the methods, no detailed elaboration will be made here.

[0101] It is further understood that "connected" can include the presence of one or more intervening components, as opposed to being directly connected.

[0102] It is further understood that, unless otherwise specified, "connected" can include the presence of one or more intervening components, as opposed to being directly connected.

[0103] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the application being indicated by the following claims.

[0104] It should be understood that the application is not limited to the precise construction and compositions that have been described above and shown in the accompanying drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the application. The scope of the application is indicated only by the appended claims.

[0105] The above embodiments are only used to illustrate the technical solutions of the present application, not limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A smart home energy optimization management system, characterized in that, include: The data acquisition unit is used to collect real-time home environment data and device operation data from various smart home devices deployed in the home environment, and to build a dynamic environment model; The prediction unit is used to identify user behavior patterns based on the dynamic environment model and historical data, calculate indoor heat load, and predict the renewable energy supply of energy supply equipment within a specified period of time based on weather data. The generation unit is used to generate a collaborative optimization scheme and parameter adjustment instructions that match the user behavior pattern, the indoor heat load, and the renewable energy supply. The processing unit is used to dynamically adjust the device operation strategy through the Deep Deterministic Strategy Gradient (DDPG) algorithm; identify the current scenario based on the user's current behavior data, and readjust the device collaborative optimization scheme, the parameter adjustment instructions, and the device operation strategy according to the current scenario; Control commands to the smart home devices are issued via the Zigbee / Wi-Fi protocol, and the execution effect of the devices is monitored in real time.

2. A smart home energy optimization management method, characterized in that, Includes the following steps: By deploying various smart home devices in the home environment, real-time data on the home environment and device operation are collected, and a dynamic environment model is built. Based on the dynamic environment model, user behavior patterns are identified through historical data, indoor heat load is calculated, and the renewable energy supply of energy supply equipment within a specified time period is predicted based on weather data. Generate equipment collaborative optimization schemes and parameter adjustment instructions that match the user behavior pattern, the indoor heat load and the renewable energy supply, and dynamically adjust the equipment operation strategy through the deep deterministic strategy gradient (DDPG) algorithm. The current scenario is identified based on the user's current behavior data, and the device collaborative optimization scheme, the parameter adjustment instructions, and the device operation strategy are adjusted again according to the current scenario. Control commands to the smart home devices are issued via the Zigbee / Wi-Fi protocol, and the execution effect of the devices is monitored in real time.

3. The method according to claim 2, characterized in that, The prediction of renewable energy supply from energy supply equipment within a specified future timeframe based on weather data includes: The weather data is input into a machine learning algorithm so that the machine learning algorithm can determine and output the weather type and sunshine duration within the specified future time period; The weather type, sunshine duration, and historical power generation data of the energy supply equipment are used as input parameters and input into a linear regression model to obtain the renewable energy supply of the energy supply equipment within a specified future time period predicted by the linear regression model.

4. The method according to claim 2, characterized in that, The device collaboration optimization scheme is used to define the collaboration methods between smart home devices; The parameter adjustment instruction is determined based on the current operating status of each smart home device and the inter-device collaboration method defined by the device collaboration optimization scheme. It is used to define the parameter adjustment method that each smart home device needs to perform to meet the device collaboration optimization scheme. The device operation strategy is used to define the optimal device operation sequence that conforms to the device collaborative optimization scheme, as well as the power allocation for each smart home device.

5. The method according to claim 2 or 4, characterized in that, For the state space of the DDPG algorithm, the state variables include the current energy storage capacity, the current grid electricity price, the outdoor temperature, the user behavior pattern, the current device operating status and device energy consumption requirements of each smart home device; The action space of the DDPG algorithm includes predefined device startup priorities, predefined power allocation schemes, and energy storage scheduling strategies. The reward function of the DDPG algorithm is configured according to the reward objective, which includes at least one of the following: minimizing energy cost within a specified time period, prioritizing the use of renewable energy, avoiding the current energy storage capacity from reaching its maximum value, and avoiding device conflicts.

6. The method according to claim 5, characterized in that, The reward objective is to minimize energy costs within a specified time period. The penalty value of the reward function is calculated using the following formula: AND penalty =And cost ×W energy ; Among them, E penalty E is the penalty value of the reward function. cost For energy usage costs, W energy Weighted by energy cost; The energy cost is calculated using the following formula: E cost =P grid ×U grid +D cost ×D power ; Among them, E cost For energy usage costs, P grid U is the current grid electricity price. grid D represents the electrical energy purchased from the grid during the current period. cost D represents the energy cost coefficient for smart home devices. power This represents the total operating power of smart home devices.

7. The method according to claim 5, characterized in that, When the reward objective is to prioritize the use of renewable energy, the reward value of the reward function is calculated using the following formula: R bonus =R renewable ×W renewable ×100; Among them, R bonus R is the reward value of the reward function. renewable For the utilization rate of renewable energy, W renewable Weighting for renewable energy; The utilization rate of renewable energy is calculated using the following formula: Among them, R renewable For the utilization rate of renewable energy, E renewable E represents the amount of renewable energy used during the current period. total This represents the total energy consumption for the current period.

8. The method according to claim 5, characterized in that, The reward objective is to avoid the energy storage capacity reaching its maximum value. The penalty value of the reward function is calculated using the following formula: SOCIETY penalty =SOC constraint ×W SOC ; Where SOC is the energy storage capacity, SOC penalty SOC is the penalty value of the reward function. constraint W is the energy storage capacity constraint value. SOC This represents the energy storage capacity constraint value. The energy storage capacity constraint value satisfies the following constraint relationship:

Images