Household electricity utilization optimization system

Abstract

The invention provides a household electricity utilization optimization system, and the system comprises a data collection module which is used for collecting a current / voltage signal in real time; the signal processing module is used for carrying out harmonic detection on the current / voltage signal to obtain first electric energy data, fusing historical operation data, and identifying a power utilization mode and predicting a harmonic trend through an AI analysis model based on a deep learning algorithm; the inversion compensation module is used for generating an IGBT compensation strategy for counteracting power grid distortion, purifying electric energy and outputting the electric energy for household appliances to use; the electric energy acquisition module is used for acquiring second electric energy data including active probability, reactive power and electric energy consumption in real time; the carbon emission metering module is used for dynamically calling a carbon emission factor of a local power grid and calculating household carbon emission data by fusing the second electric energy data and the carbon emission factor; and the carbon emission optimization module is used for generating a carbon emission optimization strategy according to the household carbon emission data, sending the carbon emission optimization strategy to the user side, and controlling various household appliances to operate according to the carbon emission optimization strategy.

Description

Technical Field

[0001] This invention relates to the field of household electricity technology, and more specifically to a household electricity optimization system. Background Technology

[0002] Driven by the global energy transition and the "dual carbon" goal, optimizing household electricity consumption has become a crucial part of building a smart energy ecosystem. On the one hand, the electrification level of residential life continues to rise, with a surge in electricity consumption scenarios such as air conditioning, smart appliances, and electric vehicle charging. The proportion of household electricity consumption in total social energy consumption is constantly expanding. This extensive electricity consumption model not only wastes energy but also exacerbates the load pressure on the power grid during peak hours. On the other hand, distributed photovoltaic and energy storage devices are gradually entering households. How to coordinate the "source-storage-use" links to achieve surplus electricity consumption, off-peak electricity use, and reduce electricity costs has become a focus of attention for both households and the power grid. From the popularization of smart meters to the interconnection of smart homes, the technological foundation has been initially established. However, problems such as insufficient awareness of optimizing electricity consumption behavior on the user side, complex multi-device collaborative scheduling algorithms, and imperfect energy interaction mechanisms between the power grid and households still restrict the transformation of household electricity consumption from "passive consumption" to "active optimization." Optimizing household electricity consumption urgently requires technological breakthroughs and model innovation to achieve a win-win situation of improved energy efficiency, reduced user costs, and stable power grid operation. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art and proposes a household electricity optimization system.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] According to a first aspect of the present invention, the present invention provides a household electricity optimization system, the system comprising:

[0006] A data acquisition module connected to the home power grid is used to acquire current / voltage signals in real time;

[0007] The signal processing module connected to the data acquisition module is used to perform harmonic detection on the current / voltage signal to obtain first power data containing harmonic component data and current / voltage distortion parameters; and to fuse the first power data with historical operating data to identify power consumption patterns and / or predict harmonic trends through an AI analysis model based on deep learning algorithms.

[0008] The inverter compensation module connected to the signal processing module is used to generate an IGBT compensation strategy to offset grid distortion based on the power consumption mode and / or harmonic trend, and to purify the power output based on the IGBT compensation strategy to supply home appliances.

[0009] The power acquisition module connected to the data acquisition module is used to acquire second power data, including active power probability, reactive power and / or power consumption, in real time.

[0010] A carbon emission metering module that accesses the carbon intensity API and is connected to the power acquisition module is used to dynamically retrieve the carbon emission factor of the local power grid and calculate household carbon emission data by integrating the second power data and the carbon emission factor.

[0011] A carbon emission optimization module, which is connected to the carbon emission metering module and the signal processing module respectively, is used to generate a carbon emission optimization strategy based on the household carbon emission data, send the carbon emission optimization strategy to the user terminal, and / or control the operation of various home appliances based on the carbon emission optimization strategy.

[0012] The household carbon emission data includes individual device carbon emission data and / or total household carbon emission data.

[0013] The present invention, by adopting the above technical solution, has at least the following beneficial effects:

[0014] This invention proposes a household carbon emission detection and harmonic optimization scheme that integrates AI intelligent algorithms. By combining real-time and historical data for predictive analysis, it achieves dynamic adaptive harmonic compensation, combining energy efficiency optimization and fault diagnosis, and greatly reduces carbon emissions from electricity.

[0015] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0016] 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, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic diagram of a household electricity optimization system according to an embodiment of the present invention is shown;

[0018] Figure 2 A schematic diagram of a household electricity optimization system according to another embodiment of the present invention is shown;

[0019] Figure 3 A simplified schematic diagram of a multi-source data fusion algorithm provided in an embodiment of the present invention is shown.

[0020] Figure 4 A schematic diagram of the power consumption pattern recognition process provided in an embodiment of the present invention is shown;

[0021] Figure 5 A schematic diagram of the harmonic trend prediction process provided by an embodiment of the present invention is shown;

[0022] Figure 6 A schematic diagram of the predictive inverter compensation process provided in an embodiment of the present invention is shown.

[0023] Figure 7 A schematic diagram of the process for predicting household carbon emissions data according to an embodiment of the present invention is shown;

[0024] Figure 8 A schematic diagram of a household electricity optimization process provided by an embodiment of the present invention is shown. Detailed Implementation

[0025] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0026] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0027] This invention provides a home electricity optimization system, such as... Figure 1 As shown, the system may include: a data acquisition module 110, a signal processing module 120, an inverter compensation module 130, an energy acquisition module 140, a carbon emission metering module 150, and a carbon emission optimization module 160. Specifically, the data acquisition module 110 is connected to both the signal processing module 120 and the energy acquisition module 140; the signal processing module 120 is connected to both the inverter compensation module 130 and the carbon emission optimization module 160; and the carbon emission metering module 150 is connected to both the energy acquisition module 130 and the carbon emission optimization module 160.

[0028] In this embodiment of the invention, the data acquisition module 110 is connected to the household power grid and can be used to acquire current / voltage signals in real time. In practical applications, the data acquisition module 110 can be equipped with a high-precision Hall sensor to acquire electrical signals in real time, with a sampling rate ≥10kHz, to ensure the capture of high-frequency harmonic characteristics.

[0029] The signal processing module 120 can be used to perform harmonic detection on current / voltage signals to obtain first power data containing harmonic component data and current / voltage distortion parameters; it can also fuse the first power data with historical operating data to identify power consumption patterns and / or predict harmonic trends through an AI analysis model based on deep learning algorithms.

[0030] like Figure 2 As shown, another home electricity optimization system proposed in this embodiment of the invention may further include a cloud platform 200 and a user terminal 300. Specifically, the signal processing module 120 may include a harmonic detection unit 121, an interactive control unit 122, and an AI analysis unit 123 connected in sequence.

[0031] The harmonic detection unit 121 has a DSP harmonic detection function. It extracts harmonic component data (e.g., 2nd-50th harmonics) and voltage / current distortion parameters by initially analyzing the current / voltage signals acquired by the data acquisition module 110. The interactive control unit 122 serves as the interface between the traditional control logic and the AI ​​analysis unit 123, coordinating data transmission and instruction execution. Specifically, the interactive control unit 122 wirelessly communicates with both the cloud platform 200 and the user terminal 300, receiving user instructions from the user terminal 300 and historical operating data from the cloud platform 200. It also coordinates data transmission between the harmonic detection unit 121 and the AI ​​analysis unit 123. The historical operating data includes peak and off-peak periods for household electricity consumption and seasonal load variations. The AI ​​analysis unit 123, as the core functional unit of this embodiment, integrates real-time acquired first-stage power data and historical operating data from the cloud platform 200, using deep learning algorithms (such as LSTM and Transformer) to identify electricity consumption patterns and predict harmonic trends.

[0032] In practical applications, the AI ​​analysis unit 123 employs a multi-source data fusion algorithm to perform feature fusion processing on the first power energy data, historical operation data, and / or household carbon emission data, obtaining key feature factors with corresponding weights. It then constructs an AI analysis model that integrates a Bi-LSTM sub-model, a Transformer sub-model, and / or a GCN sub-model. The Bi-LSTM sub-model can be used to identify electricity consumption patterns using key feature factors, obtaining pattern recognition results, including harmonic characteristics of various household appliances connected under periodic patterns, such as current surges during air conditioner startup and harmonic changes during the connection of new energy devices. The Transformer sub-model can be used to process high-frequency data in parallel, predict harmonic trends, and obtain harmonic spectrum results for a preset future time period. The GCN sub-model can be used to construct the topological relationships of various household appliance nodes and perform carbon emission-harmonic correlation analysis based on key feature factors to predict the carbon emissions of various household appliances.

[0033] The specific implementation method of the AI ​​analysis unit 123 to achieve the above functions is described below:

[0034] like Figure 3 The diagram shows a simplified schematic of the multi-source data fusion algorithm. In this embodiment, the multi-source data fusion algorithm employs a sliding window mechanism to perform spatiotemporal alignment processing on the first electrical energy data, historical operating data, and / or household carbon emission data, and performs cross-domain feature extraction to obtain key feature factors including harmonic spectrum, voltage fluctuations, historical load curves, and / or environmental temperature and humidity. In practical applications, a sliding window mechanism is used to unify timestamps, setting a 10ms window for high-frequency data (such as harmonics / voltage) and a 1s window for low-frequency data (such as equipment status). Examples of cross-domain feature extraction methods are shown in Table 1 below:

[0035] Table 1 (Cross-domain feature extraction methods)

[0036] Data types Feature extraction methods Output Dimension Harmonic spectrum FFT + wavelet packet decomposition (extracting 2nd-25th harmonic energy) 24D voltage fluctuation Standard deviation + kurtosis calculation 2D Historical load curve Piecewise Aggregation Approximation (PAA) Dimensionality Reduction 32-dimensional Ambient temperature and humidity First-order difference + moving average 4D

[0037] Furthermore, attention-based feature importance assessments can be performed on each key feature factor to dynamically allocate weights. In practical applications, real-time adjustment factors can also be used for adjustment; for example, when grid stability is poor, the harmonic feature weight can be increased by 40%.

[0038] It should be noted that the large AI models, such as the Bi-LSTM sub-model, Transformer sub-model, and / or GCN sub-model in AI analysis unit 123, can be partially deployed on the local control processor to reduce cloud dependence and ensure real-time response (response latency <50ms). Table 2 below shows the basis for model architecture selection:

[0039] Table 2 (Criteria for Model Architecture Selection)

[0040]

[0041]

[0042] like Figure 4 The diagram shown is a flowchart of the power consumption pattern recognition process.

[0043] The loss function for the Bi-LSTM sub-model is:

[0044]

[0045] in, represents the loss function, characterizing the degree of error in predicting electricity consumption habits. It combines prediction probability and regularization terms to balance accuracy and model complexity; T represents the time window length, usually an integer (e.g., 72), which can be understood as the number of consecutive observation points. One point corresponds to 15 minutes, and 72 points represent an 18-hour time window; t represents the time index, t∈[1,T], which is a specific moment in the time series data, used to locate the observation value at each time point; y t Indicates the actual load condition, y t ∈{0,1}, where 0 represents off-peak hours and 1 represents peak hours, reflecting the status label of actual electricity load; Indicates the predicted probability. That is, the "probability of peak period occurrence" output by the Bi-LSTM sub-model is used to measure the confidence of the model's prediction for each time point; θ represents the model parameter vector, ||θ|| 2 λ represents the squared L2 norm, which is the sum of the squared magnitudes of the parameter vectors. This reduces model complexity and avoids overfitting by constraining parameter sizes; λ represents the regularization coefficient, typically 10. -3 ~10 -5 Hyperparameters are used to control model complexity and prevent overfitting by penalizing excessively large model parameters.

[0046] like Figure 5 The diagram shown is a flowchart of harmonic trend prediction.

[0047] The Transformer sub-model can be used to input current waveform sequences, such as a 200ms current waveform (sampling rate 10kHz → 2000 points), and then perform position encoding on the current waveform sequence to predict the harmonic spectrum results within a preset time period, such as the harmonic spectrum in the future 50ms (THD prediction error <3%). The mathematical expression for position encoding is:

[0048]

[0049] Among them, PE (pos,2i)Indicates the encoding position; i is the dimension table; pos represents the sequence position; d model Indicates the encoding dimension.

[0050] The GCN sub-model in AI analysis unit 123 is essentially a dynamic carbon emission-harmonic correlation model, and its physical constraint loss function is:

[0051]

[0052] in, The physical constraint loss is used to comprehensively measure the loss value between carbon emission prediction error and harmonic safety, ensuring that the model output conforms to physical constraints; α represents the carbon emission loss weight, generally 0.5 to 1.0, used to adjust the importance of carbon emission prediction accuracy. A larger weight indicates that the model focuses more on fitting the actual carbon emissions; β represents the harmonic loss weight, generally 0.2 to 0.5, used to adjust the importance of harmonic safety constraints. A larger weight indicates that the model strictly adheres to the grid harmonic limit requirements; MSE represents the mean square error function, used to calculate the squared mean of the difference between the actual and predicted values, quantifying the accuracy of carbon emission prediction. The calculation formula is... E carbon This represents actual carbon emission data, expressed in g / kWh. It indicates the measured CO2 emissions per unit of electricity, reflecting the carbon footprint in actual electricity consumption. This represents the predicted carbon emission data, i.e., the estimated CO2 emission per unit of electricity output by the model, which should be as close as possible to the actual value; ReLU represents the linear rectified function, with the activation function max(0,x). It outputs directly when the input is positive and 0 when it is negative, highlighting the loss due to harmonic exceedances; THD pred The predicted harmonic distortion rate (THD) represents the total harmonic distortion rate of the power system output by the model, reflecting the degree of distortion of the current or voltage waveform. This embodiment of the invention sets a 5% safety threshold, which is the maximum allowable THD limit of the power grid. When the predicted value exceeds this threshold, the ReLU function introduces additional losses to constrain the model.

[0053] Furthermore, the inverter compensation module 130 in this embodiment of the invention can be used to generate an IGBT compensation strategy to offset grid distortion based on the power consumption pattern and / or harmonic trend, and to purify the output power for use by household appliances based on the IGBT compensation strategy. Specifically, the inverter compensation module 130 includes a strategy generation unit 131 and an IGBT optimization unit 132 connected to each other. The strategy generation unit 131 can be used to calculate an anti-phase harmonic compensation strategy, including compensation time, phase offset and / or compensation current, based on the pattern recognition result, the current grid load and the current IGBT parameters, using a reinforcement learning algorithm; the IGBT optimization unit 132 can be used to determine the IGBT compensation strategy based on the anti-phase harmonic compensation strategy according to parameter mapping rules; the IGBT compensation strategy includes switching frequency and conduction time.

[0054] like Figure 6 The diagram shows a flowchart of the predictive inverter compensation proposed in this embodiment of the invention. Taking the compensation for sudden changes in air conditioner starting current as an example, the predictive inverter compensation has a three-level architecture of "power consumption pattern recognition - compensation strategy generation - IGBT parameter optimization". The power consumption pattern recognition can utilize the pattern recognition results obtained from the AI ​​analysis unit 123 mentioned above. Specifically, LSTM is used to capture time-series dependencies (e.g., the daily pattern of air conditioner starting at 7 AM), and CNN is used to extract current waveform features (e.g., the spike pulse during startup). In this application scenario, historical current data (15 minutes / point), ambient temperature, and time features are input, and the probability of air conditioner starting within the next 30 minutes (e.g., ...) is output. This indicates a high probability of startup and predicts the peak current at startup (e.g., a predicted sudden change to 50A). The core logic of the compensation strategy generation algorithm proposed in this embodiment is to generate a compensation strategy based on the pattern recognition result using a reinforcement learning (RL) algorithm or a rule engine. Example of an RL reinforcement learning algorithm:

[0055] State: Current grid load, predicted current surge value, and current IGBT parameters;

[0056] Action: The magnitude, phase, and duration of the compensation current;

[0057] Reward: The magnitude of current fluctuation after compensation (the smaller the fluctuation, the higher the reward).

[0058] When it is predicted that the air conditioner will start at t=30 (with a sudden change in current to 50A), RL reinforcement learning can generate a compensation policy 5 minutes in advance and inject a -30A compensation current at the compensation time t=25 to offset the startup shock. In practical applications, the mathematical expression for generating the policy vector is:

[0059] S=[ΔI,Δφ,t start ]

[0060] Where S represents the strategy vector; ΔI represents the compensation current. like Δφ represents the phase shift; t start Indicates the moment of compensation.

[0061] Additionally, it should be noted that this embodiment of the invention can also extend harmonic compensation to new energy equipment. During the harmonic feature extraction process, the current signal is decomposed into the fundamental wave and various harmonics using Fourier transform, and CNN identifies the harmonic distortion rate (e.g., THD = 8%). Based on this, the compensation strategy can be: when the predicted THD exceeds the limit (e.g., a 5% threshold), a strategy for injecting anti-phase harmonic current is generated (e.g., injecting the anti-phase component of the 3rd harmonic). After constraint compensation, THD ≤ 5%.

[0062] In this embodiment of the invention, the IGBT optimization unit 132 maps the anti-phase harmonic compensation strategy output by the strategy generation unit 131 into an IGBT compensation strategy according to the parameter mapping rule. Taking inverter compensation during air conditioner startup as an example, the compensation strategy S is mapped into IGBT control parameters through the parameter mapping rule F, as follows:

[0063] S=[ΔI=-30A,Δφ=180°,t start =t=25]

[0064] Wherein, the switching frequency = f1(ΔI, rate of change of current), and the larger ΔI is, the higher the frequency is to ensure the response speed; the conduction time = t start +Predicted startup time: For example, if startup lasts 10 seconds, then the conduction time = 10 seconds.

[0065] The objective function of the parameter mapping rule is:

[0066]

[0067] Among them, I 理想 The target current without sudden changes is represented by γ, and the loss weight is represented by γ. The constraints include: switching frequency ≤ IGBT maximum allowable frequency, such as 30kHz; driving voltage is within a safe range, such as 10V to 18V.

[0068] In practical applications, gradient descent is used to adjust IGBT parameters to minimize the error between the compensated current waveform and the target waveform, while reducing switching losses, such as by adjusting the dead time to reduce device heating.

[0069] The predictive inverter compensation proposed in this invention offers significant advantages over traditional solutions. Traditional solutions initiate compensation only after detecting a current surge, resulting in a delay of approximately 50ms and low compensation efficiency. In contrast, the predictive inverter compensation proposed in this invention generates the strategy and optimizes IGBT parameters 50ms in advance, improving efficiency through the following synergistic effects: adjusting the switching frequency to a higher frequency range (e.g., from 10kHz to 20kHz) in advance, shortening the rise time of the compensation current (from 20μs to 10μs); and dynamically adjusting the drive voltage based on the predicted surge amplitude (e.g., setting the drive voltage to 17V for a 50A surge and 14V for a 30A surge), avoiding over-compensation. This invention trains and identifies the harmonic characteristics of different types of home appliances, dynamically adjusting compensation parameters to avoid energy loss caused by a "one-size-fits-all" compensation approach, achieving a compensation efficiency of over 98%.

[0070] Furthermore, the power acquisition module 140 can be used to acquire second power data, including active power probability, reactive power, and / or power consumption, in real time. In practical applications, a high-precision metering chip (such as RN8209) can be added to the existing sensor to acquire the second power data in real time, so that the accuracy is ≤0.5%.

[0071] Furthermore, the carbon emission metering module 150 can be used to dynamically retrieve the carbon emission factor from the local power grid and integrate the second power data with the carbon emission factor to calculate household carbon emission data. In practical applications, the carbon emission factor of the local power grid (such as the CO2 emissions per kWh of electricity) can be dynamically retrieved through the carbon intensity API, and the factor is updated according to time periods (such as when the proportion of photovoltaic power generation is high during the day, the carbon factor decreases). When calculating household carbon emission data, real-time power and carbon emission factor can be integrated, and the household electricity carbon footprint can be calculated according to the formula carbon emission amount = power × time × carbon emission factor, and it supports device-specific metering (such as separate metering for air conditioners and water heaters).

[0072] In an optional embodiment, the carbon emission metering module 150 can also identify various types of household appliances through current characteristics and establish a device power consumption mode-carbon emission correlation model to predict household carbon emission data. Specifically, based on the first electrical energy data, multimodal features in the current waveform can be extracted to construct feature vectors characterizing the operating status and / or environmental characteristics of various types of household appliances; based on the feature vectors, the operating status of various types of household appliances can be identified using HMM or LSTM algorithms; based on the operating status of various types of household appliances, a device power consumption mode-carbon emission correlation model can be constructed, with the following mathematical expression:

[0073]

[0074] Among them, E carbon This indicates carbon emission data for equipment or household use; P i(t) represents the real-time power of device i; α(t) represents the carbon emission factor; P i This represents the equipment type correction factor; n is 1 or the total number of equipment. Furthermore, based on the equipment electricity consumption pattern-carbon emission correlation model, household carbon emission data during peak and off-peak electricity consumption periods are calculated.

[0075] like Figure 7 The diagram shows a flowchart for predicting household carbon emissions data. The device power consumption pattern-carbon emission correlation model in this embodiment is created based on a three-layer architecture of "device characteristics - operating status - carbon emissions". First, for device characteristics, according to a multi-modal feature fusion algorithm, time-domain features (mean, variance, kurtosis), frequency-domain features (harmonic content), and time-series features (start-stop time patterns) are extracted from the current waveform to construct a feature vector. The mathematical expression is:

[0076] x t =[I mean ,I var ,THD,f1,f2,...,f n [time_of_day]

[0077] Where, x t The feature vector at time t is used to comprehensively characterize the operating state and environmental characteristics of electrical equipment at a certain moment, and is used for tasks such as equipment identification and load forecasting; I mean This represents the average current over a period of time, reflecting the basic electrical load of the equipment (such as the steady-state current of an air conditioner during operation); I var The variance of the current represents a quantitative indicator of the degree of current fluctuation. A larger variance indicates poorer current stability (e.g., sudden current changes during motor startup). THD represents Total Harmonic Distortion, a metric characterizing the harmonic content in current / voltage. When nonlinear devices are connected, a higher THD indicates poorer power quality. f1, f2, ..., f n It represents the frequency components of each harmonic (such as the fundamental wave, the third harmonic, the fifth harmonic, etc.), characterizes the amplitude of harmonic components at different frequencies, and is used to analyze the nonlinear characteristics of equipment (such as the generation of specific harmonic frequencies by frequency converters); time_of_day is a time feature, such as the hour or date type in a day, which is used to capture the periodicity of electricity consumption patterns (such as the morning and evening peak patterns of residential electricity consumption).

[0078] Secondly, for identifying the operating status of equipment, a Hidden Markov Model (HMM) or a Long Short-Term Memory (LSTM) network can be used to identify the operating status of the equipment, represented as:

[0079] P(S t |x 1:t ) = LSTM(x1:t )

[0080] Wherein, P(S) t |x 1:t ) represents the situation at time t based on the historical feature sequence x. 1:t Predicting the device is in state S t The conditional probability, P(S) t |x 1:t )∈[0,1], representing the probability that the equipment is currently in a specific operating state given historical electricity consumption characteristics (such as current, harmonics, etc.), the closer the value is to 1, the higher the prediction confidence; S t Discrete variables representing the operating state of equipment at time t, such as {standby, off, start-up, stable operation, full load operation, fault, overload}, characterize the current operating mode of the equipment, which can be defined through expert annotation or data clustering (e.g., "cooling mode" and "heating mode" of an air conditioner); x 1:t The first part represents the feature vector sequence from time 1 to time t, which contains time series data containing multi-dimensional electricity consumption characteristics (such as mean current, harmonic components, time characteristics, etc.) and is used to capture the dynamic changes in equipment operation; LSTM(·) represents the Long Short-Term Memory network model, which has a recurrent neural network structure and processes long-term dependencies in time series data through the mechanism of "forget gate", "input gate" and "output gate". Here it is used to learn the mapping relationship between feature sequences and equipment states.

[0081] Furthermore, the mathematical expression for constructing the correlation model between equipment power consumption patterns and carbon emissions is as follows:

[0082]

[0083] Among them, E carbon This indicates carbon emission data for equipment or household use; P i (t) represents the real-time power of device i, which can be calculated using the current-power conversion formula (e.g., P = I). 2 R); α(t) represents the carbon emission factor, which changes dynamically over time (e.g., the α value is large when the proportion of thermal power is high during the day); β i This represents the equipment type correction factor (e.g., 1.2 for air conditioners, 0.8 for LED lights), reflecting the differences in energy conversion efficiency between equipment; n is 1 or the total number of equipment. This formula allows calculation of carbon emission data for individual equipment or total household carbon emission data.

[0084] Taking air conditioners as an example, carbon emissions are related to the compressor's start-stop frequency and the indoor-outdoor temperature difference ΔT:

[0085]

[0086] Where P0 is the basic power consumption; k is the start-stop loss coefficient; and γ is the temperature sensitivity coefficient.

[0087] Therefore, based on the device electricity consumption pattern-carbon emission correlation model, household carbon emission data during peak and off-peak electricity consumption periods can be calculated. Specifically, by combining device electricity consumption patterns and carbon emission factors, carbon emissions for the next few steps can be predicted using an attention mechanism for weighted estimation.

[0088]

[0089] in, M(x) represents a sequence vector of predicted carbon emissions over the next T time steps, used to guide low-carbon scheduling decisions (such as choosing low-emission periods for charging); Attention represents the attention mechanism function, a weighted aggregation function that selectively focuses on key information (such as high-emission periods) by calculating the importance weights of each element in the input sequence; t-T:t ) represents historical features x t-T:t The extracted electricity consumption pattern vector can be calculated using the Bi-LSTM sub-model described above. It can be understood as an abstract representation of historical electricity consumption characteristics, capturing equipment operating patterns (such as air conditioner start-stop patterns); x t-T:t This represents a historical feature vector sequence matrix from time t- to time t, containing historical observation data of electrical features such as current, voltage, and harmonics, as well as time codes, used to infer future electricity consumption behavior; α t-T:t This represents the historical carbon emission factor sequence vector from time t- to time t, reflecting the carbon intensity of electricity sources at different times (e.g., the α value is low when the proportion of wind power at night is high).

[0090] The carbon emission optimization module 160 can be used to generate carbon emission optimization strategies based on household carbon emission data, send these strategies to the user terminal 300, and / or control the operation of various home appliances according to the carbon emission optimization strategies. The carbon emission optimization strategy aims to minimize carbon emissions while meeting user needs; its mathematical expression is as follows:

[0091]

[0092] The constraints of carbon emission optimization strategies can include comfort constraints and equipment lifespan constraints, where u is a control variable (such as air conditioning temperature setting and charging time selection).

[0093] Specifically, the carbon emission optimization module 160 may include a carbon emission optimization unit 161 connected to the carbon emission metering module 150 and a load control unit 162 connected to the carbon emission metering module 150 and the harmonic detection unit 121, respectively. The carbon emission metering module 150 can generate carbon emission optimization strategies based on the differences in household carbon emission data between peak and off-peak electricity consumption periods; for example, replacing high-harmonic appliances, charging during off-peak hours, delaying the start-up time of non-emergency equipment to off-peak hours, and prioritizing the use of clean energy by linking photovoltaic charging devices. The load control unit 162 can be used to monitor the first electrical energy data and equipment carbon emission data in real time. When it detects that the start-up of a device causes a simultaneous surge in harmonic component data and equipment carbon emission data, it generates a load control command for synchronously adjusting IGBT compensation and device switching, which is sent to the inverter compensation module 130 for execution, achieving dual optimization of "harmonic suppression + carbon emission reduction".

[0094] Furthermore, the household electricity optimization system proposed in this embodiment of the invention also includes a quality detection module 170 connected to the inverter compensation module 130 and the carbon emission optimization module 160, respectively. This module can be used to monitor the third energy data after inverter compensation and / or carbon emission optimization in real time, and feed the third energy data back to the harmonic detection unit 121 for closed-loop control; it can also feed the data back to the AI ​​analysis unit 123 for iterative training, dynamically optimizing the scheduling strategy to form a "monitoring-analysis-control" closed loop. The third energy data includes THD harmonic distortion rate, equipment power factor, and equipment carbon emission data.

[0095] Furthermore, the signal processing module 120 in this embodiment of the invention may further include: an anomaly detection unit 124 connected to the harmonic detection unit 121, which can be used to identify power grid fault conditions based on the first power data, generate early warning information, and feed the early warning information back to the cloud platform 200 and / or the user terminal 300 to assist users or maintenance personnel in quickly locating problems. Power grid fault conditions include abnormal equipment harmonics, three-phase imbalance, and voltage drops.

[0096] Furthermore, the user terminal 300 in this embodiment of the invention may be equipped with an interactive page for real-time display of third-party power data and carbon emission visualization data. The carbon emission visualization data includes household carbon emission data, device carbon emission data and their percentages (e.g., air conditioners account for 35%, water heaters for 25%), a comparison curve between regional carbon factor trends and household carbon emission, and carbon emission optimization strategies (e.g., "Today's carbon factor is low from 14:00 to 16:00; it is recommended to use an electric water heater during this period"). Thus, users can remotely monitor power quality, receive AI warnings (e.g., abnormal device harmonics), and issue control commands (e.g., manually adjusting the compensation mode) through the user terminal 300. Additionally, it can support user-defined "low-carbon mode," where the AI ​​analysis unit 123 automatically shuts down unnecessary high-carbon devices based on carbon emission data, or coordinates with the photovoltaic charging system to prioritize the use of clean energy.

[0097] In practical applications, the model parameters in the AI ​​analysis unit 123 can also be updated regularly through the cloud platform 200 to adapt to changes in new home appliances or power grid environments, forming a closed loop of "local execution + cloud optimization".

[0098] This invention proposes a household electricity optimization system, including a data acquisition module connected to the household power grid for real-time acquisition of current / voltage signals; a signal processing module connected to the data acquisition module for harmonic detection of the current / voltage signals to obtain first power data containing harmonic component data and current / voltage distortion parameters; fusing the first power data with historical operating data to identify power consumption patterns and / or predict harmonic trends using an AI analysis model based on deep learning algorithms; and an inverter compensation module connected to the signal processing module for generating an IGBT compensation strategy to offset grid distortion based on the power consumption pattern and / or harmonic trends. The T-compensation strategy purifies the output of electrical energy for use by home appliances; the power acquisition module, connected to the data acquisition module, is used to collect second power data in real time, including active power probability, reactive power, and / or power consumption; the carbon emission metering module, connected to the carbon intensity API and the power acquisition module, is used to dynamically retrieve the carbon emission factor of the local power grid, and calculate the household carbon emission data by integrating the second power data and the carbon emission factor; the carbon emission optimization module, connected to the carbon emission metering module and the signal processing module respectively, is used to generate a carbon emission optimization strategy based on the household carbon emission data, send the carbon emission optimization strategy to the user terminal, and / or control the operation of various home appliances according to the carbon emission optimization strategy. Through this invention, the system is endowed with "learning-prediction-optimization" capabilities based on a large-scale AI model of multiple types, which is especially suitable for scenarios with increasing nonlinear loads (such as variable frequency air conditioners and smart devices) in smart homes, improving power quality while achieving more refined energy-saving / low-carbon management.

[0099] To enhance the understanding of the household electricity optimization system proposed in the embodiments of the present invention by those skilled in the art, it can be combined with, for example... Figure 8The flowchart shown outlines the connection and implementation logic of each module / unit for household electricity optimization. Below, a specific application example of the household electricity optimization system provided in this embodiment of the invention is presented:

[0100] 1. Home appliance operation optimization and carbon footprint tracking:

[0101] 1.1 Application Example: Real-time monitoring of energy consumption patterns of major home appliances such as refrigerators, washing machines, dryers, ovens, and dishwashers. AI algorithms analyze usage habits (such as washing frequency and peak-hour usage), equipment efficiency status, and real-time electricity prices / grid carbon intensity.

[0102] 1.2 Data Analysis and Energy Conservation and Emission Reduction:

[0103] 1.2.1 Provide real-time and historical carbon emission reports for each household appliance.

[0104] 1.2.2 Identify high-energy-consuming or inefficient equipment and provide users with suggestions for repair or replacement.

[0105] 1.2.3 Intelligent Dispatch: During periods of low grid carbon intensity (such as when solar power generation is sufficient during the day) or low electricity prices (such as when time-of-use pricing is available), non-immediate tasks (such as washing dishes, laundry, drying, and charging electric vehicles) are automatically or suggested to be initiated. This directly reduces household carbon emissions and electricity bills.

[0106] 1.2.4 Optimize equipment operating parameters (such as refrigerator temperature setting and washing program selection) to achieve the lowest energy consumption target.

[0107] 2. Intelligent control of HVAC (Heating, Ventilation and Air Conditioning) systems:

[0108] 2.1 Application Example: By combining indoor and outdoor temperature and humidity sensors, human presence sensors, user preference settings, and weather forecast data, AI can precisely control the operation of air conditioning, heating, and fresh air systems.

[0109] 2.2 Data Analysis and Energy Conservation & Emission Reduction:

[0110] 2.2.1 It learns users' daily routines and automatically adjusts the temperature setting to energy-saving mode when leaving home or sleeping, restoring a comfortable temperature before returning home or getting out of bed.

[0111] 2.2.2 Dynamically adjust equipment power based on indoor and outdoor temperature difference, humidity, and personnel density to avoid over-cooling / heating.

[0112] 2.2.3 Analyze changes in the system's energy efficiency ratio to provide early warnings of problems such as filter blockage and insufficient refrigerant, maintain efficient system operation, and reduce ineffective energy consumption and carbon emissions.

[0113] To ensure that those skilled in the art can clearly and completely understand the operational process of this step, the calculation and diagnostic analysis of the coefficient of performance (COP) are introduced as follows:

[0114] 1) The input data source is real-time equipment parameters (1-minute intervals), including compressor input power (accuracy ±1%), refrigerant flow rate (calculated from a flow meter or by evaporator temperature difference), evaporator / condenser inlet and outlet water temperatures (PT100 temperature sensor, ±0.1℃), indoor and outdoor temperature and humidity (BME680 environmental sensor), and fan speed (PWM feedback signal). The formula for calculating the energy efficiency ratio in cooling mode is:

[0115]

[0116] Among them, Q cool Cooling capacity refers to the amount of cooling equipment provides to a space (such as a room or cold storage), i.e., the rate at which heat is removed; W comp The power consumption of the compressor represents the electrical energy consumed by the compressor in transporting and compressing the refrigerant; it is a major energy-consuming component of the refrigeration system. (W) fan The power consumption of the fan refers to the electrical energy consumed by the fan (such as the evaporator fan or condenser fan) to promote airflow and heat exchange; h eavp,in Enthalpy of the refrigerant at the evaporator inlet is a thermodynamic state parameter of the refrigerant when it enters the evaporator. Enthalpy includes internal energy and kinetic energy, reflecting its "heat-carrying capacity"; h eavp,out The enthalpy of the refrigerant at the evaporator outlet represents the enthalpy of the refrigerant leaving the evaporator, and is the difference (h) between the enthalpy at the inlet and outlet. eavp,in -h eavp,out This reflects the "heat absorption" of the evaporator; P is the refrigerant mass flow rate (kg / s); h is the refrigerant specific enthalpy (obtained from the R410A property table using temperature and pressure); P elec This represents the total input power (kW) of the compressor and fan.

[0117] The energy efficiency ratio threshold is generally set as follows: new machine rated COP: 3.8 (provided by the manufacturer); warning threshold: COP<2.5 (triggered when energy efficiency decreases by 34%); fault threshold: COP<1.8 (immediate shutdown for maintenance).

[0118] 2) Furthermore, when a decrease in COP is detected, the AI ​​analysis unit can perform multi-dimensional diagnosis, as shown in Table 3 below, which provides an example of multi-dimensional diagnosis:

[0119] Table 3 (Example of Multidimensional Diagnosis)

[0120]

[0121]

[0122] 3) In addition, harmonic optimization linkage control can be performed when the following conditions are met simultaneously: real-time COP < 2.5; current THD > 7% (IEEE 519 standard tolerance limit); compressor is in steady state operation (non-start-stop transient).

[0123] 2.2.4 Harmonic Optimization Function: Motor loads such as air conditioner compressors and fans are among the main sources of harmonics. The harmonic optimization module can effectively filter out the harmonic currents generated by these loads, reduce line losses and transformer heating, indirectly improve the overall energy efficiency of the HVAC system, and protect equipment from harmonic damage.

[0124] To ensure that those skilled in the art can clearly and completely understand the operational process of this step, the harmonic optimization process is described below:

[0125] 1) First, the formula for calculating the amplitude of harmonic current is:

[0126]

[0127] Among them, I h_comp The detected load harmonic current amplitude (decomposed by FFT); φ h is the initial phase of the load harmonics (extracted through Hilbert transform); k is the attenuation coefficient (default 0.95, to avoid overcompensation); j is the imaginary unit.

[0128] 2) Dynamic adjustment strategy for pulse interval

[0129] ① Adaptive Interval Based on Load Sudden Change Detection

[0130] Normal state: fixed interval T base =1 / (10f) h (e.g., 200μs for the 50th harmonic)

[0131] Transient response: When ΔI is detected load When / Δt>100A / s

[0132]

[0133] For example, when the load suddenly increases by 300%, the interval is shortened to half of the original value.

[0134] Among them, f h The frequency of the h-th harmonic; ΔI load Δt represents the change in load current, indicating the magnitude of the change in load current value over a very short period of time. It reflects the "amplitude" of current fluctuation and is a key physical quantity for measuring the severity of current changes during transient processes. Δt represents the change in time, the time interval, indicating the length of time it takes for the load current to change, reflecting the "speed" of the change. The shorter the time, the more rapid the transient process.rated ΔI represents the rated current of the equipment, which is the standard value of the current allowed when the equipment is working normally; ΔI represents the change in load current, reflecting the magnitude of the current change.

[0135] ② Dead time optimization, the mathematical expression is:

[0136] (IGBT switching time + margin)

[0137] Among them, t rise The IGBT turn-on rise time is the time it takes for the current / voltage to rise during the IGBT's transition from turn-off to turn-on. It reflects the "speed" of turn-on; a shorter rise time means faster switching speed, but may also increase the risk of interference. fall The IGBT turn-off fall time is the time it takes for the current / voltage to decrease during the IGBT's turn-off process. It reflects the "speed" of turn-off. A short fall time can reduce turn-off losses, but it may also introduce electromagnetic interference. When using SiCMOSFET, the dead time can be compressed to below 100ns.

[0138] 3) Pulse Width Modulation (PWM) Parameter Design

[0139] ① Switching frequency selection

[0140] The fundamental frequency can be expressed as:

[0141] f sw =20×f highest_harmonic

[0142] Among them, f highest-harmonic The highest harmonic frequency is the frequency of the highest harmonic among the harmonics that need to be compensated. This determines the highest frequency component that the system needs to "focus on," and is the key basis for selecting the switching frequency. For example, when the 50th harmonic (2.5kHz) needs to be compensated, we take (which satisfies the Nyquist sampling theorem).

[0143] Frequency conversion optimization can be represented as using graded switching frequencies for harmonics in different frequency bands.

[0144] ② Pulse width calculation

[0145] The PWM duty cycle is generated using a quasi-proportional resonant (PR) controller, and the mathematical expression is as follows:

[0146] D h (t)=K p ·I h_err +K r ·∫I h_err ·sin(hwt)dt

[0147] Among them I h_err =Ih_load -I h_comp , representing harmonic current error, K p The proportional gain is used to quickly respond to the current harmonic current error, providing an "instant correction" effect; a larger proportional gain results in a faster response. h (t) represents the dimensionless (between 0 and 1, or scaled according to actual time) PWM duty cycle corresponding to the h-th harmonic, which determines the on / off time ratio of power devices (such as IGBTs), directly controls the output compensation current, and is the "execution command" of the PR controller to the main circuit; I h _err represents the current error of the h-th harmonic, reflecting the deviation between the "actual load harmonic current" and the "compensated expected harmonic current." It is the "input deviation signal" of the PR controller, driving the controller to output a correction value. K r Resonant Gain (hwt) represents the resonant coefficient, providing high gain at the resonant frequency (hw), accurately tracking and compensating for harmonic currents of a specific order (h), achieving "resonant amplification correction" of the target harmonic; h represents the harmonic order, used to identify the harmonic order targeted by the current controller. Different harmonic orders require independent design of PR controllers or multiplexing of structural adjustment parameters; sin(hwt) represents the dimensionless sinusoidal fundamental wave of the h-th harmonic (amplitude -1 to 1), serving as a "frequency selection factor" in the resonant integral stage, allowing the controller to integrate and amplify only the error component of the h-th harmonic frequency, achieving "precise compensation for specific harmonics"; I h_load The h-th harmonic current of the load represents the h-th harmonic current generated by nonlinear loads (such as frequency converters and rectifiers) in the power system, which is an "interference source" that needs to be controlled; I h_comp The h-th harmonic compensation current output by the compensation device, and the current output by compensation equipment such as active power filters (APF), are used to offset I. h_load This makes the system-side harmonic current approach 0. Optionally, K p =0.5,K r =200 is a typical value and requires on-site debugging.

[0148] 3. Intelligent management of lighting systems:

[0149] 3.1 Application Example: The brightness and on / off status of LED lights are automatically adjusted based on natural light intensity, room occupancy, and time settings.

[0150] 3.2 Data Analysis and Energy Conservation and Emission Reduction:

[0151] 3.2.1 Statistically analyze the lighting energy consumption and carbon emissions of each area, and identify areas that are always lit but unoccupied.

[0152] 3.2.2 Provide an “on-demand lighting” strategy to maximize the use of natural light.

[0153] To ensure that those skilled in the art can clearly and completely understand the operational process of this step, the on-demand lighting optimization process is described below:

[0154] 1) First, multimodal sensor data fusion is performed. The core data sources and processing are as follows:

[0155] For the illuminance sensing matrix, multi-point illuminance sensors (such as TSL2591, 0–88 klux accuracy ±3%) are deployed to construct an indoor illuminance distribution heat map; for human presence detection, millimeter-wave radar + WIFI sensing dual verification is performed to dynamically update the human location heat map with an accuracy of ±15 cm; for natural light prediction, LSTM is used to predict the natural light change trend in the next 30 minutes by combining BIM model window-to-wall ratio data and real-time cloud cover forecasts from the meteorological bureau.

[0156] 2) Adaptive dimming algorithm

[0157] The control model is represented as follows:

[0158] L target =max(L req -α·L natural ,L min )

[0159] Among them, L req Standard illuminance for the area (e.g., 500 lx for an office); L nature The measured natural light illuminance is used; α is the light attenuation compensation coefficient (default 0.9, 0.8 for south-facing glass); L min Emergency lighting threshold (50 lx).

[0160] 3) Harmonic Co-optimization Scheme

[0161] LED driver power supply management: Real-time monitoring of current THD for harmonic detection.

[0162] Dynamic compensation strategy: When THD > 15%, the active power filter (APF) is activated to calculate the compensation current.

[0163]

[0164] Among them, I comp The compensation current output by the APF is used to offset the harmonic current generated by the LED load, so that the current on the system side approaches a sine wave. It is the "compensation command current" that the APF applies to the power grid. This indicates that compensation components are calculated separately for harmonics from h=3 to h=25, and then summed to cover typical LED harmonics (3rd / 5th / 7th, etc.) and higher harmonics, comprehensively suppressing THD; h is the harmonic order, indicating the harmonic order currently being calculated, focusing on the 3rd, 5th, and 7th harmonics (typical harmonics of LED nonlinear loads), while also considering higher harmonics for deep THD suppression; ω represents the fundamental angular frequency in rad / s, used to construct the frequency characteristics of the harmonics; t is time, reflecting the "dynamic tracking" characteristic of the compensation current, adjusting the compensation amount in real time as it changes to counteract the time-varying characteristics of the load harmonics; φ h The phase angle (in radians or degrees) of the h-th harmonic current represents the phase information of the h-th harmonic current of the LED load. During compensation, it must be phase-matched (usually out of phase) with the harmonic current to ensure effective cancellation. The negative (-) phase compensation sign makes the compensation current output by the APF opposite in phase to the LED load harmonic current, achieving "current cancellation"—the load harmonic current is: I h sin(hwt+φ h When the compensation current is reversed, the superposition can cancel out harmonics.

[0165] The results of the energy-saving benefit calculation are shown in Table 4 below:

[0166] Table 4 (Energy Saving Benefit Calculation Results)

[0167] Before treatment After treatment Improvement effect THD = 28% THD = 5% 82％↓ Line loss 0.8W / m Line loss 0.3W / m 62.5％↓ The lamp has a lifespan of 32,000 hours. Lifespan 45,000 hours +40％

[0168] 4) Spatial-temporal two-dimensional optimization, using a dawn / dusk ray tracing strategy.

[0169] For the eastern area in the morning: prioritize reducing the brightness of the lights on the east side (natural light contribution rate > 60%); maintain higher supplemental lighting on the west side (natural light contribution rate < 20%). For dynamic response on cloudy days, when the outdoor lux < 10k and cloud cover > 80%, activate global supplemental lighting to 120% of the standard illuminance (to compensate for natural light attenuation), and use a flexible dimming curve (5-minute gradual change to avoid visual discomfort).

[0170] 3.2.3 Harmonic Optimization: Low-quality LED driver power supplies generate harmonics. System harmonic optimization not only reduces power grid pollution but also avoids additional line losses (heat generation) caused by harmonics, thus saving energy and extending the life of the lamps.

[0171] To ensure that those skilled in the art can clearly and completely understand the operational process of this step, the lighting harmonic optimization process is described below:

[0172] 1) Time-domain and frequency-domain analysis of harmonic currents: The harmonic currents generated by the LED driver power supply can be expressed as:

[0173]

[0174] Where: I1 represents the fundamental current (50 / 60Hz); I h φ represents the amplitude of the h-th harmonic current (h = 3, 5, 7, ...); h This indicates harmonic phase shift.

[0175] Furthermore, the time-domain waveform is sampled and acquired, and the LED driving input current i is collected using a current sensor (such as the Hall effect sensor ACS712). (t) The sampling frequency must meet the following requirements:

[0176] f s ≥2×h max ×f1

[0177] Among them, h max This indicates the highest harmonic order that needs to be detected, and identifies the highest harmonic order that the system is interested in (e.g., the LED load needs to detect the 25th harmonic). It is a key basis for the design of the sampling frequency. For example, to measure the 25th harmonic, fs≥2.5kHz is required.

[0178] 2) The operational status correlation model and the mapping relationship between harmonic characteristics and operating conditions are shown in Table 5 below: Table 5 (Mapping relationship between harmonic characteristics and operating conditions)

[0179] LED working status Typical harmonic characteristics (THD range) Dominant harmonic components Cold start instantaneous 35％～50％ 3, 5, 7 times Constant current dimming 15％～25％ 3, 5 times PWM dimming 20％～30％ Higher harmonic groups

[0180] The empirical formula for dynamic harmonic current is:

[0181]

[0182] Among them, K h Represents harmonic coefficients (e.g., K3 = 0.25, K5 = 0.15 for inferior drives); P LED Represents real-time power (calculated by the product of voltage and current); P rated This indicates the rated power of the LED, which is the nominal rated operating power of the LED device. It is a reference power value that defines its normal operating condition and serves as a benchmark for comparison with the actual power. rated This indicates the LED's rated current, which is the current when the LED operates at its rated power. It reflects the LED's basic electrical parameters and is used to calculate the relationship between power and harmonic current, helping to determine the relative magnitude of harmonic current.

[0183] 3) Real-time harmonic estimation algorithm, using the pq method based on instantaneous power theory.

[0184] ① The formulas for calculating instantaneous active power q and reactive power q are as follows:

[0185] (After orthogonal transformation)

[0186] Where p is the instantaneous active power, reflecting the "useful power" actually consumed by the LED load, that is, the work done in converting electrical energy into light energy, heat energy, etc. The rate is a core parameter in the power system that needs to be accurately measured and controlled; q represents instantaneous reactive power, reflecting the rate of "energy exchange" between the LED load and the power grid (it does no work, but affects voltage quality and increases line losses), and needs to be compensated to improve power quality; v grid This refers to the grid voltage, representing the instantaneous voltage value on the grid side. It reflects the real-time magnitude and phase of the supply voltage and serves as the "voltage reference" for power calculation; i LED is the instantaneous value of the LED load current, including active, reactive, and harmonic components, which directly determines the power the load draws from the grid; is the instantaneous value of the orthogonal component of the LED load current, obtained through "orthogonal transformation" (such as the Parker transformation, or the αβ transformation in instantaneous reactive power theory), and is related to v. grid Orthogonal (90° phase difference) is used to separate reactive power components.

[0187] ② The formula for calculating harmonic current components is as follows:

[0188]

[0189] Among them, i h The instantaneous value of the harmonic current component represents the harmonic components (e.g., 3rd, 5th, and 7th harmonics) contained in the LED load current, which are the target quantities that need to be detected and compensated for; p represents the instantaneous active power, reflecting the active power exchange between the load and the power grid, and is calculated using the following formula: p = v grid ·i LED

[0190] q represents instantaneous reactive power, and the calculation formula is as follows:

[0191] Among them, v grid The instantaneous value of the grid voltage serves as the "reference signal" for power calculation and harmonic separation; The instantaneous value of the orthogonal component of the grid voltage, obtained through an orthogonal transformation (such as the Hilbert transform or the αβ transform in instantaneous reactive power theory), is compared with v. grid Orthogonal (90° phase difference) power is used to separate reactive and harmonic power; |v grid | 2 V is the square of the grid voltage amplitude; 2 This is used as a denominator normalization to ensure the correct dimensions of harmonic current calculations and to reflect the "constraint" of voltage on current.

[0192] 4. Harmonic pollution control and energy efficiency improvement:

[0193] 4.1 Application Example: The system continuously monitors the voltage and current waveforms and harmonic content (especially typical harmonics such as the 3rd, 5th, and 7th harmonics) on the home power grid bus.

[0194] 4.2 Data Analysis and Energy Conservation and Emission Reduction:

[0195] 4.2.1 Active Harmonic Suppression: When the detected harmonic distortion exceeds the set threshold, the built-in active power filter or other harmonic mitigation device will automatically start to inject reverse harmonic current into the power grid to cancel it out, so that the power grid current is close to a sine wave.

[0196] To ensure that those skilled in the art can clearly and completely understand the operational procedures of this step, the harmonic pollution control process is described below:

[0197] 1) The formula for calculating Total Harmonic Distortion (THD) is:

[0198]

[0199] Where I1 represents the fundamental current (50 / 60Hz); I h The value represents the amplitude of the h-th harmonic current (h = 3, 5, 7); h represents the harmonic order, which is used as an index for summation, taking values ​​of 2, 3, ..., n, indicating that calculations are performed on harmonics of the 2nd order and above; n represents the highest harmonic order involved in the calculation, representing the range of harmonics to be considered. For example, when analyzing the 25th harmonic, n = 25, covering harmonics from the 2nd to the nth order.

[0200] 2) The real-time detection process for harmonic data is as follows: current sensor signal - anti-aliasing filtering (cutoff frequency 1kHz) - ADC sampling (4kHz, 16bit) - sliding bed DFT operation - harmonic amplitude extraction (I1, I3, I5...) - THD calculation - over-limit alarm.

[0201] 3) Discrete Fourier Transform (DFT) implementation: Perform the following operations on the sampled sequence x[n] (N = 64 points):

[0202]

[0203] Among them, I h The amplitude of the h-th harmonic current is represented by f(x), which is the magnitude of the h-th harmonic current calculated by the DFT and used for harmonic detection and analysis; N is the number of sampling points, representing the length of the sampling sequence for one DFT operation, which determines the frequency resolution Δf = f(x). s / N,f s The sampling frequency is h; h is used to identify the current harmonic order, corresponding to the frequency index f of the DFT. h=hΔf; n is the sampling sequence index, which is used to accumulate and calculate the harmonic components by traversing each point of the sampling sequence; x[n] represents the nth sample value of the sampling sequence, that is, the instantaneous current value collected by AD, which is the input signal of DFT operation; cos(2πhn / N) and cos(2πhn / N) represent the basis functions of DFT, which are multiplied by x[n] and accumulated to extract the in-phase component (cosine term) and quadrature component (sine term) of the hth harmonic, respectively.

[0204] 4.2.2 Reducing Harmonic Losses: Harmonics can cause conductor heating, increased iron and copper losses in transformers, and decreased motor efficiency. Mitigating harmonics directly reduces this energy waste and improves the efficiency of the household power grid, making it an important energy-saving measure (typically achieving 3%-8%).

[0205] Table 6 below shows the flowchart for calculating energy saving due to harmonic loss:

[0206]

[0207]

[0208] The following is a detailed explanation of the key components in Table 6 above: 1) Harmonic loss sources (contributing 25% to 70% of the loss increase) ① The loss formula for copper loss in conductors and transformers is:

[0209]

[0210] Wherein, ΔP cu The copper loss increment represents the additional copper loss caused by harmonics (the increase in losses in conductors / transformer windings due to harmonics); I rms This represents the effective value of the total current containing the fundamental and harmonic waves, which can be expanded in the formula as follows:

[0211]

[0212] Wherein, I1 is the effective value of the fundamental current (50Hz or 60Hz component), which is the core component of an ideal sinusoidal current. THD I Total Harmonic Distortion (THD) of current reflects the degree of distortion in the current waveform. I 2 The amplification effect of quantization harmonics on copper losses. For example, THD. I =30% → Copper loss increases by 9%.

[0213] ② Regarding transformer / motor iron losses: High-frequency harmonics (e.g., 11th order / 550Hz) → Eddy current losses ∝f 2 → The amplification can reach 10 to 100 times that of the fundamental frequency; f is the harmonic frequency.

[0214] ③ Regarding additional losses in the motor: Harmonic magnetic fields induce eddy currents in the rotor / frame, leading to a significant increase in stray losses (>20% when THD). I >35%

[0215] 2) System loss baseline (accounting for 8% to 15% of total household electricity consumption)

[0216] The loss percentage of distribution transformers is 1.5% to 3% (allocated to users).

[0217] The inherent loss rate of electric motor systems (accounting for 25% to 40% of household electricity consumption) is 8% to 12%, which translates to 2% to 4.8% of total electricity consumption.

[0218] Long-distance low-voltage lines entering homes: losses account for 0.8% to 1.5%;

[0219] Power conversion losses for other devices, such as electronic equipment, are 0.5% to 1.2%.

[0220] 4.2.3 Protect equipment: Reduce the risk of harmonic damage to sensitive electronic equipment (computers, televisions, smart home controllers), extend equipment life, and indirectly reduce the implicit carbon emissions caused by premature equipment scrapping.

[0221] 4.2.4 Improve measurement accuracy: Harmonics may cause meter reading errors (usually resulting in overcharging). Optimization will make electricity bills more accurate.

[0222] 5. Comprehensive data analysis and user guidance:

[0223] 5.1 Application Example: The system backend integrates all energy consumption (electricity, possibly gas, water), carbon emissions, harmonic data, and equipment status information.

[0224] 5.2 Data Analysis and Energy Conservation and Emission Reduction:

[0225] 5.2.1 Visual dashboard: Provides clear total household carbon emissions, breakdown of carbon emissions (home appliances, lighting, HVAC, etc.), real-time / historical harmonic distortion rate, and energy-saving potential reports.

[0226] 5.2.2 Energy Efficiency Rating and Recommendations: Efficiency ratings are given for the entire household or individual appliances, and customized, actionable energy-saving and carbon-reduction recommendations are provided (e.g., "Raising the air conditioner setting temperature by 1°C is expected to reduce carbon emissions by XX kg per month and save XX yuan in electricity bills").

[0227] 5.2.3 Target Setting and Tracking: Allows users to set monthly / annual carbon emission reduction or energy saving targets, and the system tracks progress and provides feedback.

[0228] 5.2.4 Abnormal alarm: Provides immediate alarms for abnormally high energy consumption, excessive harmonics, and equipment malfunctions.

[0229] 5.2.5 Benchmark Comparison: (While protecting privacy) Compare with the average level of similar households in the same region to incentivize users.

[0230] Based on the above specific examples, the core value of this invention lies in the following points:

[0231] 1) Panoramic monitoring: Visible (carbon emissions, energy consumption, harmonics).

[0232] 2) Deep insights: Using AI to understand the patterns and problems behind the data.

[0233] 3) Proactive optimization: Automatically or suggest the execution of optimal strategies (scheduling equipment, mitigating harmonics).

[0234] 4) Safety guarantee: Improve power quality and protect home appliances.

[0235] 5) User empowerment: Guide users to develop more environmentally friendly lifestyle habits through clear data and suggestions.

[0236] Furthermore, the implementation architecture of the household electricity optimization system proposed in the embodiments of the present invention is described in the actual execution process:

[0237] 1) System Overview

[0238] The household electricity optimization system proposed in this invention achieves energy-saving operation of home appliances and carbon footprint tracking through intelligent monitoring, AI analysis and adaptive optimization, thereby reducing household energy consumption and carbon emissions.

[0239] 2) System Architecture

[0240] ① Hardware layer

[0241] Smart meters and sensor networks:

[0242] - Distributed wireless current sensors (such as CT sensors) monitor the real-time energy consumption of various household appliances;

[0243] - Temperature, humidity, and air quality sensors monitor environmental parameters;

[0244] - Equipment operating status sensors (such as door switch and operation indicator light monitoring).

[0245] Edge computing gateway:

[0246] - Local data preprocessing and caching;

[0247] - Execution of equipment control commands;

[0248] - Local autonomous decision-making capability when the network is down.

[0249] ② Data Layer

[0250] Real-time data acquisition module:

[0251] - Device-level energy consumption data (power, power factor, harmonics, etc.);

[0252] - Grid data (real-time electricity price, regional carbon intensity index API);

[0253] -Weather forecast data API.

[0254] Data storage:

[0255] - Time-series databases (such as InfluxDB) store high-frequency energy consumption data;

[0256] - Relational database storage device properties and user settings;

[0257] - Data lakes store historical data for long-term analysis.

[0258] ③Analysis layer

[0259] Device fingerprint recognition:

[0260] - Device identification algorithm based on VI trajectory and transient response characteristics;

[0261] - Non-invasive load decomposition (NILM) technology identifies composite loads;

[0262] - Detection of abnormal power consumption patterns based on deep learning.

[0263] Carbon footprint accounting engine:

[0264] -Dynamic carbon footprint factor = grid carbon intensity × equipment energy efficiency coefficient;

[0265] -Life Cycle Assessment (LCA) models integrate the implicit carbon emissions of major household appliances;

[0266] - The Scope 2 classification method of the WRI greenhouse gas accounting system is adopted.

[0267] ④ Optimization layer

[0268] Multi-objective optimization scheduler:

[0269] - Objective function: min(α×electricity cost + β×carbon emissions + γ×user inconvenience); - Constraints: equipment physical limitations, user preferences, power grid constraints;

[0270] - Uncertainty is handled using an improved Model Predictive Control framework. Parameter optimization engine:

[0271] -Reinforcement learning-based adaptive adjustment of device parameters;

[0272] Federated learning optimizes the model while protecting privacy.

[0273] ⑤ Application Layer

[0274] - User interface (mobile / web)

[0275] -Automation control interface;

[0276] - Third-party system integration API.

[0277] 3) Core technologies

[0278] ① New equipment identification and status monitoring algorithm

[0279] Deep Convolutional Residual Networks (DCRN) are used for load decomposition:

[0280] - Input: High-frequency sampled data of voltage and current waveforms;

[0281] - Architecture: 1D convolution combined with attention mechanism;

[0282] - Output: Operating status and energy consumption percentage of each device;

[0283] -Innovation: Introducing equipment aging factors as dynamic weights;

[0284] - Device health assessment based on Fréchet distance:

[0285] By comparing the similarity between the current VI characteristic curve and the factory baseline curve, the degree of energy efficiency degradation of the equipment can be quantified.

[0286] ② Carbon-sensing scheduling algorithm

[0287] Time-dependent carbon intensity prediction models:

[0288] - Ridge regression model combining ARIMA time series analysis and weather sensitivity - Input: historical carbon intensity, renewable energy share forecast, weather forecast - Output: confidence interval of carbon intensity curve for the next 24 hours

[0289] Robust optimization scheduling framework:

[0290] minΣ[t=1..T](λ_t^carbon×E_t+λ_t^price×E_t)st

[0291]

[0292] Wherein, λ_t^carbon and λ_t^price are time-varying weighting factors that are dynamically adjusted according to user preferences.

[0293] ③ Optimization algorithm for runtime parameters

[0294] Deep deterministic policy gradient (DDPG) is used for continuous parameter optimization:

[0295] -State space: Device status + environmental parameters + user historical behavior;

[0296] -Action space: continuous parameters such as temperature setpoint and operating mode;

[0297] - Reward function: R = -(Energy consumption + ω × User dissatisfaction).

[0298] Federated Transfer Learning Framework:

[0299] Each household's local model regularly uploads gradient updates to the cloud for aggregation, protecting privacy while leveraging collective intelligence to optimize the global model.

[0300] The following section uses refrigerator optimization as an example to introduce the power consumption optimization process:

[0301] 1) Dynamic temperature setting:

[0302] -Predict future access behavior based on door opening and closing frequency;

[0303] - Increase the set temperature appropriately during periods of low activity (e.g., from 4°C to 6°C);

[0304] -Based on the weather forecast, the high temperatures are expected to drop before the day.

[0305] 2) Optimized defrosting cycle:

[0306] -Based on a frost layer thickness estimation model (through compressor current ripple analysis);

[0307] - Trigger the defrosting cycle during low-carbon periods.

[0308] 3) Edge computing implementation:

[0309] ①Calculate resource allocation:

[0310] Main control cycle: 60 seconds;

[0311] Calculate task priority:

[0312] Real-time temperature control (interrupt-driven); defrost decision (hourly evaluation); behavioral model update (daily low-load periods).

[0313] ② Power outage emergency plan:

[0314] Maintain the last effective temperature setting; degrade to timed defrost mode (once every 24 hours); cache critical operating data locally (up to 7 days).

[0315] ③ Safety protection mechanism: soft limit of temperature setpoint (2℃~8℃); minimum interval protection of compressor (interval between two starts >3 minutes); automatic reset in abnormal state (switches to safety mode after 3 hours of continuous over-temperature).

[0316] Based on the above examples, it is further confirmed that the home power optimization system provided by this invention has the following innovative features:

[0317] 1. Pareto's cutting-edge visualization of carbon-electricity prices: helps users make quantitative trade-offs between saving on electricity bills and reducing their carbon footprint;

[0318] 2. Device-level carbon footprint labeling system: Generates dynamic carbon labels for each home appliance, similar to nutrition labels;

[0319] 3. Cross-device collaborative optimization: For example, using waste heat from a refrigerator compressor to assist in heating a water heater;

[0320] 4. Blockchain Evidence Storage: Key emission reduction activities generate tamper-proof records that can be used for carbon credit trading.

[0321] Those skilled in the art will clearly understand that the specific working process of the systems, devices, modules and units described above can be referred to the corresponding process in the foregoing method embodiments. For the sake of brevity, it will not be repeated here.

[0322] It should be noted that the functional units in the various embodiments of the present invention can be physically independent of each other, or two or more functional units can be integrated together, or all functional units can be integrated into one processing unit. The integrated functional units described above can be implemented in hardware, or in software or firmware.

[0323] Those skilled in the art will understand that if the integrated functional unit is implemented in software and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or all or part of it, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computing device (e.g., a personal computer, server, or network device) to execute all or part of the steps of the methods described in the embodiments of the present invention when running the instructions. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0324] Alternatively, all or part of the steps of the foregoing method embodiments can be implemented by hardware (such as a computing device, personal computer, server, or network device) related to program instructions. The program instructions can be stored in a computer-readable storage medium. When the program instructions are executed by the processor of the computing device, the computing device executes all or part of the steps of the methods described in the various embodiments of the present invention.

[0325] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that within the spirit and principles of the present invention, modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the corresponding technical solutions to depart from the protection scope of the present invention.

Claims

1. A household electricity optimization system, characterized in that, The system includes: A data acquisition module connected to the home power grid is used to acquire current / voltage signals in real time; The signal processing module connected to the data acquisition module is used to perform harmonic detection on the current / voltage signal to obtain first power data containing harmonic component data and current / voltage distortion parameters; and to fuse the first power data with historical operating data to identify power consumption patterns and / or predict harmonic trends through an AI analysis model based on deep learning algorithms. The inverter compensation module connected to the signal processing module is used to generate an IGBT compensation strategy to offset grid distortion based on the power consumption mode and / or harmonic trend, and to purify the power output based on the IGBT compensation strategy to supply home appliances. The power acquisition module connected to the data acquisition module is used to acquire second power data, including active power probability, reactive power and / or power consumption, in real time. A carbon emission metering module that accesses the carbon intensity API and is connected to the power acquisition module is used to dynamically retrieve the carbon emission factor of the local power grid and calculate household carbon emission data by integrating the second power data and the carbon emission factor. A carbon emission optimization module, which is connected to the carbon emission metering module and the signal processing module respectively, is used to generate a carbon emission optimization strategy based on the household carbon emission data, send the carbon emission optimization strategy to the user terminal, and / or control the operation of various home appliances based on the carbon emission optimization strategy. The household carbon emission data includes individual device carbon emission data and / or total household carbon emission data.

2. The system according to claim 1, characterized in that, The system also includes: a cloud platform and the user terminal; the signal processing module includes a harmonic detection unit, an interactive control unit, and an AI analysis unit connected in sequence. The interactive control unit wirelessly communicates with both the cloud platform and the user terminal to receive user commands from the user terminal and historical operating data from the cloud platform; it also coordinates data transmission between the harmonic detection unit and the AI ​​analysis unit; wherein the historical operating data includes at least one of peak and off-peak periods of household electricity consumption and seasonal load changes. The AI ​​analysis unit is also connected to the carbon emission metering module and is used to perform feature fusion processing on the first electrical energy data, the historical operation data and / or the household carbon emission data using a multi-source data fusion algorithm to obtain key feature factors with corresponding weights; it is also used to construct an AI analysis model that integrates Bi-LSTM sub-model, Transformer sub-model and / or GCN sub-model. The Bi-LSTM sub-model is used to identify electricity consumption patterns using the key feature factors to obtain pattern recognition results; the Transformer sub-model is used to process high-frequency data in parallel, perform harmonic trend prediction, and obtain harmonic spectrum results within a preset future time period; the GCN sub-model is used to construct the topological relationship of various types of home appliance nodes and perform carbon emission-harmonic correlation analysis based on the key feature factors to predict the carbon emissions of various types of home appliances. The pattern recognition results include harmonic characteristics of various home appliances connected under periodic patterns.

3. The system according to claim 2, characterized in that, The inverter compensation module includes a strategy generation unit and an IGBT optimization unit connected together; The strategy generation unit is used to calculate an anti-phase harmonic compensation strategy, including compensation time, phase offset and / or compensation current, based on the pattern recognition results, the current grid load and the current IGBT parameters, using an RL reinforcement learning algorithm. The IGBT optimization unit is used to determine the IGBT compensation strategy based on the anti-phase harmonic compensation strategy and according to the parameter mapping rules; the IGBT compensation strategy includes switching frequency and conduction time.

4. The system according to claim 2, characterized in that, The carbon emission metering module is used for: Based on the first electrical energy data, multimodal features are extracted from the current waveform to construct feature vectors characterizing the operating status and / or environmental features of various household appliances. The mathematical expression is as follows: x t =[I mean ,I var ,THD,f1,f2,...,f n ,time_of_day] Where, x t Let I be the eigenvector at time t; mean I represents the average current; var The variance of the current is represented by f1, f2, ..., f2. n Represents the frequency components of each harmonic; time_of_day represents the time characteristics; Based on the feature vectors, the operating status of the various types of home appliances is identified using the HMM algorithm or LSTM algorithm. Based on the operating status of various household appliances, a correlation model between appliance power consumption patterns and carbon emissions is constructed, with the following mathematical expression: Among them, E carbon This indicates carbon emission data for equipment or household use; P i (t) represents the real-time power of device i; α(t) represents the carbon emission factor; β i This represents the equipment type correction factor; n is 1 or the total number of equipment. Based on the aforementioned device electricity consumption pattern-carbon emission correlation model, household carbon emission data during peak and off-peak electricity consumption periods are calculated.

5. The system according to claim 4, characterized in that, The carbon emission optimization module includes: The carbon emission optimization unit connected to the carbon emission metering module is used to generate a carbon emission optimization strategy based on the difference between household carbon emission data during peak and off-peak periods of household electricity consumption. The carbon emission optimization strategy includes delaying the start-up time of non-emergency equipment to off-peak hours and prioritizing the use of at least one of the following in conjunction with photovoltaic charging devices: The load control unit, which is connected to the carbon emission metering module and the harmonic detection unit respectively, is used to monitor the first electrical energy data and the equipment carbon emission data in real time. When it is detected that the start-up of a certain device causes a simultaneous surge in the harmonic component data and the equipment carbon emission data, a load control command is generated for synchronously adjusting the IGBT compensation and the device switch, and sent to the inverter compensation module for execution.

6. The system according to claim 2, further comprising: The quality detection module, which is connected to the inverter compensation module and the carbon emission optimization module respectively, is used to monitor the third electrical energy data after inverter compensation and / or carbon emission optimization in real time, and feed the third electrical energy data back to the harmonic detection unit for closed-loop control, and / or feed it back to the AI ​​analysis unit for iterative training; The third electrical energy data includes at least one of the following: THD harmonic distortion rate, equipment power factor, and equipment carbon emission data.

7. The system according to claim 2, characterized in that, The multi-source data fusion algorithm includes: A sliding window mechanism is used to perform spatiotemporal alignment processing on the first electrical energy data, the historical operating data, and / or the household carbon emission data, and cross-domain feature extraction is performed to obtain key feature factors including harmonic spectrum, voltage fluctuation, historical load curve, and / or ambient temperature and humidity. The key feature factors are weighted using an attention-based feature importance evaluation algorithm to obtain key feature factors with corresponding weights. And / or, The loss function of the Bi-LSTM sub-model is: in, The loss function is represented by T; the time window length is represented by t; the time index is represented by t∈[1,T]; y t Indicates the actual load condition, y t ∈{0,1}; Indicates the predicted probability. θ represents the model parameter vector, ||θ|| 2 λ represents the squared L2 norm; λ represents the regularization coefficient. And / or, The Transformer sub-model is used to input a current waveform sequence, perform position encoding on the current waveform sequence, and predict the harmonic spectrum results within a preset time period. The mathematical expression for the position encoding is: Among them, PE (pos,2i) Indicates the encoding position; i is the dimension table; pos represents the sequence position; d model Indicates the encoding dimension; And / or, The loss function of the GCN sub-model is: in, Represents physical constraint loss; α represents carbon emission loss weight; β represents harmonic loss weight; MSE represents the mean square error function; E carbon This represents actual carbon emission data; Represents predicted carbon emissions data; ReLU represents the linear rectified function; THD pred This indicates the predicted harmonic distortion rate.

8. The system according to claim 3, characterized in that, The policy generation unit is also used to generate policy vectors, the mathematical expression of which is: S=[ΔI,Δφ,t start ] Where S represents the strategy vector; ΔI represents the compensation current. Δφ represents the phase shift; t start Indicates the time of compensation; And / or, The objective function of the parameter mapping rule is: Among them, I 理想 The target current is represented by γ, which represents the loss weight. The constraints include: switching frequency ≤ IGBT maximum allowable frequency and driving voltage within the safe range.

9. The system according to any one of claims 1 to 8, characterized in that, The signal processing module further includes: An anomaly detection unit connected to the harmonic detection unit is used to identify power grid faults based on the first power data, generate early warning information, and feed the early warning information back to the cloud platform and / or the user terminal. The power grid fault conditions include at least one of the following: abnormal equipment harmonics, three-phase imbalance, and voltage drop.

10. The system according to claim 9, characterized in that, The user terminal is equipped with an interactive page for displaying the third electrical energy data and carbon emission visualization data in real time; The carbon emission visualization data includes at least one of the following: household carbon emission data, device carbon emission data and percentage, regional carbon factor trend and household carbon emission comparison curve, and carbon emission optimization strategy.

Images