Non-invasive Load Monitoring Method, Device, Computer Equipment and Storage Medium

Abstract

The present application relates to a non-invasive load monitoring method, device, computer device, and storage medium. The method includes: extracting an active power sequence and a reactive power sequence from meter data, using the active power sequence and the reactive power sequence as inputs, and calling a trained hybrid non-invasive load monitoring model to obtain a state detection result of an electrical device and an energy injection sequence prediction result. When the hybrid non-invasive load monitoring model is called, the following steps are performed: extracting and splicing the active power features in the active power sequence and the reactive power features in the reactive power sequence to obtain a comprehensive feature sequence, capturing the long-term dependencies in the comprehensive feature sequence to obtain a multi-scale feature sequence, and based on the multi-scale feature sequence, predicting the state of the electrical device and the energy injection of the micro-inverter respectively to obtain the state detection result of the electrical device and the energy injection sequence prediction result. Using this method can achieve accurate monitoring of the power consumption of electrical devices.

Description

Technical Field

[0001] The present application relates to the technical field of load monitoring, and particularly to a non-intrusive load monitoring method, device, computer device, computer-readable storage medium, and computer program product. Background Art

[0002] Non-Intrusive Load Monitoring (NILM) technology is a method of identifying and estimating the electricity consumption of each electrical device by analyzing the total electricity meter data. The traditional NILM task is to identify the electricity consumption of each electrical appliance based on the total electricity data. From the perspective of signal processing, it belongs to the blind source separation problem. Blind Source Separation (BSS) refers to the process of separating the original source signals from the observed mixed signals only when the source signals and the mixing process are unknown.

[0003] With the popularization of distributed energy, the electricity consumption situation in modern households has become more complex, which undoubtedly increases the difficulty of signal separation. Therefore, for the modern smart home environment containing distributed energy, the current NILM task is difficult to accurately monitor the electricity consumption of electrical devices. Summary of the Invention

[0004] Based on this, in view of the above technical problems, it is necessary to provide a non-intrusive load monitoring method, device, computer device, computer-readable storage medium, and computer program product that can accurately monitor the electricity consumption of electrical devices.

[0005] In a first aspect, the present application provides a non-intrusive load monitoring method, including:

[0006] Extract the active power sequence and the reactive power sequence from the electricity meter data;

[0007] Using the active power sequence and the reactive power sequence as inputs, call the trained hybrid non-intrusive load monitoring model to obtain the state detection result of the electrical device and the energy injection sequence prediction result;

[0008] When the hybrid non-intrusive load monitoring model is called, it performs the following steps:

[0009] Extract the active power feature in the active power sequence and the reactive power feature in the reactive power sequence, and splice the active power feature and the reactive power feature to obtain a comprehensive feature sequence;

[0010] Capture the long-term dependence relationship in the comprehensive feature sequence to obtain a multi-scale feature sequence;

[0011] Based on the multi-scale feature sequence, the states of the electrical equipment and the energy injection of the micro-inverter are predicted respectively, and the state detection result of the electrical equipment and the energy injection sequence prediction result are obtained;

[0012] Among them, the hybrid non-intrusive load monitoring model is trained based on the historical active power sequence and the historical reactive power sequence.

[0013] In one embodiment, the hybrid non-intrusive load monitoring model includes an encoder, a converter model, a sequence-to-point module, and a sequence-to-sequence module.

[0014] In one embodiment, the encoder includes a first convolutional encoder and a second convolutional encoder; the extraction of the active power feature in the active power sequence and the reactive power feature in the reactive power sequence includes:

[0015] Call the first convolutional encoder to perform one-dimensional convolution on the active power sequence, and extract the active power feature in the active power sequence;

[0016] Call the second convolutional encoder to perform one-dimensional convolution on the reactive power sequence, and extract the reactive power feature in the reactive power sequence feature.

[0017] In one embodiment, the capturing of the long-term dependencies in the comprehensive feature sequence to obtain the multi-scale feature sequence includes:

[0018] Input the comprehensive feature sequence into the converter model, and the converter model captures the long-term dependencies in the comprehensive feature sequence through the multi-head attention mechanism to obtain the multi-scale feature sequence.

[0019] In one embodiment, the predicting of the states of the electrical equipment and the energy injection of the micro-inverter respectively based on the multi-scale feature sequence to obtain the state detection result of the electrical equipment and the energy injection sequence prediction result includes:

[0020] Input the multi-scale feature sequence output by the converter model into the sequence-to-point module, and the sequence-to-point module maps the last time step of the multi-scale feature sequence to the electrical equipment state prediction task through a fully connected layer to obtain the state detection result of the electrical equipment;

[0021] Input the multi-scale feature sequence output by the converter model into the sequence-to-sequence module, and the sequence-to-sequence module maps the multi-scale features of each time step in the multi-scale feature sequence to the energy injection prediction task through a fully connected layer to obtain the energy injection sequence prediction result.

[0022] In one embodiment, after obtaining the state detection result of the electrical device and the energy injection sequence prediction result, the method further includes:

[0023] Perform normalization processing on the energy injection sequence prediction result to obtain a normalized energy injection sequence prediction result;

[0024] Perform adaptive filtering on the normalized energy injection sequence prediction result, map the normalized energy injection sequence prediction result to a predefined power output gear, and determine the output power level of the micro-inverter, where the output gear corresponds to the output power level one by one.

[0025] In one embodiment, the extracting the active power sequence and the reactive power sequence from the electricity meter data includes:

[0026] Preprocess the electricity meter data;

[0027] Extract the active power sequence and the reactive power sequence with a fixed length from the preprocessed electricity meter data through a sliding window method.

[0028] In a second aspect, the present application further provides a non-intrusive load monitoring device, including:

[0029] A data extraction module for extracting the active power sequence and the reactive power sequence from the electricity meter data;

[0030] A data processing module for using the active power sequence and the reactive power sequence as inputs and calling a trained hybrid non-intrusive load monitoring model to obtain the state detection result of the electrical device and the energy injection sequence prediction result;

[0031] When the hybrid non-intrusive load monitoring model is called, it performs the following steps:

[0032] Extract the active power feature in the active power sequence and the reactive power feature in the reactive power sequence, splice the active power feature and the reactive power feature to obtain a comprehensive feature sequence;

[0033] Capture the long-term dependence relationship in the comprehensive feature sequence to obtain a multi-scale feature sequence;

[0034] Based on the multi-scale feature sequence, respectively predict the state of the electrical device and the energy injection of the micro-inverter to obtain the state detection result of the electrical device and the energy injection sequence prediction result;

[0035] Wherein, the hybrid non-intrusive load monitoring model is trained based on the historical active power sequence and the historical reactive power sequence.

[0036] In a third aspect, the present application also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps in any one of the above non-intrusive load monitoring method embodiments are implemented.

[0037] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in any one of the above non-intrusive load monitoring method embodiments are implemented.

[0038] In a fifth aspect, the present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps in any one of the above non-intrusive load monitoring method embodiments are implemented.

[0039] The above non-intrusive load monitoring method, device, computer device, computer-readable storage medium, and computer program product are pre-trained with a hybrid non-intrusive load monitoring model. This hybrid non-intrusive load monitoring model has a multi-task learning framework and can simultaneously perform device state prediction and energy injection estimation. In practical applications, considering data in two dimensions of active power and reactive power, an active power sequence and a reactive power sequence are extracted from the meter data, which can comprehensively reflect the working state of electrical appliances and the operation of the power system. Subsequently, the active power sequence and the reactive power sequence are input into the trained hybrid non-intrusive load monitoring model. The model concatenates the features in the active power sequence and the features in the reactive power sequence to obtain a comprehensive feature sequence. Then, the long-term dependencies in the comprehensive feature sequence are captured, and complex time dependencies are modeled to obtain a multi-scale feature sequence. Finally, based on the multi-scale feature sequence, the state of the electrical appliance and the energy injection of the micro-inverter are predicted simultaneously, obtaining the state detection result of the electrical appliance and the prediction result of the energy injection sequence. The entire solution only needs to collect meter data to identify the state of the electrical appliance and estimate the energy injection, obtaining comprehensive and accurate monitoring results, and can be applied to modern smart home environments containing distributed energy. Moreover, the above solution has stability and robustness in data from different families, different time periods, and different types of electrical appliances. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other related drawings can be obtained without creative efforts based on these drawings.

[0041] Figure 1It is an application environment diagram of a non-intrusive load monitoring method in an embodiment;

[0042] Figure 2 It is a schematic flow diagram of a non-intrusive load monitoring method in an embodiment;

[0043] Figure 3 It is a schematic diagram of the data processing process involved in a model in an embodiment;

[0044] Figure 4 It is a schematic diagram of the data processing process involved in a model in another embodiment;

[0045] Figure 5 It is a schematic flow diagram of a non-intrusive load monitoring method in another embodiment;

[0046] Figure 6 It is a structural block diagram of a non-intrusive load monitoring device in an embodiment;

[0047] Figure 7 It is an internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0048] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0049] The non-intrusive load monitoring method provided by the embodiments of the present application can be applied to an application environment as Figure 1 shown. Among them, the electricity meter 100 and the terminal 102 communicate with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed in the cloud or on other network servers. A hybrid non-intrusive load monitoring model is deployed on the server 104.

[0050] Specifically, the electricity meter 100 may transmit the collected electricity meter data to the server 104. When there is a load monitoring requirement, the operator sends a load monitoring message to the server 104 through the terminal 102. The server 104 responds to this message, extracts the active power sequence and the reactive power sequence from the electricity meter data, uses the active power sequence and the reactive power sequence as inputs, and calls the trained hybrid non-intrusive load monitoring model. This hybrid non-intrusive load monitoring model extracts the active power features in the active power sequence and the reactive power features in the reactive power sequence, splices the active power features and the reactive power features to obtain a comprehensive feature sequence, captures the long-term dependencies in the comprehensive feature sequence to obtain a multi-scale feature sequence, and based on the multi-scale feature sequence, predicts the states of the electrical devices and the energy injection of the micro-inverters respectively, to obtain the state detection results of the electrical devices and the prediction results of the energy injection sequence.

[0051] Among them, the electricity meter 100 may be, but is not limited to, an intelligent electricity meter. The terminal 102 may be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices may be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0052] In an exemplary embodiment, as Figure 2 shown, a non-intrusive load monitoring method is provided. Taking the method applied to the Figure 1 server 104 as an example for illustration, it includes the following steps (hereinafter simply referred to as S) 200 to S400. Among them:

[0053] S200, extract the active power sequence and the reactive power sequence from the electricity meter data.

[0054] The electricity meter data refers to the power consumption information recorded by intelligent electricity meters (including main electricity meters and sub-electricity meters), and usually includes parameters such as voltage, current, and frequency. For non-intrusive load monitoring, the focus is on active power and reactive power. The active power sequence represents the electrical energy actually consumed by electrical devices over a period of time, in kilowatt-hours (kWh) or watts (W), and reflects the part of the energy converted into useful work. The reactive power sequence represents the part used to establish a magnetic field or other forms of energy storage but not doing work.

[0055] In practical applications, taking the home scenario as an example, a smart meter is installed in the user's home. This meter can measure and record the total power consumption in the home (i.e., meter data) in real time, including active power and reactive power, and send the recorded data to the server through the network. The user sends a load monitoring message to the server through the terminal. The server can perform data preprocessing on the meter data, including cleaning, missing value and outlier processing, and data alignment, etc. Subsequently, the active power and reactive power are extracted from the preprocessed meter data to obtain the active power sequence and the reactive power sequence.

[0056] S400, taking the active power sequence and the reactive power sequence as inputs, calls the trained hybrid non-intrusive load monitoring model to obtain the state detection result of the electrical equipment and the prediction result of the energy injection sequence. Among them, the hybrid non-intrusive load monitoring model is trained based on the historical active power sequence and the historical reactive power sequence.

[0057] The state detection result of the electrical equipment refers to the predicted on, off or running state of each electrical equipment. The prediction result of the energy injection sequence refers to the predicted energy injection amount from general energy injection or renewable energy within a period of time, such as the energy injected by solar panels through a micro-inverter.

[0058] The hybrid non-intrusive load monitoring model (i.e., HybridNILM) is a deep learning model trained based on a large amount of historical active power and reactive power data, aiming to identify the working state of individual appliances and predict the energy injection situation by analyzing the active power and reactive power.

[0059] In practical applications, the model training methods include training from scratch and fine-tuning based on a pre-trained model. Specifically, the learning rate adjustment strategy can adopt a cosine annealing learning rate scheduling strategy with warm restarts, which converges quickly in the initial stage of training and then jumps out of local optima through periodic restarts in the later stage. Specifically, in the model training stage, a large learning rate is used at the beginning. Generally, the learning rate of 0.001 defaulted by the Adam optimizer is used, and then manual adjustment is made according to the training loss and performance results of a few epochs. In the second half of the training or after the loss function stops decreasing, a small learning rate, a large number of training epochs, and storing the training evaluation metrics at each step are used. The optimal checkpoint model is selected as the model weights according to the performance metrics. In the model fine-tuning stage, the pre-trained model weights are loaded, some layers are frozen, and only specific layers (the decoder and the output layer) are trained. Moreover, a smaller learning rate is used for fine-tuning, generally a learning rate of about 0.0001, and adjustments are made according to the actual situation. Finally, it is found that around epoch 30, both the test and training performances tend to be stable, indicating that the model parameters have converged. At this time, the optimal checkpoint saved can be selected as the model weights, or the learning rate can be further decreased to observe the change in the model performance. If the test performance metric of the model decreases after the learning rate is decreased, it indicates overfitting. Through the above training and fine-tuning strategies, the HybridNILM model can effectively learn complex electrical appliance usage patterns and energy injection patterns and predict the states of electrical appliances.

[0060] After training the hybrid non-intrusive load monitoring model, the hybrid non-intrusive load monitoring model is deployed in the server. After the server extracts the active power sequence and the reactive power sequence, the active power sequence and the reactive power sequence are input into the trained hybrid non-intrusive load monitoring model (hereinafter referred to as the model), and the model predicts the states of each electrical device and the energy injection situation to obtain the state detection results of the electrical devices and the energy injection sequence prediction results.

[0061] As Figure 3 shown, specifically, when the hybrid non-intrusive load monitoring model is called, the following steps are executed:

[0062] S420, extract the active power features in the active power sequence and the reactive power features in the reactive power sequence, and splice the active power features and the reactive power features to obtain a comprehensive feature sequence.

[0063] The active power feature refers to specific attributes extracted from the active power sequence that can reflect the operating mode or state of electrical appliances, such as instantaneous power changes, average power levels, power fluctuation frequencies, etc. The reactive power feature refers to patterns related to the electromagnetic characteristics in the power system that can be revealed by extracting from the reactive power sequence, such as starting current, harmonic components, etc. The comprehensive feature sequence refers to combining the active power feature and the reactive power feature together to form a new feature vector or sequence, which can more comprehensively describe the power consumption at each time point.

[0064] In specific implementation, the model can extract the active power feature from the active power sequence and the reactive power feature from the reactive power sequence through an encoder. Then, the extracted active power feature and reactive power feature are combined one by one in chronological order to ensure that each pair of features corresponds to the data at the same time point, resulting in a new sequence (comprehensive feature sequence) containing all features.

[0065] S440, capture the long-term dependencies in the comprehensive feature sequence to obtain a multi-scale feature sequence.

[0066] The long-term dependency refers to the correlation between different time points within a long time span, especially those relationships that span a long time interval but still maintain a significant association. The multi-scale feature sequence is a series of feature representations generated by capturing patterns and trends at different time scales, used to adapt to the identification of different types of electrical appliances and electricity consumption behaviors.

[0067] In specific implementation, the model can divide the concatenated feature sequence into time windows of different lengths, call corresponding models for capturing long-term dependencies such as long short-term memory networks or transformer models, process the data within each time window, learn how to identify important patterns at different time scales, and integrate the feature representations from different time windows to form a multi-scale feature sequence.

[0068] S460, based on the multi-scale feature sequence, respectively predict the state of the electrical equipment and the energy injection of the micro-inverter to obtain the state detection result of the electrical equipment and the prediction result of the energy injection sequence.

[0069] After obtaining the multi-scale feature sequence, the model can use the multi-scale feature sequence to infer which electrical appliances are currently in use and their operating states, such as on / off states, and predict the amount of electricity that the distributed power generation system (such as solar panels) will supply to the power grid or local load in the future based on the multi-scale feature sequence.

[0070] The above non-intrusive load monitoring method is pre-trained with a hybrid non-intrusive load monitoring model. The hybrid non-intrusive load monitoring model has a multi-task learning framework and can simultaneously perform device status prediction and energy injection estimation. In practical applications, considering data in both the active power and reactive power dimensions, the active power sequence and reactive power sequence are extracted from the meter data, which can comprehensively reflect the working status of electrical appliances and the operation of the power system. Subsequently, the active power sequence and reactive power sequence are input into the trained hybrid non-intrusive load monitoring model. The model concatenates the features in the active power sequence and the features in the reactive power sequence to obtain a comprehensive feature sequence. Then, the long-term dependencies in the comprehensive feature sequence are captured, and the complex time dependencies are modeled to obtain a multi-scale feature sequence. Finally, based on the multi-scale feature sequence, the status of the electrical equipment and the energy injection of the micro-inverter are predicted simultaneously to obtain the status detection result of the electrical equipment and the prediction result of the energy injection sequence. The entire solution only needs to collect meter data to identify the status of the electrical equipment, estimate the energy injection, and obtain comprehensive and accurate monitoring results, which can be applied to modern smart home environments containing distributed energy. Moreover, the above solution is stable and robust for data from different families, different time periods, and different types of electrical appliances.

[0071] In an exemplary embodiment, the hybrid non-intrusive load monitoring model includes an encoder, a Transformer model, a sequence-to-point module, and a sequence-to-sequence module.

[0072] In this embodiment, the overall architecture of the HybridNILM model includes the following main components:

[0073] (1) Encoders, such as CNN encoders (CNNEncoders, convolutional neural network encoders, simply referred to as convolutional encoders) and RNN encoders (recurrent neural networks), etc.

[0074] (2) The Transformer model (i.e., the Transformer model), where the Transformer model includes a Transformer encoder (TransformerEncoder) and a Transformer decoder (TransformerDecoder).

[0075] (3) Task-specific projection layers, including the Seq2Point (sequence-to-point) module and the Seq2Seq (sequence-to-sequence) module. The Seq2Point module is used to process the status detection of electrical equipment, and the Seq2Seq module is used to predict the energy injection amount.

[0076] Such as Figure 4As shown, in some embodiments, S420 includes S422, which calls a first convolutional encoder to perform one-dimensional convolution on the active power sequence to extract the active power features in the active power sequence, and calls a second convolutional encoder to perform one-dimensional convolution on the reactive power sequence to extract the reactive power features in the reactive power sequence features.

[0077] In this embodiment, the encoder is taken as a CNN encoder for illustration. Each input feature channel of the CNN encoder has a separate one-dimensional CNN encoder for extracting local time patterns. Each CNN encoder contains multiple one-dimensional convolutional layers, and there is a LayerNorm operation after each layer. The design of the CNN encoder allows the model to learn local patterns specific to each feature, enhancing the model's ability to capture the unique characteristics of different electrical measurement features.

[0078] The mathematical expression of the CNN encoder is as follows:

[0079]

[0080] Where, is the th feature channel, is the corresponding encoded representation.

[0081] Specifically in implementation, the server calls a first convolutional encoder to perform one-dimensional convolution on the active power sequence to extract the active power features in the active power sequence, and calls a second convolutional encoder to perform one-dimensional convolution on the reactive power sequence to extract the reactive power features in the reactive power sequence features.

[0082] Among them, each CNN encoder consists of multiple one-dimensional convolutional layers, layer normalization layers, and activation functions. The specific processing process is as follows:

[0083] One-dimensional convolution operation: For each power sequence, the CNN encoder first performs a one-dimensional convolution operation on the input data to extract the local features of the sequence. The size of the convolutional kernel is usually set to 5, and the number of channels of the convolutional layer is set to 64. For the input sequence ( is the sequence length), the convolution operation is calculated as:

[0084]

[0085] Where, is the convolutional kernel weight of the th layer, is the bias term, represents the convolution operation.

[0086] Layer Normalization: The output after convolution undergoes Layer Normalization to accelerate training and improve the stability of the model. The mathematical expression of Layer Normalization is:

[0087]

[0088] where and are the mean and standard deviation of respectively, and and are learnable parameters.

[0089] Activation Function: After each convolutional layer, the ReLU activation function is used to perform a non-linear transformation on the output. The mathematical expression of the ReLU activation function is:

[0090]

[0091] After the above processing, the feature tensors of active power and reactive power can be obtained, with the shape of (batch size, number of channels, sequence length). Then, these two feature tensors are concatenated in the feature dimension (i.e., the number of channels dimension) to form a comprehensive feature representation, namely the comprehensive sequence feature, with the shape of (batch size, 2 * number of channels, sequence length).

[0092] In this embodiment, two designed one-dimensional convolutional encoders are called to process the active power and reactive power sequences respectively, which can not only efficiently extract features but also significantly improve the performance of the system. Moreover, this architecture can adjust the number of convolutional layers, the size of convolutional kernels, and other hyperparameters according to specific task requirements, providing great flexibility.

[0093] As Figure 4 shown, in an exemplary embodiment, S440 includes: S442, inputting the comprehensive feature sequence into the Transformer model, and the Transformer model captures the long-term dependencies in the comprehensive feature sequence through the multi-head attention mechanism to obtain the multi-scale feature sequence.

[0094] Continuing from the above embodiment, the Transformer model includes a Transformer encoder and a Transformer decoder. The structure of the Transformer decoder is similar to that of the encoder but includes an additional cross-attention layer. After concatenating to obtain the comprehensive sequence feature, the comprehensive sequence feature can be input into the Transformer encoder, and the Transformer encoder uses the multi-head self-attention mechanism to model the temporal dependencies in the sequence. The specific process is as follows:

[0095] Multi-Head Self-Attention Mechanism: The core of the Transformer encoder is the multi-head self-attention mechanism, which allows the model to focus on information from other time steps in the sequence when calculating the representation at the current time step. Through multiple attention heads, the model can learn the dependencies between sequence elements from different subspaces. The calculation process of self-attention is as follows:

[0096]

[0097] Among them, are the query, key, and value matrices respectively, is the dimension of the key.

[0098] Position Encoding: Since the Transformer model itself does not contain position information, position encoding needs to be added to retain the temporal information of the sequence. The calculation method of position encoding is as follows:

[0099]

[0100] Among them, is the position, is the dimension index, is the model dimension.

[0101] Encoding Process: The Transformer encoder processes the input comprehensive feature sequence to generate a new sequence representation, namely the multi-scale feature sequence.

[0102] In this embodiment, through the multi-head attention mechanism, global temporal dependencies can be captured, enabling the model to understand long-term patterns in the sequence. Through the processing of the Transformer encoder, the model can better capture the hidden features in the active power and reactive power sequences, providing a more effective feature representation for subsequent prediction tasks.

[0103] As Figure 4 shown, in an exemplary embodiment, S460 includes:

[0104] S462, input the multi-scale feature sequence output by the converter model into the sequence-to-point module, and the sequence-to-point module maps the last time step of the multi-scale feature sequence to the power consumption device state prediction task through a fully connected layer to obtain the state detection result of the power consumption device.

[0105] S464, input the multi-scale feature sequence output by the converter model into the sequence-to-sequence module, and the sequence-to-sequence module maps the multi-scale features of each time step in the multi-scale feature sequence to the energy injection prediction task through a fully connected layer to obtain the energy injection sequence prediction result.

[0106] In this embodiment, the model has a multi-task learning framework. In order to enable the model to simultaneously handle the tasks of predicting the state of electrical equipment and estimating energy injection, a hybrid method combining the sequence-to-point (Seq2Point) and sequence-to-sequence (Seq2Seq) strategies is adopted. Among them, the Seq2Point method is mainly used to predict the state of an electrical device at a certain time point from a multi-scale feature sequence, and the Seq2Seq method is used to predict energy injection, generating an output sequence that matches the length of the input sequence, that is, the prediction result of the energy injection sequence. The prediction result of the energy injection sequence may include the predicted energy injection amount within a preset time period (such as every minute), as well as a possible confidence interval.

[0107] Specifically, when implementing, the multi-scale feature sequence output by the Transformer decoder is input into the Seq2Point module and the Seq2Seq module. Since the goal of Seq2Point is to predict a single output point based on the entire input sequence, only the information of the last time step of the output of the Transformer decoder is used. Specifically, the features of the last time step of the multi-scale feature sequence are mapped to the task of predicting the state of electrical equipment through a fully connected layer, that is, mapped to a low-dimensional space, and the states of each electrical appliance at the current moment are predicted, directly obtaining the prediction data of the state of the electrical equipment.

[0108] Specifically, in the Seq2Point module, the feature vector of the last time step output by the Transformer decoder is input into the fully connected layer for projection to obtain the state prediction of each device. The mathematical expression is:

[0109] The mathematical expression is:

[0110]

[0111] Among them, is the last time step of and are learnable parameters, is the sigmoid activation function. Since the state prediction of the electrical equipment is a probability value representing the on / off state of the device (0 or 1). Using the Sigmoid function can map the output to between (0, 1), which is suitable for probability prediction of binary classification problems.

[0112] At the same time, the Seq2Seq module processes the entire multi-scale feature sequence. Through the fully connected layer and the sigmoid activation function, the multi-scale features of each time step in the multi-scale feature sequence are mapped to the energy injection prediction task, predicting the future energy injection sequence, and obtaining the normalized prediction result of the energy injection sequence. The mathematical expression is:

[0113]

[0114] Among them, is the predicted energy injection sequence, and are learnable parameters.

[0115] In the Seq2Seq module, after the output of the Transformer decoder passes through the fully connected layer, the Sigmoid activation function is used to compress the predicted value between 0 and 1. The mathematical expression for this step is:

[0116]

[0117] Among them, is the Sigmoid activation function, and are the weights and biases of the fully connected layer.

[0118] Through the Sigmoid function, the output of the model can be mapped between (0, 1), which is convenient for subsequent mapping of continuous predicted values to predefined energy levels.

[0119] In this embodiment, by combining the Seq2Point and Seq2Seq strategies, the model can handle both the power consumption device state detection and energy injection prediction tasks simultaneously, improving the overall performance of the model.

[0120] In the NILM project, for the integration of multiple tasks, it is necessary to evaluate the performance of the power consumption device state detection task and energy injection estimation and optimize it by designing a loss function. Therefore, in some embodiments, a series of evaluation metrics applicable to the power consumption device state detection task (classification task) and energy injection decomposition (regression task) can also be adopted, and a corresponding multi-task loss function is designed. Specifically, for the power consumption device state detection task metrics, the following classification evaluation metrics are used:

[0121] 1. Accuracy:

[0122]

[0123] 2. Precision:

[0124]

[0125] 3. Recall:

[0126]

[0127] 4. F1 score:

[0128]

[0129] 5. False Alarm Rate (FAR):

[0130]

[0131] Among them, TP is the true positive, TN is the true negative, FP is the false positive, and FN is the false negative. Among them, from a numerical perspective, 1 represents the on state and 0 represents the off state.

[0132] These metrics provide a comprehensive performance evaluation of the model when identifying the electrical appliance status. Among them, the newly added false alarm rate metric is used to evaluate the reliability of the model in practical applications. In this way, the possibility of unnecessary intervention or incorrect energy management strategies caused by incorrect detection of the electrical appliance on state can be reduced.

[0133] For the energy injection estimation metric, the following regression evaluation metrics are adopted:

[0134] 1. Root Mean Square Error (RMSE):

[0135]

[0136] 2. Mean Absolute Error (MAE):

[0137]

[0138] 3. Normalized Root Mean Square Error (NRMSE):

[0139]

[0140] Among them, is the time step is the true value, is the corresponding predicted value, is the total number of time steps, is the average value of the true values.

[0141] To optimize both the electrical appliance status detection task and the energy injection estimation task simultaneously, in this embodiment, a multi-task loss function is also designed:

[0142]

[0143] Among them, and are weight coefficients used to balance the importance of the two tasks.

[0144] For the task of detecting the state of electrical equipment, which is a multi-classification task, the most commonly used metric is the cross-entropy loss function. For a single sample, the cross-entropy loss is defined as follows:

[0145]

[0146] where, is the number of classes, is the one-hot encoding of the true label, is the predicted probability of the model for class

[0147] For a batch of data, the average cross-entropy loss is usually used:

[0148]

[0149] where, is the batch size.

[0150] The cross-entropy loss is effective in dealing with multi-classification problems because it directly optimizes the difference between the predicted probability distribution and the true distribution.

[0151] In addition, in some other embodiments, the Dice loss function can also be selected, which performs well in dealing with imbalanced classes:

[0152]

[0153] where, is the predicted probability of the model for the th sample, is the corresponding true label (0 or 1), is the number of samples, is a small constant used to prevent the denominator from being zero.

[0154] For the task of energy injection estimation, a combination of the mean squared error (MSE) loss and the mean absolute error (MAE) loss can be used:

[0155]

[0156]

[0157]

[0158] where, is a weight coefficient between 0 and 1, used to balance the contributions of MSE and MAE. MSE is more sensitive to large errors and helps with fast convergence, while MAE is more robust to outliers.

[0159] ​In the above embodiments, through the design of such a multi-task loss function, the model can optimize the electrical state recognition and energy injection estimation tasks simultaneously, improving the generalization ability and robustness of the model while maintaining high accuracy.

[0160] As Figure 5 shown, in an exemplary embodiment, after S400, the method further includes:

[0161] S600, normalize the predicted energy injection sequence results to obtain the normalized predicted energy injection sequence results, perform adaptive filtering on the normalized predicted energy injection sequence results, map the normalized predicted energy injection sequence results to predefined power output levels, and determine the output power level of the micro-inverter.

[0162] Adaptive filtering methods are a class of algorithms that can dynamically adjust their parameters (such as filter coefficients) according to the statistical characteristics of the input signal.

[0163] In this embodiment, to solve the problem of numerical instability in energy injection prediction and make the prediction results more consistent with the actual output mode of the micro-inverter, an adaptive filtering algorithm is introduced. The core idea of this algorithm is to map continuous energy injection prediction values to predefined discrete output levels and eliminate short-term fluctuations through adaptive smoothing.

[0164] Specifically, when implementing, after obtaining the predicted energy injection sequence results, the following processing is performed:

[0165] Normalization: First, divide the output prediction value of the micro-inverter (the predicted energy injection amount, hereinafter referred to as the prediction value) by the preset maximum power (determined by the nameplate parameters of the micro-inverter) to make it fall within the range of [0, 1]. This not only solves the problem of numerical instability but also enables the prediction value to directly correspond to different power output levels.

[0166] Quantization: Map the continuous prediction values to predefined discrete levels. This reflects the characteristic that actual micro-inverters usually have several fixed output power levels. The power output levels correspond one-to-one with the output power levels.

[0167] Adaptive smoothing: Identify the abnormal points in the sequence (points with too large a difference from adjacent points) and adjust them to the average value of adjacent points. Iteratively perform this process until no more abnormal points are identified or the maximum number of iterations is reached.

[0168] Re-quantization: After each smoothing operation, the prediction result is re-quantized to the nearest bin to ensure that the output always conforms to the predefined output bins. The power output bins can be [0, 0.2, 0.3, 0.4, 0.7, 0.8, 1], corresponding to 7 different output power levels of the micro-inverter. In this way, discrete energy injection estimates are obtained. For example, bin "0.2" can correspond to an energy injection estimate of 100W, and bin "0.3" can correspond to an energy injection estimate of 500W.

[0169] In this embodiment, adaptive filtering of the energy injection sequence prediction result can effectively eliminate short-term fluctuations and outliers, improve the reliability of the prediction result, and also enable the energy injection prediction value to directly correspond to different power output bins of the micro-inverter, providing important support for the application of the NILM system in smart homes containing renewable energy.

[0170] Data preprocessing is a key step to ensure data quality and consistency. As Figure 5 shown, in an exemplary embodiment, S200 includes:

[0171] S220, preprocess the meter data, and extract the active power sequence and reactive power sequence with a fixed length from the preprocessed meter data through a sliding window method.

[0172] In practical applications, after obtaining the meter data, data preprocessing will be performed on the meter data. The preprocessing includes the following aspects:

[0173] 1. Data cleaning: Handle missing values and outliers, solve the problem of discontinuous timestamps, and handle negative value problems (check for micro-inverter injection or reversed line connections).

[0174] 2. Time alignment: Ensure that the timestamps of the main meter and each sub-meter are aligned, and use interpolation methods to handle the problem of inconsistent sampling rates.

[0175] 3. Loadtype encoding: Use the encode_new_load_type_C function to convert the Loadtype string into the binary states of each electrical appliance.

[0176] 4. Data resampling: Resample the data to a 2-second interval to ensure data consistency.

[0177] Subsequently, through feature engineering, active power and reactive power are extracted from the preprocessed meter data to obtain the active power sequence and reactive power sequence.

[0178] To capture the dynamic characteristics of time series, a sliding window method is used to create input sequences of fixed length. For example, the sequence lengths of the active power sequence and the reactive power sequence are set to 300 time steps (corresponding to 10 minutes of data), and the step size is usually set to 1, indicating that only one data point is moved forward each time. Subsequently, the sliding window is applied to create an active power sequence and an active power sequence with a fixed length of 300 time steps.

[0179] In this embodiment, the fixed-length active power sequence and active power sequence obtained through the sliding window can help better capture the power consumption patterns and their changes.

[0180] In some embodiments, it also includes detecting abnormal events according to abnormal rules and identifying abnormal events.

[0181] Specifically, the following abnormal rules may be included:

[0182] (1) Excessive energy consumption peak: Observe whether the real-time energy consumption of the electrical appliance exceeds 1.1 times the historical maximum;

[0183] (2) Excessive cumulative energy consumption: Statistically determine whether the daily energy consumption exceeds 1.02 times the historical value;

[0184] (3) Excessive energy consumption fluctuation: Statistically determine whether the daily energy consumption fluctuation range exceeds 1.05 times the historical value;

[0185] (4) Excessive single-time startup duration: Observe whether the real-time single-use duration of the electrical appliance exceeds 1.5 times the historical maximum;

[0186] (5) Excessive cumulative startup duration: Statistically determine whether the daily usage duration exceeds 1.2 times the historical value;

[0187] (6) Abnormal startup time: Observe whether the startup time of the electrical appliance is earlier than the earliest in history, or the shutdown time is later than the latest in history;

[0188] (7) Excessive On - off frequency: Statistically determine whether the daily On - off frequency exceeds 2 times the historical value.

[0189] Specifically, based on the above abnormal rules, the power consumption time of each electrical device can be detected to check for the existence of abnormal events, and the abnormal event detection result can be obtained. If an abnormal event is detected, a corresponding prompt message can be pushed to notify the operation and maintenance personnel to handle it in a timely manner.

[0190] To provide a clearer description of the non - intrusive load monitoring method provided in this application, a specific embodiment is described below. This specific embodiment includes the following content:

[0191] S1, Obtain the total electricity meter data.

[0192] The total electricity meter data includes the power consumption data of an air purifier, an electric heater, multiple light bulb groups, an air compressor, and an air conditioner. These data reflect the different usage frequencies of various electrical appliances in the dataset.

[0193] S2. Preprocess the total electricity meter data. By using the sliding window method, extract the active power sequence and reactive power sequence with a fixed length from the preprocessed electricity meter data.

[0194] S3. Using the active power sequence and reactive power sequence as inputs, call the trained HybridNILM model to obtain the state detection results of electrical appliances and the prediction results of the energy injection sequence.

[0195] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these steps or stages is not necessarily sequential either, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0196] Based on the same inventive concept, the embodiments of the present application also provide a non-intrusive load monitoring device for implementing the non-intrusive load monitoring method described above. The implementation solutions provided by this device to solve problems are similar to the implementation solutions described in the above method. Therefore, the specific limitations in one or more embodiments of the non-intrusive load monitoring device provided below can refer to the limitations on the non-intrusive load monitoring method in the above text, and will not be elaborated here.

[0197] In an exemplary embodiment, as Figure 6 shown, a non-intrusive load monitoring device 600 is provided, including: a data extraction module 610 and a data processing module 620, where:

[0198] The data extraction module 610 is used to extract the active power sequence and reactive power sequence from the electricity meter data.

[0199] The data processing module 620 is used to use the active power sequence and reactive power sequence as inputs, call the trained hybrid non-intrusive load monitoring model, and obtain the state detection results of electrical appliances and the prediction results of the energy injection sequence.

[0200] When the hybrid non-intrusive load monitoring model is called, the data processing module 620 performs the following steps: extracting the active power features in the active power sequence and the reactive power features in the reactive power sequence, concatenating the active power features and the reactive power features to obtain a comprehensive feature sequence, capturing the long-term dependencies in the comprehensive feature sequence to obtain a multi-scale feature sequence, and based on the multi-scale feature sequence, predicting the state of the electrical equipment and the energy injection of the micro-inverter respectively to obtain the state detection result of the electrical equipment and the prediction result of the energy injection sequence; wherein, the hybrid non-intrusive load monitoring model is trained based on the historical active power sequence and the historical reactive power sequence.

[0201] In an exemplary embodiment, the encoder includes a first convolutional encoder and a second convolutional encoder; the data processing module 620 is further configured to call the first convolutional encoder to perform one-dimensional convolution on the active power sequence to extract the active power features in the active power sequence, and call the second convolutional encoder to perform one-dimensional convolution on the reactive power sequence to extract the reactive power features in the reactive power sequence features.

[0202] In an exemplary embodiment, the data processing module 620 is further configured to input the comprehensive feature sequence into a transformer model, and the transformer model captures the long-term dependencies in the comprehensive feature sequence through a multi-head attention mechanism to obtain a multi-scale feature sequence.

[0203] In an exemplary embodiment, the data processing module 620 is further configured to input the multi-scale feature sequence output by the transformer model into a sequence-to-point module, and the sequence-to-point module maps the last time step of the multi-scale feature sequence to the electrical equipment state prediction task through a fully connected layer to obtain the state detection result of the electrical equipment, input the multi-scale feature sequence output by the transformer model into a sequence-to-sequence module, and the sequence-to-sequence module maps the multi-scale features of each time step in the multi-scale feature sequence to the energy injection prediction task through a fully connected layer to obtain the prediction result of the energy injection sequence.

[0204] In an exemplary embodiment, the data processing module 620 is further configured to perform normalization processing on the prediction result of the energy injection sequence to obtain a normalized prediction result of the energy injection sequence, perform adaptive filtering on the normalized prediction result of the energy injection sequence, map the normalized prediction result of the energy injection sequence to a predefined power output gear, and determine the output power level of the micro-inverter, and the power output gear corresponds to the output power level one by one.

[0205] In an exemplary embodiment, the data extraction module 610 is further configured to preprocess the meter data, and extract a fixed-length active power sequence and a reactive power sequence from the preprocessed meter data through a sliding window method.

[0206] Each module in the above non-intrusive load monitoring device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to each of the above modules.

[0207] In an exemplary embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 7 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store electric meter data, power consumption device status detection data, and energy injection prediction sequence result data, etc. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it implements a non-intrusive load monitoring method.

[0208] Those skilled in the art can understand that Figure 7 the structure shown in

[0209] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0210] In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps in any one of the above non-intrusive load monitoring method embodiments.

[0211] In an embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, it implements the steps in any one of the above non-intrusive load monitoring method embodiments.

[0212] It should be noted that the user information involved in this application (including but not limited to user device information such as electricity meter data, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0213] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0214] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this application.

[0215] The above-described embodiments only express several implementation manners of this application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of this application. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several deformations and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this application should be subject to the appended claims.

Claims

1. A non-intrusive load monitoring method, characterized in that, The method includes: Extracting the active power sequence and the reactive power sequence from the electricity meter data; Using the active power sequence and the reactive power sequence as inputs, calling the trained hybrid non-intrusive load monitoring model to obtain the state detection result of the electrical equipment and the energy injection sequence prediction result; When the hybrid non-intrusive load monitoring model is called, the following steps are executed: Extracting the active power feature in the active power sequence and the reactive power feature in the reactive power sequence, and splicing the active power feature and the reactive power feature to obtain a comprehensive feature sequence; Capturing the long-term dependencies in the comprehensive feature sequence to obtain a multi-scale feature sequence; Inputting the multi-scale feature sequence into a sequence-to-point module to predict the state of the electrical equipment to obtain the state detection result of the electrical equipment, and inputting the multi-scale feature sequence into a sequence-to-sequence module to predict the energy injection of the micro-inverter to obtain the energy injection sequence prediction result; Performing normalization processing on the energy injection sequence prediction result, and mapping the normalized energy injection sequence prediction result to a predefined power output gear to determine the output power level of the micro-inverter, where the output gear corresponds one-to-one to the output power level; Among them, the hybrid non-intrusive load monitoring model is trained based on the historical active power sequence and the historical reactive power sequence.

2. The method according to claim 1, characterized in that The hybrid non-intrusive load monitoring model further includes an encoder and a transformer model.

3. The method according to claim 2, wherein The encoder includes a first convolutional encoder and a second convolutional encoder; the extracting the active power feature in the active power sequence and the reactive power feature in the reactive power sequence includes: Calling the first convolutional encoder to perform one-dimensional convolution on the active power sequence to extract the active power feature in the active power sequence; Calling the second convolutional encoder to perform one-dimensional convolution on the reactive power sequence to extract the reactive power feature in the reactive power sequence feature.

4. The method according to claim 2, characterized in that The capturing the long-term dependencies in the comprehensive feature sequence to obtain a multi-scale feature sequence includes: Inputting the comprehensive feature sequence into the transformer model, and the transformer model captures the long-term dependencies in the comprehensive feature sequence through the multi-head attention mechanism to obtain a multi-scale feature sequence.

5. The method according to claim 2, wherein The inputting the multi-scale feature sequence into a sequence-to-point module to predict the state of the electrical equipment to obtain the state detection result of the electrical equipment, and inputting the multi-scale feature sequence into a sequence-to-sequence module to predict the energy injection of the micro-inverter to obtain the energy injection sequence prediction result includes: Inputting the multi-scale feature sequence output by the transformer model into the sequence-to-point module, and the sequence-to-point module maps the last time step of the multi-scale feature sequence to the electrical equipment state prediction task through a fully connected layer to obtain the state detection result of the electrical equipment; Inputting the multi-scale feature sequence output by the transformer model into the sequence-to-sequence module, and the sequence-to-sequence module maps the multi-scale features of each time step in the multi-scale feature sequence to the energy injection prediction task through a fully connected layer to obtain the energy injection sequence prediction result.

6. The method according to any one of claims 1 to 5, characterized in that After normalizing the prediction result of the energy injection sequence, the method further includes: Performing adaptive filtering on the normalized prediction result of the energy injection sequence, mapping the filtered prediction result of the energy injection sequence to a predefined power output gear, and determining the output power level of the micro-inverter.

7. The method according to any one of claims 1 to 5, characterized in that, The extracting the active power sequence and the reactive power sequence from the meter data includes: Preprocessing the meter data; By using a sliding window method, extracting the active power sequence and the reactive power sequence with a fixed length from the preprocessed meter data.

8. A non-invasive load monitoring device, characterized in that, The device includes: A data extraction module for extracting the active power sequence and the reactive power sequence from the meter data; A data processing module for using the active power sequence and the reactive power sequence as inputs, calling a trained hybrid non-intrusive load monitoring model, and obtaining the state detection result of the electrical equipment and the prediction result of the energy injection sequence; When the hybrid non-intrusive load monitoring model is called, the following steps are executed: Extracting the active power feature in the active power sequence and the reactive power feature in the reactive power sequence, and splicing the active power feature and the reactive power feature to obtain a comprehensive feature sequence; Capturing the long-term dependencies in the comprehensive feature sequence to obtain a multi-scale feature sequence; Inputting the multi-scale feature sequence into a sequence-to-point module to predict the state of the electrical equipment, obtaining the state detection result of the electrical equipment, and inputting the multi-scale feature sequence into a sequence-to-sequence module to predict the energy injection of the micro-inverter, obtaining the prediction result of the energy injection sequence; Normalizing the prediction result of the energy injection sequence, and mapping the normalized prediction result of the energy injection sequence to a predefined power output gear, and determining the output power level of the micro-inverter, where the output gear corresponds one-to-one to the output power level; Wherein, the hybrid non-intrusive load monitoring model is trained based on the historical active power sequence and the historical reactive power sequence.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images