Load identification method and system

Abstract

The invention discloses a load identification method and system. The method comprises the following steps: acquiring parameter data of a household appliance in steady-state operation; constructing a first composite feature based on the parameter data; and carrying out load identification according to the first composite feature in combination with a preset first load identification model. Through fusion of time domain and frequency domain features, operation features of different loads are comprehensively represented, and the identification precision is improved; forward and reverse features of load data can be effectively captured by adopting the BiLSTM model, and the processing capability of complex time series data is enhanced; a multi-head attention mechanism (MA) is introduced, a higher weight is given to key load features in the feature selection process, and the stability and generalization ability of the model are improved. The method is lightweight in design, is suitable for an edge computing environment, and can guarantee the recognition precision and reduce the computing overhead under the condition of limited resources.

Description

Technical Field

[0001] The present invention relates to the technical field of non-invasive load identification, and in particular to a load identification method and system. Background Art

[0002] In the power system, load identification technology can help power companies and users monitor the operating status of power equipment, optimize power management, and achieve energy efficiency improvement and energy conservation and emission reduction. At present, non-intrusive load identification (NILM) technology has become a research hotspot. This technology analyzes the overall power consumption signals of power users and infers the working status and power consumption of each power device, avoiding the installation complexity and high cost of traditional invasive methods. However, since power load data often has highly nonlinear, strong noise and multi-scale characteristics, traditional load identification methods are limited in recognition accuracy and generalization ability.

[0003] In recent years, analysis methods based on the combination of time domain and frequency domain features have been widely used in the field of load identification. Time domain features can directly describe the dynamic process of load changes, while frequency domain features can reflect the working mode and periodic characteristics of the equipment. The effective fusion of time and frequency domain features can more comprehensively characterize the power consumption behavior of the equipment, thereby improving the accuracy of identification. In addition, deep learning models, especially long short-term memory networks (LSTM) and their improved forms, have been proven to have significant advantages in processing time series data. Bidirectional LSTM (BiLSTM) can capture the global information of time series data through forward and backward learning, effectively improving the accuracy of load identification.

[0004] Nevertheless, there is still room for improvement in the recognition efficiency and accuracy of existing methods in edge computing environments. Summary of the invention

[0005] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0006] In view of the above existing problems, the present invention is proposed.

[0007] Therefore, the present invention provides a load identification method and system, which can solve the problems mentioned in the background technology.

[0008] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0009] In a first aspect, the present invention provides a load identification method, comprising:

[0010] Obtain parameter data of household appliances during steady-state operation;

[0011] constructing a first composite feature based on the parameter data;

[0012] Load identification is performed based on the first composite feature and in combination with a preset first load identification model.

[0013] As a preferred solution of the load identification method of the present invention, the constructing of the first composite time-frequency domain feature based on the parameter data includes:

[0014] The parameter data at least includes voltage data and current data of the household appliance when it is in steady-state operation;

[0015] Calculate a first feature, a second feature, and a third feature according to the parameter data;

[0016] The first feature, the second feature and the third feature are integrated to obtain a first composite feature.

[0017] As a preferred solution of the load identification method of the present invention, the first load identification model includes:

[0018] The first load identification model is any model whose input is the first composite feature and whose output is the load identification result or load identification related parameters;

[0019] The load identification related parameters are any parameters that can be used to directly or indirectly obtain load identification results.

[0020] As a preferred solution of the load identification method of the present invention, the calculation of the first feature, the second feature and the third feature according to the parameter data includes:

[0021] The first feature is any one of a time domain feature, a frequency domain feature, and a time-frequency domain feature;

[0022] When the first feature is a time domain feature, the time domain feature includes a current average value, a current peak value, a current standard deviation, a current rectified average value, a current effective value and active power;

[0023] The time domain characteristics are calculated based on the parameter data.

[0024] As a preferred solution of the load identification method of the present invention, the calculation of the first feature, the second feature and the third feature according to the parameter data further includes:

[0025] The second feature is any one of a time domain feature, a frequency domain feature, and a time-frequency domain feature, and is not repeated with the first feature;

[0026] When the second feature is a frequency domain feature, the frequency domain feature obtains several sub-frequency domain components by performing Fourier decomposition on the current signal;

[0027] The frequency domain features are calculated based on the parameter data.

[0028] As a preferred solution of the load identification method of the present invention, the calculation of the first feature, the second feature and the third feature according to the parameter data further includes:

[0029] The third feature is any one of a time domain feature, a frequency domain feature, and a time-frequency domain feature, and is not repeated with the first feature or the second feature;

[0030] When the third feature is a time-frequency domain feature, the time-frequency domain feature includes a wavelet packet energy contribution factor obtained by wavelet packet transformation;

[0031] The time-frequency domain features are calculated based on the parameter data.

[0032] As a preferred solution of the load identification method of the present invention, the step of obtaining parameter data of the household appliance during steady-state operation includes:

[0033] Use inductive voltage transformers, inductive current transformers and oscilloscopes to obtain parameter data of household appliances during steady-state operation;

[0034] The inductive voltage transformer and the inductive current transformer have an accuracy level not lower than level 1.0, and the sampling frequency of the oscilloscope should be not less than 10KHz.

[0035] In a second aspect, the present invention provides a load identification system, comprising:

[0036] A data acquisition module, used to acquire parameter data of household appliances during steady-state operation;

[0037] A feature acquisition module, used for constructing a first composite feature based on the parameter data;

[0038] The load identification module is used to perform load identification according to the first composite feature in combination with a preset first load identification model.

[0039] In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned method when executing the computer program.

[0040] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method described above when executed by a processor.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention proposes a load identification method and system to obtain parameter data of household appliances during steady-state operation; construct a first composite feature based on the parameter data; and perform load identification based on the first composite feature in combination with a preset first load identification model. By integrating time domain and frequency domain features, the operating characteristics of different loads are comprehensively characterized to improve the recognition accuracy; the BiLSTM model can effectively capture the forward and reverse features of load data, and enhance the processing capability of complex time series data; the multi-head attention mechanism (MA) is introduced to give higher weights to key load features during feature selection, thereby improving the stability and generalization capability of the model. The method is lightweight in design and suitable for edge computing environments. It can ensure recognition accuracy and reduce computing overhead under limited resources. The method has good real-time performance, can quickly respond to load identification needs, and meet the requirements of the power system for real-time monitoring and response, providing efficient and accurate technical support for power grid management and operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0043] Figure 1 A method flow chart of a load identification method and system provided by an embodiment of the present invention;

[0044] Figure 2 An internal structural diagram of a computer device of a load identification method and system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0046] Example 1

[0047] Reference Figure 1-Figure 2 , which is the first embodiment of the present invention, provides a load identification method and system, including:

[0048] There are some shortcomings in the existing related technologies, such as low recognition accuracy, limited ability to process complex time series data, and insufficient model generalization ability.

[0049] The present application provides a method that can effectively solve the above-mentioned problems. Next, how to implement the load identification method will be described in detail in combination with multiple embodiments.

[0050] Figure 1 A method flow chart of a load identification method and system is shown, including:

[0051] S101, obtaining parameter data of the household appliance during steady-state operation;

[0052] In an optional embodiment, the parameter data of the household appliance in steady-state operation may include voltage value, current value, power factor, etc. These parameter data may reflect the electrical characteristics of the household appliance in a stable working state.

[0053] In an optional embodiment, the parameter data of household appliances during steady-state operation can be obtained by using an inductive voltage transformer, an inductive current transformer, and an oscilloscope to obtain the parameter data of household appliances during steady-state operation. It can also be obtained by other means, for example, installing a smart meter in a distribution box at home or business. Smart meters collect voltage, current and other data in real time through built-in sensors. The data is transmitted to a central server or cloud platform via wired (such as power line communication PLC) or wireless (such as Wi-Fi, LoRa, NB-IoT, etc.).

[0054] In an optional embodiment, the parameter data of household appliances during steady-state operation can also be obtained by installing a PLC module in the appliance or circuit to be monitored. The PLC module collects data such as voltage and current through built-in sensors, and transmits the data to a central controller or a cloud platform through power lines.

[0055] In an optional embodiment, the parameter data of household appliances during steady-state operation can also be obtained by deploying wireless sensor nodes near the appliances or circuits that need to be monitored. The sensor nodes collect data such as voltage and current through built-in sensors. The data is transmitted to a central node or gateway through wireless communication technology (such as ZigBee, Bluetooth, Wi-Fi, etc.), and then the gateway transmits the data to a central server or cloud platform.

[0056] In an optional embodiment, the parameter data of the household appliance during steady-state operation may also be obtained through an embedded system, a cloud platform, or the like.

[0057] In the embodiment of the present application, obtaining parameter data of the household appliance during steady-state operation includes:

[0058] Use inductive voltage transformers, inductive current transformers and oscilloscopes to obtain parameter data of household appliances during steady-state operation;

[0059] In the embodiment of the present application, the inductive voltage transformer and the inductive current transformer have an accuracy level not lower than level 1.0, and the sampling frequency of the oscilloscope should be not less than 10KHz.

[0060] It should be noted that the inductive voltage transformer and the inductive current transformer have an accuracy level of no less than 1.0. The sampling frequency of the oscilloscope should be no less than 10KHz to ensure that the collected parameter data has sufficient accuracy and reliability. The accuracy level of the inductive voltage transformer and the inductive current transformer is set to no less than 1.0. Such an accuracy level can ensure that the errors of the voltage and current values ​​are controlled within a small range during the measurement process, thereby providing accurate basic data for subsequent load identification. At the same time, the sampling frequency of the oscilloscope is set to no less than 10KHz, because a higher sampling frequency can capture the rapid changes of household appliances during steady-state operation and ensure the time domain resolution of the data, which is crucial for analyzing and identifying load characteristics. Through such parameter settings, the accuracy and efficiency of load identification can be effectively improved.

[0061] S102, constructing a first composite feature based on parameter data;

[0062] In the embodiment of the present application, constructing the first composite time-frequency domain feature based on the parameter data includes:

[0063] The parameter data at least includes voltage data and current data of the household appliance when it is in steady-state operation;

[0064] Calculate the first feature, the second feature and the third feature according to the parameter data;

[0065] The first feature, the second feature and the third feature are integrated to obtain a first composite feature.

[0066] In an optional embodiment, the first composite time-frequency domain feature may include a time domain feature, a frequency domain feature, and any two or more feature combinations of time-frequency domain features. For example, a combination of time domain features and frequency domain features is selected, wherein the first feature and the second feature are selected as time domain features, and the third feature is selected as a frequency domain feature. Alternatively, the first feature and the third feature may be selected as time domain features, and the second feature may be selected as a frequency domain feature. The time domain feature may reflect the changing pattern of household appliances in a time series, while the frequency domain feature may reveal the frequency distribution characteristics of household appliances during steady-state operation. By combining these features, the operating characteristics of household appliances may be captured more comprehensively, providing richer information for load identification.

[0067] In an optional embodiment, when the first composite time-frequency domain feature needs to include three features: time domain feature, frequency domain feature and time-frequency domain feature, the first feature can be any one of the time domain feature, frequency domain feature and time-frequency domain feature, and the second feature can be any one of the time domain feature, frequency domain feature and time-frequency domain feature, and it is not repeated with the first feature, and the third feature can be any one of the time domain feature, frequency domain feature and time-frequency domain feature, and it is not repeated with the first feature and the second feature.

[0068] In the embodiment of the present application, calculating the first feature, the second feature, and the third feature according to the parameter data includes:

[0069] The first feature is any one of the time domain feature, the frequency domain feature and the time-frequency domain feature;

[0070] When the first feature is a time domain feature, the time domain feature includes a current average value, a current peak value, a current standard deviation, a current rectified average value, a current effective value and active power;

[0071] The time domain characteristics are calculated based on the parameter data.

[0072] In the embodiment of the present application, the calculation formula of the current average value is:

[0073]

[0074] Among them, I avg is the average current, T is the sampling period, i t is the instantaneous value of the current at time t;

[0075] The calculation formula for the peak current is:

[0076] I peak =max(i t ),t∈[0,T]

[0077] Among them, I peak is the current peak value, max() is the maximum value function;

[0078] The formula for calculating the current standard deviation is:

[0079]

[0080] Where σ is the current standard deviation, is the average current value;

[0081] The formula for calculating the average value of current rectification is:

[0082]

[0083] Among them, I rectavg is the average value of the rectified current, || is the symbol for taking the absolute value;

[0084] The formula for calculating the effective value of current is:

[0085]

[0086] Among them, I rms is the effective value of current;

[0087] The calculation formula for active power is

[0088]

[0089] Among them, v t is the instantaneous value of voltage at time t.

[0090] It should be noted that the time domain features selected, such as current average value, current peak value, current standard deviation, current rectified average value, current effective value and active power, can accurately reflect the current changes of household appliances at different time points. The current average value can provide the average current level during steady-state operation, while the current peak value can indicate the maximum instantaneous value of the current, which is very important for identifying the peak power consumption of the appliance. The current standard deviation, as an indicator of current volatility, can reveal the stability of household appliances. The current rectified average value can reflect the average effective value of the current, which is also important for evaluating the power consumption of appliances. Through the calculation of these time domain features, key current data can be provided for subsequent load identification and analysis.

[0091] In the embodiment of the present application, calculating the first feature, the second feature, and the third feature according to the parameter data further includes:

[0092] The second feature is any one of the time domain feature, the frequency domain feature and the time-frequency domain feature, and is not repeated with the first feature;

[0093] When the second feature is a frequency domain feature, the frequency domain feature obtains several sub-frequency domain components by performing Fourier decomposition on the current signal;

[0094] The frequency domain features are calculated based on the parameter data.

[0095] Specifically, the frequency domain component obtained after Fourier decomposition of the current signal is:

[0096]

[0097] in, is the DC component, A km cos(kω1t+φ k ) is the kth component after Fourier decomposition, and its amplitude is A km , the frequency is kω1t, the phase angle is φ k , the fundamental frequency is ω1.

[0098] It should be noted that after data collection and analysis, the frequency domain components of 7th order and above are much smaller than the frequency domain components of 6th order and below. Therefore, the frequency domain features in this application use fundamental frequency components, secondary components, cubic components, quartic components, quintic components, and sextic components.

[0099] It should be noted that the frequency domain characteristics can more accurately describe the frequency characteristics of household appliances in steady-state operation by performing Fourier decomposition on the current signal to obtain several frequency domain components. The fundamental frequency component reflects the basic operating frequency of the appliance, while the secondary component, tertiary component, etc. reveal the harmonic phenomena that may occur during the operation of the appliance. By analyzing these frequency domain components, we can have a deeper understanding of the working status and possible problems of the appliance. For example, an increase in the secondary component may indicate that the appliance has a nonlinear load, while an abnormality in the tertiary component may indicate problems such as rotor imbalance in motor-type appliances. Therefore, the analysis of frequency domain characteristics is of great significance for load identification and fault diagnosis.

[0100] In an optional embodiment, in addition to performing Fourier decomposition on the current signal to obtain several frequency domain components, the frequency domain characteristics can also be obtained by calculating the harmonic content of the current signal. By calculating the harmonic content, the relative strength of the harmonics in the current signal can be evaluated, which is very useful for identifying and diagnosing nonlinear load problems in electrical appliances.

[0101] In the embodiment of the present application, calculating the first feature, the second feature, and the third feature according to the parameter data further includes:

[0102] The third feature is any one of the time domain feature, the frequency domain feature and the time-frequency domain feature, and is not repeated with the first feature or the second feature;

[0103] When the third feature is a time-frequency domain feature, the time-frequency domain feature includes a wavelet packet energy contribution factor obtained by wavelet packet transformation;

[0104] The time-frequency domain features are calculated based on the parameter data.

[0105] Specifically, the current signal i is decomposed into n layers of wavelet packets, and the wavelet packet coefficient of the jth band of the i-th layer is recorded as d i,j , where j = 0, 1, 2, ..., 2 i -1;

[0106] Calculate the energy of each wavelet packet in the i-th layer. The calculation formula is:

[0107]

[0108] Normalized calculation, the calculation formula is:

[0109]

[0110] Among them, P i,j is d i,j The energy probability density of

[0111] The calculation formula for wavelet packet energy contribution factor is:

[0112] H i,j =-P i,j log(P i,j )

[0113] Among them, H i,j is the calculated wavelet packet energy contribution factor.

[0114] It should be noted that the time-frequency domain features, including the wavelet packet energy contribution factor obtained by wavelet packet transform, can provide the energy distribution of the current signal on different time scales. Wavelet packet transform is an analysis tool that can provide time and frequency information at the same time. It can decompose the signal into a series of wavelet packets with different frequency ranges, thereby obtaining the energy distribution of the signal in different frequency bands. By analyzing these energy contribution factors, we can have a more detailed understanding of the dynamic characteristics of household appliances during operation. For example, a sudden increase in energy in a certain frequency band may indicate that the appliance is in an abnormal working state within the frequency band. Therefore, the analysis of time-frequency domain features is also of great reference value for load identification and fault diagnosis.

[0115] In an optional embodiment, the time-frequency domain features can also be obtained in other ways, such as analyzing the time-frequency characteristics of the current signal by calculating the time-frequency spectrum of the current signal. Time-frequency spectrum analysis can reveal the frequency distribution of the current signal over time, which is very helpful for identifying the characteristics of the appliance at different operating stages. For example, the transient process when the appliance is started and the frequency change during steady-state operation can be observed through the time-frequency spectrum.

[0116] S103: Perform load identification according to the first composite feature and a preset first load identification model.

[0117] In the embodiment of the present application, the first load identification model includes:

[0118] The first load identification model is any model whose input is the first composite feature and whose output is the load identification result or load identification related parameters;

[0119] The load identification related parameters are any parameters that can be used to directly or indirectly obtain the load identification results.

[0120] In an optional embodiment, the design of the first load identification model can use a variety of machine learning algorithms, such as support vector machines (SVM), random forests, neural networks, etc. These algorithms can be trained based on the composite features of the input to identify different types of loads. For example, SVM distinguishes different types of loads by constructing a hyperplane, while random forests improve the accuracy and robustness of identification by constructing multiple decision trees. Neural networks can learn and identify complex nonlinear relationships by simulating the structure and function of neurons in the human brain.

[0121] In an optional embodiment, the design of the first load identification model can also use deep learning techniques, such as convolutional neural networks (CNN) or recurrent neural networks (RNN), to process and analyze time series data. These deep learning models can automatically extract features from the data and learn complex patterns of loads through multiple layers of nonlinear transformations. For example, CNN can capture spatial features in current signals, while RNN is good at processing time series data and can identify patterns of load changes over time. Through these advanced models, the accuracy and efficiency of load identification can be further improved.

[0122] In an embodiment of the present application, load identification is performed according to the first composite feature in combination with a preset first load identification model, wherein the first load identification model is a MA-BiLSTM model, and the basic model of the MA-BiLSTM model is a bidirectional long short-term memory (BiLSTM) model. The long short-term memory model is an improved recursive neural network model. The LSTM model consists of four network layers: a forget gate, an input gate, a cell state update gate, and an output gate. The input information is updated to the current cell state through the input gate and the forget gate, and then the output is obtained through the output gate. The calculation formula is:

[0123]

[0124] Among them, h t-1 and h t Respectively represent the unit state update gate parameters before and after the update, c t Update gate parameters for cell states, W f , W i , W c , W g They represent the weight parameters of the forget gate, input gate, unit state update gate and output gate, respectively, and b f , b i , b c , b g They represent the bias parameters of the forget gate, input gate, unit state update gate and output gate respectively.

[0125] It should be noted that the bidirectional long short-term memory (BiLSTM) model, based on the LSTM model, uses a combination of forward LSTM and backward LSTM to capture and encode forward information and backward information, which is conducive to solving the problem of gradient disappearance or gradient explosion and further improving model performance.

[0126] In the embodiment of the present application, the composite time-frequency domain features are input into the MA-BiLSTM model for load identification model. The MA-BiLSTM model adds the MA mechanism. The MA mechanism extracts the input information into several subsequences based on the attention mechanism, thereby extracting multiple attention features.

[0127] The MA mechanism comprises the following steps:

[0128] Extract the input sequence x and get x i ,i=1,2,...,n, where the extraction length is fixed to L.

[0129] Furthermore, for the input sequence x i Perform linear operations to calculate the query vector Query, key vector Key, and value vector Value. The calculation formula is:

[0130] Query = W Q ·x i

[0131] Key=W K ·x i

[0132] Value = W V ·x i

[0133] Among them, W Q , W K , W V is the weight matrix of the query vector, key vector and value vector.

[0134] Furthermore, the dot product of the query vector and the value vector is calculated to calculate the correlation between the two vectors, namely:

[0135]

[0136] Furthermore, the correlations are concatenated to obtain a vector h of the same size as the input sequence x.

[0137] Furthermore, based on the label value and the predicted value, a cross entropy loss function is constructed, and the loss function is:

[0138]

[0139] Among them, y i is the actual label value, y i is the predicted label value.

[0140] In an optional embodiment, after the model is trained, reasoning and testing are performed based on the trained MA-BiLSTM model. The method constructs a composite time-frequency domain feature integrating multiple parameters of time, frequency and time-frequency domains, and realizes effective distinction of loads with similar current signals.

[0141] In an optional embodiment, the load identification related parameters are any parameters that can be used to directly or indirectly obtain the load identification results. For example, the load identification related parameters that can directly obtain the load identification results can include the current effective value, current peak value, active power, etc. These parameters are obtained through specific calculation formulas. The load identification related parameters that can indirectly obtain the load identification results can include the shape parameters of the current waveform, the harmonic content of the current waveform, etc. These indirect parameters are obtained by analyzing the characteristics of the current waveform. They can provide additional information about the quality of the current signal and help to more accurately identify and distinguish different types of loads. For example, the shape parameters of the current waveform can reflect the degree of distortion of the current signal, while the harmonic content can reveal the frequency components contained in the signal. By combining these direct and indirect parameters, the load identification system can more comprehensively understand the characteristics of the current signal, thereby improving the accuracy and reliability of identification.

[0142] In summary, the present invention proposes a load identification method, which obtains parameter data of household appliances during steady-state operation; constructs a first composite feature based on the parameter data; and performs load identification according to the first composite feature in combination with a preset first load identification model. By integrating time domain and frequency domain features, the operating characteristics of different loads are comprehensively characterized to improve the recognition accuracy; the BiLSTM model can effectively capture the forward and reverse features of load data, and enhances the processing capability of complex time series data; the multi-head attention mechanism (MA) is introduced to give higher weights to key load features during feature selection, thereby improving the stability and generalization capability of the model. The method is lightweight in design and suitable for edge computing environments. It can ensure recognition accuracy and reduce computing overhead under limited resources. The method has good real-time performance, can quickly respond to load identification needs, and meet the requirements of the power system for real-time monitoring and response, providing efficient and accurate technical support for power grid management and operation.

[0143] This embodiment also provides a load identification system, including:

[0144] A data acquisition module, used to acquire parameter data of household appliances during steady-state operation;

[0145] A feature acquisition module, used for constructing a first composite feature based on parameter data;

[0146] The load identification module is used to perform load identification according to the first composite feature in combination with a preset first load identification model.

[0147] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0148] This embodiment also provides a computer device, which may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 2 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. 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 and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a load identification method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0149] This embodiment further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0150] Obtain parameter data of household appliances during steady-state operation;

[0151] constructing a first composite feature based on the parameter data;

[0152] Load identification is performed based on the first composite feature and in combination with a preset first load identification model.

[0153] Example 2

[0154] Reference Figure 1-Figure 2 , which is an embodiment of the present invention, provides a load identification method and system. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0155] A classical deep learning algorithm was used for comparison and a comparative test was conducted with the method of the present invention, and the test results were compared by means of scientific demonstration to verify the real effect of the method.

[0156] Classic deep learning algorithms are used for comparison, and the algorithms used are LSTM and BiLSTM. Table 1 is the test environment table, and Table 2 is the experimental result comparison table.

[0157] Test environment: As shown in Table 1, the test device 1 deploys the classic deep learning model, and the test device 2 deploys the method of the present invention. The comparison index is accuracy.

[0158] Table 1 Test environment

[0159] Test Group Test equipment 1 Test equipment 2 Test equipment 3 Algorithm Model LSTM BiLSTM This method

[0160] Table 2 Test results

[0161] Comparison indicators Test equipment 1 Test equipment 2 Test equipment 3 Accuracy / % 68.74 84.94 96.76

[0162] The experimental results show that the method of the present invention helps to improve the accuracy of abnormal power consumption behavior and further improve the recognition efficiency. The experimental results are shown in Table 2. The method of the present invention is significantly better than the traditional LSTM and BiLSTM algorithms in terms of accuracy, with an increase of 18.22% and 11.82% respectively. This shows that the method of the present invention can more accurately identify abnormal power consumption behavior when processing complex load data, providing strong technical support for the stable operation of the power grid. In addition, in practical applications, the method of the present invention can also dynamically adjust the recognition model according to the real-time data of the power grid operation to adapt to the changes in the power grid load, ensuring the flexibility and adaptability of the recognition system.

[0163] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0164] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0165] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0166] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0167] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0168] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0169] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A load identification method, characterized in that: include: Obtain parameter data of household appliances during steady-state operation; constructing a first composite feature based on the parameter data; Load identification is performed based on the first composite feature and in combination with a preset first load identification model.

2. The load identification method according to claim 1, characterized in that: The constructing the first composite time-frequency domain feature based on the parameter data comprises: The parameter data at least includes voltage data and current data of the household appliance when it is in steady-state operation; Calculate a first feature, a second feature, and a third feature according to the parameter data; The first feature, the second feature and the third feature are integrated to obtain a first composite feature.

3. The load identification method according to claim 2, characterized in that: The first load identification model includes: The first load identification model is any model whose input is the first composite feature and whose output is the load identification result or load identification related parameters; The load identification related parameters are any parameters that can be used to directly or indirectly obtain load identification results.

4. The load identification method according to claim 3, characterized in that: The calculating the first feature, the second feature and the third feature according to the parameter data comprises: The first feature is any one of a time domain feature, a frequency domain feature, and a time-frequency domain feature; When the first feature is a time domain feature, the time domain feature includes a current average value, a current peak value, a current standard deviation, a current rectified average value, a current effective value and active power; The time domain characteristics are calculated based on the parameter data.

5. The load identification method according to claim 4, characterized in that: The calculating the first feature, the second feature and the third feature according to the parameter data further comprises: The second feature is any one of a time domain feature, a frequency domain feature, and a time-frequency domain feature, and is not repeated with the first feature; When the second feature is a frequency domain feature, the frequency domain feature obtains several sub-frequency domain components by performing Fourier decomposition on the current signal; The frequency domain features are calculated based on the parameter data.

6. The load identification method according to claim 5, characterized in that: The calculating the first feature, the second feature and the third feature according to the parameter data further comprises: The third feature is any one of a time domain feature, a frequency domain feature, and a time-frequency domain feature, and is not repeated with the first feature or the second feature; When the third feature is a time-frequency domain feature, the time-frequency domain feature includes a wavelet packet energy contribution factor obtained by wavelet packet transformation; The time-frequency domain features are calculated based on the parameter data.

7. The load identification method according to claim 6, characterized in that: The method of obtaining parameter data of the household appliance during steady-state operation includes: Use inductive voltage transformers, inductive current transformers and oscilloscopes to obtain parameter data of household appliances during steady-state operation; The inductive voltage transformer and the inductive current transformer have an accuracy level not lower than level 1.0, and the sampling frequency of the oscilloscope should be not less than 10KHz.

8. A load identification system, characterized in that: include: A data acquisition module, used to acquire parameter data of household appliances during steady-state operation; A feature acquisition module, used for constructing a first composite feature based on the parameter data; The load identification module is used to perform load identification according to the first composite feature in combination with a preset first load identification model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: 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 a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images