Non-intrusive dormitory electricity safety monitoring method, system and storage medium

Abstract

The present application discloses a non-invasive dormitory electricity safety monitoring method, system and storage medium, the method comprising: sampling the total voltage, total current and total switch state. Multiply the total voltage and total current to obtain the total useful power. The power decomposition characteristics in the total useful power and the switch state are extracted through a power decomposition model, and a number of target electrical appliance types and corresponding decomposed powers are determined according to the power decomposition characteristics. If the target electrical appliance type and the corresponding decomposed power interval exist in the disabled electrical appliance table, an illegal electricity use warning is issued. By performing load power decomposition on the total power consumption of the main line, the power-consuming electrical appliances in the dormitory are predicted, and the disabled electrical appliances are determined. The electricity consumption data of the dormitory can be obtained without physical intervention in the circuit, and the electrical appliances in each dormitory are analyzed to improve the safety of dormitory electricity use.

Description

Technical Field

[0001] The present application relates to the field of power grid technology, and in particular to a non-intrusive dormitory electricity safety monitoring method, system and storage medium. Background Art

[0002] The safety of dormitory electricity use is related to the safety of life and property, and is extremely important. The existing dormitory electricity safety management method generally adopts the method of indoor total power control. When the total power in the dormitory exceeds the preset power threshold, the power is automatically cut off. The existing method can only effectively control the use of prohibited electrical appliances in the dormitory by setting a very small power threshold, which causes great inconvenience to residents and users. Moreover, the dormitory electricity use can only be controlled according to the power size, but the type of electrical appliances cannot be analyzed, and it is impossible to determine whether there are really prohibited electrical appliances in the dormitory. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present application proposes a non-invasive dormitory electricity safety monitoring method, system and storage medium, which solves the problem that the dormitory electricity safety monitoring method is likely to cause inconvenience to users and cannot analyze electrical appliances.

[0004] In order to achieve the above purpose, the technical solution of this application is as follows:

[0005] A non-intrusive dormitory electricity safety monitoring method, comprising:

[0006] S1. Collecting electricity consumption data: sampling the total voltage, total current and total switch status of the current indoor electricity consumption at the distribution box at a preset sampling frequency.

[0007] S2. Preprocessing of power consumption data: If the total current exceeds the preset current range, the current total current and the corresponding total voltage and total switch status are deleted, and the total voltage, total current and total switch status are collected again.

[0008] S3. Calculate the total power: multiply the total voltage and the total current to obtain the total useful power.

[0009] S4. Determine the target electrical appliances and power: input the total useful power and the total switch state into a preset power decomposition model, extract the power decomposition features in the total useful power and the switch state through the power decomposition model, and determine several target electrical appliance types and the corresponding decomposed powers based on the power decomposition features.

[0010] S5. Warning of illegal electricity usage: traverse the preset prohibited electrical appliance table. If the target electrical appliance type and the corresponding decomposed power range exist in the prohibited electrical appliance table, a warning of illegal electricity usage is issued.

[0011] Optionally, step S2 includes the following steps S21-S24:

[0012] S21. Obtain a total voltage set and a total current set.

[0013] The total current set includes the total current of a plurality of consecutive time points in the room collected at the distribution box at a preset sampling frequency.

[0014] S22. Calculate the average value and standard deviation corresponding to the total current set.

[0015] S23. Determine the current range according to the mean value and the standard deviation.

[0016] S24. If the current total current exceeds the current range, the current total current and the corresponding total voltage and total switch state are deleted, and the total voltage, total current and total switch state are collected again.

[0017] Alternatively, the expression for the mean value μ is:

[0018]

[0019] The expression of standard deviation σ is:

[0020]

[0021] Among them, I i represents the i-th total current collected, and n represents the number of samples of the total current.

[0022] Optionally, the power decomposition model power decomposition expression is:

[0023]

[0024] Among them, y t represents the total power at time t; represents the useful power of the i-th target appliance at time t; ε t Indicates the noise composed of non-target electrical appliances; m indicates the number of target electrical appliances.

[0025] Optionally, before step S4, the following steps S041-S044 are also included:

[0026] S041. Obtain a total useful power sequence sample, a useful power sequence sample of a first target electrical appliance, and a switch state sequence sample.

[0027] The first target electrical appliance is any target electrical appliance in the target electrical appliance set.

[0028] Target appliances include prohibited appliances and permitted appliances.

[0029] S042. Label the total useful power sequence samples and the switch state sequence samples of the first target electrical appliance with the useful power sequence samples and types of the first target electrical appliance in chronological order to obtain a sample training set.

[0030] S043. Use a neural network model to establish a power decomposition model to be trained corresponding to the total useful power, the total switch state, and the useful power and type of the target electrical appliance.

[0031] S044. Take the total useful power sequence samples and the switch state sequence samples of the first target electrical appliance as input, and the useful power sequence samples and types of the first target electrical appliance as output, to train the power decomposition model to be trained; similarly, take the total useful power sequence samples and the switch state sequence samples of each target electrical appliance in the target electrical appliance set as input, and the useful power sequence samples and types of each target electrical appliance in the target electrical appliance set as output, to train the power decomposition model to be trained, until the loss function of the power decomposition model to be trained converges, thereby obtaining the power decomposition model.

[0032] Optionally, the loss function L is expressed as:

[0033]

[0034] Among them, represents the total time length of the sequence, x t represents the sample label at time t, Represents the output of the model at time t.

[0035] Optionally, the prohibited electrical appliances include electric irons, electric soldering irons, electric kettles, electric blankets, and induction cookers.

[0036] Based on the same technical concept, the present application also provides a non-intrusive dormitory electricity safety monitoring system, including:

[0037] The acquisition module is used to sample the total voltage, total current and total switch status of the current indoor power consumption at the distribution box at a preset sampling frequency.

[0038] The processing module is used to delete the current total current and the corresponding total voltage and total switch state if the total current exceeds a preset current range, and re-collect the total voltage, total current and total switch state; multiply the total voltage and the total current to obtain the total useful power; input the total useful power and the total switch state into a preset power decomposition model, extract the power decomposition characteristics in the total useful power and the switch state through the power decomposition model, and determine a number of target electrical appliance types and corresponding decomposed powers based on the power decomposition characteristics; traverse the preset disabled electrical appliance table, and if the target electrical appliance type and the corresponding decomposed power range exist in the disabled electrical appliance table, issue an illegal electricity usage warning.

[0039] Optionally, the power decomposition model power decomposition expression is:

[0040]

[0041] Among them, y t represents the total power at time t; represents the useful power of the i-th target appliance at time t; ε t Indicates the noise composed of non-target electrical appliances; m indicates the number of target electrical appliances.

[0042] Based on the same technical concept, the present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the non-intrusive dormitory electricity safety monitoring method described in the above-mentioned embodiments.

[0043] The beneficial effects of the present application are as follows: by performing load power decomposition on the total power consumption of the main line, predicting power-consuming appliances in the dormitory, and determining the disabled appliances, a simple, economical, reliable and complete electricity safety monitoring solution is provided. The electricity consumption data of the dormitory can be obtained without physical intervention in the circuit, and the electrical appliances in each dormitory can be analyzed, which helps to timely discover the disabled appliances in the dormitory and improve the electricity safety of the dormitory. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Schematic diagram of the process of the non-intrusive dormitory electricity safety monitoring method in the embodiment of the present application.

[0045] Figure 2 Schematic diagram of the structure of the non-intrusive dormitory electricity safety monitoring system in an embodiment of the present application.

[0046] in, Figure 2 The figures in the figure are marked as: 1-acquisition module, 2-processing module. DETAILED DESCRIPTION

[0047] 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.

[0048] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, procedures, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, procedures, steps, operations, elements, components and / or groups thereof.

[0049] Figure 1This is a flow chart of a non-intrusive dormitory electricity safety monitoring method in some implementations of the present application, such as Figure 1 As shown, the following steps S1-S5 may be included:

[0050] S1. Collecting electricity consumption data: sampling the total voltage, total current and total switch status of the current indoor electricity consumption at the distribution box at a preset sampling frequency.

[0051] The main switch state includes the main circuit disconnection or conduction state of the circuit, as well as the time point and duration of the main circuit conduction.

[0052] The distribution box is a circuit distribution box for all users. A collector can be installed at the distribution box to collect the voltage and current data of dormitory electricity. Generally, the total voltage of electricity consumption is the mains voltage. With the help of different types of current transformers, the collection needs of different current levels can be met, and they can also be installed and deployed without power outages. When there are electrical appliances in a dormitory that are powered on, current is generated at the distribution box, that is, on the main road. At this time, the main road switch state is considered to be in the on state, and the time point and duration of the main road on are recorded. Some electrical appliances are in standby mode and consume very little current. They can be considered not in working state. In order to eliminate such influences, a current threshold is preset. When the total current is greater than the current threshold, the main switch is considered to be in the on state, that is, it is considered that an electrical appliance is in use. In other words, the on state of the main switch is the superposition of the on states of the switches of each electrical appliance in the room.

[0053] S2. Preprocessing of power consumption data: If the total current exceeds the preset current range, the current total current and the corresponding total voltage and total switch status are deleted, and the total voltage, total current and total switch status are collected again.

[0054] In some implementations, step S2 includes the following steps S21-S24:

[0055] S21. Obtain a total voltage set and a total current set.

[0056] The total current set includes the total current of a plurality of consecutive time points in the room collected at the distribution box at a preset sampling frequency.

[0057] S22. Calculate the average value and standard deviation corresponding to the total current set.

[0058] Optionally, the expressions for the mean μ and standard deviation σ are:

[0059]

[0060]

[0061] Among them, I i represents the i-th total current collected, and n represents the number of samples of the total current.

[0062] S23. Determine the current range according to the mean value and the standard deviation.

[0063] Optionally, the current interval is set to (μ-2σ, μ+2σ)

[0064] S24. If the current total current exceeds the current range, the current total current and the corresponding total voltage and total switch state are deleted, and the total voltage, total current and total switch state are collected again.

[0065] In the above embodiment, the influence of abnormal data caused by electric sparks or switch glitches is eliminated by setting the current interval. If abnormal data is identified, the current data is deleted and resampled to ensure the accuracy of the data.

[0066] S3. Calculate the total power: multiply the total voltage and the total current to obtain the total useful power.

[0067] The total active power is the sum of the useful powers of the target electrical appliances currently in use in the room. The total active power sequence can be expressed as y = (y1, y2, ..., y t ), the useful power sequence of the i-th target appliance can be expressed as i∈m, m represents the number of target electrical appliances.

[0068] S4. Determine the target electrical appliances and power: input the total useful power and the total switch state into a preset power decomposition model, extract the power decomposition features in the total useful power and the switch state through the power decomposition model, and determine several target electrical appliance types and the corresponding decomposed powers based on the power decomposition features.

[0069] Optionally, the power decomposition model power decomposition expression is:

[0070]

[0071] Among them, y t It represents the total power at time t, that is, the sum of the total power consumption of all target electrical appliances; represents the useful power of the i-th target appliance at time t; ε t Indicates the noise composed of non-target electrical appliances; m indicates the number of target electrical appliances.

[0072] The power consumption of the target appliance is taken as the original data, and the power consumption of other non-target appliances is taken as noise. The active power consumption y is aggregated at time t. t It is a noisy data.

[0073] S5. Warning of illegal electricity usage: traverse the preset prohibited electrical appliance table. If the target electrical appliance type and the corresponding decomposed power range exist in the prohibited electrical appliance table, a warning of illegal electricity usage is issued.

[0074] The list of prohibited electrical appliances includes electrical appliances that are prone to danger and have high power. Electrical appliances in the list of prohibited electrical appliances can be added or deleted according to actual management needs.

[0075] Optionally, the banned electrical appliances include electric irons, electric soldering irons, electric kettles, electric blankets, induction cookers, electric hot pots, microwave ovens, and hot pot machines, etc.

[0076] After determining the target appliance type and the corresponding decomposed power in step S4, the target appliance type and decomposed power are matched with the preset disabled appliance table. If the target appliance type appears in the disabled appliance table, and the decomposed power of the target appliance identified in step S4 is within the decomposed power interval corresponding to the appliance in, the target appliance is determined to be a disabled appliance.

[0077] This application can be combined with total power limitation to manage dormitory electricity safety. Within the tolerance range of line power supply, a larger total power threshold can be set for the dormitory. When the total useful power of the dormitory exceeds the total power threshold, the dormitory is directly powered off. When the total useful power of the dormitory does not exceed the total power threshold, it can ensure that the residents have a wider range of electricity use, and the types of electrical appliances in the dormitory can be monitored and analyzed in real time. When it is predicted that prohibited electrical appliances are in use, an early warning is issued in time to manage electricity use.

[0078] This application provides a simple, economical, reliable and complete electricity safety monitoring solution by performing load power decomposition of the total power consumption of the main line, predicting power-consuming appliances in the dormitory and determining the disabled appliances. The electricity consumption data of the dormitory can be obtained without physical intervention in the circuit, and the electrical appliances in each dormitory can be analyzed, which helps to timely discover the disabled appliances in the dormitory and improve the safety of electricity use in the dormitory.

[0079] In some implementations, before step S4, the following steps S041-S044 are also included:

[0080] S041. Obtain a total useful power sequence sample, a useful power sequence sample of a first target electrical appliance, and a switch state sequence sample.

[0081] The first target electrical appliance is any target electrical appliance in the target electrical appliance set.

[0082] Target appliances are appliances that need to be identified during electricity use, including prohibited appliances and permitted appliances. Permitted appliances include daily appliances such as lights, mobile phone chargers, computers, televisions, air conditioners, etc.

[0083] The total useful power sequence samples are obtained by arranging a plurality of total useful power samples according to time points. Each total useful power sample is the sum of the powers of several target electrical appliances in the target electrical appliance set under the simulated power consumption condition.

[0084] For example, the target electrical appliance set includes electric irons, electric soldering irons, electric kettles, electric blankets, electric lamps, computers, and air conditioners.

[0085] For example, at time t1, the first gear of the electric iron, the electric kettle, and the electric light are simulated to be in use. Then the value of the total useful power sample at time t1 is the power of the first gear of the electric iron, the electric kettle, and the electric light when they are in use, that is, the sum of the rated power. At time t1, the second gear of the electric iron, the electric light, and the computer are simulated to be in use. For example, at time t2, the value of the total useful power sample is the power of the second gear of the electric iron, the electric kettle, and the electric light when they are in use, that is, the sum of the rated power.

[0086] And according to the simulated usage of each target appliance, the corresponding switch state can be determined. For example, if the light is in use at all times t1-t5, then the switch state of the light in the time period t1-t5 is on, and the on time is determined accordingly. The same is true for other target appliances. The switch state of each target appliance is also arranged according to time, forming a switch state sequence sample of each target appliance.

[0087] It is understandable that the power usage of each electrical appliance is known and can be obtained or set according to the nameplate of the product.

[0088] S042. Label the total useful power sequence samples and the switch state sequence samples of the first target electrical appliance with the useful power sequence samples and types of the first target electrical appliance in chronological order to obtain a sample training set.

[0089] For example, the total useful power of the total useful power sequence samples at time t1 is marked with the useful power of the first target electrical appliance at time t1. Similarly, the total useful power of the total useful power sequence samples at each time is marked with the useful power sequence samples of other target electrical appliances.

[0090] If in the simulation state, the total current of the target electrical appliances in the working state at time t1 exceeds the current threshold, the total switch state is deemed to be in the on state at time t1. In this way, based on the total main current of the circuit, it is more accurate to identify whether there are high-power-consuming electrical appliances in use indoors.

[0091] S043. Use a neural network model to establish a power decomposition model to be trained corresponding to the total useful power, the total switch state, and the useful power and type of the target electrical appliance.

[0092] Deep learning is a branch of machine learning. Its core idea is to use multi-layer neural networks to learn the representation and features of data, extract the features of external input information by simulating the hierarchical structure of the human brain, and learn a deep nonlinear expression relationship to achieve approximation of complex functions.

[0093] The neural network model of this application adopts a multi-task learning strategy based on a fully convolutional denoising autoencoder, and trains the model with a joint loss function. The fully convolutional denoising autoencoder (FCN-DAE) only uses convolutional neural networks in the encoder, latent space layer, and decoder, which can better retain time information than the denoising autoencoder (DAE). By training the DAE with a supervised learning method, the encoder can learn to extract the most important features into a low-dimensional latent space feature vector h, and the decoder can learn to reconstruct the feature vector h into a value that is consistent with the true value x. t The closest reconstruction data The weights of the convolutional neural network are updated through the loss function. The model learning rate is set to 0.001, the input window length batch size is set to 256, and the input window length is 256.

[0094] S044. Take the total useful power sequence samples and the switch state sequence samples of the first target electrical appliance as input, and the useful power sequence samples and types of the first target electrical appliance as output, to train the power decomposition model to be trained; similarly, take the total useful power sequence samples and the switch state sequence samples of each target electrical appliance in the target electrical appliance set as input, and the useful power sequence samples and types of each target electrical appliance in the target electrical appliance set as output, to train the power decomposition model to be trained, until the loss function of the power decomposition model to be trained converges, thereby obtaining the power decomposition model.

[0095] Optionally, the loss function L is expressed as:

[0096]

[0097] Among them, represents the total time length of the sequence, x t represents the sample label at time t, Represents the output of the model at time t.

[0098] The power decomposition model decomposes the total useful power and the total switch state at a certain moment, so as to identify which electrical appliances are running in the room, when they are turned on and when they are turned off, and obtain the operating status, power and other power consumption information of the electrical appliances, so as to achieve the purpose of non-invasive load monitoring. For example, assuming that the working power of the induction cooker is 1000W, the decomposition power corresponding to the induction cooker is set between 900W and 1000W. According to habit, the induction cooker is usually turned on for about 25 minutes at around 11:30 noon, so the total switch state corresponding to the induction cooker is that the opening time point is between 11:00 and 12:00 noon, and the conduction time is 30 minutes. In other words, when the total power exceeds 1000W between 11:00 and 12:00 noon, and the total switch state is turned on for more than 30 minutes, the induction cooker is likely to be in use. Assuming that the working power of the hair dryer is 500W, the decomposition power corresponding to the induction cooker is set between 500W and 600W. According to habit, the hair dryer is usually turned on for about 3 minutes around 7:30 in the morning. The corresponding main switch state of the induction cooker is that the turn-on time point is between 7 and 8 in the morning, and the conduction time is less than 4 minutes.

[0099] In some implementations, the main regression subtask and the auxiliary classification subtask networks are constructed through a neural network model, and the model is trained with a joint loss function to achieve information interaction between subtasks.

[0100] The main regression subtask network only learns the mapping relationship between the total input power and the active power of the target appliance, and is a regression network. The auxiliary classification subtask network only learns the relationship between the total input power and the on / off state of the target appliance, and is a binary classification network.

[0101] The model consists of two parts: the channel attention module and the spatial attention module. It can learn "what to pay attention to" and "where to pay attention to" from the channel and space respectively, so as to pay attention to important features and suppress unimportant features. The calculation process of the convolutional block attention model can be expressed as:

[0102]

[0103]

[0104] In the formula It is represented as element-by-element multiplication, F is the input of the convolutional block attention model, M C (F) is the channel attention feature matrix, F C is the input of the spatial attention module, M S (F C ) is the spatial attention feature matrix. CS is the output of the convolutional block attention model, M C (F) and M S (F C ) is calculated as follows:

[0105]

[0106] M s (F c )=σ(M conv )

[0107] M conv =f Conv1D ([f AvgPool (F c );f MaxPool (F c )])

[0108] In some implementations, the total useful power sequence samples are labeled with the useful power sequence samples and types of the target electrical appliance in chronological order to obtain a sample training set. A power decomposition model to be trained corresponding to the total useful power and the useful power and type of the target electrical appliance is established using a neural network model. The total useful power sequence samples are used as input, and the useful power sequence samples and types of the target electrical appliance are used as output. The power decomposition model to be trained is trained until the loss function of the power decomposition model to be trained converges to obtain a power decomposition model. The power decomposition model of this embodiment only takes the total useful power as input, decomposes the total useful power, and obtains the target electrical appliance type and decomposed power.

[0109] Based on the same technical concept, the present application also provides a non-intrusive dormitory electricity safety monitoring system, which can be used for dormitory electricity safety monitoring. Figure 1 The steps of the method for non-intrusive dormitory electricity safety monitoring performed in the corresponding embodiment. The functions implemented by the system can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions, and the modules can be software and / or hardware. Figure 2 As shown, the system includes an acquisition module 1 and a processing module 2. The functions of the processing module 2 and the acquisition module 2 can be realized by referring to Figure 1 The operations performed in the corresponding embodiments are not described in detail here. The processing module 2 can be used to control the sending and receiving operations of the acquisition module 1 .

[0110] The acquisition module 1 is used to sample the total voltage, total current and total switch state of the current indoor power consumption at the distribution box at a preset sampling frequency.

[0111] Processing module 2 is used to delete the current total current and the corresponding total voltage and total switch state if the total current exceeds the preset current range, and re-collect the total voltage, total current and total switch state; multiply the total voltage and the total current to obtain the total useful power; input the total useful power and the total switch state into a preset power decomposition model, extract the power decomposition characteristics in the total useful power and the switch state through the power decomposition model, and determine several target electrical appliance types and corresponding decomposed powers according to the power decomposition characteristics; traverse the preset disabled electrical appliance table, and if the target electrical appliance type and the corresponding decomposed power range exist in the disabled electrical appliance table, issue an illegal electricity usage warning.

[0112] Optionally, the power decomposition model power decomposition expression is:

[0113]

[0114] Among them, y t represents the total power at time t; represents the useful power of the i-th target appliance at time t; ε t Indicates the noise composed of non-target electrical appliances; m indicates the number of target electrical appliances.

[0115] This application provides a simple, economical, reliable and complete electricity safety monitoring solution by performing load power decomposition of the total power consumption of the main line, predicting power-consuming appliances in the dormitory and determining the disabled appliances. The electricity consumption data of the dormitory can be obtained without physical intervention in the circuit, and the electrical appliances in each dormitory can be analyzed, which helps to timely discover the disabled appliances in the dormitory and improve the safety of electricity use in the dormitory.

[0116] Based on the same technical concept, the present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the non-intrusive dormitory electricity safety monitoring method described in the above-mentioned embodiments.

[0117] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM) and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0118] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the inspiration of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims. All equivalent structures or equivalent process changes made using the contents of the specification and drawings of this application, or directly or indirectly used in other related technical fields, are protected by the present application.

Claims

1. A non-intrusive dormitory electricity safety monitoring method, characterized in that: include: S1. Sample the total voltage, total current and total switch status of the current indoor power consumption at the distribution box at a preset sampling frequency; S2. If the total current exceeds the preset current range, the current total current and the corresponding total voltage and total switch state are deleted, and the total voltage, total current and total switch state are collected again; S3, multiply the total voltage and the total current to obtain the total useful power; S4, inputting the total useful power and the total switch state into a preset power decomposition model, extracting the power decomposition features in the total useful power and the switch state through the power decomposition model, and determining a number of target electrical appliance types and corresponding decomposed powers according to the power decomposition features; wherein, before step S4, the following steps S041-S044 are also included: S041, obtaining a total useful power sequence sample, a useful power sequence sample of a first target electrical appliance, and a switch state sequence sample; the first target electrical appliance is any target electrical appliance in the target electrical appliance set; S042, marking the total useful power sequence sample and the switch state sequence sample of the first target electrical appliance with the useful power sequence sample and type of the first target electrical appliance in chronological order to obtain a sample training set; S043 , using a neural network model to establish a power decomposition model to be trained corresponding to the total useful power, the total switch state and the useful power and type of the target electrical appliance; S044, taking the total useful power sequence samples and the switch state sequence samples of the first target electrical appliance as input, and the useful power sequence samples and types of the first target electrical appliance as output, the power decomposition model to be trained is trained; similarly, taking the total useful power sequence samples and the switch state sequence samples of each target electrical appliance in the target electrical appliance set as input, and the useful power sequence samples and types of each target electrical appliance in the target electrical appliance set as output, the power decomposition model to be trained is trained until the loss function of the power decomposition model to be trained converges, and a power decomposition model is obtained; wherein the expression of the loss function L is: Where T represents the total time length of the sequence, x t represents the sample label at time t, Represents the output of the model at time t; S5. Traverse the preset prohibited electrical appliance table, and if the target electrical appliance type and the corresponding decomposed power range exist in the prohibited electrical appliance table, issue an illegal electricity usage warning.

2. The non-intrusive dormitory electricity safety monitoring method according to claim 1 is characterized in that: Step S2 includes the following steps S21-S24: S21, obtaining a total voltage set and a total current set; The total current set includes the total current at multiple consecutive time points collected indoors at the distribution box at a preset sampling frequency; S22, calculating the average value and standard deviation corresponding to the total current set; S23, determining a current range according to the average value and the standard deviation; S24. If the current total current exceeds the current range, the current total current and the corresponding total voltage and total switch state are deleted, and the total voltage, total current and total switch state are collected again.

3. The non-intrusive dormitory electricity safety monitoring method according to claim 2 is characterized in that: The expression of the average value is: The expression of the standard deviation is: Among them, I i represents the i-th total current collected, and n represents the number of samples of the total current.

4. The non-intrusive dormitory electricity safety monitoring method according to claim 1 is characterized in that: Power decomposition model The expression of power decomposition is: Among them, y t represents the total power at time t; represents the useful power of the i-th target appliance at time t; ε t Indicates the noise composed of non-target electrical appliances; m indicates the number of target electrical appliances.

5. The non-intrusive dormitory electricity safety monitoring method according to claim 1 is characterized in that: Prohibited electrical appliances include electric irons, soldering irons, electric kettles, electric blankets and induction cookers.

6. A non-intrusive dormitory electricity safety monitoring system, characterized in that: include: An acquisition module is used to sample the total voltage, total current and total switch status of the current indoor power consumption at the distribution box at a preset sampling frequency; A processing module is used to delete the current total current and the corresponding total voltage and total switch state if the total current exceeds a preset current range, and re-collect the total voltage, total current and total switch state; multiply the total voltage and the total current to obtain the total useful power; input the total useful power and the total switch state into a preset power decomposition model, extract the power decomposition characteristics in the total useful power and the switch state through the power decomposition model, and determine a number of target electrical appliance types and corresponding decomposed powers according to the power decomposition characteristics; traverse the preset disabled electrical appliance table, and if the target electrical appliance type and the corresponding decomposed power range exist in the disabled electrical appliance table, issue an illegal electricity use warning; wherein, before the total useful power and the total switch state are input into the preset power decomposition model, the following steps are also included: S041, obtain a total useful power sequence sample, a useful power sequence sample and a switch state sequence sample of a first target electrical appliance; the first target electrical appliance is any target electrical appliance in the target electrical appliance set; S04 2. In chronological order, use the useful power sequence samples and types of the first target electrical appliance to mark the total useful power sequence samples and the switch state sequence samples of the first target electrical appliance to obtain a sample training set; S043. Use the neural network model to establish a power decomposition model to be trained corresponding to the total useful power, the total switch state and the useful power and type of the target electrical appliance; S044. Take the total useful power sequence samples and the switch state sequence samples of the first target electrical appliance as input, and the useful power sequence samples and types of the first target electrical appliance as output to train the power decomposition model to be trained; similarly, take the total useful power sequence samples and the switch state sequence samples of each target electrical appliance in the target electrical appliance set as input, and the useful power sequence samples and types of each target electrical appliance in the target electrical appliance set as output to train the power decomposition model to be trained until the loss function of the power decomposition model to be trained converges to obtain a power decomposition model; wherein the expression of the loss function L is: Where T represents the total time length of the sequence, x t represents the sample label at time t, Represents the output of the model at time t.

7. The non-intrusive dormitory electricity safety monitoring system according to claim 6 is characterized in that: Power decomposition model The expression of power decomposition is: Among them, y t represents the total power at time t; represents the useful power of the i-th target appliance at time t; ε t Indicates the noise composed of non-target electrical appliances; m indicates the number of target electrical appliances.

8. A storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps in the non-intrusive dormitory electricity safety monitoring method as described in any one of claims 1 to 5.

Images