Intelligent electricity consumption supervision method and supervision system based on Internet of Things

Abstract

The invention discloses an intelligent power utilization supervision method based on the Internet of Things, and the method comprises the steps: building a multi-modal data collaborative collection strategy at a power utilization collection end, and deploying an intelligent monitoring terminal and at least one dynamic environment sensing node; carrying out real-time data fusion processing on the collected electrical parameters and environmental parameters, carrying out data synchronization by adopting a space-time alignment algorithm, ensuring space-time consistency among the parameters, and extracting environmental characteristics; an edge side abnormity real-time diagnosis strategy is constructed, and classification decision making is carried out in combination with the miniature circuit breaker; the method comprises the steps of generating an electricity price reference value through a dynamic game, simulating the game process of a supply and demand power supply party and an electricity utilization party in combination with historical electricity utilization data and a real-time load condition, and generating an optimal electricity price strategy. A space-time alignment algorithm is adopted to perform data synchronization, and an electricity price reference value is generated through a dynamic game, so that the electricity consumption supervision level is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of Internet of Things, in particular to a smart power consumption supervision method and system based on Internet of Things. BACKGROUND

[0002] With the rapid development of science and technology, intelligent Internet of Things power management system is gradually becoming a key force in the field of energy management, not only for household electricity, industrial electricity, shopping mall electricity and other scenarios cannot do without the application of intelligent Internet of Things.

[0003] In the prior art, the intelligent Internet of Things power management system has many advantages, such as real-time monitoring, data analysis, remote control, etc. During the power consumption peak, the system can automatically coordinate various power consumption equipment, arrange peak-shaving power consumption, reduce the load pressure of the power grid, and effectively reduce the risk of large-scale power failure. At the same time, for distributed energy generation such as solar and wind power generation system, the intelligent Internet of Things power management system can optimize its access and consumption, further improve the stability and flexibility of the power grid, and promote the transformation of the energy supply system to a more green and sustainable direction.

[0004] However, the existing power consumption supervision system only relies on the data of the electric meter, cannot detect device-level abnormalities, and the traditional demand response mechanism relies on the fixed price model, cannot real-time match the changes of supply and demand, and the device maintenance relies on regular manual inspection, cannot predict hidden failures, resulting in delayed emergency response, which are all problems that need to be solved urgently. SUMMARY

[0005] The present application aims to at least solve one of the above technical problems in the art.

[0006] To achieve the above purpose, the first aspect of the present application provides a smart power consumption supervision method based on Internet of Things, comprising: Establishing a multi-modal data collaborative collection strategy at the power consumption collection end, and deploying an intelligent monitoring terminal and at least one dynamic environment perception node, wherein the intelligent monitoring terminal collects electrical parameters, and the dynamic environment perception node collects environmental parameters; Performing real-time data fusion processing on the collected electrical parameters and environmental parameters, and using a space-time alignment algorithm for data synchronization to ensure the space-time consistency between the parameters, and extracting environmental features; Building an edge-side real-time anomaly diagnosis strategy, and combining a miniature circuit breaker for classification decision; Generating a price reference value through dynamic game, combining historical power consumption data and real-time load conditions, simulating the game process between the power supply and demand sides, and generating an optimal price strategy.

[0007] A life prediction model is established, a physical information neural network is adopted, device operation parameters and environmental data are fused, the remaining life of the device is accurately evaluated, and a multi-level warning mechanism is set to respond.

[0008] Further, the intelligent monitoring terminal comprises a high-frequency current sensor for monitoring current changes in real time, a voiceprint acquisition module for capturing voiceprint features during device operation, and a visual acquisition device, which is an infrared thermal imager.

[0009] Further, the dynamic environment perception node is formed by fusing a plurality of environment sensor arrays, and combines signal conditioning, a microcontroller, a communication module, an edge gateway, and a local PLC to realize environment parameter monitoring.

[0010] Further, after collecting electrical parameters and environmental parameters, a space-time alignment algorithm is used for data synchronization, which includes: obtaining the time series of electrical parameters and environmental parameters respectively, determining the main reference signal and the signal to be aligned, and obtaining the weight coefficient through signal-to-noise ratio normalization; obtaining the regularization strength according to the sensor sampling interval, controlling the smoothness of the time offset compensation amount, and smoothing the time offset compensation amount through second-order difference to prevent jumping; taking the main reference signal and the signal to be aligned as the data alignment item, and taking the regularization strength and the second-order difference smoothing constraint as the smoothing constraint item to calculate the time offset compensation amount; and obtaining the minimum time offset compensation amount through the minimum value function.

[0011] Further, the value of the regularization strength is determined according to the sampling interval, and the specific formula is regularization strength = 1 / sampling interval, and the second-order difference formula is Δ 2 x(t) = x(t+1) - 2x(t) + x(t-1), where x(t) is the compensation amount at the current time, and Δ 2 x(t) represents the rate of change of the compensation amount at adjacent times.

[0012] Further, the calculation formula of the time offset compensation amount is: time offset compensation amount = (main reference signal - signal to be aligned) * weight coefficient + regularization strength * Δ 2 x(t), where the difference between the main reference signal and the signal to be aligned multiplied by the weight coefficient reflects the synchronization requirement between the data, and the product of the regularization strength and the second-order difference ensures the smoothness of the compensation amount.

[0013] Further, the edge side abnormal real-time diagnosis strategy includes: deploying a lightweight hybrid neural network, which includes a one-dimensional convolutional neural network layer, a long short-term memory network layer and an attention mechanism layer; wavelet denoising the input current signal, and generating a multi-dimensional feature vector through the lightweight hybrid neural network; identifying the fault type through an SVM classifier, the fault type including arc, short circuit and overload; and according to the fault type, automatically triggering a miniature circuit breaker for power-off protection to prevent the fault from expanding, and sending an alarm information to a monitoring center.

[0014] Further, the one-dimensional convolutional neural network layer is used to extract local features of the current waveform, the long short-term memory network layer is used to capture time sequence features of the voiceprint signal, and the attention mechanism layer is used to dynamically weight the environmental data weight, so as to identify the abnormal mode through multi-level feature fusion.

[0015] Further, the process of generating the electricity price reference value includes: constructing a game model through a user utility function and a power grid cost function, wherein the user utility function is related to the electricity consumption and the electricity price, and the power grid cost function is related to the power supply, and solving the Nash equilibrium point of the game model through an iterative optimization algorithm to obtain the optimal electricity price reference value.

[0016] On the other hand, the application also proposes a smart power consumption supervision system based on the Internet of Things, which includes: A multi-modal data collaborative collection strategy establishment module is used to establish a multi-modal data collaborative collection strategy at a power consumption collection end, and deploy an intelligent monitoring terminal and at least one dynamic environment perception node, wherein the intelligent monitoring terminal collects electrical parameters, and the dynamic environment perception node collects environmental parameters; A data fusion processing module is used to perform real-time data fusion processing on the collected electrical parameters and environmental parameters, and perform data synchronization through a space-time alignment algorithm to ensure the space-time consistency between the parameters, and extract environmental features; An edge side abnormal diagnosis module is used to construct an edge side abnormal real-time diagnosis strategy, and make a classification decision in combination with a miniature circuit breaker; An electricity price reference value generation module is used to generate an electricity price reference value through a dynamic game, simulate the game process of the power supply and demand sides and the power consumption side in combination with historical power consumption data and real-time load conditions, and generate an optimal electricity price strategy.

[0017] A life prediction module is used to establish a life prediction model, fuse device operating parameters and environmental data through a physical information neural network, accurately evaluate the remaining life of the device, and set a multi-level warning mechanism for response.

[0018] According to the present invention, a smart electricity consumption monitoring method and system based on the Internet of Things can improve the level of electricity consumption monitoring by establishing a multimodal data collaborative acquisition strategy, collecting electrical and environmental parameters, performing data fusion processing, using a spatiotemporal alignment algorithm for data synchronization, and generating electricity price reference values ​​through dynamic game theory.

[0019] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating a smart electricity monitoring method based on the Internet of Things according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the module structure of an Internet of Things-based smart electricity monitoring system according to an embodiment of the present invention. Detailed Implementation

[0021] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0022] The following description, in conjunction with the accompanying drawings, illustrates an embodiment of the present invention: a smart electricity consumption monitoring method and system based on the Internet of Things (IoT).

[0023] Example 1: As Figure 1 As shown in the figure, an embodiment of the present invention provides a smart electricity consumption monitoring method based on the Internet of Things. This method is applied to electricity consumption monitoring in various scenarios, such as household electricity consumption, shopping mall electricity consumption, and industrial electricity consumption. The method includes the following steps: Step 101: Establish a multimodal data collaborative acquisition strategy at the power consumption acquisition terminal, and deploy an intelligent monitoring terminal and at least one dynamic environment sensing node, wherein the intelligent monitoring terminal acquires electrical parameters and the dynamic environment sensing node acquires environmental parameters. When the Internet of Things technology is applied to the electricity supervision, hardware deployment is needed, and in the embodiment, a multi-modal data collaborative collection strategy is established at the electricity collection end, which is realized by embedding an intelligent monitoring terminal in the electricity circuit, wherein the intelligent monitoring terminal comprises a high-frequency current sensor for monitoring the current change in real time, a voiceprint collection module for capturing the voiceprint features of the equipment during operation, specifically arc voiceprint, and a visual collection device, which is an infrared thermal imager, for monitoring the temperature change of the equipment surface. The intelligent monitoring terminal is mainly responsible for collecting the electrical parameters of the equipment. In order to collect the environmental parameters in the electricity consumption scene in real time, at least one dynamic environment perception node needs to be established, which is formed by fusing a plurality of environmental sensor arrays and combining signal conditioning, microcontrollers, communication modules, edge gateways, local PLCs, etc. to realize environmental parameter monitoring. It not only has the advantage of low power consumption, but also can synchronously monitor multiple types of data, and the accuracy of multi-modal data is higher.

[0024] Step 102: Real-time data fusion processing is performed on the collected electrical parameters and environmental parameters, and a time-space alignment algorithm is used for data synchronization to ensure the time-space consistency between the parameters, while extracting environmental features; After collecting the electrical parameters and environmental parameters, due to the problems of time delay, sampling frequency difference, etc. of different sensors, the data needs to be synchronized by a time-space alignment algorithm to ensure the matching of each parameter in time and space. Specifically, the data synchronization using the time-space alignment algorithm includes: The time series of electrical parameters and environmental parameters are obtained respectively, and the main reference signal and the signal to be aligned are determined, and the weight coefficient is obtained by signal-to-noise ratio normalization; In the synchronization process, the time series of electrical parameters and environmental parameters need to be obtained first to determine the parameters to be optimized and the main reference signal and the signal to be aligned. In the embodiment, the main reference signal is the current waveform, because the collection sensor used by the current is usually a high sampling rate sensor, and the current waveform takes the effective value sequence with a sampling rate of 10 kHz. The signal to be aligned is taken as an example of an environmental or auxiliary sensor signal with time delay, such as voiceprint. The weight coefficient also needs to be obtained. In the embodiment, the weight coefficient is obtained by the signal-to-noise ratio normalization method. The signal-to-noise ratio normalization method can reflect the reliability of the data collected by each sensor. The specific calculation method can refer to the signal-to-noise ratio calculation formula in the prior art, which is not described in detail in the embodiment. The regularization strength is obtained according to the sensor sampling interval to control the smoothness of the time offset compensation amount, and the time offset compensation amount is smoothed by second-order difference to prevent jumping; In the process of time offset compensation, the final time offset compensation amount needs to meet the smoothness requirement to avoid data distortion caused by overcompensation or interference, so the regularization strength is obtained according to the sensor sampling interval to ensure that the time offset compensation amount changes continuously with time. Specifically, the value of the regularization strength is determined according to the sampling interval, and the specific formula is regularization strength = 1 / sampling interval; At the same time, the time offset compensation amount also needs to be smoothed by second-order difference to prevent data distortion caused by jump, and to ensure smooth transition of data. Specifically, the second-order difference formula is Δ 2 x(t) = x(t+1) - 2x(t) + x(t-1), where x(t) is the compensation amount at the current time, Δ 2 x(t) represents the change rate of the compensation amount at adjacent time, which can effectively suppress the mutation of the compensation amount and ensure the continuity and stability of the data; The main reference signal and the signal to be aligned are taken as the data alignment item, and the regularization strength and the second-order difference smoothing constraint are taken as the smoothing constraint item to calculate the time offset compensation amount. The specific formula is: Time offset compensation amount = (main reference signal - signal to be aligned) * weight coefficient + regularization strength * Δ 2 x(t), where the difference between the main reference signal and the signal to be aligned multiplied by the weight coefficient reflects the synchronization requirement between the data, and the product of the regularization strength and the second-order difference ensures the smoothness of the compensation amount; After calculating the time offset compensation amount, the minimum time offset compensation amount needs to be obtained to meet the accuracy and stability of data synchronization and ensure the consistency of sensor signals in the time dimension. The minimum time offset compensation amount is obtained by a minimum value function, that is, the minimum value in the calculation result is taken as the final compensation amount to realize the precise alignment of sensor signals in time; At the same time, since the above process of obtaining the minimum time offset compensation amount involves not only the first-order derivative (regularization strength) and the second-order derivative (Δ 2 x(t)) of the time offset compensation amount, the influence of mechanical vibration and random noise on data synchronization in the data acquisition process can be avoided, and the accuracy of time offset compensation is further improved; Step 103: Construct an edge side abnormal real-time diagnosis strategy, and make a classification decision in combination with the miniature circuit breaker; In intelligent power consumption supervision, anomaly diagnosis and decision-making are indispensable links. By monitoring parameters such as current and voltage in real time, combined with the state information of miniature circuit breakers, abnormal conditions can be quickly identified and classified for processing, ensuring the safe and stable operation of the power consumption system. At the same time, edge computing technology is used to improve the speed and accuracy of diagnosis, reduce misdiagnosis and missed diagnosis, and ensure the continuity and reliability of power supply. Specifically, the edge-side real-time anomaly diagnosis strategy includes: A lightweight hybrid neural network is deployed, which includes a one-dimensional convolutional neural network layer, a long short-term memory network layer, and an attention mechanism layer. The one-dimensional convolutional neural network layer is used to extract local features of the current waveform, the long short-term memory network layer is used to capture the time sequence characteristics of the voiceprint signal, and the attention mechanism layer is used to dynamically weight the environmental data weight. Through multi-level feature fusion, abnormal patterns can be accurately identified. The input current signal is denoised using wavelet denoising, and a multi-dimensional feature vector is generated using a lightweight hybrid neural network. Since the input current signal may contain high-frequency noise and random fluctuations, wavelet denoising can effectively filter out interference and improve signal quality. Through the multi-layer processing of the lightweight hybrid neural network, a higher-dimensional feature vector is extracted, which contains multi-dimensional information such as current amplitude, frequency, and phase in this embodiment. The SVM classifier is used to identify the fault type, which includes arc, short circuit, and overload. The SVM classifier is a high-efficiency classification algorithm based on support vector machine, which can accurately distinguish between arc, short circuit, and overload fault types. Through comprehensive analysis of the multi-dimensional feature vector, the accuracy and real-time performance of fault identification are ensured. According to the fault type, the miniature circuit breaker is automatically triggered for power-off protection to prevent the expansion of the fault, and an alarm message is sent to the monitoring center. The miniature circuit breaker is a key protection device in the smart grid, which effectively isolates the fault area through its fast response mechanism to ensure the overall safety of the system.

[0025] Step 104: Generate the electricity price reference value through dynamic game, combine historical electricity consumption data and real-time load conditions, simulate the game process between supply and demand power supply and power consumption sides, and generate the optimal electricity price strategy. Traditional electricity pricing often lacks flexibility and is difficult to adapt to market changes. The dynamic game model simulates the interaction between supply and demand, combines historical data and real-time load, and adjusts the electricity price flexibly to achieve supply and demand balance. Specifically, a game model is constructed through a user utility function and a power grid cost function. The user utility function is related to electricity consumption and electricity price, and the power grid cost function is related to power supply. The Nash equilibrium point of the game model is solved through an iterative optimization algorithm to obtain the optimal electricity price reference value. Specifically, a maximum value objective function is used to construct a profit objective function, the function being profit = max (p * qi - Ci (qi)), wherein qi is the power generation of power generator i, in units of megawatt hours, p is the unified clearing price of the power market, in units of yuan per megawatt hour, and Ci (qi) is the power generation cost function of power generator i, representing the cost required to produce qi amount of power. The optimal power generation under the maximum profit is obtained according to the profit objective function, wherein the cost function is Ci (qi) = aqi^ + bqi + c), wherein a is the quadratic term cost coefficient of power generation, representing fuel efficiency, b is the first-order term cost function of power generation, such as labor cost, and c is the fixed cost, such as equipment depreciation. By constructing the cost function, the cost information of the power plant can be accurately obtained. For example, assuming that the power generation cost function of a thermal power plant is C(qi)=0.3qi^+50qi+1000, if the market electricity price is 200 yuan / MWh, the optimal power generation qi can maximize the profit; Real-time power consumption data and user preference data are collected through the grid side and the user side, the game model parameters are dynamically updated to ensure the real-time and accuracy of the electricity price strategy, and the pricing strategy network is trained through deep reinforcement learning, wherein the input of the trained pricing strategy network includes but is not limited to load rate, renewable energy ratio, and user response history, and the output is the optimal electricity price and confidence interval. The electricity price strategy is continuously optimized to improve the supply-demand matching degree and reduce the system operation cost, thereby enhancing the stability and economy of the power grid operation.

[0026] Step 105: Establish a life prediction model, use a physical information neural network, fuse device operating parameters and environmental data, accurately evaluate the remaining life of the device, and set a multi-level warning mechanism for response.

[0027] The physical information neural network includes data items and physical constraint items, wherein the data items are various types of data collected in real time, and the physical constraint items are device heating equations. The device heating equations can refer to the principles and formulas in thermodynamics, which are not described in detail in this embodiment. The multi-level warning mechanism is used to trigger warnings at different stages of the remaining life of the device to perform maintenance or replacement in advance. In this embodiment, the multi-level warning mechanism is divided into three levels, wherein the response corresponding to the first-level warning mechanism is an APP push maintenance reminder, the second-level warning mechanism triggers manual inspection, and the third-level warning mechanism forces the switching of a backup circuit to ensure the safety of device operation.

[0028] Embodiment Two As shown in Figure 2 The present application also proposes a smart power consumption supervision system based on the Internet of Things, which runs the smart power consumption supervision method in Embodiment One. The system comprises: The multi-modal data collaborative collection strategy establishment module is configured to establish a multi-modal data collaborative collection strategy at a power consumption collection end, and deploy an intelligent monitoring terminal and at least one dynamic environment perception node, wherein the intelligent monitoring terminal collects electrical parameters, and the dynamic environment perception node collects environmental parameters; The data fusion processing module is configured to perform real-time data fusion processing on the collected electrical parameters and environmental parameters, and perform data synchronization using a space-time alignment algorithm to ensure space-time consistency between the parameters, and extract environmental features; The edge side anomaly diagnosis module is configured to construct an edge side anomaly real-time diagnosis strategy, and make a classification decision in combination with a miniature circuit breaker. The electricity price reference value generation module is configured to generate an electricity price reference value through dynamic game playing, simulate a game playing process of power supply and demand sides and power consumption sides in combination with historical power consumption data and real-time load conditions, and generate an optimal electricity price strategy.

[0029] The life prediction module is configured to establish a life prediction model, adopt a physical information neural network, fuse device operation parameters and environmental data, accurately evaluate the remaining life of the device, and set a multi-level early warning mechanism to respond.

[0030] In the description of the present specification, the terms "first", "second" are only used for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, such as two, three, etc., unless otherwise specifically limited.

[0031] In the description of the present specification, the description of the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, different embodiments or examples described in the present specification and the features of different embodiments or examples can be combined and combined by those skilled in the art without contradiction.

[0032] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A smart power consumption monitoring method based on Internet of Things, characterized in that, The application relates to a power consumption collection terminal and a dynamic environment perception node, and relates to a method for establishing a multi-modal data collaborative collection strategy, deploying an intelligent monitoring terminal and at least one dynamic environment perception node, performing real-time data fusion processing on collected electrical parameters and environmental parameters, adopting a time-space alignment algorithm for data synchronization to ensure time-space consistency between parameters, extracting environmental characteristics, constructing an edge-side abnormality real-time diagnosis strategy, generating a power price reference value through dynamic game playing, simulating a game process between power supply and power consumption parties based on historical power consumption data and real-time load conditions to generate an optimal power price strategy, and setting up a life prediction model to accurately assess the remaining life of equipment by adopting a physical information neural network and fusing equipment operation parameters and environmental data. The intelligent monitoring terminal comprises a high-frequency current sensor for monitoring current changes in real time, a voiceprint collection module for capturing voiceprint characteristics during equipment operation, and a visual collection device, which is an infrared thermal imager. The dynamic environment perception node is formed by fusing a plurality of environmental sensor arrays and combining signal conditioning, a microcontroller, a communication module, an edge gateway and a local PLC to realize environmental parameter monitoring. After collecting electrical parameters and environmental parameters, a time-space alignment algorithm is adopted for data synchronization, which comprises the following steps: time series of electrical parameters and environmental parameters are respectively acquired, a main reference signal and a signal to be aligned are determined, and a weight coefficient is acquired through signal-to-noise ratio normalization; the regularization strength is acquired according to the sensor sampling interval, the smoothness of the time offset compensation amount is controlled, the time offset compensation amount is smoothed and constrained through second-order difference, and jumping is prevented; the main reference signal and the signal to be aligned are taken as data alignment items, the regularization strength and the second-order difference smoothing constraint are taken as smoothing constraint items, and the time offset compensation amount is calculated; and the minimum time offset compensation amount is acquired through a minimum value function. The edge-side abnormality real-time diagnosis strategy comprises the following steps: a lightweight hybrid neural network is deployed, the neural network comprises a one-dimensional convolutional neural network layer, a long short-term memory network layer and an attention mechanism layer; a wavelet denoising operation is performed on an input current signal, and a multi-dimensional feature vector is generated through the lightweight hybrid neural network; a fault type is identified through an SVM classifier, the fault type comprises arc, short circuit and overload; and an action response is performed according to the fault type, a miniature circuit breaker is automatically triggered to perform power-off protection, the fault is prevented from being enlarged, and alarm information is sent to a monitoring center. The one-dimensional convolutional neural network layer is used for extracting current waveform local features, the long short-term memory network layer is used for capturing voiceprint signal time sequence characteristics, and the attention mechanism layer is used for dynamically weighting environmental data weights; through multi-level feature fusion, an abnormal mode is identified. 2.The smart power utilization supervision method based on the Internet of Things according to claim 1, characterized in that: ​ 3.The smart power utilization supervision method based on the Internet of Things according to claim 1, characterized in that: ​ 4.The smart power utilization supervision method based on the Internet of Things according to claim 1, characterized in that: ​ 5. The smart power utilization monitoring method based on the Internet of Things according to claim 4, characterized in that: The value of the regularization strength is determined according to the sampling interval, and the specific formula is regularization strength = 1 / sampling interval, and the second-order difference formula is Δ 2 x(t) = x(t+1) - 2x(t) + x(t-1), wherein x(t) is the compensation amount at the current moment, and Δ 2 x(t) represents the rate of change of the compensation amount at adjacent moments.

6. The smart power utilization monitoring method based on the Internet of Things according to claim 5, characterized in that: The calculation formula of the time offset compensation amount is: time offset compensation amount = (main reference signal - signal to be aligned) * weight coefficient + regularization strength * Δ 2 x(t), wherein the difference between the main reference signal and the signal to be aligned multiplied by the weight coefficient reflects the synchronization requirement between data, and the product of the regularization strength and the second-order difference ensures the smoothness of the compensation amount. 7.The smart power utilization supervision method based on the Internet of Things according to claim 1, characterized in that: ​ 8.The smart power utilization supervision method based on the Internet of Things according to claim 7, characterized in that: ​ 9.The smart power utilization supervision method based on the Internet of Things according to claim 1, characterized in that: The process of generating the electricity price reference value comprises: constructing a game model through a user utility function and a power grid cost function, wherein the user utility function is related to electricity consumption and electricity price, and the power grid cost function is related to power supply; solving a Nash equilibrium point of the game model through an iterative optimization algorithm to obtain an optimal electricity price reference value.

10. An Internet of Things based smart power utilization supervision system for implementing the Internet of Things based smart power utilization supervision method according to any one of claims 1 to 9, characterized in that: Comprise: A multi-modal data collaborative acquisition strategy establishment module is configured to establish a multi-modal data collaborative acquisition strategy at an electricity acquisition end, deploy an intelligent monitoring terminal and at least one dynamic environment perception node, wherein the intelligent monitoring terminal acquires electrical parameters, and the dynamic environment perception node acquires environmental parameters; A data fusion processing module is configured to perform real-time data fusion processing on the acquired electrical parameters and environmental parameters, and perform data synchronization using a space-time alignment algorithm to ensure space-time consistency between the parameters, and extract environmental features; An edge-side abnormality diagnosis module is configured to construct an edge-side real-time abnormality diagnosis strategy, and make a classification decision in combination with a miniature circuit breaker; An electricity price reference value generation module is configured to generate an electricity price reference value through a dynamic game, simulate a game process of a power supply side and a power consumption side in combination with historical electricity consumption data and real-time load conditions, and generate an optimal electricity price strategy; A life prediction module is configured to establish a life prediction model, fuse device operating parameters and environmental data using a physical information neural network, evaluate a remaining life of the device, and set a multi-level warning mechanism to respond.

Images