A power consumption load prediction method, device and equipment for smart power protection

Abstract

The application provides a power consumption load prediction method, device and equipment for smart power protection, and relates to the technical field of power system load monitoring and power supply protection. The method comprises the following steps: collecting actual power consumption data of a load line. Based on the actual power consumption data, a load prediction strategy is determined. According to the load prediction strategy, the actual power consumption data is subjected to load prediction, and a prediction result is output; and the prediction result is sent to a user terminal. The method is used to achieve the effect of quickly and accurately diagnosing and predicting user load with less construction.

Description

Technical Field

[0001] This application relates to the field of power system load monitoring and power supply guarantee technology, specifically to a method, device and equipment for predicting power load for smart power supply guarantee. Background Technology

[0002] Driven by the new energy security strategy, refined management of customer load has become a core breakthrough for energy efficiency services and customer-side power supply security. Currently, most power users still use extensive bus metering, and the load inside the user's premises remains in a "black box" state for a long time, making it difficult to conveniently and accurately depict energy consumption and accurately perceive the power consumption status of terminal equipment.

[0003] In traditional and large-scale electricity consumption scenarios, such as university dormitories and sports venues, safe electricity use is of paramount importance, as accidents such as fires caused by overload can have extremely serious consequences. Therefore, it is necessary to achieve the analysis and early warning of electricity consumption data.

[0004] In existing technologies, a common approach is to make customized modifications for specific scenarios, such as installing a large number of metering devices such as electricity meters at various monitoring nodes and monitoring electricity consumption.

[0005] However, the above-mentioned power monitoring scheme is an intrusive monitoring method, which requires the installation of fixed metering equipment at each monitoring node, resulting in problems such as excessive modification costs and long modification cycles. Summary of the Invention

[0006] The purpose of this application is to provide a method, apparatus, and equipment for predicting power load for smart power supply, in order to solve the above-mentioned problems existing in the prior art, which can greatly reduce construction costs and ensure the accuracy of load prediction.

[0007] Firstly, a method for predicting electricity load for intelligent power supply is provided, applied to a portable energy consumption diagnostic terminal, wherein the portable energy consumption diagnostic terminal is installed externally to the load line in a non-intrusive manner; the method may include: Collect the actual power consumption data of the load line; Based on the actual electricity consumption data, a load forecasting strategy is determined; According to the load forecasting strategy, load forecasting is performed on the actual electricity consumption data, and the forecasting results are output; the forecasting results are then sent to the user terminal.

[0008] Secondly, a power load prediction device for intelligent power supply is provided, applied to a portable energy consumption diagnostic terminal, wherein the portable energy consumption diagnostic terminal is installed outside the load line in a non-intrusive manner; the device may include: The data acquisition module is used to collect the actual power consumption data of the load line; The determination module is used to determine a load forecasting strategy based on the actual electricity consumption data; The prediction module is used to perform load prediction on the actual electricity consumption data according to the load prediction strategy, output the prediction result, and send the prediction result to the user terminal.

[0009] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0010] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0011] This application provides a method, apparatus, and equipment for predicting electricity load for smart power supply, which collects actual electricity consumption data of load lines. Based on the actual electricity consumption data, a load prediction strategy is determined. According to the load prediction strategy, load prediction is performed on the actual electricity consumption data, and the prediction results are output; the prediction results are then sent to the user terminal. In this solution, the portable energy consumption diagnostic terminal is plug-and-play, requiring no complex installation to quickly acquire electricity consumption data; according to the load prediction strategy, it quickly identifies electricity consumption risks, accurately predicts load changes, improves prediction accuracy, and achieves rapid and accurate diagnosis and prediction of user load with minimal construction work. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating a power load forecasting method for smart power supply protection provided in this application embodiment; Figure 2 A flowchart illustrating another power load forecasting method for smart power supply provided in this application embodiment; Figure 3 A load forecasting flowchart of an assisted load forecasting strategy is provided in an embodiment of this application; Figure 4A flowchart illustrating another power load forecasting method for smart power supply provided in this application embodiment; Figure 5 A schematic diagram of the structure of a power load prediction device for intelligent power supply provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art. The words "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are only used to distinguish different components. The words "comprising" or "including," etc., mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but do not exclude other elements or objects. The words "connected," "coupled," or "connected," etc., are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up," "down," "left," "right," etc., are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0015] Currently, in traditional and large-scale electricity consumption scenarios, such as university dormitories and sports venues, safe electricity use is of paramount importance, as accidents such as fires caused by overload can have extremely serious consequences. Therefore, it is necessary to achieve the analysis and early warning of electricity consumption data.

[0016] In one example, a common approach is to customize modifications for specific scenarios, such as installing numerous metering devices like electricity meters at various monitoring nodes to monitor electricity consumption. However, this electricity monitoring solution is an invasive method, requiring the installation of fixed metering devices at each monitoring node, which presents problems such as excessive modification costs and long modification cycles.

[0017] In one example, current load sensing models incorporating AI algorithms often target larger load nodes, such as those relying on building meters or transformer data to detect larger electricity consumption units. General-purpose metering devices, similar to portable diagnostic tools, targeting smaller electricity consumption units, consist only of smart meters and upper-level data acquisition software. However, systems for detecting data from larger electricity consumption units are limited by the precision of the hardware and the challenges of hardware-software coordination, making it difficult to detect and control finer nodes. Metering devices for smaller electricity consumption units lack the ability to collaborate with upper-level software for intelligent analysis and control, and can only be controlled through relatively crude methods such as circuit breaking, resulting in a limited control approach.

[0018] The power load forecasting method for smart power supply provided in this application can be applied to electronic devices, terminal devices, power load forecasting devices or equipment for smart power supply, or other devices or equipment that can execute this embodiment, and there are no limitations on this. In this embodiment, the execution subject is described as an electronic device.

[0019] The terminal can be a user equipment (UE) such as a mobile phone, smartphone, laptop computer, digital broadcast receiver, personal digital assistant (PDA), or tablet computer (PAD), handheld device, in-vehicle device, wearable device, computing device, or other processing device connected to a wireless modem, mobile station (MS), or mobile terminal. This terminal has the ability to communicate with one or more core networks via a radio access network (RAN).

[0020] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0021] Figure 1 This is a flowchart illustrating a power load prediction method for intelligent power supply protection provided in an embodiment of this application. This application applies to a portable energy consumption diagnostic terminal, which is installed non-intrusively outside the load line; such as... Figure 1 As shown, the method may include: Step S101: Collect the actual power consumption data of the load line.

[0022] For example, the portable energy consumption diagnostic terminal is installed externally to the load line in a non-intrusive manner. For instance, the current transformer of the portable energy consumption diagnostic terminal is clipped onto the load line to acquire the user's electricity consumption information through inductive means; the device is plug-and-play. The portable energy consumption diagnostic terminal is used to preliminarily determine whether the load status requires an early warning through its own data processing capabilities.

[0023] In this step, after examining the on-site electrical topology and meter installation, a portable energy consumption diagnostic instrument is mounted on high-power equipment or sensitive electrical nodes.

[0024] Step S102: Determine the load forecasting strategy based on actual electricity consumption data.

[0025] For example, the portable energy consumption diagnostic terminal first determines whether the local computing resources can process the actual electricity consumption data, and generates a load forecasting strategy based on the determination result. The load forecasting strategy includes a local load forecasting strategy, a cloud load forecasting strategy, and an assisted load forecasting strategy. For example, if the local system can process the data, the load forecasting strategy is a local load forecasting strategy; if the local system cannot process the data, the load forecasting strategy is a cloud load forecasting strategy; if the local system can process part of the data or if other idle portable energy consumption diagnostic terminals are detected locally, the load forecasting strategy is an assisted load forecasting strategy.

[0026] For example, for data that its own hardware can process, after simple processing / comparison, the obtained abnormal information is pushed to the user terminal through the communication module, so that the administrator can go to the site of the node judged to be abnormal for inspection; for datasets that it cannot process or for which the fault has not yet been determined, the cloud load prediction strategy is triggered, and the data is reported to the preset data processing system for calculation through the communication module.

[0027] Step S103: Based on the load forecasting strategy, perform load forecasting on the actual electricity consumption data, output the forecasting results, and send the forecasting results to the user terminal.

[0028] For example, load forecasting is performed on actual electricity consumption data according to the type of load forecasting strategy, the forecasting results are output, and the forecasting results are sent to the user terminal.

[0029] The method provided in this application collects actual power consumption data of the load line. Based on the actual power consumption data, a load forecasting strategy is determined. According to the load forecasting strategy, load forecasting is performed on the actual power consumption data, and the forecasting results are output; the forecasting results are then sent to the user terminal. In this solution, the portable energy consumption diagnostic terminal is plug-and-play, requiring no complex installation to quickly acquire power consumption data; based on the load forecasting strategy, it quickly identifies power consumption risks, accurately predicts load changes, and improves forecasting accuracy, achieving rapid and accurate diagnosis and prediction of user load with minimal construction work.

[0030] Figure 2 A flowchart illustrating another power load forecasting method for smart power supply protection provided in this application is shown below. Figure 2 As shown, in this embodiment... Figure 1 Based on the embodiments, the method is described in detail below, and the method includes: Step S201: Collect actual power consumption data of the load line.

[0031] In one example, S201 includes: sending a model acquisition instruction to a preset data processing system; the model acquisition instruction is used to acquire a power supply network structure model, which includes topology information of the load lines; acquiring the power supply network structure model sent by the data processing system; determining a collection strategy based on the topology information indicated in the power supply network structure model; and collecting actual power consumption data of the load lines according to the collection strategy.

[0032] For example, a pre-built data processing system has a power supply network structure model, which includes topology information of the load lines. A portable energy consumption diagnostic terminal sends a model acquisition command to the pre-built data processing system. Upon receiving the command, the data processing system sends the power supply network structure model to the portable energy consumption diagnostic terminal; this can be the power supply network structure model itself, or a portion thereof. The portable energy consumption diagnostic terminal acquires the power supply network structure model sent by the data processing system and, based on the topology information indicated in the model, determines a data collection strategy. According to this strategy, it collects the actual power consumption data of the load lines, thus enabling the collection of actual power consumption data according to the topology information.

[0033] For example, after the portable energy consumption diagnostic terminal is installed and powered on for the first time, it sends a registration request to the data processing system via a pre-set communication module. The registration request includes the portable energy consumption diagnostic terminal's device identification number and installation location information. Upon receiving the registration request, the data processing system queries the power supply network structure model based on the device identification number and installation location information. This power supply network structure model, pre-stored in the data processing system, contains the topological relationships between branches and main lines in the entire power consumption scenario. The data processing system then sends the topological relationship information corresponding to the installation location of the portable energy consumption diagnostic terminal to the portable energy consumption diagnostic terminal via the communication module. Upon receiving the topological relationship information, the portable energy consumption diagnostic terminal parses the node type and topological level indicated in the information. The topological relationship information includes: whether the node is a main line node or a branch node, the identifier of the superior node, the type and number of subordinate loads, and the positional relationships of adjacent nodes.

[0034] The portable energy consumption diagnostic terminal determines the data collection strategy based on the topology information according to the following rules: If this node is a backbone node, it will be determined to collect total current, total voltage, total power, and total power factor at a higher sampling frequency, and continuously collect these parameters to monitor the overall load change trend. If this node is a front-end node for high-power equipment, it will be determined to collect the current, voltage, power, and power factor of this equipment at a higher sampling frequency, and extend the duration of each sampling to capture the power consumption characteristics throughout the equipment's startup, operation, and shutdown processes. If this node is a general branch node, it will be determined to collect the branch's current, voltage, power, and power factor at a conventional sampling frequency, and perform intermittent sampling according to a set period to reduce terminal power consumption. If this node is a sensitive electrical node, it will be determined to continuously sample current and voltage at high speed, and additionally collect secondary key parameters such as harmonic current, harmonic voltage, and harmonic power to capture transient abnormal signals.

[0035] Based on the aforementioned data acquisition strategy, the portable energy consumption diagnostic terminal non-intrusively couples to the load line via its self-deployed mutual inductance acquisition unit. It acquires current and voltage signals from the load line at a set sampling frequency and duration. Based on the acquired current and voltage signals, it calculates active power, reactive power, apparent power, and power factor to generate actual electricity consumption data. Alternatively, the portable energy consumption diagnostic terminal can initially acquire and store a complete power supply network structure model, and subsequently query topology information locally.

[0036] Furthermore, during the data collection process, the portable energy consumption diagnostic terminal also performs auxiliary calibration on the collected data based on the hierarchical relationships indicated in the topology information. Specifically, the power supply network structure model also includes a topological constraint relationship where the sum of the electricity consumption data of each branch equals the total value of the main line. Based on this topological constraint relationship, the portable energy consumption diagnostic terminal cross-compares the electricity consumption data collected by itself with the electricity consumption data collected by the upper-level node terminal. If the actual deviation of the ratio between the branch electricity consumption data collected by itself and the main line electricity consumption data of the upper-level node exceeds a set deviation threshold compared to the theoretical ratio in the topological constraint relationship, it is determined that the collected data has a deviation. The terminal then calibrates the collected data according to the topological constraint relationship to ensure that the sum of the electricity consumption data of each branch equals the total value of the main line, thus satisfying the topological constraint relationship.

[0037] Therefore, the portable energy consumption diagnostic terminal determines the data acquisition strategy based on the topological relationship information in the power supply network structure model obtained from the data processing system, achieving the following beneficial effects: The acquisition strategy is adaptively matched with the node type: different sampling frequencies and acquisition durations are used for terminals in different topology locations, which not only ensures the monitoring accuracy of key nodes (backbone, high-power equipment) but also reduces the power consumption burden on ordinary branch nodes. Data self-calibration: The topological constraint relationship (the sum of branches equals the total value of the trunk) is used to cross-compare and calibrate the collected data, which improves the accuracy of non-contact mutual inductance acquisition. This provides a data foundation for subsequent multi-node collaborative analysis: data collected by each terminal according to a unified topology can be comprehensively analyzed in the cloud based on the power supply network structure model to achieve precise location of fault points.

[0038] Step S202: Determine the load forecasting strategy based on actual electricity consumption data.

[0039] For example, a load forecasting strategy is determined based on the length of time-series data of actual electricity consumption. Specifically, the length of time-series data processed by the portable energy consumption diagnostic terminal can be set to a base value of no more than 1 hour. If local processing is possible, the load forecasting strategy is a local load forecasting strategy. If local computing power is insufficient, data acquisition should be the primary focus, and the load forecasting strategy becomes a cloud-based load forecasting strategy, which involves transferring the computational task to the data processing system. If other portable energy consumption diagnostic terminals with sufficient computing power exist within the communication network, the load forecasting strategy becomes an assisted load forecasting strategy. Requests can be sent to other nearby portable energy consumption diagnostic terminals to assist this portable energy consumption diagnostic terminal in completing local computations (transferring some computational steps to other idle portable energy consumption diagnostic terminals and returning results, but all records are reported to the data processing system).

[0040] Step S203: Based on the load forecasting strategy, perform load forecasting on the actual electricity consumption data and output the forecast results.

[0041] In one example, S203 includes: if the load forecasting strategy is a local load forecasting strategy, then the actual electricity consumption data is input into a preset local model for forecasting, and a first forecast result is output in the forecast results; if the load forecasting strategy is a cloud-based load forecasting strategy, then the actual electricity consumption data is sent to a preset data processing system; the data processing system is used to generate a second forecast result in the forecast results; if the load forecasting strategy is an assisted load forecasting strategy, then some or all of the actual electricity consumption data is sent to other portable energy consumption diagnostic terminals locally; the other portable energy consumption diagnostic terminals are used to assist the portable energy consumption diagnostic terminals in generating a third forecast result in the forecast results.

[0042] In one example, if the load forecasting strategy is a local load forecasting strategy, the actual electricity consumption data is input into a preset local model for forecasting, and the first forecast result is output. This includes: if the load forecasting strategy is an assisted load forecasting strategy, the actual electricity consumption data is classified into states according to time series, and a time-series electricity consumption dataset with state labels is output; the state labels include steady-state labels, transient labels, and operating state labels; the time-series electricity consumption dataset is input into the local time-series model in the preset local model for trend analysis, and the first load forecasting result is output; the electricity consumption parameters and corresponding state labels at the current moment are extracted from the time-series electricity consumption dataset, and the electricity consumption parameters and corresponding state labels at the current moment are input into the local non-time-series model in the local model for parameter extrapolation, and the first fault prediction result is output; the first load forecasting result and the first fault prediction result are fused together to output the first forecasting result.

[0043] In one example, a hybrid prediction model is built within the data processing system. This model includes a cloud-based time-series model and a cloud-based non-time-series model. The data processing system generates a second prediction result, which includes: classifying the received actual electricity consumption data into states and outputting a time-series electricity consumption dataset with state labels; the state labels include steady-state labels, transient labels, and operating state labels; inputting the time-series electricity consumption dataset into the cloud-based time-series model within the hybrid prediction model for trend analysis and outputting a second load prediction result; extracting the current electricity consumption parameters and corresponding state labels from the time-series electricity consumption dataset, inputting both the current electricity consumption parameters and corresponding state labels into the cloud-based non-time-series model for parameter extrapolation, and outputting a second fault prediction result; and finally, fusing the second load prediction result and the second fault prediction result to output a second prediction result.

[0044] In one example, the data processing system includes an external interference model and an anomaly case library. The data processing system inputs the time-series electricity consumption dataset into the cloud-based time-series model in the hybrid prediction model for trend analysis and outputs a second load prediction result, including: the data processing system identifies the changing trend characteristics of the actual electricity consumption data through the cloud-based time-series model; if the changing trend characteristics match the anomaly trend characteristics within a preset time before the occurrence of a known fault in the anomaly case library, a second load prediction result is generated; the second load prediction result includes early warning information; the early warning information is sent to the user terminal; the early warning information includes the remaining time from the current moment to the predicted fault occurrence and the predicted fault type.

[0045] In one example, the data processing system extracts the current electricity consumption parameters and corresponding status labels from the time-series electricity consumption dataset, inputs the current electricity consumption parameters and corresponding status labels into a cloud-based non-time-series model for parameter estimation, and outputs a second fault prediction result. This includes: the data processing system performs cluster analysis on the actual electricity consumption data through the cloud-based non-time-series model, compares the obtained clustering results with historical abnormal cases in the abnormal case library, and generates a second fault prediction result; the second fault prediction result includes the location of the abnormal fault; and the location of the abnormal fault is sent to the user terminal.

[0046] In one example, an external disturbance model is built within the data processing system; the method also includes: the data processing system acquiring non-electrical factor data, inputting the non-electrical factor data into the external disturbance model, and outputting a quantization correction coefficient; correcting the quantization correction coefficient with the second prediction result, and outputting a multi-dimensional load prediction result.

[0047] In one example, if the load forecasting strategy is an assisted load forecasting strategy, then some or all of the actual electricity consumption data is sent to other portable energy consumption diagnostic terminals locally. This includes: if the load forecasting strategy is an assisted load forecasting strategy, then based on the actual computing power resources of the portable energy consumption diagnostic terminals, generating an assistance request and a local forecasting instruction; the assistance request includes the amount of actual electricity consumption data and / or the information of the calculation steps to be allocated; broadcasting or sending the assistance request to other portable energy consumption diagnostic terminals within a preset communication network; the other portable energy consumption diagnostic terminals generating response instructions to the assistance request based on the actual computing power status; if it is determined that the received response instruction is a confirmation instruction, then allocating the amount of data and / or the information of the calculation steps to the other portable energy consumption diagnostic terminals so that the other portable energy consumption diagnostic terminals can perform forecasting calculations and obtain assisted forecasting results; receiving the assisted forecasting results returned by the other portable energy consumption diagnostic terminals; executing the local forecasting instruction to obtain local forecasting results; and determining the assisted forecasting results and the local forecasting results as a third forecasting result.

[0048] For example, the preset local model includes a local time-series model and a local non-time-series model. The data processing system includes a hybrid prediction model and an external disturbance model. The hybrid prediction model includes a cloud-based time-series model and a cloud-based non-time-series model, used for load forecasting and fault diagnosis of electricity consumption data, respectively. The local time-series model is a lightweight version of the cloud-based time-series model, and the local non-time-series model is a lightweight version of the cloud-based non-time-series model. The local time-series model and the cloud-based time-series model are collectively referred to as time-series models, and the local non-time-series model and the cloud-based non-time-series model are collectively referred to as non-time-series models. A time-series model is a model for a continuous set of data. For example, it can substitute 480 current values ​​(collected at 15 seconds per point) collected between 1 PM and 3 PM today to estimate the current values ​​between 1 PM and 3 PM tomorrow. A non-time-series model, on the other hand, is an estimation of parameter values. For example, if today's power factor is 'a', and the current and voltage values ​​at node 1 are I1 and U1, it can estimate the current and voltage values ​​when the power factor changes to 'a'. Or, if today's temperature is T1, corresponding to I2 and U2, it can estimate the values ​​at T2.

[0049] Because pure time-series models are significantly affected by installation time and load characteristics in real-world engineering environments, non-time-series models, such as tree models, are introduced and stacked with time-series models. Introducing additional feature parameters into non-time-series models significantly enhances prediction performance.

[0050] The data processing system also maintains an anomaly case database containing historical anomaly cases. These cases include the dataset corresponding to the anomaly occurrence time t1, and the dataset within a preset time period (e.g., t0~t1) prior to the anomaly. A time-series model can be used to identify anomaly trends, allowing for early detection and warnings during load transitions from normal to abnormal states. Alternatively, a non-time-series model can be used, incorporating all relevant parameters into a machine learning-based model and employing cluster analysis for judgment. Besides analyzing and identifying fault nodes, the anomaly case database also needs to analyze the safety thresholds of each parameter. Below these thresholds, the system considers the user / power consumption scenario safe; however, once a parameter reaches its threshold, the system predicts a fault or safety risk within a defined timeframe, and this timeframe must be less than or equal to the time it takes for administrators or maintenance personnel to arrive on-site for maintenance.

[0051] Optionally, if the load forecasting strategy is a local load forecasting strategy, the actual electricity consumption data is input into a preset local model for forecasting, and the first forecast result is output. If the load forecasting strategy is a cloud-based load forecasting strategy, the actual electricity consumption data is sent to a preset data processing system; the data processing system forecasts the received actual electricity consumption data and generates a second forecast result. If the load forecasting strategy is an assisted load forecasting strategy, some or all of the actual electricity consumption data is sent to other portable energy consumption diagnostic terminals locally, and these other portable energy consumption diagnostic terminals assist the portable energy consumption diagnostic terminals in generating a third forecast result.

[0052] Optionally, the energy consumption diagnostic terminal mainly collects actual electricity consumption data such as current and voltage from users or loads. Both the energy consumption diagnostic terminal and the preset data processing system can further subdivide the actual electricity consumption data according to judgment weights, as shown in Table 1 below. The subdivision process is as follows: Table 1

[0053] Each data point included in the above parameters has a corresponding timestamp. Taking current as an example, the current dataset for a specific time period can be represented as: (1) Where k represents the number of sampling points; This represents the current data at the first sampling point. , The corresponding first timestamp; and so on, This represents the current data at the k-th sampling point. , The corresponding k-th timestamp.

[0054] The dataset can be divided into the following three categories of labels: 1. Steady-state label: If the fluctuation range of electrical data is within a preset value (e.g., ±5%) within a certain period of time, it is considered to be in a steady state. For example, between 3 and 4 o'clock, the current data collected by the portable energy consumption diagnostic terminal remains stable between 9.5A and 10.5A. The above current dataset (1) can be extended as follows: (2) 2. Transient Tag: Electrical data (instantaneous / continuous) exhibits significant transient rises and falls within a short period of time (i.e.,... K_th can be simply understood as fluctuation; among which... This represents the change in current. The value represents the change over time, and K_th represents the preset threshold. The dataset (1) above can be extended to: (3) 3. Operating State Labels: Due to the operating mode of the portable energy consumption diagnostic terminal, the electrical data repeatedly or periodically exhibits a certain regularity, which may be a mixture of steady-state and transient states, or an alternation of steady-state and transient states due to the switching of device operating modes. The above dataset (1) can be expanded as follows: (4) Optionally, if the load forecasting strategy is a local load forecasting strategy, the actual electricity consumption data is classified into states according to time series, and a time-series electricity consumption dataset with state labels is output; the state labels include steady-state labels, transient labels, and operating state labels. The time-series electricity consumption dataset is input into the local time-series model in the preset local model for trend analysis, and the first load forecast result is output. The first load forecast result includes the predicted fault time and the duration between the current time and the predicted fault time. The electricity consumption parameters and corresponding state labels at the current time are extracted from the time-series electricity consumption dataset, and the electricity consumption parameters and corresponding state labels at the current time are input into the local non-time-series model in the local model for parameter extrapolation, and the first fault forecast result is output. Finally, the first load forecast result and the first fault forecast result are fused to output the first forecast result.

[0055] Optionally, if the load forecasting strategy is a cloud-based load forecasting strategy, the portable energy consumption diagnostic instrument sends actual electricity consumption data to the data processing system via a communication module. The data processing system classifies the received actual electricity consumption data by state and outputs a time-series electricity consumption dataset with state labels; the state labels include steady-state labels, transient labels, and operating state labels. The time-series electricity consumption dataset is input into the cloud-based time-series model in the hybrid forecasting model for trend analysis. The cloud-based time-series model identifies the changing trend characteristics of the actual electricity consumption data. If the changing trend characteristics match the abnormal trend characteristics within a preset time before the occurrence of known faults in the abnormal case library, a second load forecast result is generated; the second load forecast result includes warning information; the warning information is sent to the user terminal; the warning information includes the remaining time from the current moment to the predicted fault occurrence, the predicted fault type, and the predicted electricity consumption at the future forecast time, etc. The actual electricity consumption data is clustered using a cloud-based non-time-series model, and the obtained clustering results are compared with historical abnormal cases in the abnormal case library to generate a second fault prediction result; the second fault prediction result includes the location of the abnormal fault; the location of the abnormal fault is sent to the user terminal. Finally, the second load prediction result and the second fault prediction result are merged to output the second prediction result.

[0056] The data processing system acquires non-electrical factor data, inputs it into an external interference model, outputs a quantified correction coefficient, and then corrects the second prediction result using this coefficient, outputting a multi-dimensional load prediction result. This multi-dimensional load prediction result includes the comprehensively predicted abnormal fault type, abnormal fault location, comprehensively predicted electricity consumption trend over a preset future time period, the quantified correction coefficient, and the corrected prediction result. When the electricity consumption parameter trend in the multi-dimensional load prediction result reaches a preset safety threshold, the data processing system issues an early warning signal.

[0057] In one embodiment, the correction is as follows: the quantization correction coefficient is multiplied by the predicted electricity consumption in the second prediction result to obtain the corrected predicted electricity consumption; the safety threshold is then re-determined based on the corrected predicted electricity consumption. In another embodiment, the correction is as follows: the quantization correction coefficient is used as an additional feature and input along with the second prediction result into a preset classifier, which then outputs the final multi-dimensional prediction result. The correction method is not limited.

[0058] Optionally, if the load forecasting strategy is an assisted load forecasting strategy, i.e., when the local processing unit of the portable energy consumption diagnostic terminal determines that its own computing power is insufficient to complete the current load forecasting task, the assisted load forecasting strategy is activated: an assistance request is sent to other portable energy consumption diagnostic terminals within the communication network via the communication module, allowing other terminals to share some of the calculation steps. After receiving the assisted calculation results, the portable energy consumption diagnostic terminal reports the assisted calculation results along with the local forecasting results to the data processing system and user terminals, etc. Specifically, based on the actual computing power resources of the portable energy consumption diagnostic terminal, an assistance request and a local forecasting instruction are generated; the assistance request includes the amount of actual electricity consumption data and / or the information on the calculation steps to be allocated. The assistance request is broadcast or directed to other portable energy consumption diagnostic terminals within the preset communication network. Other portable energy consumption diagnostic terminals generate response instructions to the assistance request according to their actual computing power status and send the response instructions to the portable energy consumption diagnostic terminal. If the portable energy consumption diagnostic terminal determines that the received response command is a confirmation command, it will allocate the data volume and / or calculation process information to other portable energy consumption diagnostic terminals, so that the other portable energy consumption diagnostic terminals can perform predictive calculations and obtain assisted prediction results; and receive the assisted prediction results returned by other portable energy consumption diagnostic terminals, and based on the assisted prediction results, continue to execute the local prediction command to obtain a third prediction result.

[0059] For example, a portable energy consumption diagnostic terminal has a local processing unit internally deployed. This unit has a built-in resource monitoring submodule that monitors the current CPU utilization and memory usage in real time. When the computational complexity of the data to be processed exceeds the local processing capacity (e.g., current CPU utilization ≥ 85% for more than 3 seconds), and / or, the length of the time-series data to be processed in the current load forecasting task exceeds a preset 1-hour threshold, the local processing unit determines that the current computing power is insufficient to independently complete the load forecasting task. It then sends a computing power detection signal to other portable energy consumption diagnostic terminals within the communication network via the communication module. Upon receiving the detection signal, each other portable energy consumption diagnostic terminal checks its own current computing power status (CPU utilization, memory usage, current task queue length, etc.), generates status information, and returns this status information to the local portable energy consumption diagnostic terminal. The status information carries information about the currently available computing power resources and the types of computational tasks that can be undertaken.

[0060] Based on the returned status information, this portable power consumption diagnostic terminal determines whether there are other portable power consumption diagnostic terminals within the communication network with available computing power exceeding a set threshold. If such terminals exist, an assisted load forecasting strategy is triggered, selecting one or more other portable power consumption diagnostic terminals as assisted computing nodes according to either the "proximity priority" principle (highest signal strength priority) or the "computing power priority" principle (highest available computing power priority). If none exist, the assisted load forecasting strategy is not triggered; instead, a cloud-based load forecasting strategy can be triggered, and the data to be processed is reported to the data processing system for in-depth calculations.

[0061] After triggering the load forecasting assistance strategy, this portable energy consumption diagnostic terminal generates an assistance request and sends it to the assisting computing node, assigning computing tasks accordingly. The assistance request carries the following information: the device identifier of this portable energy consumption diagnostic terminal, the current task type identifier, the amount of data to be processed, a description of the required computing procedure (e.g., "Trend analysis of the currently collected time-series electricity consumption data is required"), and the latest response time limit. For example, this portable energy consumption diagnostic terminal sends longer data packets (e.g., current time-series data exceeding 30 minutes) of time-series electricity consumption data that require trend analysis to the assisting computing node, which then uses these packets in its time-series model for calculation. This portable energy consumption diagnostic terminal retains non-time-series model calculation tasks (e.g., correlation estimation of various electricity consumption parameters at the current moment) for local processing.

[0062] After receiving the allocated data, the assisting computing node invokes its built-in time-series model to perform trend analysis and calculations, outputting an assisting prediction result (e.g., "The current will continue to rise to 25A in the next 30 minutes"). Upon completion of the calculation, the assisting computing node returns the prediction result to the portable energy consumption diagnostic terminal via the communication module. After receiving the prediction result, the portable energy consumption diagnostic terminal correlates and merges it with the results of other calculations performed by its local processing unit (i.e., local prediction results) to generate a complete analysis report containing both the local prediction results and the assisting calculation results. The portable energy consumption diagnostic terminal then submits the complete analysis report to the data processing system via the communication module. The data processing system receives the report and performs unified archiving and comprehensive analysis.

[0063] For example, Figure 3 A load forecasting flowchart for an assisted load forecasting strategy provided in this application embodiment is shown below. Figure 3 As shown, it includes: terminal A (i.e., this portable energy consumption diagnostic terminal) and terminal B (i.e., an assisting computing node with idle computing power resources). Terminal A requests assistance from terminal B. Terminal A sends part A of the obtained calculation result to the data processing system, and terminal B sends part B of the obtained calculation result to the data processing system.

[0064] Optionally, regarding the priority of cloud-based load forecasting strategies and assisted load forecasting strategies, the assisted load forecasting strategy can be triggered first. If no other portable energy consumption diagnostic terminal can be found to assist or assistance fails, then the cloud-based load forecasting strategy can be executed. Alternatively, the cloud-based load forecasting strategy can be triggered first, without any limitation on this.

[0065] Optionally, the model itself is subject to non-time-series characteristics such as equipment installation time limits and load changes, requiring periodic model iteration to ensure model accuracy meets operational and maintenance requirements. Iteration methods include autonomous iteration and expert-guided iteration. Autonomous iteration in this solution mainly includes autoregressive model training, timed model retraining, and timed retraining of multiple model stacking weights. Expert-guided iteration in this solution mainly includes modifying model feature parameters, correcting false alarm causes, correcting model post-processing parameters, and updating the model stacking structure. The models requiring iteration include: The power supply network structure model helps terminals collect electricity consumption data more accurately and improves calculation accuracy.

[0066] Load analysis models (time-series + non-time-series) can more accurately identify fault points and predict the likelihood of faults occurring earlier.

[0067] External disturbance model is used to quantify how external influences cause users to change their electricity consumption behavior / habits in a scenario. For example, high temperatures in summer lead to high electricity consumption during the day in office scenarios, high electricity consumption at night in residential scenarios, and a surge in electricity consumption at sports venues on weekends or when events are held.

[0068] It should be noted that if the requirements for predictive performance are not high, or the load type is relatively simple, a traditional industrial computer can be used in conjunction with a host computer artificial intelligence algorithm for life prediction and fault detection.

[0069] Step S204: Send the prediction results to the user terminal.

[0070] For example, the prediction results are sent to the user terminal.

[0071] The method provided in this application collects actual power consumption data of the load line. Based on the actual power consumption data, a load forecasting strategy is determined. According to the load forecasting strategy, load forecasting is performed on the actual power consumption data, and the forecasting results are output; the forecasting results are then sent to the user terminal. In this solution, the portable energy consumption diagnostic terminal is plug-and-play, requiring no complex installation to quickly acquire power consumption data; according to the load forecasting strategy, it quickly identifies power consumption risks, accurately predicts load changes, and improves forecasting accuracy, achieving rapid and accurate diagnosis and forecasting of user load with minimal construction. Furthermore, by combining electrical factors (i.e., actual power consumption data) and non-electrical factors (non-electrical factor data), it comprehensively identifies faults and predicts risks, improving forecasting accuracy and providing timely warnings; through time-series models, non-time-series models, and anomaly case libraries, it accurately identifies fault information and achieves long-term prediction, issuing warnings at the initial stage of risk occurrence, without waiting for critical points such as overheating or overload.

[0072] In one embodiment, this application also provides a flowchart illustrating another method for predicting electricity load for smart power supply protection, such as... Figure 4 As shown, it includes: (1) Portable energy consumption diagnostic terminal: It obtains the user's electricity consumption information by mutual inductance. The device is plug-and-play and can preliminarily determine whether the load status needs to be warned through its own data processing capabilities.

[0073] (2) Communication Module: The portable energy consumption diagnostic terminal reports the actual electricity consumption data collected to a powerful data processing system, and simultaneously synchronizes the updated models (time-series, non-time-series, etc.) from the data processing system to the portable energy consumption diagnostic terminal. Furthermore, since the portable energy consumption diagnostic terminal uses a contactless method to acquire electricity consumption data, errors may exist. It is necessary to use the power supply network structure model to correct these errors and to communicate and compare data with the gateway terminal to prevent significant deviations. Simply put, the sum of the electricity consumption data of each branch should be less than or equal to the total value of the main line. If a significant deviation occurs, the fault location can be preliminarily inferred using the communication module and the topological relationship information in the power supply network structure model.

[0074] (3) Data processing system: Build mathematical models (hybrid prediction model, external interference model, power supply network structure model, etc.) and perform calculations on the data collected by the portable energy consumption diagnostic terminal to obtain optimized results.

[0075] (4) It also includes the power grid, users, and user data (i.e. actual electricity consumption data), with the power grid used to supply power to users.

[0076] Corresponding to the above method, this application also provides an electricity load prediction device for intelligent power supply, applied to a portable energy consumption diagnostic terminal, which is non-intrusively installed outside the load line; such as Figure 5 As shown, the device includes: Acquisition module 41 is used to collect actual power consumption data of the load line; Module 42 is used to determine the load forecasting strategy based on actual electricity consumption data; The prediction module 43 is used to perform load prediction on the actual electricity consumption data according to the load prediction strategy, output the prediction results, and send the prediction results to the user terminal.

[0077] The functions of each functional unit of the power load forecasting device for smart power supply provided in the above embodiments of this application can be implemented through the above methods and steps. Therefore, the specific working process and beneficial effects of each unit in the power load forecasting device for smart power supply provided in the embodiments of this application will not be repeated here.

[0078] This application also provides an electronic device, such as... Figure 6 As shown, it includes a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540.

[0079] Memory 530 is used to store computer programs; The processor 510 performs the above steps when executing the program stored in the memory 530.

[0080] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0081] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0082] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0083] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0084] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 1 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0085] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform any of the above embodiments of the power load prediction method for smart power supply protection.

[0086] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the above embodiments of the power load prediction method for smart power supply protection.

[0087] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take 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.) containing computer-usable program code.

[0088] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0089] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0090] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0091] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0092] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims in this application and their equivalents, then this application also intends to include these modifications and variations.

Claims

1. A method for predicting electricity load for intelligent power supply, characterized in that, An application to a portable energy consumption diagnostic terminal, wherein the portable energy consumption diagnostic terminal is disposed externally to the load line in a non-invasive manner; the method includes: Collect the actual power consumption data of the load line; Based on the actual electricity consumption data, a load forecasting strategy is determined; According to the load forecasting strategy, load forecasting is performed on the actual electricity consumption data, and the forecasting results are output; the forecasting results are then sent to the user terminal.

2. The method as described in claim 1, characterized in that, Based on the load forecasting strategy, load forecasting is performed on the actual electricity consumption data, and the forecasting results are output, including: If the load forecasting strategy is a local load forecasting strategy, then the actual electricity consumption data is input into a preset local model for forecasting, and the first forecast result in the forecast results is output. If the load forecasting strategy is a cloud-based load forecasting strategy, then the actual electricity consumption data is sent to a preset data processing system; the data processing system is used to generate a second forecast result in the forecast results; If the load forecasting strategy is an assistive load forecasting strategy, then some or all of the actual electricity consumption data will be sent to other portable energy consumption diagnostic terminals in the local area; the other portable energy consumption diagnostic terminals are used to assist the portable energy consumption diagnostic terminals in generating the third forecast result in the forecast results.

3. The method as described in claim 2, characterized in that, If the load forecasting strategy is a local load forecasting strategy, then the actual electricity consumption data is input into a preset local model for forecasting, and the first forecast result in the forecast results is output, including: If the load forecasting strategy is a local load forecasting strategy, the actual electricity consumption data is classified into states according to the time series, and a time series electricity consumption dataset with state labels is output; the state labels include steady-state labels, transient labels and operating state labels; The time-series electricity consumption dataset is input into the local time-series model in the preset local model for trend analysis, and the first load forecast result is output. Extract the current power consumption parameters and corresponding status labels from the time-series power consumption dataset, input the current power consumption parameters and corresponding status labels into the local non-time-series model in the local model for parameter estimation, and output the first fault prediction result. The first load prediction result and the first fault prediction result are fused together to output the first prediction result.

4. The method as described in claim 2, characterized in that, The data processing system incorporates a hybrid prediction model, which includes a cloud-based time-series model and a cloud-based non-time-series model. The data processing system generates a second prediction result from the prediction results, including: The data processing system classifies the received actual electricity consumption data into states and outputs a time-series electricity consumption dataset with state labels; the state labels include steady-state labels, transient labels, and operating state labels. The data processing system inputs the time-series electricity consumption dataset into the cloud-based time-series model in the hybrid prediction model for trend analysis and outputs a second load prediction result. The data processing system extracts the current power consumption parameters and corresponding status labels from the time-series power consumption dataset, inputs the current power consumption parameters and corresponding status labels into the cloud-based non-time-series model for parameter estimation, and outputs the second fault prediction result. The data processing system merges the second load prediction result and the second fault prediction result to output a second prediction result.

5. The method as described in claim 4, characterized in that, The data processing system includes an external interference model and an anomaly case library. The system inputs the time-series electricity consumption dataset into the cloud-based time-series model within the hybrid prediction model for trend analysis, and outputs the second load prediction result, including: The data processing system identifies the changing trend characteristics of the actual electricity consumption data through the cloud-based time-series model; if the changing trend characteristics match the abnormal trend characteristics within a preset time period before the occurrence of known faults in the abnormal case database, then the second load prediction result is generated; the second load prediction result includes early warning information; the early warning information is sent to the user terminal; the early warning information includes the remaining time from the current moment to the predicted fault occurrence and the predicted fault type; The data processing system extracts the current electricity consumption parameters and corresponding status labels from the time-series electricity consumption dataset, inputs the current electricity consumption parameters and corresponding status labels into the cloud-based non-time-series model for parameter estimation, and outputs the second fault prediction result, including: The data processing system performs cluster analysis on the actual electricity consumption data using the cloud-based non-time-series model, compares the obtained clustering results with historical abnormal cases in the abnormal case library, and generates the second fault prediction result; the second fault prediction result includes the location of the abnormal fault; and sends the location of the abnormal fault to the user terminal.

6. The method as described in claim 4, characterized in that, The data processing system incorporates an external interference model; the method further includes: The data processing system acquires non-electrical factor data, inputs the non-electrical factor data into the external disturbance model, and outputs a quantization correction coefficient; the quantization correction coefficient is then used to correct the second prediction result, and a multi-dimensional load prediction result is output.

7. The method as described in claim 2, characterized in that, If the load forecasting strategy is an assistive load forecasting strategy, then some or all of the actual electricity consumption data will be sent to other portable energy consumption diagnostic terminals locally, including: If the load forecasting strategy is an assisted load forecasting strategy, then based on the actual computing power resources of the portable energy consumption diagnostic terminal, an assistance request and a local forecasting instruction are generated; the assistance request includes the amount of data of the actual electricity consumption data and / or the information of the calculation process to be allocated. The assistance request is broadcast or directed to other portable energy consumption diagnostic terminals within a preset communication network; the other portable energy consumption diagnostic terminals are used to generate response instructions to the assistance request based on the actual computing power status. If the received response instruction is determined to be a confirmation instruction, the data volume and / or calculation process information are allocated to the other portable energy consumption diagnostic terminals so that the other portable energy consumption diagnostic terminals can perform predictive calculations and obtain assisted prediction results. Receive the assisted prediction results returned by the other portable energy consumption diagnostic terminals; Execute the local prediction instruction to obtain the local prediction result; determine the assisted prediction result and the local prediction result as the third prediction result.

8. The method as described in claim 1, characterized in that, Collect the actual power consumption data of the load line, including: A model acquisition instruction is sent to a preset data processing system; the model acquisition instruction is used to acquire a power supply network structure model, the power supply network structure model including the topology information of the load lines; Obtain the power supply network structure model sent by the data processing system; Based on the topological relationship information indicated in the power supply network structure model, a data acquisition strategy is determined; According to the acquisition strategy, the actual power consumption data of the load line is collected.

9. A power load prediction device for intelligent power supply, characterized in that, A portable energy consumption diagnostic terminal is applied to a device that is non-invasively installed outside the load circuit; the device includes: The data acquisition module is used to collect the actual power consumption data of the load line; The determination module is used to determine a load forecasting strategy based on the actual electricity consumption data; The prediction module is used to perform load prediction on the actual electricity consumption data according to the load prediction strategy, output the prediction result, and send the prediction result to the user terminal.

10. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-8.

Images