A power distribution load monitoring system in a power grid

Abstract

The present invention provides a power distribution load monitoring system within a power grid, which relates to the field of power network load monitoring. A load monitoring server is communicatively connected to an electricity user monitoring terminal, which is provided with a current transformer, a voltage transformer, an electric energy meter, a time module, a storage device, a communication module, and a data processor. The electricity user monitoring terminal collects electricity user load information in real time and, based on control instructions from the load monitoring server, uploads the collected electricity user load information and the load collection time corresponding to the load information to the load monitoring server. The load monitoring server stores the collected information and displays it according to monitoring instructions from monitoring personnel. It also predicts the collected information and predicts the load status of the electricity user, and issues an alarm if an abnormality occurs. The load of each power supply area is effectively monitored, and abnormalities can be discovered in a timely manner and promptly notified to the monitoring personnel.

Description

Technical Field

[0001] The present invention relates to the field of power network load monitoring, and in particular to a power distribution load monitoring system within a power grid. Background Art

[0002] Electric load is the total amount of electrical power drawn from the power system by electricity users' electrical equipment at a given moment. Based on the different load characteristics of electricity users, electric load can be categorized into various types, such as industrial load, agricultural load, transportation load, and household load. Electric load can be expressed in terms of electrical power. In AC circuits, electrical power consists of active power, also known as active load, and reactive power, also known as reactive load or reactive power.

[0003] Power load is a crucial parameter for the stable operation of the power grid. Excessive increases in load in a particular area can lead to overloaded power equipment. Prolonged overloads can easily cause equipment failures, impacting the stable operation of the power grid. This necessitates effective monitoring of the power load in each power supply area to ensure power grid stability and the normal functioning of people's lives and work. Therefore, how to effectively monitor the power load in each power supply area, detect anomalies promptly, and promptly alert monitoring personnel to address them before they escalate and impact the entire power grid is a pressing technical challenge. Summary of the Invention

[0004] The present invention provides a power distribution load monitoring system within a power grid, which can effectively monitor the power load of each power supply area, detect abnormalities in time, and promptly prompt monitoring personnel.

[0005] The power distribution load monitoring system in the power grid includes: load monitoring server and power user monitoring terminal;

[0006] The load monitoring server is connected to the electricity user monitoring terminal and sends control instructions to the electricity user monitoring terminal;

[0007] The user monitoring terminal is equipped with a current transformer, a voltage transformer, an electric energy meter, a time module, a storage device, a communication module, and a data processor, which are installed in the user. The user monitoring terminal collects user load information in real time and stores it in the storage device. The user monitoring terminal also uploads the collected user load information and the load collection time corresponding to the load information to the load monitoring server based on the control instructions of the load monitoring server.

[0008] The load monitoring server stores the acquired information and displays it according to the monitoring instructions of the monitoring personnel. It also predicts the acquired information and the load status of the electricity user, and issues an alarm if any abnormality occurs.

[0009] Preferably, the load monitoring server is configured with an electricity user load monitoring model, and compares the load collection time of the collected electricity user load information with the electricity user load monitoring model to obtain first load state comparison information;

[0010] The load monitoring server predicts the load state of the electricity user from the first load state comparison information;

[0011] The load monitoring server compares the user load information with the predicted user load information to obtain second load status comparison information. If the second load status comparison information exceeds a preset power load threshold, a load status abnormality warning is generated.

[0012] Preferably, the load monitoring server calculates the real-time user load information and the corresponding predicted user load information of the same user, and obtains the second load state comparison information by subtracting the corresponding predicted user load information from the real-time user load information;

[0013] If the second load status comparison information is a positive number and is greater than a preset power load threshold, a load status abnormality warning is generated.

[0014] Preferably, the load monitoring server retrieves historical electricity user load information and obtains historical normal electricity user load information from the historical electricity user load information;

[0015] Obtain load collection time period information from historical normal electricity user load information;

[0016] The load monitoring server obtains the load state set to be trained based on the historical normal power user load information and load collection time period information;

[0017] After training the load state set to be trained, the electricity user load monitoring model is obtained.

[0018] Preferably, the load monitoring server is further used to divide the load time points to be processed from the load collection time period information; filter out repeated power load data from each load time point to be processed; and perform load information extension processing on the power load data to obtain a load state set to be trained.

[0019] Preferably, the load monitoring server normalizes the load state set to be trained to obtain a training set to be divided; and divides the training set to be divided according to a preset load monitoring rule to obtain a training set to be regressed;

[0020] The load monitoring server uses a BP neural network including an input layer and several hidden layers to perform regression prediction on the regressive electric load training set to obtain the load monitoring model of the electricity user;

[0021] Based on the random generation module, the weight matrices of the hidden layer and the output layer are initialized; the mean square error is used as the loss function, and gradient descent is used to gradually optimize the load monitoring model of electricity users.

[0022] Preferably, the load monitoring server reorders the processed data, i.e., the training set to be divided, and then performs data segmentation processing to form a load information training set and a load information test set based on the training model and the test model, and then randomly divides the load information training set and the load information to obtain the electric load training set to be regressed.

[0023] Preferably, the load monitoring server obtains the divided training model. In each training round, it first performs feedforward calculation, then uses MSE to calculate the error, and then uses the error to perform back propagation calculation to update the parameters. After each training, the error between the prediction module and the training model is calculated using MSE, and the training model is adjusted according to the error result.

[0024] The training model is tested using the load monitoring model of electricity users that meets the requirements, the corresponding prediction module is obtained, the error between the prediction module and the training model is calculated, and adjustments are made according to the electricity load fitting requirements until the requirements are met.

[0025] Preferably, the load monitoring server sends a control instruction to the electricity user monitoring terminal according to the monitoring call instruction to obtain the electricity user load information and the load collection time corresponding to the load information;

[0026] The load monitoring server uses the electricity user load monitoring model and combines the average value of the electricity user load information in the current polling period as the predicted electricity user load information.

[0027] Preferably, after obtaining the monitoring retrieval instruction, the load monitoring server determines whether the network delay between the load monitoring server and the electricity user monitoring terminal exceeds the preset delay threshold; if so, the historical electricity user load information is determined based on the data type identification information in the control instruction; the electricity user load monitoring model is used in combination with the historical electricity user load information to predict the average value of the electricity user load information in the current polling cycle to obtain the predicted electricity user load information.

[0028] It can be seen from the above technical solutions that the present invention has the following advantages:

[0029] The load monitoring server in the power distribution load monitoring system provided by the present invention can predict whether the power load exceeds the maximum power load during the load collection period based on the load information of the power users. This can effectively monitor the power load of each power supply area, detect any anomalies in a timely manner, and promptly notify the monitoring personnel.

[0030] By pre-training the load monitoring model of the electricity user, the present invention can use the load monitoring model of the electricity user to detect the load information of the electricity user after the monitoring terminal located in the electricity user obtains the load information of the electricity user, and then can judge the current electricity load of a certain electricity user according to the detection result and compare it with the corresponding prediction result. The present invention obtains whether the electricity user has exceeded the threshold status through comparison, and then generates an abnormal load status warning, which enables the monitoring personnel to monitor the electricity user and check and handle it in time.

[0031] This invention leverages existing user load information and a user load monitoring model to address communication delays during polling cycles. When the load monitoring server obtains actual user load information, it stores it in a local database. This invention ensures timely and consistent response from the load monitoring server, prevents network issues from impacting user monitoring, and ensures stable operation of the power network. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for the description. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 This is a schematic diagram of the distribution load monitoring system in the power grid;

[0034] Figure 2 This is a schematic diagram of the electricity user monitoring terminal. DETAILED DESCRIPTION

[0035] like Figure 1 and 2 The diagrams provided in the power distribution load monitoring system within the power grid provided by the present invention are only used to illustrate the basic concept of the present invention in a schematic manner. Therefore, the diagrams only show modules related to the present invention rather than the number and functions of modules in actual implementation. In actual implementation, the functions, quantity and effects of each module may be changed at will, and the functions and uses of the modules may also be more complex.

[0036] The power distribution load monitoring system within the power grid can acquire and process relevant data based on artificial intelligence technology. The system utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, and to develop theories, methods, technologies, and application devices for perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0037] Power distribution load monitoring systems within power grids incorporate both hardware and software technologies. These fundamental technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. Software technologies for intelligent diagnostic methods for CNC machine tools primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0038] The power distribution load monitoring system within the power grid also has a machine learning function, wherein the machine learning and deep learning in the method of the present invention generally include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and formula-based learning technologies.

[0039] The server of the power distribution load monitoring system within the power grid may also include network devices and / or user devices. The network devices include, but are not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of hosts or network servers.

[0040] The network where the distribution load monitoring system in the power grid is located includes but is not limited to the Internet, wide area network, metropolitan area network, local area network, virtual private network (VPN), etc.

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] The power distribution load monitoring system provided by the present invention comprises: a load monitoring server and a power user monitoring terminal;

[0043] The load monitoring server is connected to the electricity user monitoring terminal and sends control instructions to the electricity user monitoring terminal to obtain data information.

[0044] The electricity user monitoring terminal is equipped with a current transformer, a voltage transformer, an electricity meter, a time module, a storage, a communication module and a data processor installed in the electricity user; the electricity user monitoring terminal collects the electricity user's load information in real time and stores it in the storage. The electricity user monitoring terminal also uploads the collected electricity user load information and the load collection time corresponding to the load information to the load monitoring server based on the control instructions of the load monitoring server.

[0045] Electricity users can be industrial plants, certain critical equipment, office buildings, office districts, residential electricity users in large areas, or commercial areas, among others. Load collection time refers to the time when user load information is acquired. Real-time user load information refers to the data on the user's load consumption and load flow within the user load information.

[0046] Furthermore, corresponding sensors can be set up as needed through current transformers, voltage transformers, electricity meters, time modules, etc. to monitor the load information of electricity users in real time and record the current load collection time as the load information of electricity users within the load collection time.

[0047] The load monitoring server configures an electricity user load monitoring model, compares the load collection time of the collected electricity user load information with the electricity user load monitoring model, and obtains first load status comparison information.

[0048] In an exemplary embodiment, the load monitoring server configures a user load monitoring model based on preset conditions and a preset algorithm for monitoring power load to determine whether the power load of each user exceeds a preset load threshold of the user load monitoring model during the collection time. The first load status comparison information is based on time data in the user load monitoring model that corresponds to the load collection time in the user load information. Using the user load monitoring model, the load collection time of the user load information can be input into the user load monitoring model to obtain the first load status comparison information. The load monitoring server predicts the user load status from the first load status comparison information.

[0049] In this way, the load monitoring server can predict whether the power load exceeds the maximum power load during the load collection time based on the power user's load information.

[0050] Furthermore, the time data corresponding to the load collection time of the user load monitoring model is retrieved from the first load status comparison information, and the corresponding predicted user load information is obtained from the user load monitoring model based on the time data.

[0051] Since the electricity user monitoring terminals are installed in different electricity users, the electricity loads of different electricity users under normal circumstances may also be different. Therefore, the predicted electricity load data corresponding to each electricity user monitoring terminal may also be different.

[0052] The load monitoring server compares the user load information with the predicted user load information to obtain second load status comparison information. If the second load status comparison information exceeds a preset power load threshold, a load status abnormality warning is generated.

[0053] The second load status comparison information of the present invention is a comparison result formed by comparing the user load information with the predicted user load information. The load monitoring server can pre-set the load threshold value to determine whether the current power load reaches or exceeds the load status abnormality warning.

[0054] In the present invention, the real-time user load information and the corresponding predicted user load information of the same user are calculated, and the second load state comparison information is obtained by subtracting the corresponding predicted user load information from the real-time user load information.

[0055] Furthermore, if the second load status comparison information is a positive number and is greater than a preset power load threshold, a load status abnormality warning is generated.

[0056] The abnormal load status warning is sent to the monitoring terminal of the monitoring personnel, so that the monitoring personnel can monitor the electricity users to monitor whether there are abnormal power consumption.

[0057] In an embodiment of the present invention, by pre-training the electricity user load monitoring model, after the electricity user load information is obtained at the monitoring terminal located in the electricity user, the electricity user load monitoring model can be used to detect the electricity user load information, and then the current electricity load of a certain electricity user can be judged according to the detection result and compared with the corresponding prediction result. The present invention obtains whether the electricity user exceeds the threshold status through comparison, and then generates an abnormal load status alert, which enables the monitoring personnel to monitor the electricity user and check and handle it in time.

[0058] In the present invention, the load monitoring server retrieves historical electricity user load information and obtains historical normal electricity user load information from the historical electricity user load information.

[0059] The historical user load information is the user load information of each user in the past. The historical normal user load information is the normal load data of each user within a preset range of users.

[0060] Specifically, the load monitoring server retrieves historical user load information from a database storing historical user load information, and selects data with abnormal load that is not marked as abnormal from the historical user load information as historical normal user load information.

[0061] The load monitoring server obtains load collection time period information from historical normal power user load information.

[0062] In the present invention, load collection time period information is set according to the electricity consumption time of the electricity user to be monitored or the time period to be monitored.

[0063] The load monitoring server obtains the load state set to be trained based on the historical normal power user load information and load collection time period information.

[0064] In an exemplary embodiment, the load state set to be trained refers to a set of historical normal load conditions stored and recorded within a load collection period.

[0065] Specifically, the historical electricity user load information corresponding to each monitoring device is matched with the load collection time period information to form a load state set to be trained.

[0066] After training the load state set to be trained, a user load monitoring model is obtained. The present invention can use a BP neural network to train the load state set to be trained, so that the trained user load monitoring model can detect whether the current user load exceeds a threshold value according to different time periods.

[0067] In an exemplary embodiment, the load monitoring server is further configured to divide the load collection time period information into unprocessed load time points. The unprocessed load time points are used to divide the load collection time period information and filter out time periods for different users from historical normal electricity user load information. Repeated electricity load data is filtered from each unprocessed load time point.

[0068] Specifically, the load monitoring server uses the load information deduplication module to filter out repeated load information from the load time point to be processed, that is, to filter out repeated power load information and only retain the first load information of the repeated value, that is, to retain the repeated power load information only once.

[0069] The load monitoring server applies load information extension processing to the power load data to obtain a load state set to be trained.

[0070] In the present invention, the current information, voltage information, electric energy information, voltage fluctuation information, flow information, maximum current, and maximum voltage of electricity users can be used as training sample sets, and the load information of electricity users corresponding to each collection time can be used as the load state set to be trained.

[0071] In the present invention, to enhance the stability of the process of training a user load monitoring model, a preprocessing function is used to normalize the data. After the load monitoring server trains the load state set to be trained, the user load monitoring model is obtained. The steps include the following: S2041: The load monitoring server normalizes the load state set to be trained to obtain a training set to be divided.

[0072] The load monitoring server divides the training set to be divided according to the preset load monitoring rules to obtain the training set to be regressed.

[0073] To elaborate, the processed data, i.e., the training set to be divided, is reordered and then segmented. Based on the training model and the test model, a load information training set and a load information test set are formed. The load information training set and the load information are then randomly divided to obtain the load training set to be regressed.

[0074] The load monitoring server uses a BP neural network including an input layer and several hidden layers to perform regression prediction on a training set of regressive electric loads to obtain a load monitoring model for electric users.

[0075] In the present invention, the electricity user load monitoring model includes a BP neural network with an electricity load input layer, multiple electricity load hidden layers and an electricity load output layer for regression prediction. The number of nodes in the output layer in the algorithm is determined by the number of features obtained based on the status of the electricity user, and the number of nodes in the output layer is 1.

[0076] The load monitoring server is based on randomly generated modules and initialized weight matrices of hidden and output layers. The mean square error is used as the loss function, and gradient descent is used to gradually optimize the load monitoring model for electricity users.

[0077] The load monitoring server adopts a variable adaptive learning rate to automatically set different learning rates at different stages of network training and preset the number of iteration rounds.

[0078] In order to reduce the error of the electricity user load monitoring model obtained by training, the present invention obtains a divided training model. In each training round, feedforward calculation is first performed, then the error is calculated using MSE, and then the error is back-propagated to update the parameters. After each training, the error between the prediction module and the training model is calculated using MSE, and the training model is adjusted according to the error result.

[0079] The training model is tested using the load monitoring model of electricity users that meets the requirements, the corresponding prediction module is obtained, the error between the prediction module and the training model is calculated, and adjustments are made according to the electricity load fitting requirements until the requirements are met.

[0080] The electric load fitting requirements are set according to the electricity consumption characteristics, electricity consumption reserve, load threshold and other conditions of the electricity user load monitoring model, so that the load situation detected by the electricity user load monitoring model meets the requirements.

[0081] The load monitoring server sends control instructions to the user monitoring terminal according to the monitoring call instruction to obtain the user's load information and the load collection time corresponding to the load information.

[0082] The load monitoring server uses the electricity user load monitoring model and combines the average value of the electricity user load information in the current polling period as the predicted electricity user load information.

[0083] Specifically, the control instruction includes the electricity user identification information and the load collection time.

[0084] The load monitoring server is further used to determine the current time and combine it with the average value of the power user load information in the current polling period as the predicted power user load information.

[0085] The load monitoring server stores the acquired load information of the electricity users and the load collection time corresponding to the load information, and also displays it according to the instructions.

[0086] After the actual power user load information of the monitored object is acquired, the predicted power user load information is overwritten with the actual power user load information.

[0087] The present invention solves the problem that the system cannot obtain the electricity user load information uploaded by the electricity user monitoring terminal in real time due to network communication, electricity user distribution and other reasons, resulting in the user not receiving a response to the request for obtaining the electricity user load information after the monitoring retrieval instruction is initiated, thereby affecting monitoring.

[0088] After receiving the monitoring call instruction, the load monitoring server determines whether the network delay between the load monitoring server and the user monitoring terminal exceeds the preset delay threshold; if so, it determines the historical load information of the user according to the data type identification information in the control instruction; uses the load monitoring model of the user to combine the historical load information of the user to predict the average load information of the user in the current polling cycle, and obtains the predicted load information of the user.

[0089] This invention leverages existing user load information and a user load monitoring model to address communication delays during polling cycles. When the load monitoring server obtains actual user load information, it stores it in a local database. This invention ensures timely and consistent response from the load monitoring server, prevents network issues from impacting user monitoring, and ensures stable operation of the power network.

[0090] The units and algorithm steps of each example described in the disclosed embodiments of the power distribution load monitoring system within the power grid provided by the present invention can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0091] In the power distribution load monitoring system within the power grid, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or can be an electrical, mechanical or other form of connection.

[0092] In the power distribution load monitoring system provided by the present invention, the computer program code for performing the operations disclosed herein can be written in one or more programming languages ​​or a combination thereof, and the above-mentioned programming languages ​​include but are not limited to object-oriented programming languages ​​- such as Java, Smalltalk, C++, and also include conventional procedural programming languages ​​- such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or power server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (exemplarily using an Internet service provider to connect through the Internet).

[0093] In the description and claims of the present invention, as well as in the accompanying drawings, the terms "first," "second," "third," "fourth," and so forth (if any) are used to distinguish similar items and are not necessarily used to describe a particular order or sequential sequence. It should be understood that the terms used in this manner are interchangeable where appropriate, such that the embodiments of the present invention described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "including," "comprising," and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions.

[0094] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A power distribution load monitoring system in a power grid, characterized in that: include: Load monitoring server and electricity user monitoring terminal; The load monitoring server is connected to the electricity user monitoring terminal and sends control instructions to the electricity user monitoring terminal; The electricity user monitoring terminal is equipped with a current transformer, a voltage transformer, an electric energy meter, a time module, a storage, a communication module and a data processor which are arranged in the electricity user; The user monitoring terminal collects user load information in real time and stores it in a storage device. The user monitoring terminal also uploads the collected user load information and the load collection time corresponding to the load information to the load monitoring server based on the control instructions of the load monitoring server. The load monitoring server stores the acquired information and displays it according to the monitoring instructions of the monitoring personnel. It also predicts the acquired information and the load status of the electricity user, and issues an alarm if any abnormality occurs; The load monitoring server configures a user load monitoring model, compares the load collection time of the collected user load information with the user load monitoring model, and obtains first load status comparison information; The load monitoring server predicts the load state of the electricity user from the first load state comparison information; The load monitoring server compares the load information of the electricity user with the predicted load information of the electricity user to obtain second load status comparison information. If the second load status comparison information exceeds a preset electricity load threshold, a load status abnormality warning is generated; The load monitoring server calculates the real-time user load information and the corresponding predicted user load information of the same user, and obtains the second load status comparison information by subtracting the corresponding predicted user load information from the real-time user load information; If the second load status comparison information is a positive number and is greater than the preset power load threshold, a load status abnormality warning is generated; The load monitoring server retrieves historical electricity user load information and obtains historical normal electricity user load information from the historical electricity user load information; Obtain load collection time period information from historical normal electricity user load information; The load monitoring server obtains the load state set to be trained based on the historical normal power user load information and load collection time period information; After training the load state set to be trained, the electricity user load monitoring model is obtained.

2. The power distribution load monitoring system in the power grid according to claim 1, characterized in that: The load monitoring server is further used to divide the load time points to be processed from the load collection time period information; and to filter out the repeated power load data from each load time point to be processed; The load information extension processing is performed on the power load data to obtain the load state set to be trained.

3. The power distribution load monitoring system in the power grid according to claim 1, characterized in that: The load monitoring server normalizes the training load state set to obtain a training set to be divided; and divides the training set to be divided according to a preset load monitoring rule to obtain a training set to be regressed; The load monitoring server uses a BP neural network including an input layer and several hidden layers to perform regression prediction on the regressive electric load training set to obtain the load monitoring model of the electricity user; Based on the random generation module, the weight matrices of the hidden layer and the output layer are initialized; the mean square error is used as the loss function, and gradient descent is used to gradually optimize the load monitoring model of electricity users.

4. The power distribution load monitoring system in the power grid according to claim 3, characterized in that: The load monitoring server reorders the processed data, i.e., the training set to be divided, and performs data segmentation processing. Based on the training model and the test model, a load information training set and a load information test set are formed. The load information training set and the load information are then randomly divided to obtain the load training set to be regressed.

5. The power distribution load monitoring system in the power grid according to claim 1, characterized in that: The load monitoring server obtains the divided training model. In each training round, it first performs feedforward calculation, then uses MSE to calculate the error, and then uses the error to perform back propagation calculation to update the parameters. After each training, the error between the prediction module and the training model is calculated using MSE, and the training model is adjusted based on the error result. The training model is tested using the load monitoring model of electricity users that meets the requirements, the corresponding prediction module is obtained, the error between the prediction module and the training model is calculated, and adjustments are made according to the electricity load fitting requirements until the requirements are met.

6. The power distribution load monitoring system in the power grid according to claim 1, characterized in that: The load monitoring server sends a control instruction to the user monitoring terminal according to the monitoring call instruction to obtain the user's load information and the load collection time corresponding to the load information; The load monitoring server uses the electricity user load monitoring model and combines the average value of the electricity user load information in the current polling period as the predicted electricity user load information.

7. The power distribution load monitoring system in the power grid according to claim 1, characterized in that: After receiving the monitoring call instruction, the load monitoring server determines whether the network delay between the load monitoring server and the power user monitoring terminal exceeds the preset delay threshold; If yes, determine the historical electricity user load information according to the data type identification information in the control instruction; The average value of the electricity user load information in the current polling cycle is predicted by combining the electricity user load monitoring model with historical electricity user load information to obtain the predicted electricity user load information.

Images