CN116108736A - Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system - Google Patents

Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system Download PDF

Info

Publication number
CN116108736A
CN116108736A CN202211412133.0A CN202211412133A CN116108736A CN 116108736 A CN116108736 A CN 116108736A CN 202211412133 A CN202211412133 A CN 202211412133A CN 116108736 A CN116108736 A CN 116108736A
Authority
CN
China
Prior art keywords
line loss
neural network
training
model
area
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211412133.0A
Other languages
Chinese (zh)
Inventor
张淞珲
徐新光
刘涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
Marketing Service Center of State Grid Shandong Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
Marketing Service Center of State Grid Shandong Electric Power Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, Marketing Service Center of State Grid Shandong Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN202211412133.0A priority Critical patent/CN116108736A/en
Publication of CN116108736A publication Critical patent/CN116108736A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Train Traffic Observation, Control, And Security (AREA)

Abstract

The present disclosure provides a method and a system for predicting line loss of a station area based on cross attention and a deep learning algorithm, comprising collecting line loss data of the station area; classifying the acquired transformer area line loss data; dividing a training sample and a test sample; constructing a BP neural network model; training the BP neural network by using a training sample to obtain a transformer area line loss analysis calculation model; and inputting the test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result. And identifying key factors of the line loss of the transformer area by the BP neural network, determining an optimized input variable set of the line loss calculation model of the transformer area, and improving the efficiency and accuracy of the line loss calculation of the transformer area.

Description

Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system
Technical Field
The disclosure belongs to the field of station area line loss prediction, and particularly relates to a station area line loss prediction method and system based on cross attention and a deep learning algorithm.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The line loss rate of the power grid is an important economic and technical index of the power enterprise, the line loss of a platform region directly reflects the power grid management level of a certain area, the line loss rate is analyzed by taking the platform region as a unit, the planning design and the operation management level of the power distribution network can be directly reflected, and the prediction of the reasonable line loss of the platform region is the premise and key for realizing the lean management of the line loss. However, the complexity of the line loss management of the transformer area is increased due to the large number of users in the low-voltage transformer area, various loads, uneven management level of a power grid base layer and the network frame construction mechanism, imperfect management of the transformer accounts and complex and various line distribution. Based on the current situation, the accurate and rapid calculation of the line loss rate of the cell becomes a problem to be solved.
The traditional calculation of the theoretical line loss of the low-voltage transformer area mostly adopts engineering approximation algorithm and forward-push substitution algorithm, the simplifying assumption made by the methods is more, and the consideration of the complicated topological structure of the low-voltage distribution network, the unbalanced running state of the three-phase load and other factors is insufficient, so that the calculation accuracy of the method is not high, and the requirement of the line loss lean analysis of the power company cannot be well met.
With the promotion and the deep promotion of intelligent ammeter and the construction work of an electricity consumption information acquisition system, the data of a platform area and user information which can be acquired by an electric company are increased exponentially, and on the basis of accumulating more comprehensive and finer mass data related to the platform area, the management of the line loss of the platform area is developed by using a data driving method, so that the problem of overlarge dependence on the topological structure and parameters of the platform area in the traditional calculation can be solved, the change rule of the line loss of the platform area can be excavated, the weak link in the line loss management of the platform area can be found in time, and the accuracy of loss reduction measures is improved.
Although the machine learning algorithm adopted in the prior art has great improvement on the calculation or prediction accuracy of the line loss, certain limitations exist, such as a calculation method based on kmeans-LightGBM, because the initialization center of the clustering algorithm is randomly selected, different initial values can lead to different classification results, and when the classification is inaccurate, the defect of large parameter reconstruction error is easily caused; therefore, how to accurately classify the transformer area and identify key factors and optimize input variables for the subsequent transformer area line loss calculation model is a key for improving the efficiency and accuracy of transformer area line loss calculation.
Disclosure of Invention
In order to overcome the defects in the prior art, the present disclosure provides a method and a system for predicting the line loss of a platform region based on cross attention and a deep learning algorithm, which identify key factors of the line loss of the platform region and the platform region through a BP neural network, determine an optimized input variable set of a calculation model of the line loss of the platform region, and improve the efficiency and the accuracy of the calculation of the line loss of the platform region.
To achieve the above object, one or more embodiments of the present disclosure provide a method for predicting a line loss of a region based on a cross-over attention and a deep learning algorithm:
a cross-attention and deep learning algorithm based prediction of line loss of a region, comprising:
constructing a BP neural network model;
training the BP neural network by using a training sample to obtain a transformer area line loss analysis calculation model;
and inputting the test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result.
Further, before constructing the BP neural network model, classifying the acquired region line loss data.
Further, the BP neural network is a method based on reverse error transfer, and the learning process of the BP neural network consists of two parts of forward propagation of input data and reverse propagation of errors.
Further, the BP neural network includes: input layer, hidden layer and output layer.
Further, the electrical characteristic index with strong correlation and the corresponding line loss rate of the transformer area are selected to be respectively used as the input and the output of the model.
Further, the training sample is used for training the BP neural network, and normalization processing of training data is performed first.
Furthermore, the LM algorithm is selected for the learning algorithm setting of the BP neural network.
In another aspect, one or more embodiments of the present disclosure provide a cluster and discretized sampling based area line loss prediction system:
a cross-attention and deep learning algorithm based station line loss prediction system, comprising:
a model building module configured to: constructing a BP neural network model;
a model training module configured to: training the BP neural network by using a training sample to obtain a transformer area line loss analysis calculation model;
a prediction module configured to: and inputting the test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result.
In other embodiments, the following technical solutions are adopted:
a terminal device comprising a processor and a memory, the processor being configured to implement instructions; the memory is used for storing a plurality of instructions adapted to be loaded by the processor and to perform the steps in the above-described cross-attention and deep learning algorithm-based region line loss prediction method.
In other embodiments, the following technical solutions are adopted:
a computer readable storage medium having stored therein a plurality of instructions, wherein the program when executed by a processor performs the steps in the cross-attention and deep learning algorithm based method for predicting line loss in a region.
The one or more of the above technical solutions have the following beneficial effects:
firstly, classifying line loss data of different areas, constructing BP neural networks and training to obtain an area line loss analysis calculation model; the BP neural network is utilized to identify key factors of the line loss of the transformer area, an optimized input variable set of the line loss calculation model of the transformer area is determined, the efficiency and accuracy of the line loss calculation of the transformer area are improved, and the method not only can rapidly analyze and process massive power information of the transformer area, but also has higher identification accuracy.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate and explain the exemplary embodiments of the disclosure and together with the description serve to explain the disclosure, and do not constitute an undue limitation on the disclosure.
FIG. 1 is an algorithm flow diagram of a BP neural network of an embodiment of the present disclosure;
FIG. 2 is a block diagram of a BP neural network model according to an embodiment of the present disclosure;
FIG. 3 is a graph showing BP neural network behavior under different numbers of input variables according to an embodiment of the present disclosure;
fig. 4 is a diagram illustrating a first type of line loss prediction result according to an embodiment of the present disclosure.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments in accordance with the present disclosure. As used herein, the singular is also intended to include the plural unless the context clearly indicates otherwise, and furthermore, it is to be understood that the terms "comprises" and/or "comprising" when used in this specification are taken to specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof.
Embodiments of the present disclosure and features of embodiments may be combined with each other without conflict.
The general idea presented by the present disclosure:
the method and the device identify the key factors of the line loss of the transformer area based on the BP neural network, determine the optimized input variable set of the line loss calculation model of the transformer area, and improve the efficiency and accuracy of the line loss calculation of the transformer area.
Example 1
The embodiment discloses a method for predicting the line loss of a platform region based on a cross attention and deep learning algorithm, which comprises the following steps:
(1) Constructing a BP neural network model;
(2) Training the BP neural network by using a training sample to obtain a transformer area line loss analysis calculation model;
(3) And inputting the test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result.
Because the line loss characteristics of different areas have significant differences, when a neural network is adopted to build an area line loss calculation model, the original sample data is generally classified firstly, the different areas are used as classification indexes, and after classification, each area corresponds to one type of sample data with corresponding characteristics.
The BP neural network is a method based on reverse error transfer, and the learning process of the BP network consists of two parts of forward propagation of input data and reverse propagation of errors. Forward propagation refers to input samples from an input layer being processed layer by layer through hidden layers to an output layer. If the output layer output result does not reach the expected value, the back propagation of the error is turned to. The error back propagation is to reversely propagate the output error layer by layer through an implicit layer, and adjust the weight and the threshold value of each neuron. The continuous adjustment process of the weight and the threshold value is the network learning training process until the error reaches the expected range or the set learning times. An algorithm flow chart of the BP neural network is shown in figure 1. The whole process can be divided into three steps of model construction, training and prediction.
The step (1) is to construct a BP neural network model:
the BP neural network structure schematic diagram is shown in figure 2, and comprises an input layer, an hidden layer and an output layer.
The strongly-correlated electrical characteristic indexes and the corresponding line loss rates of the transformer areas are selected to be respectively used as the input and output of the model, the number of nodes of an input layer depends on the number of the electrical characteristic indexes, and the number of nodes of an hidden layer can be approximately determined according to the formula (1):
Figure BDA0003939107950000051
where m and n are the number of neurons in the output layer and input layer, respectively, and a is a constant between [0,10 ].
The learning rate determines the amount of weight change that occurs during each cycle of training. Too high a learning rate may lead to instability of the system, but too low a learning rate may lead to longer training times, may converge very slowly, but can ensure that the error value of the network jumps out of the valley of the error surface and eventually tends to a minimum error value. In general, a smaller learning rate tends to be chosen to ensure system stability. The learning rate is selected to be 0.01-0.8.
And (5) selecting an expected error. The desired error value should also be determined by comparing the training to a suitable value during the training of the network. The so-called "fit" is determined with respect to the number of nodes of the hidden layer required, since the smaller expected error is obtained by increasing the nodes of the hidden layer, as well as the training time.
The step (2) is to train the BP neural network by using training samples to obtain a platform area line loss analysis and calculation model:
BP neural network training: first, normalization processing of training data is performed. The normalization processing of the data is to convert all data points into constants among the data points, and the purpose of the normalization processing is to cancel the order-of-magnitude difference among the dimensional data and avoid network training errors caused by large order-of-magnitude difference of input and output data. The data normalization method mainly comprises the following two steps:
(1) Maximum minimum method:
x k ＝(x k -x min )/(x max -x min ) (2)
(2) Mean variance method:
x k ＝(x k -x mean )/x var (3)
wherein x is mean Mean value of data sequence, x var For numbers of digitsAccording to the variance of the sequence.
The method adopts a first data normalization method, and the normalization function adopts MATLAB self-contained function mapmin max. At the same time, the selection of hidden layer and output layer functions (node transfer functions) has a greater impact on the accuracy of network predictions. The implicit and output layer functions selected here are tan sig and purelin, respectively. The calculation formula is as follows:
Figure BDA0003939107950000061
purelin(n)＝n (5)
in the learning algorithm setting of the BP neural network, an LM algorithm is often selected. The LM algorithm has high calculation efficiency, and a second derivative is adopted at the part with updated weight, as shown in a formula (6):
Δw＝[J T (w)J(w)+μI] -1 J T (w)e(w) (6)
wherein I is an identity matrix, mu is a user-defined learning rate, and J (w) is a jacobian matrix.
The step (3) is to input a test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result:
inputting the test and sample data into a learned and trained BP neural network station area line loss analysis and calculation model, reasonably predicting the station area line loss, and analyzing prediction errors.
And (3) carrying out calculation analysis:
1) According to the classification result, training and learning three types of samples by using a BP neural network model, calculating the line loss rate of the three types of transformer area samples, and carrying out error analysis. The network behavior under different numbers of input variables is shown in fig. 3, with the upper curve representing the sweep Loss and the lower curve representing the Validation Loss.
As can be seen from fig. 3, the network performs best at four inputs, and as the number of input variables increases continuously, the network will suffer from over-fitting phenomenon, which affects the learning efficiency of the network.
Taking a class of areas as an example, taking 4 electrical characteristic indexes with highest association degree as input, taking corresponding line loss rate as output, selecting 120 samples as training sets, and taking the rest 20 samples as test sets. The true value, estimated value and relative error of the predicted samples are shown in fig. 4. As can be seen from fig. 4, the foregoing classification of the area and identification of the key factors effectively improve the accuracy of predicting the line loss of the area.
From the above, it can be determined that the optimized input variable set of the area line loss calculation model is the 4 electrical characteristic indexes with the highest association degree in a certain type of area.
In the embodiment, the BP neural network is used for identifying the key factors of the line loss of the area, so that the optimized input variable set of the line loss calculation model of the area is determined, and the efficiency and accuracy of the line loss calculation of the area are improved.
Example two
The embodiment discloses a platform district line loss prediction system based on cross attention and deep learning algorithm, including:
a model building module configured to: constructing a BP neural network model;
a model training module configured to: training the BP neural network by using a training sample to obtain a transformer area line loss analysis calculation model;
a prediction module configured to: and inputting the test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result.
Example III
An object of the present embodiment is to provide a terminal device.
A terminal device comprising a processor and a memory, the processor being configured to implement instructions; the memory is used for storing a plurality of instructions adapted to be loaded by the processor and to perform the steps in the above-described cross-attention and deep learning algorithm-based region line loss prediction method.
Example IV
An object of the present embodiment is to provide a computer-readable storage medium.
A computer readable storage medium having stored therein a plurality of instructions, wherein the program when executed by a processor performs the steps in the cross-attention and deep learning algorithm based method for predicting line loss in a region.
The steps involved in the second, third and fourth embodiments correspond to the first embodiment of the method, and the detailed description of the second embodiment refers to the relevant description of the first embodiment. The term "computer-readable storage medium" should be taken to include a single medium or multiple media including one or more sets of instructions; it should also be understood to include any medium capable of storing, encoding or carrying a set of instructions for execution by a processor and that cause the processor to perform any one of the methods of the present disclosure.
It will be appreciated by those skilled in the art that the modules or steps of the disclosure described above may be implemented by general-purpose computer means, alternatively they may be implemented by program code executable by computing means, so that they may be stored in storage means and executed by computing means, or they may be fabricated separately as individual integrated circuit modules, or a plurality of modules or steps in them may be fabricated as a single integrated circuit module. The present disclosure is not limited to any specific combination of hardware and software.
The foregoing description of the preferred embodiments of the present disclosure is provided only and not intended to limit the disclosure so that various modifications and changes may be made to the present disclosure by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure should be included in the protection scope of the present disclosure.
While the specific embodiments of the present disclosure have been described above with reference to the drawings, it should be understood that the present disclosure is not limited to the embodiments, and that various modifications and changes can be made by one skilled in the art without inventive effort on the basis of the technical solutions of the present disclosure while remaining within the scope of the present disclosure.

Claims (10)

1. A method for predicting the line loss of a station area based on clustering and discretization sampling is characterized by comprising the following steps:
constructing a BP neural network model;
training the BP neural network by using a training sample to obtain a transformer area line loss analysis calculation model;
and inputting the test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result.
2. The method for predicting line loss of a region based on clustering and discretizing sampling as recited in claim 1, wherein the collected line loss data of the region is classified into regions before constructing the BP neural network model.
3. The method for predicting line loss of a region based on clustering and discretizing sampling as recited in claim 1, wherein the BP neural network is a method based on backward error transfer, and the learning process is composed of two parts of forward propagation of input data and backward propagation of errors.
4. The method for predicting line loss of a region based on clustering and discretizing sampling as recited in claim 1, wherein the BP neural network comprises: input layer, hidden layer and output layer.
5. The method for predicting line loss of a cell based on clustering and discretizing sampling as recited in claim 4, wherein the electrical characteristic index of strong correlation and the corresponding line loss rate of the cell are selected as input and output of a model, respectively.
6. The method for predicting line loss of a region based on clustering and discretizing sampling as recited in claim 1, wherein training the BP neural network with the training sample first performs normalization processing of training data.
7. The method for predicting line loss of a region based on clustering and discretizing sampling as recited in claim 1, wherein the learning algorithm of the BP neural network is set to select an LM algorithm.
8. A platform district line loss prediction system based on cluster and discretization sampling is characterized by comprising:
a model building module configured to: constructing a BP neural network model;
a model training module configured to: training the BP neural network by using a training sample to obtain a transformer area line loss analysis calculation model;
a prediction module configured to: and inputting the test sample into a platform area line loss analysis and calculation model to obtain a platform area line loss prediction result.
9. A terminal device comprising a processor and a memory, the processor being configured to implement instructions; the memory is used for storing a plurality of instructions adapted to be loaded by the processor and to perform the steps in the method of claims 1-7.
10. A computer readable storage medium having stored therein a plurality of instructions, characterized in that the program when executed by a processor performs the steps in the method of claims 1-7.
CN202211412133.0A 2022-11-11 2022-11-11 Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system Pending CN116108736A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211412133.0A CN116108736A (en) 2022-11-11 2022-11-11 Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211412133.0A CN116108736A (en) 2022-11-11 2022-11-11 Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system

Publications (1)

Publication Number Publication Date
CN116108736A true CN116108736A (en) 2023-05-12

Family

ID=86266360

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211412133.0A Pending CN116108736A (en) 2022-11-11 2022-11-11 Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system

Country Status (1)

Country Link
CN (1) CN116108736A (en)

Similar Documents

Publication Publication Date Title
CN108520357B (en) Method and device for judging line loss abnormality reason and server
CN112149873B (en) Low-voltage station line loss reasonable interval prediction method based on deep learning
CN104809658B (en) A kind of rapid analysis method of low-voltage distribution network taiwan area line loss
CN109324291B (en) Prediction method for predicting service life of proton exchange membrane fuel cell
CN110335168B (en) Method and system for optimizing power utilization information acquisition terminal fault prediction model based on GRU
CN110135635B (en) Regional power saturated load prediction method and system
CN112733417B (en) Abnormal load data detection and correction method and system based on model optimization
CN109740859A (en) Transformer condition evaluation and system based on Principal Component Analysis and support vector machines
CN113077020B (en) Transformer cluster management method and system
CN105023042A (en) User electricity stealing suspicion analyzing device and method based on big data neural network algorithm
CN113139596A (en) Optimization algorithm of low-voltage transformer area line loss neural network
CN111242161A (en) Non-invasive non-resident user load identification method based on intelligent learning
CN110866366A (en) XGboost algorithm-based island detection method for photovoltaic microgrid containing PHEV
CN113112090A (en) Space load prediction method based on principal component analysis of comprehensive mutual information degree
CN116821832A (en) Abnormal data identification and correction method for high-voltage industrial and commercial user power load
CN113988558B (en) Power grid dynamic security assessment method based on blind area identification and electric coordinate system expansion
CN110059938B (en) Power distribution network planning method based on association rule driving
CN113033898A (en) Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network
CN112508363A (en) Deep learning-based power information system state analysis method and device
CN116108736A (en) Cross-attention and deep learning algorithm-based transformer area line loss prediction method and system
CN114389265B (en) Power grid emergency load optimization method and system based on proliferation and reduced evolution
CN115186882A (en) Clustering-based controllable load spatial density prediction method
CN111092425B (en) Net rack clustering analysis method and system based on topological characteristics
CN112465253A (en) Method and device for predicting links in urban road network
CN117706377B (en) Battery inconsistency identification method and device based on self-adaptive clustering

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination