CN111639783A - Line loss prediction method and system based on LSTM neural network - Google Patents
Line loss prediction method and system based on LSTM neural network Download PDFInfo
- Publication number
- CN111639783A CN111639783A CN202010309179.4A CN202010309179A CN111639783A CN 111639783 A CN111639783 A CN 111639783A CN 202010309179 A CN202010309179 A CN 202010309179A CN 111639783 A CN111639783 A CN 111639783A
- Authority
- CN
- China
- Prior art keywords
- line loss
- neural network
- lstm neural
- value
- loss rate
- 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
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 58
- 238000012545 processing Methods 0.000 claims abstract description 46
- 230000015654 memory Effects 0.000 claims abstract description 39
- 230000007246 mechanism Effects 0.000 claims abstract description 11
- 210000002569 neuron Anatomy 0.000 claims abstract description 11
- 238000004364 calculation method Methods 0.000 claims description 27
- 238000012360 testing method Methods 0.000 claims description 19
- 230000002159 abnormal effect Effects 0.000 claims description 12
- 230000004913 activation Effects 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 10
- 238000012886 linear function Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 7
- 238000009826 distribution Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000012512 characterization method Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 15
- 238000004590 computer program Methods 0.000 description 7
- 210000004027 cell Anatomy 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000004134 energy conservation Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- General Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Software Systems (AREA)
- Game Theory and Decision Science (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Educational Administration (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a line loss prediction method based on an LSTM neural network, which comprises the following steps: collecting characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors; processing the characteristic data of the line loss rate; inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and extracting to obtain a line loss prediction value; the line loss prediction model of the LSTM neural network takes the line loss rate influence factors as an input layer of the model, replaces common neurons of a hidden layer with a memory module containing a gating mechanism, and obtains a line loss prediction value after training; the line loss prediction model based on the LSTM neural network is suitable for the massive line loss data in the user data acquisition system, and the line loss value is predicted more accurately.
Description
Technical Field
The invention belongs to the technical field of power system automation, and relates to a line loss prediction method and system based on an LSTM neural network.
Background
The line loss rate of the power grid is an important economic and technical index of a power enterprise, and the line loss rate prediction is an important work of the power grid enterprise. Line losses are power and power losses and other losses generated by individual components or devices in the power grid during the transmission and distribution of electrical energy. The line loss rate refers to the percentage of the power line loss load lost in the power network to the load supplying the power network. The line loss rate plays an important guiding role in energy conservation and development planning of the power system, reflects the design, operation and management level of the power system, is an important technical and economic index for assessing power supply enterprises, and can help the power supply enterprises to formulate reasonable loss reduction and energy conservation targets by predicting the line loss rate.
As an important link in the line loss management of the power enterprise, the determination of the theoretical line loss has important significance for improving the lean level of the line loss management, and the traditional calculation method for the theoretical line loss is mainly based on an energy consumption model or a load flow calculation method. With the wide application of artificial intelligence technology in the field of electric power, the theoretical line loss calculation method is also developed to mainly use artificial neural networks, support vector machines and other improved algorithms. However, in the past, the research on the theoretical line loss adopts standard arithmetic examples to design experiments, the number of samples is small, and the actual line loss condition in the actual production is difficult to be comprehensively reflected. In recent years, with the overall construction of power consumption information acquisition systems and the wide application of big data processing technology, it has become possible to process line loss data of the whole transformer area by using a data mining method. The method is characterized in that a large amount of line loss data of the transformer area are analyzed, modeled and predicted from the angle of a statistical theory, and further potential information and association are found out, so that the method is an important task for an electric power marketing department.
Disclosure of Invention
Aiming at the defects that the conventional research on theoretical line loss adopts standard examples to design experiments, the number of samples is small, and the actual line loss condition in actual production is difficult to reflect comprehensively, the invention provides a line loss prediction method and a line loss prediction system based on an LSTM neural network, which specifically comprise the following steps:
collecting characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors;
processing the characteristic data of the line loss rate;
inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and extracting to obtain a line loss prediction value;
the line loss prediction model of the LSTM neural network takes the line loss rate influence factors as an input layer of the model, replaces ordinary neurons of a hidden layer with a memory module containing a gating mechanism, and obtains a line loss prediction value after training.
Preferably, the line loss rate influencing factors include: power supply, distribution capacity, line length, power factor, air temperature and holidays.
Preferably, the processing the characteristic data of the line loss rate includes:
determining a missing value of the feature data of the line loss rate acquired in advance based on the influence factors of the line loss rate determined in advance, filling the missing value, and performing consistent processing;
replacing the abnormal characteristic data with the average value of adjacent normal line loss for the abnormal characteristic data in the characteristic data of the line loss rate acquired in advance and correcting errors;
normalization processing is performed based on the normal, filled and error corrected feature data.
Preferably, the constructing of the LSTM neural network line loss prediction model includes:
acquiring historical data of the line loss rate containing a time sequence based on the line loss rate influence factors, and processing the historical data;
dividing the processed historical data into a training data set and a testing data set;
taking the line loss rate influence factors in the training data set as an input layer, taking the line loss value in the training data set as an output layer, replacing common neurons of a hidden layer with a memory module containing a gating mechanism, and training to obtain an initial LSTM neural network line loss prediction model;
substituting the line loss rate influence factor value of the test data set into the initial LSTM neural network line loss prediction model to obtain a line loss prediction value;
and verifying the initial LSTM neural network line loss prediction model based on the line loss prediction value and the actual line loss value in the test data set to obtain a trained LSTM neural network line loss prediction model.
Preferably, the verifying the initial LSTM neural network line loss prediction model based on the line loss prediction value and the actual line loss value in the test data set to obtain a trained LSTM neural network line loss prediction model includes:
inputting the test data set into the trained LSTM neural network prediction model;
based on the predicted line loss value and the actual line loss value of the test data set, carrying out accuracy evaluation by adopting a root mean square error index and an average absolute percentage error index;
and when the accuracy does not reach the preset threshold value, continuing training the LSTM neural network line loss prediction model based on the time sequence characteristic until the preset threshold value is reached to obtain the trained LSTM neural network line loss prediction model.
Preferably, the inputting the processed feature data into a line loss prediction model of the LSTM neural network trained in advance, and obtaining a line loss prediction value after extraction processing includes:
sending the training data set, the state parameters of the previous-time hidden layer and the state parameters of the previous-time memory module to an input gate of a memory module for calculation to obtain a calculation result of a one-dimensional column vector, and sending the training data set to a forgetting gate of the memory module after keeping current input information in the training data set;
calculating the training data set by using the forgetting gate to obtain a calculation result of a one-dimensional column vector, and sending the training data set to an output gate of a memory module after historical information needing to be reserved in the updated memory module is reserved;
calculating the training data set by using the output gate to obtain a calculation result of a one-dimensional column vector, and outputting the state parameters of the memory module at the current moment and the state parameters of the hidden layer at the current moment after output information needing to be reserved in the training data set is reserved;
and inputting the calculation result of the one-dimensional column vector into the linear function for calculation based on an activation formula of the linear function to obtain a line loss prediction value.
Preferably, the activation formula of the linear function is as follows:
y＝actv(w·a+b)
where a is a calculation result of a one-dimensional column vector, y is a line loss prediction value, actv () is an activation function, w is a weight parameter of input data, and b is a constant.
Based on the same conception, the invention provides a line loss prediction system based on an LSTM neural network, which comprises the following components: the device comprises an acquisition module, a processing module and a prediction value module;
the acquisition module is used for acquiring characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors;
the processing module is used for processing the characteristic data of the line loss rate;
the prediction module is used for inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and obtaining a line loss prediction value after extraction processing;
the line loss prediction model of the LSTM neural network takes the line loss rate influence factors as an input layer of the model, replaces ordinary neurons of a hidden layer with a memory module containing a gating mechanism, and obtains a line loss prediction value after training.
Preferably, the processing module includes: a fill submodule, a modify submodule and a normalize submodule;
the filling submodule is used for determining a missing value of the characteristic data of the line loss rate acquired in advance based on a predetermined line loss rate influence factor, filling the missing value and carrying out consistent processing;
the correction submodule is used for replacing the abnormal characteristic data with the average value of adjacent normal line loss for the abnormal characteristic data in the characteristic data of the line loss rate acquired in advance and performing error correction;
and the normalization submodule is used for performing normalization processing on the basis of the normal, filled and error corrected characteristic data.
Preferably, the system further comprises: a model building module;
and the model construction module is used for training the LSTM neural network to obtain an LSTM neural network line loss prediction model.
Compared with the prior art, the invention has the beneficial effects that:
1. the invention provides a line loss prediction method based on an LSTM neural network, which comprises the following steps: collecting characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors; processing the characteristic data of the line loss rate; inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and extracting to obtain a line loss prediction value; the line loss prediction model of the LSTM neural network takes the line loss rate influence factors as an input layer of the model, replaces common neurons of a hidden layer with a memory module containing a gating mechanism, and obtains a line loss prediction value after training; the line loss prediction model based on the LSTM neural network is suitable for the massive line loss data in the user data acquisition system, and the line loss value is predicted more accurately.
Drawings
FIG. 1 is a flow chart of a method provided by the present invention;
FIG. 2 is a flow chart of data transmission according to an embodiment of the present invention;
FIG. 3 is a diagram illustrating integrated data according to an embodiment of the present invention;
FIG. 4 is a diagram of the LSTM neural network architecture provided by the present invention;
fig. 5 is a system configuration diagram provided by the present invention.
Detailed Description
The embodiments of the present invention will be further described with reference to the accompanying drawings.
Example 1:
the invention aims to provide a line loss prediction method based on an LSTM neural network, aims to solve the problem that the influence of external environmental factors on the line loss prediction precision is considered in line loss prediction, is introduced by combining with a method structure diagram of a figure 1, and specifically comprises the following steps:
step 1: collecting characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors;
step 2: processing the characteristic data of the line loss rate;
and step 3: inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and extracting to obtain a line loss prediction value;
wherein, the step 1: based on predetermined line loss rate influence factors, collecting characteristic data of the line loss rate according to a time sequence, specifically comprising:
determining the influencing factors of the line loss rate
Various factors influencing the line loss rate are analyzed and compared, six influencing factors including power supply quantity, distribution and transformation capacity, line length, power factor, air temperature and holiday type are finally determined, the six influencing factors are introduced by combining the integrated data schematic diagram of the figure 3, the determined influencing factors fully consider the factors of the power distribution network and the external environment factors, and the influenced situation of line loss rate prediction can be scientifically and comprehensively reflected.
Step 2: processing the characteristic data of the line loss rate, specifically comprising:
six different influence factors have different dimensions and magnitude levels, and in order to facilitate calculation and improve the accuracy of a prediction result, the influence factor index collected data is preprocessed as follows: (1) filling missing values in the acquired data, uniformly filling the missing values to be-1 for the missing parts of the data so as to distinguish normal numerical values, and uniformly processing samples which are inconsistent before and after; (2) carrying out smoothing processing and error correction on abnormal samples (abnormal characteristic data, namely characteristic data which fluctuates up and down violently in a short period, characteristic data for counting negative values and the like), and replacing the negative values for counting errors by average values of adjacent normal line losses; (3) all sample data were Z-score normalized.
Where x represents the raw data, μ represents the mean of the raw data, σ represents the standard deviation of the raw data, and
representing data after Z-score standardized transformation, n is the number of data in the data set, xiIs the ith original data in the original data set. Step three, data set division
And step 3: inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and obtaining a line loss prediction value after extraction processing, wherein the line loss prediction value specifically comprises the following steps:
and dividing the data set (the data sets sequentially arranged according to the time sequence) formed after the data preprocessing in the second step according to the proportion of 9: 1, constructing and training the LSTM network by taking the first 90% of the time sequence (training data set) of the data set as the training set, and evaluating the accuracy of the prediction result by taking the last 10% of the time sequence (test data set) of the data set as the test set.
Constructing a line loss prediction model based on an LSTM neural network, and introducing by combining a data transmission flow chart of fig. 2:
an LSTM neural network composed of an input layer, an output layer and an implied layer is constructed, introduction is carried out by combining the LSTM neural network structure diagram of fig. 4, and compared with the traditional recurrent neural network, the implied layer of the recurrent neural network is provided with a memory module containing a gating mechanism to replace a common neuron. The network is composed of 1 input layer, a plurality of hidden layers (hidden layers) and 1 output layer, wherein the output of the previous hidden layer is used as the input of the next hidden layer, the input layer and the hidden layers jointly realize the extraction of input data characteristics, the output of the last hidden layer is a one-dimensional column vector, and the predicted value of the processed data is obtained through a linear regression function.
The activation formula for the linear function is as follows:
y＝actv(w·a+b)
where a is a calculation result of a one-dimensional column vector, y is a line loss prediction value, actv () is an activation function, w is a weight parameter of input data, and b is a constant.
The state at time t is denoted as ciContains the long-term memory information of the sequence, and correspondingly, the state h of the hidden layer at the time tiThe short-term memory information containing the sequence comprises a forgetting gate, an input gate and an outputThe reading and modification of the information of the gate and the memory unit (memory module) are realized by controlling the forgetting gate, the input gate and the output gate.
The input gate is used for calculating the input training data set, and sending the updated training data set to the forgetting gate after keeping the current input information in the training data set; the forgetting gate is used for calculating the input updated training data set, reserving historical information required to be reserved after updating and sending the current training data set to the output gate; and the output gate is used for calculating the current training data set and reserving output information to be reserved.
Assume that at time t, the inputs to the memory module in the LSTM include: sequence (feature data after preprocessing) input xt(training data set), state of memory cell at time t-1 (state parameter of memory module at last time) ct-1And the state of the hidden layer at time t-1 (the state parameter of the hidden layer at the previous time) ht-1(ii) a The output of the memory module includes: state c of memory cell at time ttState h at time t of hidden layert. In the memory module, the forgetting gate is used for determining the history information which needs to be kept by the memory unit, i.e. control ct-1To ctThe degree of influence of (c); the function of the input gate is to determine the current input information that the memory cell needs to retain, i.e. control xtTo ctThe degree of influence of (c); the output gate is used to determine the output information that the memory cell needs to retain, i.e. ctTo htThe degree of influence of (c). The calculation formula is as follows:
ft＝σ(Wf·[ht-1，xt]+bf)
it＝σ(Wi·[ht-1，xt]+bi)
ot＝σ(Wo·[ht-1，xt]+bo)
in the formula: f. oft、it、otRespectively representing the calculation results of the forgetting gate, the input gate and the output gate at the moment t; wf、Wi、WoRespectively show a forgetting gate, an input gate andoutputting a weight matrix of the gate; bf、bi、boBias terms representing a forgetting gate, an input gate and an output gate, respectively; sigma denotes a sigmoid activation function, whose role is to map variables to the interval 0, 1]In (1).
Evaluating the accuracy of the prediction result
Two indexes are selected to evaluate the prediction effect of the line loss prediction model based on the LSTM neural network, wherein the two indexes are Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE), and the calculation formula is as follows:
in the formula: y isi、
Respectively representing a true value and a predicted value of the line loss; n represents the number of the predicted verification data; i represents a predicted point sequence number.
The determined influence factors of the power supply quantity, the distribution transformer capacity, the line length, the power factor, the air temperature and the holidays fully consider the factors of the power distribution network and the external environment factors, and scientifically and comprehensively reflect the influenced situation of line loss rate prediction.
Example 2:
based on the same concept, the invention provides a line loss prediction system based on an LSTM neural network, which is introduced by combining with the system structure diagram of fig. 5, and specifically includes: the device comprises an acquisition module, a processing module and a prediction value module;
the acquisition module is used for acquiring characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors;
the processing module is used for processing the characteristic data of the line loss rate;
the prediction module is used for inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and obtaining a line loss prediction value after extraction processing;
the line loss prediction model of the LSTM neural network takes the line loss rate influence factors as an input layer of the model, replaces ordinary neurons of a hidden layer with a memory module containing a gating mechanism, and obtains a line loss prediction value after training.
The processing module comprises: a fill submodule, a modify submodule and a normalize submodule;
the filling submodule is used for determining a missing value of the characteristic data of the line loss rate acquired in advance based on a predetermined line loss rate influence factor, filling the missing value and carrying out consistent processing;
the correction submodule is used for replacing the abnormal characteristic data with the average value of adjacent normal line loss for the abnormal characteristic data in the characteristic data of the line loss rate acquired in advance and performing error correction;
and the normalization submodule is used for performing normalization processing on the basis of the normal, filled and error corrected characteristic data.
The system further comprises: a model building module;
and the model construction module is used for training the LSTM neural network to obtain an LSTM neural network line loss prediction model.
The model building module comprises: the system comprises a historical data submodule, a dividing submodule, an initial model submodule, a predicted value submodule and a checking submodule;
the historical data submodule is used for acquiring historical data containing the line loss rate of a time sequence based on the line loss rate influence factors and processing the historical data;
the dividing submodule is used for dividing the processed historical data into a training data set and a testing data set;
the initial model submodule is used for taking the line loss rate influence factors in the training data set as an input layer, taking the line loss values in the training data set as an output layer, replacing common neurons of a hidden layer with a memory module containing a gating mechanism, and training to obtain an initial LSTM neural network line loss prediction model;
the line loss rate influence factor value of the test data set is substituted into the initial LSTM neural network line loss prediction model to obtain a line loss prediction value;
and the verification submodule is used for verifying the initial LSTM neural network line loss prediction model based on the line loss prediction value and the actual line loss value in the test data set to obtain a trained LSTM neural network line loss prediction model.
The check submodule comprises: the system comprises a verification input unit, an evaluation unit and a training unit;
the verification input unit is used for inputting the test data set into the trained LSTM neural network prediction model;
the evaluation unit is used for carrying out accuracy evaluation by adopting a root mean square error index and an average absolute percentage error index based on the line loss predicted value and the line loss actual value of the test data set;
and the training unit is used for continuing training the LSTM neural network line loss prediction model based on the time sequence characteristic when the accuracy does not reach a preset threshold value until the trained LSTM neural network line loss prediction model is obtained.
The prediction module comprises: the input gate submodule, the forgetting gate submodule, the output gate submodule and the activation output submodule;
the input gate submodule is used for sending the training data set, the state parameters of the previous-time hidden layer and the state parameters of the previous-time memory module to an input gate for calculation to obtain a calculation result of a one-dimensional column vector, and sending the training data set to a forgetting gate after keeping current input information in the training data set;
the forgetting gate submodule is used for calculating the training data set by using the forgetting gate to obtain a calculation result of a one-dimensional column vector, and sending the training data set to an output gate after history information needing to be reserved in the updated memory module is reserved;
the output gate submodule is used for calculating the training data set by using the output gate to obtain a calculation result of a one-dimensional column vector, and outputting the state parameters of the memory module at the current moment and the state parameters of the hidden layer at the current moment after output information needing to be reserved in the training data set is reserved;
and the activation output sub-module is used for inputting the calculation result of the one-dimensional column vector into the linear function for calculation based on the activation formula of the linear function to obtain a line loss prediction value.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may 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, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The present invention is not limited to the above embodiments, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention are included in the scope of the claims of the present invention which are filed as the application.
Claims (10)
1. A line loss prediction method based on an LSTM neural network is characterized by comprising the following steps:
collecting characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors;
processing the characteristic data of the line loss rate;
inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and extracting to obtain a line loss prediction value;
the line loss prediction model of the LSTM neural network takes the line loss rate influence factors as an input layer of the model, replaces ordinary neurons of a hidden layer with a memory module containing a gating mechanism, and obtains a line loss prediction value after training.
2. The method of claim 1, wherein the line loss rate affecting factors comprise: power supply, distribution capacity, line length, power factor, air temperature and holidays.
3. The method of claim 2, wherein the processing the characterization data of the line loss rate comprises:
determining a missing value of the feature data of the line loss rate acquired in advance based on the influence factors of the line loss rate determined in advance, filling the missing value, and performing consistent processing;
replacing the abnormal characteristic data with the average value of adjacent normal line loss for the abnormal characteristic data in the characteristic data of the line loss rate acquired in advance and correcting errors;
normalization processing is performed based on the normal, filled and error corrected feature data.
4. The method of claim 1, wherein the constructing of the LSTM neural network line loss prediction model comprises:
acquiring historical data of the line loss rate containing a time sequence based on the line loss rate influence factors, and processing the historical data;
dividing the processed historical data into a training data set and a testing data set;
taking the line loss rate influence factors in the training data set as an input layer, taking the line loss value in the training data set as an output layer, replacing common neurons of a hidden layer with a memory module containing a gating mechanism, and training to obtain an initial LSTM neural network line loss prediction model;
substituting the line loss rate influence factor value of the test data set into the initial LSTM neural network line loss prediction model to obtain a line loss prediction value;
and verifying the initial LSTM neural network line loss prediction model based on the line loss prediction value and the actual line loss value in the test data set to obtain a trained LSTM neural network line loss prediction model.
5. The method of claim 4, wherein the verifying the initial LSTM neural network line loss prediction model based on the line loss prediction values and actual line loss values in the test dataset results in a trained LSTM neural network line loss prediction model, comprising:
inputting the test data set into the trained LSTM neural network prediction model;
based on the predicted line loss value and the actual line loss value of the test data set, carrying out accuracy evaluation by adopting a root mean square error index and an average absolute percentage error index;
and when the accuracy does not reach the preset threshold value, continuing training the LSTM neural network line loss prediction model based on the time sequence characteristic until the preset threshold value is reached to obtain the trained LSTM neural network line loss prediction model.
6. The method of claim 5, wherein the inputting the processed feature data into a line loss prediction model of the LSTM neural network trained in advance, and obtaining a line loss prediction value after extraction processing, comprises:
sending the training data set, the state parameters of the previous-time hidden layer and the state parameters of the memory module to a memory module for calculation to obtain a calculation result of a one-dimensional column vector, and sending the training data set to a forgetting gate of the memory module after keeping current input information in the training data set;
calculating the training data set by using the forgetting gate to obtain a calculation result of a one-dimensional column vector, and sending the training data set to an output gate of a memory module after historical information needing to be reserved in the updated memory module is reserved;
calculating the training data set by using the output gate to obtain a calculation result of a one-dimensional column vector, and outputting the state parameters of the memory module at the current moment and the state parameters of the hidden layer at the current moment after output information needing to be reserved in the training data set is reserved;
and calculating the calculation result of the one-dimensional column vector based on an activation formula of a linear function to obtain a line loss prediction value.
7. The method of claim 6, wherein the activation formula of the linear function is as follows:
y＝actv(w·a+b)
where a is a calculation result of a one-dimensional column vector, y is a line loss prediction value, actv () is an activation function, w is a weight parameter of input data, and b is a constant.
8. A line loss prediction system based on an LSTM neural network is characterized by comprising: the device comprises an acquisition module, a processing module and a prediction value module;
the acquisition module is used for acquiring characteristic data of the line loss rate according to a time sequence based on predetermined line loss rate influence factors;
the processing module is used for processing the characteristic data of the line loss rate;
the prediction module is used for inputting the processed characteristic data into a pre-trained LSTM neural network line loss prediction model, and obtaining a line loss prediction value after extraction processing;
the line loss prediction model of the LSTM neural network takes the line loss rate influence factors as an input layer of the model, replaces ordinary neurons of a hidden layer with a memory module containing a gating mechanism, and obtains a line loss prediction value after training.
9. The system of claim 8, wherein the processing module comprises: a fill submodule, a modify submodule and a normalize submodule;
the filling submodule is used for determining a missing value of the characteristic data of the line loss rate acquired in advance based on a predetermined line loss rate influence factor, filling the missing value and carrying out consistent processing;
the correction submodule is used for replacing the abnormal characteristic data with the average value of adjacent normal line loss for the abnormal characteristic data in the characteristic data of the line loss rate acquired in advance and performing error correction;
and the normalization submodule is used for performing normalization processing on the basis of the normal, filled and error corrected characteristic data.
10. The system of claim 9, further comprising: a model building module;
and the model construction module is used for training the LSTM neural network to obtain an LSTM neural network line loss prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010309179.4A CN111639783A (en) | 2020-04-17 | 2020-04-17 | Line loss prediction method and system based on LSTM neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010309179.4A CN111639783A (en) | 2020-04-17 | 2020-04-17 | Line loss prediction method and system based on LSTM neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111639783A true CN111639783A (en) | 2020-09-08 |
Family
ID=72328714
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010309179.4A Pending CN111639783A (en) | 2020-04-17 | 2020-04-17 | Line loss prediction method and system based on LSTM neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111639783A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112258007A (en) * | 2020-10-16 | 2021-01-22 | 国家电网有限公司 | Distribution transformer capacity-based method for quantizing synchronous line loss cost of power distribution network |
| CN112288172A (en) * | 2020-10-30 | 2021-01-29 | 国家电网有限公司 | Prediction method and device for line loss rate of transformer area |
| CN112488395A (en) * | 2020-12-01 | 2021-03-12 | 湖南大学 | Power distribution network line loss prediction method and system |
| CN112508261A (en) * | 2020-12-03 | 2021-03-16 | 广东电网有限责任公司 | Neural network-based distribution transformer load hierarchical prediction method and device |
| CN113837607A (en) * | 2021-09-24 | 2021-12-24 | 浙江中烟工业有限责任公司 | Real-time analysis method and device for abnormal loss of related cut tobacco removed from cigarette packets |
| CN114708927A (en) * | 2022-03-09 | 2022-07-05 | 北京理工大学 | High-temperature alloy fatigue performance prediction method based on grey prediction and LSTM |
| CN115796396A (en) * | 2023-01-16 | 2023-03-14 | 江苏新恒基特种装备股份有限公司 | Method and system for predicting loss of forged material through related parameters |
| CN116882589A (en) * | 2023-09-04 | 2023-10-13 | 国网天津市电力公司营销服务中心 | Online line loss rate prediction method based on Bayesian optimization deep neural network |
| CN116930667A (en) * | 2023-09-15 | 2023-10-24 | 广东电网有限责任公司 | Method, device, equipment and storage medium for testing edge of power grid of transformer area |
| CN117408167A (en) * | 2023-12-15 | 2024-01-16 | 四川省能源地质调查研究所 | Debris flow disaster vulnerability prediction method based on deep neural network |
-
2020
- 2020-04-17 CN CN202010309179.4A patent/CN111639783A/en active Pending
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112258007B (en) * | 2020-10-16 | 2023-11-10 | 国家电网有限公司 | Power distribution network synchronous line loss cost quantification method based on distribution transformer capacity |
| CN112258007A (en) * | 2020-10-16 | 2021-01-22 | 国家电网有限公司 | Distribution transformer capacity-based method for quantizing synchronous line loss cost of power distribution network |
| CN112288172A (en) * | 2020-10-30 | 2021-01-29 | 国家电网有限公司 | Prediction method and device for line loss rate of transformer area |
| CN112488395A (en) * | 2020-12-01 | 2021-03-12 | 湖南大学 | Power distribution network line loss prediction method and system |
| CN112488395B (en) * | 2020-12-01 | 2024-04-05 | 湖南大学 | Method and system for predicting line loss of power distribution network |
| CN112508261A (en) * | 2020-12-03 | 2021-03-16 | 广东电网有限责任公司 | Neural network-based distribution transformer load hierarchical prediction method and device |
| CN112508261B (en) * | 2020-12-03 | 2021-09-03 | 广东电网有限责任公司 | Neural network-based distribution transformer load hierarchical prediction method and device |
| CN113837607A (en) * | 2021-09-24 | 2021-12-24 | 浙江中烟工业有限责任公司 | Real-time analysis method and device for abnormal loss of related cut tobacco removed from cigarette packets |
| CN113837607B (en) * | 2021-09-24 | 2024-04-30 | 浙江中烟工业有限责任公司 | Real-time analysis method and device for loss abnormality of tobacco shreds related to package rejection |
| CN114708927A (en) * | 2022-03-09 | 2022-07-05 | 北京理工大学 | High-temperature alloy fatigue performance prediction method based on grey prediction and LSTM |
| CN115796396A (en) * | 2023-01-16 | 2023-03-14 | 江苏新恒基特种装备股份有限公司 | Method and system for predicting loss of forged material through related parameters |
| CN116882589A (en) * | 2023-09-04 | 2023-10-13 | 国网天津市电力公司营销服务中心 | Online line loss rate prediction method based on Bayesian optimization deep neural network |
| CN116930667A (en) * | 2023-09-15 | 2023-10-24 | 广东电网有限责任公司 | Method, device, equipment and storage medium for testing edge of power grid of transformer area |
| CN117408167A (en) * | 2023-12-15 | 2024-01-16 | 四川省能源地质调查研究所 | Debris flow disaster vulnerability prediction method based on deep neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111639783A (en) | Line loss prediction method and system based on LSTM neural network | |
| CN107967928B (en) | Flash memory chip life prediction method based on mathematical model | |
| CN112488395B (en) | Method and system for predicting line loss of power distribution network | |
| CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
| CN111950810B (en) | Multi-variable time sequence prediction method and equipment based on self-evolution pre-training | |
| CN113449919B (en) | Power consumption prediction method and system based on feature and trend perception | |
| CN112241836B (en) | Virtual load leading parameter identification method based on incremental learning | |
| CN115438726A (en) | Device life and fault type prediction method and system based on digital twin technology | |
| CN115147012A (en) | Carbon emission accounting method and device based on neural network model | |
| CN115358437A (en) | Power supply load prediction method based on convolutional neural network | |
| CN114548494A (en) | Visual cost data prediction intelligent analysis system | |
| CN112232570A (en) | Forward active total electric quantity prediction method and device and readable storage medium | |
| CN117277312A (en) | Gray correlation analysis-based power load influence factor method and equipment | |
| CN112700050A (en) | Method and system for predicting ultra-short-term 1 st point power of photovoltaic power station | |
| CN117113086A (en) | Energy storage unit load prediction method, system, electronic equipment and medium | |
| CN116542701A (en) | Carbon price prediction method and system based on CNN-LSTM combination model | |
| CN116316600A (en) | Wind power day-ahead prediction self-adaptive filtering method based on seasonal typing | |
| CN113111588B (en) | NO of gas turbine X Emission concentration prediction method and device | |
| CN114759579A (en) | Power grid active power optimization control system, method and medium based on data driving | |
| CN114784795A (en) | Wind power prediction method and device, electronic equipment and storage medium | |
| CN113792828A (en) | Power grid load prediction method, system, equipment and medium based on deep learning | |
| CN113762591A (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy | |
| Arakelyan et al. | Analysis of the DCS historical data for estimation of input signal significance | |
| CN117408394B (en) | Carbon emission factor prediction method and device for electric power system and electronic equipment | |
| Indralaksono et al. | Hierarchical Clustering and Deep Learning for Short-Term Load Forecasting with Influenced Factors |
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 |