CN111709790A - Method, device, equipment and storage medium for identifying abnormal electricity price of day-ahead market - Google Patents
Method, device, equipment and storage medium for identifying abnormal electricity price of day-ahead market Download PDFInfo
- Publication number
- CN111709790A CN111709790A CN202010565912.9A CN202010565912A CN111709790A CN 111709790 A CN111709790 A CN 111709790A CN 202010565912 A CN202010565912 A CN 202010565912A CN 111709790 A CN111709790 A CN 111709790A
- Authority
- CN
- China
- Prior art keywords
- electricity price
- abnormal
- day
- node
- preset
- 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.)
- Granted
Links
- 230000005611 electricity Effects 0.000 title claims abstract description 201
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 159
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000012360 testing method Methods 0.000 claims abstract description 101
- 238000007637 random forest analysis Methods 0.000 claims abstract description 57
- 238000013145 classification model Methods 0.000 claims abstract description 29
- 238000012549 training Methods 0.000 claims description 81
- 230000003203 everyday effect Effects 0.000 claims description 22
- 238000013527 convolutional neural network Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 17
- 238000011176 pooling Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 2
- 238000003066 decision tree Methods 0.000 description 5
- 230000004913 activation Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000002790 cross-validation Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000006467 substitution reaction 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0283—Price estimation or determination
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Tourism & Hospitality (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Biology (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
Abstract
The application discloses a method, a device, equipment and a storage medium for identifying abnormal electricity prices of a day-ahead market, which are used for acquiring electricity price data and load data generated by each node in an electricity day-ahead market at each moment in a first preset day; all the electricity price data of each day are used as a test sample to be input into a preset classification model, and the classification result of the test sample is output; when the test sample is an abnormal test sample, respectively inputting load data corresponding to the abnormal test sample to a preset random forest fractional bit regression model corresponding to each node, and outputting a condition confidence interval; when the electricity price data corresponding to the load data is outside the condition confidence interval corresponding to the load data, the electricity price data is abnormal electricity price data, and the corresponding nodes and time are abnormal nodes and abnormal time respectively, so that the technical problems that the existing method for identifying the abnormal electricity price is low in recognition rate and efficiency, and the specific node and time of the abnormal electricity price data cannot be determined to be abnormal are solved.
Description
Technical Field
The application relates to the technical field of power markets, in particular to a method, a device, equipment and a storage medium for identifying abnormal electricity prices in the market in the day ahead.
Background
The node price is applied to settle the electric energy transaction so as to reflect the space-time value of the electric energy, and the method is the direction of the development of the electric power market. At present, part of electric power market research enters a day-ahead market trial operation stage, and most of electric power markets settle accounts by adopting a node electricity price system. Compared with the unified electricity price, the clearing mechanism behind the node electricity price is complex, the node electricity price is abnormal due to the change of a physical model (such as unit parameters, network topology and the like), mistakes and omissions in the collection or transmission of calculation model parameters (such as the network topology is asynchronous with the calculation model), or artificial tampering and the like, and the abnormality is difficult to find by naked eyes, so that the difficulty is brought to supervision and market transparence. The problem that the identification rate and the efficiency are low and the specific node and the time of abnormal electricity price data cannot be determined in the prior art exists in the abnormal electricity price identification.
Disclosure of Invention
The application provides a method, a device, equipment and a storage medium for identifying abnormal electricity prices in the market at present, which are used for solving the technical problems that the existing method for identifying abnormal electricity prices is low in identification rate and efficiency, and the specific node and time electricity price data are abnormal.
In view of the above, a first aspect of the present application provides a method for identifying an abnormal electricity price in a market in the future, including:
obtaining test node data, wherein the test node data comprises electricity price data and load data generated by each node in the power day-ahead market at each moment in a first preset day;
taking the electricity price data generated by all nodes every day in the first preset day as a test sample, inputting the test sample into a preset classification model, and outputting a classification result of the test sample;
when the classification result shows that the test sample is an abnormal test sample, the load data corresponding to each node in the abnormal test sample at each moment are respectively input into a preset random forest score bit regression model corresponding to each node, a condition confidence interval of the electricity price under each load data condition is output, and the preset random forest score bit regression model is used for predicting a lower bound and an upper bound of the condition confidence interval;
when the electricity price data corresponding to the load data is outside the condition confidence interval corresponding to the load data, outputting the electricity price data outside the condition confidence interval as abnormal electricity price data, wherein the node and the time corresponding to the abnormal electricity price data are an abnormal node and an abnormal time respectively.
Optionally, the configuration process of the preset classification model includes:
acquiring training node data, wherein the training node data comprises the electricity price data and the load data generated by each node in the power day-ahead market at each moment in a second preset day;
and taking the electricity price data generated by all nodes every day in the second preset day as a training sample, determining a label of each training sample, inputting the training sample and the label corresponding to the training sample into a preset convolutional neural network for training to obtain the preset classification model, wherein the label is used for indicating whether the training sample is normal or abnormal.
Optionally, the preset convolutional neural network includes a plurality of convolutional units and a full connection layer;
the convolution unit is composed of a convolution layer, a pooling layer and a Dropout layer, and the number of the convolution units is as follows:
n is the number of nodes participating in clearing, and T is the clearing time interval of the day-ahead market every dayThe number of the first and second groups is,
is a rounded down function.
Optionally, the configuration process of the preset random forest fractional bit regression model includes:
constructing a random forest fractional bit regression model for each node;
and simultaneously inputting the load data and the electricity price data generated by each node in the second preset day into the random forest score bit regression model corresponding to each node for training to obtain the preset random forest score bit regression model.
The present application provides in a second aspect a device is identified to market unusual price of electricity day before, include:
the system comprises a first acquisition unit, a second acquisition unit and a control unit, wherein the first acquisition unit is used for acquiring test node data, and the test node data comprises electricity price data and load data which are generated by each node in the power day-ahead market at each moment within a first preset day;
the first input unit is used for taking the electricity price data generated by all nodes every day in the first preset day as a test sample, inputting the test sample into a preset classification model and outputting a classification result of the test sample;
the second input unit is used for respectively inputting the load data corresponding to each node in the abnormal test sample at each moment into a preset random forest score position regression model corresponding to each node when the test sample is displayed as an abnormal test sample according to the classification result, outputting a condition confidence interval of the electricity price under each condition of the load data, and the preset random forest score position regression model is used for predicting a lower bound and an upper bound of the condition confidence interval;
and an output unit, configured to output the electricity price data outside the condition confidence interval as abnormal electricity price data when the electricity price data corresponding to the load data is outside the condition confidence interval corresponding to the load data, where a node and a time corresponding to the abnormal electricity price data are an abnormal node and an abnormal time, respectively.
Optionally, the method further includes:
a second obtaining unit, configured to obtain training node data, where the training node data includes the electricity price data and the load data generated by each node in a power market at each time within a second preset day;
and the first training unit is used for taking the electricity price data generated by all nodes every day in the second preset day as a training sample, determining a label of each training sample, and inputting the training sample and the label corresponding to the training sample into a preset convolutional neural network for training to obtain the preset classification model, wherein the label is used for indicating whether the training sample is normal or abnormal.
Optionally, the preset convolutional neural network includes a plurality of convolutional units and a full connection layer;
the convolution unit is composed of a convolution layer, a pooling layer and a Dropout layer, and the number of the convolution units is as follows:
n is the number of nodes participating in clearing, T is the number of clearing time periods of the market every day ahead,
is a rounded down function.
Optionally, the method further includes:
the construction unit is used for constructing a random forest fractional bit regression model for each node;
and the second training unit is used for simultaneously inputting the load data and the electricity price data generated by each node in the second preset day into the random forest score bit regression model corresponding to each node for training to obtain the preset random forest score bit regression model.
A third aspect of the present application provides a day-ahead market abnormal electricity price identifying apparatus, the apparatus including a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute any one of the method for identifying an abnormal electricity price in a market in the future according to instructions in the program code.
A fourth aspect of the present application provides a computer-readable storage medium for storing program code for executing the method for identifying abnormal electricity prices in a market at present according to any one of the first aspects.
According to the technical scheme, the method has the following advantages:
the application provides a method for identifying abnormal electricity prices in the day-ahead market, which comprises the following steps: acquiring test node data, wherein the test node data comprises electricity price data and load data generated by each node in the power day-ahead market at each moment in a first preset day; taking the electricity price data generated by all nodes every day in the first preset day as a test sample, inputting the test sample into a preset classification model, and outputting a classification result of the test sample; when the classification result shows that the test sample is an abnormal test sample, respectively inputting load data corresponding to each node in the abnormal test sample at each moment into a preset random forest score bit regression model corresponding to each node, outputting a condition confidence interval of the electricity price under each load data condition, wherein the preset random forest score bit regression model is used for predicting the lower bound and the upper bound of the condition confidence interval; and when the electricity price data corresponding to the load data is out of the condition confidence interval corresponding to the load data, outputting the electricity price data out of the condition confidence interval as abnormal electricity price data, wherein the node and the time corresponding to the abnormal electricity price data are respectively an abnormal node and an abnormal time.
The method for identifying the abnormal electricity price of the day-ahead market comprises the steps of judging whether electricity price data of the whole day is abnormal or not through a preset classification model, and when the electricity price data of a certain day is abnormal, further finding out abnormal nodes and abnormal moments of the certain day through a preset random forest fractional bit regression model; the abnormal electricity price can be rapidly and accurately identified through the preset classification model, and the electric power market supervisor can be helped to rapidly judge whether the electricity price is clear or not; in addition, abnormal nodes and abnormal moments can be found out quickly and accurately through a preset random forest fractional bit regression model, and the method is helpful for power market participants to determine the key moments of topology or line parameter changes in the market operation, so that the technical problems that the existing method for identifying abnormal electricity prices is low in recognition rate and efficiency, and specific nodes and moments of abnormal electricity price data cannot be determined are solved.
Drawings
Fig. 1 is a schematic flow chart of a method for identifying an abnormal electricity price in a market in the future according to an embodiment of the present disclosure;
fig. 2 is another schematic flow chart of a method for identifying an abnormal electricity price in a market at the present day time according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a device for identifying an abnormal electricity price in a market in the future according to an embodiment of the present application.
Detailed Description
In order to make the technical solutions of the present application better understood, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
For easy understanding, referring to fig. 1, an embodiment of a method for identifying an abnormal electricity price in a market today includes:
Test node data for testing may be obtained from the power day-ahead market, the test node data including power rate data and load data generated by each node in the power day-ahead market at each time within a first preset day.
And 102, taking the electricity price data generated by all nodes every day in the first preset day as a test sample, inputting the test sample into a preset classification model, and outputting a classification result of the test sample.
Integrating the acquired electricity price data according to the day and the node, taking the electricity price data generated by all the nodes every day as a test sample, namely, one test sample is all the electricity price data of one day, inputting the test sample into a preset classification model to obtain a classification result of the test sample, wherein the preset classification model is used for judging whether the test sample is a normal test sample or an abnormal test sample, namely judging whether the electricity price data of each day is normal or abnormal.
And 103, when the classification result shows that the test sample is an abnormal test sample, respectively inputting the load data corresponding to each node in the abnormal test sample at each moment into a preset random forest fractional bit regression model corresponding to each node, and outputting a condition confidence interval of the electricity price under each load data condition.
When the classification result shows that the test sample is a normal test sample, indicating that the electricity price data of the day is normal; and when the classification result shows that the test sample is an abnormal test sample, indicating that the electricity price data of the day is abnormal. In order to further confirm the node and the time of the abnormal electricity price data of the day, in the embodiment of the application, the load data corresponding to each node in the abnormal test sample at each time is input into a preset random forest fractional bit regression model corresponding to each node, so that a condition confidence interval of the electricity price under each given load data is obtained, namely, a condition confidence interval is correspondingly output for each input load data, wherein the preset random forest fractional bit regression model is used for predicting the lower bound and the upper bound of the condition confidence interval corresponding to the load data.
And 104, when the electricity price data corresponding to the load data is out of the condition confidence interval corresponding to the load data, outputting the electricity price data out of the condition confidence interval as abnormal electricity price data, wherein the node and the time corresponding to the abnormal electricity price data are an abnormal node and an abnormal time respectively.
After the condition confidence interval is obtained, whether the electricity price data corresponding to each load data is outside the condition confidence interval corresponding to the load data or not is determined, the electricity price data outside the condition confidence interval is abnormal electricity price data, the node corresponding to the abnormal electricity price data is an abnormal node, and the time corresponding to the abnormal electricity price data is abnormal time.
The method for identifying the abnormal electricity price of the day-ahead market in the embodiment of the application comprises the steps of judging whether electricity price data of the whole day is abnormal or not through a preset classification model, and when the electricity price data of a certain day is abnormal, further finding out abnormal nodes and abnormal moments of the certain day through a preset random forest fractional bit regression model; the abnormal electricity price can be rapidly and accurately identified through the preset classification model, and the electric power market supervisor can be helped to rapidly judge whether the electricity price is clear or not; in addition, abnormal nodes and abnormal moments can be found out quickly and accurately through a preset random forest fractional bit regression model, and the method is helpful for power market participants to determine the key moments of topology or line parameter changes in the market operation, so that the technical problems that the existing method for identifying abnormal electricity prices is low in recognition rate and efficiency, and specific nodes and moments of abnormal electricity price data cannot be determined are solved.
The above is an embodiment of a method for identifying an abnormal electricity price in a day-ahead market provided by the present application, and the following is another embodiment of the method for identifying an abnormal electricity price in a day-ahead market provided by the present application.
For easy understanding, please refer to fig. 2, another embodiment of a method for identifying abnormal electricity prices in a day-ahead market provided by the present application includes:
Training node data for training is acquired, wherein the training node data comprises electricity price data and load data generated by each node in the power day-ahead market at each moment within the second preset day. Assuming the second preset number of days is K days, dayThe number of the leaving time periods of the market every day is T, the number of the nodes participating in leaving is N, and for the T time period and the N node of the k day, the corresponding load data is recorded as
Electricity price data is
Where K is 1,2, …, K, T is 1,2, …, T, N is 1,2, …, N, and the electricity price data are integrated by day and node, and the obtained electricity price data of the K day can be represented as an N × T-dimensional electricity price matrix:
all the electricity rate data for K days acquired may be expressed as P ═ P(1),P(2),…,P(k),…,P(K)]。
Aiming at the node scale of the power market, a corresponding convolutional neural network is constructed, the convolutional neural network is formed by stacking a plurality of convolutional layers, a pooling layer, a Dropout layer and a full-connection layer, and an activation function used by each convolutional layer of the preset convolutional neural network in the embodiment of the application is a ReLU function, namely:
f(x)＝max{0,x}；
where x is the input training sample.
The activation function used by the fully-connected layer may be a ReLU function or a sigmoid function, that is:
in the preset convolutional neural network in the embodiment of the application, a pooling layer and a Dropout layer are connected behind each convolutional layer, and one convolutional layer, one pooling layer and one Dropout layer are defined as a convolution unit, the preset convolutional neural network in the application is composed of a plurality of convolution units and fully-connected layers, wherein the number of the convolution units is M, the value of M is related to the number of nodes, and the calculation formula of M is as follows:
n is the number of nodes participating in clearing, T is the number of clearing time periods of the market every day ahead,
is a rounded down function.
Is provided with
The preset convolutional neural network has m +2 convolutional units, the size of convolutional kernels in convolutional layers is preferably 3 × 3, the number of convolutional kernels in the first convolutional layer is preferably 8 in the first m convolutional units, the number of convolutional kernels in the convolutional layers except the first convolutional layer is 2 times of the number of convolutional kernels in the convolutional layer in the previous layer, the number of convolutional kernels in the last 2 convolutional units is equal to the number of convolutional kernels in the convolutional layer in the previous layer, all pooling layers adopt the maximum pooling strategy, the size of a pooling layer filter of the first m convolutional units is 2 × 1, the size of the rest pooling layer filters is 2 ×. Dropout layers are used for preventing overfitting, the number of _ ratio of each Dropout layer is set to be 0.25, after m +2 convolutional units and one flaten layer, the features are expanded into a vector, the number of neurons of the first fully connected layer is consistent with the number of channels of the previous layer, the number of activation functions of the Draopout layers is a second full connected function, and the full connected functions of Dropout layers are two Dropout layers.
The electricity price data generated by all nodes in the second preset day is used as a training sample, namely, P ═ P(1),P(2),…,P(k),…,P(K)]For the training set, P in the training set P(k)For a training sample, each training sample P is determined(k)Wherein the label is used to indicate that the training sample is normal or abnormal, i.e. to determine the training sample P(k)Inputting the training samples and labels corresponding to the training samples into a preset convolutional neural network for training to obtain a preset classification model, training the optimal parameters of the preset convolutional neural network by using a cross entropy as a loss function and adopting an RMSprop algorithm and a 5-fold cross validation (5-fold cross validation) strategy in the training process, and judging whether the electricity price data of the day is abnormal or not by the trained classification model according to an input N × T-dimensional electricity price matrix.
And step 203, constructing a random forest fractional bit regression model for each node.
In order to further judge the time and the node when the abnormal electricity price occurs, a random forest score bit regression model between the load and the electricity price is constructed and trained to obtain a condition confidence interval of the electricity price under the given load. In the embodiment of the application, a random forest fraction bit regression model is constructed for each node and used for predicting a lower bound l and an upper bound u of a condition confidence interval, the random forest fraction bit regression model is marked as y (X, q) ═ f (X | s, theta, q), wherein X is model input, and during training, X is load data of each node at each moment in a second preset day
And electricity price data of the same node at the first 4 moments of the moment
The number of the input X electricity price data can be selected according to the actual situation, and the X electricity price data can be electricity price data of the first 5 moments or electricity price data of the first 3 moments; when t-j is<When 1 hour, the electricity price data is the electricity price data at the corresponding time of the last day, namely
The output y (X, q) is given XThe electricity price corresponding to the time division digit q, s is the number of decision trees contained in the random forest fractional digit regression model, theta is a decision parameter and is represented by theta1，θ2,，…，θsAnd (4) forming.
And 204, simultaneously inputting the load data and the electricity price data generated by each node in the second preset day into a random forest score bit regression model corresponding to each node for training to obtain a preset random forest score bit regression model.
The training process of the random forest fraction bit regression model y (X, q) ═ f (X | s, theta, q) is independent of q, and the random forest model y (X) ═ f containing the same decision tree and decision parameters is trained by using the sum of squared errors as a loss function0(X | s, θ), whose output y (X) is the desired electricity price y given input X. The optimal parameter s can be determined by a 5-fold cross validation method and is marked as s*Then inputting the load data and the electricity price data generated by each node simultaneously0(X | s, theta) training is carried out to obtain the optimal theta which is recorded as
Thereby obtaining a preset random forest fraction bit regression model.
Setting the number of training samples for training a random forest fractional bit regression model to be l, giving an X according to the properties of a decision tree, and outputting a weight to an ith training sample by a jth decision tree
And the comprehensive weight w obtained by the regression model of the corresponding random forest fractional bitsi(X) is s*The weighted average of the decision tree, namely:
the weight can be used to output a prediction of the cumulative distribution function for electricity price y (y ∈ R, y is a real number set) given X
Namely:
wherein the function 1{·}A Boolean function of an input condition is used, the condition is that 1 is taken when true, otherwise 0 is taken, and a random forest fraction bit regression model f (X | s)*,θ*Q) is:
wherein inf represents the lower bound of the set, the lower quantile of the condition interval set in the embodiment of the present application is 0.05, the upper quantile is 0.95, that is, the lower quantile corresponds to a 90% confidence interval, and the breakpoints of the confidence intervals are y (X,0.05) and y (X,0.95), respectively.
Test node data for testing may be obtained from the power day-ahead market, the test node data including power rate data and load data generated by each node in the power day-ahead market at each time within a first preset day. Wherein the date of the first preset day is different from the date of the second preset day.
And step 206, taking the electricity price data generated by all nodes every day in the first preset day as a test sample, inputting the test sample into a preset classification model, and outputting a classification result of the test sample.
Suppose that the first preset number of days is H, the t-th period of the H-th day and the load data of the n-th node are
Electricity price data is
Wherein H is 1,2, …, H, T is 1,2, …, T, N is 1,2, …, N, the obtained electricity price data is according to the sum of daysThe nodes are integrated, and the obtained electricity price data of the h day can be represented as an N × T-dimensional electricity price matrix:
all the electricity rate data for H days acquired may be expressed as P ═ P(1),P(2),…,P(h),…,P(H)]The electricity price data generated by all nodes every day is used as a test sample, namely
And inputting the test sample into the preset classification model obtained by the configuration in the previous step to obtain a classification result of the test sample, wherein the preset classification model is used for judging whether the test sample is a normal test sample or an abnormal test sample, namely judging whether the daily electricity price data is normal or abnormal.
And step 207, when the classification result shows that the test sample is an abnormal test sample, respectively inputting the load data corresponding to each node in the abnormal test sample at each moment into a preset random forest fractional bit regression model corresponding to each node, and outputting a condition confidence interval of the electricity price under each load data condition.
When the classification result shows that the test sample is a normal test sample, indicating that the electricity price data of the day is normal; and when the classification result shows that the test sample is an abnormal test sample, indicating that the electricity price data of the day is abnormal. In order to further confirm which node is abnormal and which time is abnormal in the abnormal electricity price data of the day, in the embodiment of the application, the load data corresponding to each node in the abnormal test sample at each time is input into the preset random forest score bit regression model corresponding to each node, so as to obtain the condition confidence interval of the electricity price under each given load data, wherein the preset random forest score bit regression model is used for predicting the lower bound and the upper bound of the condition confidence interval corresponding to the load data.
Supposing that the preset classification model outputs a test sample PhFor abnormal test samples, i.e. electricity price data on day h
If the time of day h is abnormal, and if the time of day h is abnormal, the node is abnormal, the test sample P is testedhRespectively inputting the load data corresponding to each node at each moment into a preset random forest score bit regression model corresponding to each node, that is, inputting the load data of each node on the h day corresponding to the electricity price data of each node on the h day into the preset random forest score bit regression model corresponding to each node, for example, inputting the load data of each node on the h day corresponding to the Nth node in the T time period
Inputting the data into a preset random forest fractional bit regression model corresponding to the Nth node, and outputting the data of electricity price at a given load
A conditional confidence interval of, wherein the load data for day h can be expressed as:
and step 208, when the electricity price data corresponding to the load data is out of the condition confidence interval corresponding to the load data, outputting the electricity price data out of the condition confidence interval as abnormal electricity price data, wherein the node and the time corresponding to the abnormal electricity price data are an abnormal node and an abnormal time respectively.
After the condition confidence interval of the electricity price under each given load data is obtained, whether the electricity price data corresponding to each load data is outside the condition confidence interval corresponding to the corresponding load data or not is determined, the electricity price data outside the condition confidence interval is abnormal electricity price data, a node corresponding to the abnormal electricity price data is an abnormal node, and the time corresponding to the abnormal electricity price data is abnormal time. Continuing with the above example, the electricity price data of the h day is obtained to be abnormal through the preset classification model, and then each load data of the h day is input into the preset random forest score bitModeling, obtaining each condition confidence interval of the electricity price under each given load data, and supposing that the load data of the 3 rd node of the 4 th time period of the h day
Corresponding conditional confidence interval of [ l, u]And load data of the 3 rd node in the 4 th period of the h day is found by comparison
Corresponding electricity price data
(true price of electricity) in a conditional confidence interval [ l, u]External, therefore, load data
Corresponding electricity price data
Is abnormal electricity price data
The corresponding node N is 3 which is an abnormal node, and the abnormal electricity price data
The corresponding time when the corresponding time period T is 4 is an abnormal time.
The above is an embodiment of a method for identifying an abnormal electricity price in a day-ahead market provided by the present application, and the following is an embodiment of an apparatus for identifying an abnormal electricity price in a day-ahead market provided by the present application.
For easy understanding, referring to fig. 3, an embodiment of a device for identifying abnormal electricity prices in a day-ahead market provided by the present application includes:
the first acquiring unit 301 is configured to acquire test node data, where the test node data includes electricity price data and load data generated by each node in the power market at each time within a first preset day.
The first input unit 302 is configured to use the electricity price data generated by all nodes every day in the first preset day as a test sample, input the test sample to the preset classification model, and output a classification result of the test sample.
And a second input unit 303, configured to, when the classification result shows that the test sample is an abnormal test sample, respectively input load data corresponding to each node in the abnormal test sample at each time to a preset random forest score bit regression model corresponding to each node, and output a condition confidence interval of the electricity price under each load data condition, where the preset random forest score bit regression model is used to predict a lower bound and an upper bound of the condition confidence interval.
And an output unit 304, configured to output the electricity price data outside the condition confidence interval as abnormal electricity price data when the electricity price data corresponding to the load data is outside the condition confidence interval corresponding to the load data, where a node and a time corresponding to the abnormal electricity price data are an abnormal node and an abnormal time, respectively.
As a further improvement, the method further comprises the following steps:
a second obtaining unit 305, configured to obtain training node data, where the training node data includes electricity price data and load data generated by each node in the power market at each time within a second preset day.
And the first training unit 306 is configured to use the electricity price data generated by all nodes every day in the second preset day as a training sample, determine a label of each training sample, input the training sample and the label corresponding to the training sample into a preset convolutional neural network for training, and obtain a preset classification model, where the label is used to indicate that the training sample is normal or abnormal.
As a further improvement, the preset convolution neural network comprises a plurality of convolution units and a full connection layer;
the convolution unit is composed of a convolution layer, a pooling layer and a Dropout layer, and the number of the convolution units is as follows:
n is the number of nodes participating in clearing, T is the number of clearing time periods of the market every day ahead,
is a rounded down function.
As a further improvement, the method further comprises the following steps:
and the constructing unit 307 is configured to construct a random forest fractional bit regression model for each node.
And the second training unit 308 is configured to input the load data and the electricity price data generated by each node in the second preset day to the random forest score bit regression model corresponding to each node at the same time for training, so as to obtain the preset random forest score bit regression model.
The embodiment of the application also provides equipment for identifying the abnormal electricity price of the day-ahead market, which comprises a processor and a memory;
the memory is used for storing the program codes and transmitting the program codes to the processor;
the processor is used for executing the method for identifying the abnormal electricity price of the market at present in the embodiment of the method for identifying the abnormal electricity price of the market at present according to the instructions in the program codes.
An embodiment of the present application further provides a computer-readable storage medium, which is configured to store a program code, where the program code is configured to execute the method for identifying an abnormal power price in a market at present in the embodiment of the method for identifying an abnormal power price in a market at present.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for executing all or part of the steps of the method described in the embodiments of the present application through a computer device (which may be a personal computer, a server, or a network device). And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (10)
1. A method for identifying abnormal electricity prices in the day-ahead market is characterized by comprising the following steps:
obtaining test node data, wherein the test node data comprises electricity price data and load data generated by each node in the power day-ahead market at each moment in a first preset day;
taking the electricity price data generated by all nodes every day in the first preset day as a test sample, inputting the test sample into a preset classification model, and outputting a classification result of the test sample;
when the classification result shows that the test sample is an abnormal test sample, the load data corresponding to each node in the abnormal test sample at each moment are respectively input into a preset random forest score bit regression model corresponding to each node, a condition confidence interval of the electricity price under each load data condition is output, and the preset random forest score bit regression model is used for predicting a lower bound and an upper bound of the condition confidence interval;
when the electricity price data corresponding to the load data is outside the condition confidence interval corresponding to the load data, outputting the electricity price data outside the condition confidence interval as abnormal electricity price data, wherein the node and the time corresponding to the abnormal electricity price data are an abnormal node and an abnormal time respectively.
2. The method for identifying abnormal electricity prices in the market today according to claim 1, wherein the configuration process of the preset classification model comprises:
acquiring training node data, wherein the training node data comprises the electricity price data and the load data generated by each node in the power day-ahead market at each moment in a second preset day;
and taking the electricity price data generated by all nodes every day in the second preset day as a training sample, determining a label of each training sample, inputting the training sample and the label corresponding to the training sample into a preset convolutional neural network for training to obtain the preset classification model, wherein the label is used for indicating whether the training sample is normal or abnormal.
3. The method for identifying abnormal electricity prices in the market today according to claim 2, characterized in that the pre-configured convolutional neural network comprises a plurality of convolutional units and a full connection layer;
the convolution unit is composed of a convolution layer, a pooling layer and a Dropout layer, and the number of the convolution units is as follows:
n is the number of nodes participating in clearing, T is the number of clearing time periods of the market every day ahead,
is a rounded down function.
4. The method for identifying abnormal electricity prices in the market at present according to claim 2, wherein the configuration process of the preset random forest score bit regression model comprises the following steps:
constructing a random forest fractional bit regression model for each node;
and simultaneously inputting the load data and the electricity price data generated by each node in the second preset day into the random forest score bit regression model corresponding to each node for training to obtain the preset random forest score bit regression model.
5. An abnormal electricity price identification device for the day-ahead market is characterized by comprising:
the system comprises a first acquisition unit, a second acquisition unit and a control unit, wherein the first acquisition unit is used for acquiring test node data, and the test node data comprises electricity price data and load data which are generated by each node in the power day-ahead market at each moment within a first preset day;
the first input unit is used for taking the electricity price data generated by all nodes every day in the first preset day as a test sample, inputting the test sample into a preset classification model and outputting a classification result of the test sample;
the second input unit is used for respectively inputting the load data corresponding to each node in the abnormal test sample at each moment into a preset random forest score position regression model corresponding to each node when the test sample is displayed as an abnormal test sample according to the classification result, outputting a condition confidence interval of the electricity price under each condition of the load data, and the preset random forest score position regression model is used for predicting a lower bound and an upper bound of the condition confidence interval;
and an output unit, configured to output the electricity price data outside the condition confidence interval as abnormal electricity price data when the electricity price data corresponding to the load data is outside the condition confidence interval corresponding to the load data, where a node and a time corresponding to the abnormal electricity price data are an abnormal node and an abnormal time, respectively.
6. The day-ahead market abnormal electricity price recognition device according to claim 5, further comprising:
a second obtaining unit, configured to obtain training node data, where the training node data includes the electricity price data and the load data generated by each node in a power market at each time within a second preset day;
and the first training unit is used for taking the electricity price data generated by all nodes every day in the second preset day as a training sample, determining a label of each training sample, and inputting the training sample and the label corresponding to the training sample into a preset convolutional neural network for training to obtain the preset classification model, wherein the label is used for indicating whether the training sample is normal or abnormal.
7. The day-ahead market abnormal electricity price identification device according to claim 6, wherein the preset convolutional neural network comprises a plurality of convolutional units and a full connection layer;
the convolution unit is composed of a convolution layer, a pooling layer and a Dropout layer, and the number of the convolution units is as follows:
n is the number of nodes participating in clearing, T is the number of clearing time periods of the market every day ahead,
is a rounded down function.
8. The day-ahead market abnormal electricity price recognition device according to claim 6, further comprising:
the construction unit is used for constructing a random forest fractional bit regression model for each node;
and the second training unit is used for simultaneously inputting the load data and the electricity price data generated by each node in the second preset day into the random forest score bit regression model corresponding to each node for training to obtain the preset random forest score bit regression model.
9. The device for identifying the abnormal electricity price of the market at present is characterized by comprising a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is used for executing the method for identifying the abnormal electricity price of the market in the future according to the instructions in the program codes, wherein the method for identifying the abnormal electricity price of the market in the future is defined in any one of claims 1 to 4.
10. A computer-readable storage medium for storing program code for executing the method for identifying abnormal electricity prices in a market today according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010565912.9A CN111709790B (en) | 2020-06-19 | 2020-06-19 | Method, device, equipment and storage medium for identifying abnormal electricity price of day-ahead market |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010565912.9A CN111709790B (en) | 2020-06-19 | 2020-06-19 | Method, device, equipment and storage medium for identifying abnormal electricity price of day-ahead market |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111709790A true CN111709790A (en) | 2020-09-25 |
| CN111709790B CN111709790B (en) | 2022-01-07 |
Family
ID=72541227
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010565912.9A Active CN111709790B (en) | 2020-06-19 | 2020-06-19 | Method, device, equipment and storage medium for identifying abnormal electricity price of day-ahead market |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111709790B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112507686A (en) * | 2020-12-07 | 2021-03-16 | 深圳供电局有限公司 | User electricity price file management method and system, computer device and storage medium |
| CN112819067A (en) * | 2021-01-28 | 2021-05-18 | 广东电网有限责任公司广州供电局 | Method, device, equipment and storage medium for processing bad data of power distribution network |
| CN115409296A (en) * | 2022-11-02 | 2022-11-29 | 国网江西省电力有限公司电力科学研究院 | Method for predicting net load probability of active power distribution network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104794192A (en) * | 2015-04-17 | 2015-07-22 | 南京大学 | Multi-level anomaly detection method based on exponential smoothing and integrated learning model |
| CN106204335A (en) * | 2016-07-21 | 2016-12-07 | 广东工业大学 | A kind of electricity price performs abnormality judgment method, Apparatus and system |
| CN108062616A (en) * | 2017-11-20 | 2018-05-22 | 国网福建省电力有限公司 | A kind of multiplexing electric abnormality detection model with constraints |
| CN109345786A (en) * | 2018-09-13 | 2019-02-15 | 国网上海市电力公司 | A kind of non-resident user power utilization abnormal conditions automatic alarm system of low pressure |
| CN109726731A (en) * | 2018-07-26 | 2019-05-07 | 平安科技（深圳）有限公司 | Electricity consumption method for detecting abnormality, device, equipment and readable storage medium storing program for executing |
| CN110045209A (en) * | 2019-05-10 | 2019-07-23 | 广东电网有限责任公司 | Detection method, device, equipment and the readable storage medium storing program for executing of electricity consumption data exception |
-
2020
- 2020-06-19 CN CN202010565912.9A patent/CN111709790B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104794192A (en) * | 2015-04-17 | 2015-07-22 | 南京大学 | Multi-level anomaly detection method based on exponential smoothing and integrated learning model |
| CN106204335A (en) * | 2016-07-21 | 2016-12-07 | 广东工业大学 | A kind of electricity price performs abnormality judgment method, Apparatus and system |
| CN108062616A (en) * | 2017-11-20 | 2018-05-22 | 国网福建省电力有限公司 | A kind of multiplexing electric abnormality detection model with constraints |
| CN109726731A (en) * | 2018-07-26 | 2019-05-07 | 平安科技（深圳）有限公司 | Electricity consumption method for detecting abnormality, device, equipment and readable storage medium storing program for executing |
| CN109345786A (en) * | 2018-09-13 | 2019-02-15 | 国网上海市电力公司 | A kind of non-resident user power utilization abnormal conditions automatic alarm system of low pressure |
| CN110045209A (en) * | 2019-05-10 | 2019-07-23 | 广东电网有限责任公司 | Detection method, device, equipment and the readable storage medium storing program for executing of electricity consumption data exception |
Non-Patent Citations (2)
| Title |
|---|
| PENG LI 等: "Power Network Fault Diagnosis Based on Multi - source Network Message Data", 《IOP CONFERENCE SERIES: MATERIALS SCIENCE AND ENGINEERING》 * |
| 杨学良 等: "基于随机森林及时间序列模型的电能量异常数据检测及修复方法", 《第三届 智能电网会议论文集 计算机工程与应用》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112507686A (en) * | 2020-12-07 | 2021-03-16 | 深圳供电局有限公司 | User electricity price file management method and system, computer device and storage medium |
| CN112819067A (en) * | 2021-01-28 | 2021-05-18 | 广东电网有限责任公司广州供电局 | Method, device, equipment and storage medium for processing bad data of power distribution network |
| CN115409296A (en) * | 2022-11-02 | 2022-11-29 | 国网江西省电力有限公司电力科学研究院 | Method for predicting net load probability of active power distribution network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111709790B (en) | 2022-01-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111709790B (en) | Method, device, equipment and storage medium for identifying abnormal electricity price of day-ahead market | |
| CN110473083B (en) | Tree risk account identification method, device, server and storage medium | |
| CN110888857A (en) | Data label generation method, device, terminal and medium based on neural network | |
| CN111553759A (en) | Product information pushing method, device, equipment and storage medium | |
| CN111652066A (en) | Medical behavior identification method based on multi-self-attention mechanism deep learning | |
| CN110852881B (en) | Risk account identification method and device, electronic equipment and medium | |
| CN113592019B (en) | Fault detection method, device, equipment and medium based on multi-model fusion | |
| CN110298663A (en) | Based on the wide fraudulent trading detection method learnt deeply of sequence | |
| Li et al. | Multi-factor based stock price prediction using hybrid neural networks with attention mechanism | |
| CN110471821A (en) | Abnormal alteration detection method, server and computer readable storage medium | |
| CN109886554A (en) | Unlawful practice method of discrimination, device, computer equipment and storage medium | |
| CN110264270A (en) | A kind of behavior prediction method, apparatus, equipment and storage medium | |
| CN112381019B (en) | Compound expression recognition method and device, terminal equipment and storage medium | |
| CN116579447A (en) | Time sequence prediction method based on decomposition mechanism and attention mechanism | |
| CN114970357A (en) | Energy-saving effect evaluation method, system, device and storage medium | |
| CN116340726A (en) | Energy economy big data cleaning method, system, equipment and storage medium | |
| CN105913142B (en) | Method for improving order completion period prediction accuracy by utilizing workshop RFID data | |
| CN111126629A (en) | Model generation method, system, device and medium for identifying brushing behavior | |
| CN112767190B (en) | Method and device for identifying phase sequence of transformer area based on multilayer stacked neural network | |
| CN115952928B (en) | Short-term power load prediction method, device, equipment and storage medium | |
| CN108765137A (en) | A kind of credit demand prediction technique and system, storage medium | |
| CN112541579A (en) | Model training method, poverty degree information identification method, device and storage medium | |
| CN117009785A (en) | Security monitoring method, device, server and system | |
| CN116662307A (en) | Intelligent early warning method, system and equipment based on multi-source data fusion | |
| CN111666991A (en) | Convolutional neural network-based pattern recognition method and device and computer equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |