CN112508734B - Method and device for predicting power generation capacity of power enterprise based on convolutional neural network - Google Patents
Method and device for predicting power generation capacity of power enterprise based on convolutional neural network Download PDFInfo
- Publication number
- CN112508734B CN112508734B CN202011363422.7A CN202011363422A CN112508734B CN 112508734 B CN112508734 B CN 112508734B CN 202011363422 A CN202011363422 A CN 202011363422A CN 112508734 B CN112508734 B CN 112508734B
- Authority
- CN
- China
- Prior art keywords
- neural network
- convolutional neural
- data
- training
- enterprise
- 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.)
- Active
Links
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 124
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000010248 power generation Methods 0.000 title claims description 40
- 238000012549 training Methods 0.000 claims abstract description 133
- 230000005611 electricity Effects 0.000 claims abstract description 111
- 238000012360 testing method Methods 0.000 claims abstract description 42
- 239000011159 matrix material Substances 0.000 claims description 36
- 238000004519 manufacturing process Methods 0.000 claims description 22
- 238000011176 pooling Methods 0.000 claims description 14
- 230000008901 benefit Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 10
- 230000009467 reduction Effects 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 2
- 239000013598 vector Substances 0.000 description 16
- 210000002569 neuron Anatomy 0.000 description 8
- 238000011161 development Methods 0.000 description 7
- 101100134058 Caenorhabditis elegans nth-1 gene Proteins 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000012271 agricultural production Methods 0.000 description 2
- 230000003631 expected effect Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
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
- 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
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Biophysics (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Primary Health Care (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a prediction method and a device for generating capacity of an electric power enterprise based on a convolutional neural network, wherein the prediction method comprises the following steps: acquiring historical electricity selling data and historical load of an electric power enterprise through an electric power sensor, establishing a training set based on the acquired data, and establishing a test set based on historical enterprise income data of the electric power enterprise; inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, and training the convolutional neural network based on the training result and the test set; the expected electricity selling data of the year is input into a trained convolutional neural network, the expected load is determined through the convolutional neural network, and the electricity generating capacity of the year is determined based on the expected load. The nonlinear relation between the historical electricity selling situation and the enterprise income is constructed based on the convolutional neural network, the obstacle of different index data dimensionalities can be overcome, the relation between the two can be accurately described, and the power enterprise can reasonably plan the power resources according to the prediction result.
Description
Technical Field
The invention belongs to the field of power generation amount prediction, and particularly relates to a power generation amount prediction method and device for an electric power enterprise based on a convolutional neural network.
Background
In the power industry, the income situation of the next year is usually predicted according to historical data of the past year, so that the power resource of the next year is planned, the income situation of a power enterprise is influenced by a plurality of factors such as user type change, domestic overall economic development situation and the like, index data reflecting the factors are often difficult to establish the association between the sales power quantity and the enterprise income due to non-uniform data dimension, and the accuracy of enterprise income prediction is reduced.
Disclosure of Invention
In order to solve the defects and shortcomings in the prior art, the invention provides a method for predicting the power generation capacity of an electric power enterprise based on a convolutional neural network, which comprises the following steps:
acquiring historical electricity selling data and historical load of an electric power enterprise through an electric power sensor, establishing a training set based on the acquired data,
Establishing a test set based on historical enterprise revenue data of the power enterprise;
inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, and training the convolutional neural network based on the training result and the test set;
the expected electricity selling data of the year is input into a trained convolutional neural network, the expected load is determined through the convolutional neural network, and the electricity generating capacity of the year is determined based on the expected load.
Optionally, the acquiring, by the power sensor, historical electricity selling data and historical load of the power enterprise, and establishing the training set based on the acquired data includes:
Acquiring historical electricity selling data and historical load of an electric power enterprise through an electric power sensor, wherein the historical electricity selling data comprises historical electricity selling quantity of the electric power enterprise in preset unit time, and the historical load comprises daily load density and night load density;
Calculating the electricity sales quantity increase rate in unit time according to the historical electricity sales quantity;
and taking the electricity sales volume increase rate, the historical electricity sales data and the historical load as training data, and adding the training data into a training set.
Optionally, the establishing a test set based on the historical enterprise revenue data of the power enterprise includes:
Acquiring prestored historical enterprise revenue data of the power enterprise, wherein the historical enterprise revenue data comprises the electricity fee growth rate and the electricity generation amount of the power enterprise in a preset unit time;
Acquiring a domestic production total value and a domestic production total value increase rate within a preset unit time;
calculating the ratio of the electricity charge increase rate to the domestic total production increase rate to obtain the power consumption elastic coefficient of the power enterprise;
calculating the ratio of the generated energy to the total domestic production value to obtain the electricity generation value of the power enterprise;
and taking the power consumption elastic coefficient and the electricity generation value as test data, and adding the test data into a test set.
Optionally, inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, including:
inputting training data in a training set into a convolutional neural network;
Carrying out feature extraction on input training data through a convolution layer of a convolution neural network, and outputting a feature map;
performing dimension reduction processing on the feature map through a pooling layer of the convolutional neural network to obtain a training matrix after the dimension reduction processing;
inputting the training matrix into a preset classifier, and outputting a training result through the classifier.
Optionally, the training the convolutional neural network based on the training result and the test set includes:
Calculating an average value of data in the test set;
Calculating the error of the training result of the convolutional neural network relative to the calculated average value, and ending the training of the convolutional neural network if the error does not exceed a preset threshold value.
Specifically, the calculating the error of the training result of the convolutional neural network relative to the calculated average value includes:
calculating an error L between a training result f (x) of the convolutional neural network and a calculated average value y based on a formula I, wherein the formula I is as follows:
L= (y-f (x)) 2 formula one;
wherein x is the data input into the convolutional neural network, and the value ranges of x, y and f (x) are positive numbers.
Optionally, the training of the convolutional neural network further includes a process of optimizing the convolutional neural network, and the optimizing process includes:
adjusting various parameters in the convolutional neural network based on a gradient descent algorithm comprising the formula:
Wherein, θ j' is an optimized parameter, θ j is a parameter before optimization, j is a label of the parameter, α is a preset learning rate, and L (θ j) is an objective function of a preset parameter θ j; the values of theta j'、θj and alpha are real numbers, and the value of j is a positive integer.
Optionally, the inputting the expected electricity selling data of the current year into the trained convolutional neural network, determining the expected load through the convolutional neural network, and determining the electricity generating capacity of the current year based on the expected load includes:
Determining expected electricity selling data of the year according to the income situation of enterprises in the last year;
inputting the predicted electricity selling data and the preset load data into a trained convolutional neural network, and outputting annual enterprise benefits through the convolutional neural network;
if the predicted result does not reach the preset profit threshold, the input preset load data is adjusted until the predicted result reaches the preset profit threshold;
And if the predicted result reaches the preset income limit value, taking the adjusted preset load data as the expected load, and distributing the expected load to input power generation equipment according to the expected load to obtain the annual predicted power generation amount.
The invention also provides a device for predicting the power generation capacity of the power enterprise of the convolutional neural network based on the same thought, wherein the device for predicting the power generation capacity of the power enterprise comprises the following components:
Training set unit: the power sensor is used for acquiring historical electricity selling data and historical load of an electric power enterprise, and a training set is established based on the acquired data;
Test set unit: the method comprises the steps of establishing a test set based on historical enterprise revenue data of a power enterprise;
Training unit: the training set is used for inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, and training is carried out on the convolutional neural network based on the training result and the test set;
Prediction unit: the method is used for inputting expected electricity selling data of the year into the trained convolutional neural network, determining expected load through the convolutional neural network, and determining the electricity generation amount of the year based on the expected load.
Optionally, the prediction unit is configured to:
Determining expected electricity selling data of the year according to the income situation of enterprises in the last year;
inputting the predicted electricity selling data and the preset load data into a trained convolutional neural network, and outputting annual enterprise benefits through the convolutional neural network;
if the predicted result does not reach the preset profit threshold, the input preset load data is adjusted until the predicted result reaches the preset profit threshold;
And if the predicted result reaches the preset income limit value, taking the adjusted preset load data as the expected load, and distributing the expected load to input power generation equipment according to the expected load to obtain the annual predicted power generation amount.
The technical scheme provided by the invention has the beneficial effects that:
The nonlinear relation between the electricity selling situation and the enterprise benefits is constructed based on the convolutional neural network, the obstacle of different index data dimensionalities can be overcome, the relation between the electricity selling situation and the enterprise benefits can be accurately described, the accuracy of predicting the enterprise benefits is improved, and the power enterprises can reasonably plan the power resources according to the prediction results.
Drawings
In order to more clearly illustrate the technical solutions of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for predicting the power generation capacity of an electric power enterprise based on a convolutional neural network;
FIG. 2 is a neural network architecture of a classifier;
fig. 3 is a structural block diagram of a prediction device for generating capacity of an electric power enterprise based on a convolutional neural network.
Detailed Description
In order to make the structure and advantages of the present invention more apparent, the structure of the present invention will be further described with reference to the accompanying drawings.
Example 1
As shown in fig. 1, the invention provides a method for predicting the power generation capacity of an electric power enterprise based on a convolutional neural network, which comprises the following steps:
S1: historical electricity selling data and historical load of an electric power enterprise are obtained through the electric power sensor, and a training set is established based on the obtained data.
The historical electricity selling data and the historical load are obtained through the electric power sensors such as the electric energy collecting device deployed in the electric power enterprise, in the embodiment, the historical electricity selling data comprise the electricity selling quantity of the electric power enterprise in a preset unit time, the number of users of each type, and the historical load comprises daily load density and night load density. Types of users include industrial, commercial, agricultural production, and residential class 4 users. And calculating the electricity sales volume increase rate in unit time according to the electricity sales volume, calculating the user increase rate in unit time according to the number of users of each type, and adding the electricity sales volume increase rate and the user increase rate as training data into a training set. In this embodiment, the increase rate of the sales power of each month relative to the sales power of the previous month, that is, the sales power increase rate= (nth sales power/nth-1 month sales power) ×100% and the user increase rate= (nth month user number/nth-1 month user number) ×100% is calculated in the unit time of month, and the calculated sales power increase rate is added to the training set in the form of a row vector of 1×12, that is, training data of 12 dimensions.
Besides the electricity sales quantity and the electricity sales quantity increase rate, other indexes reflecting the electricity development in the training set are specifically as follows:
(1) Total amount of monthly electricity for enterprises: the power supply sum of the power enterprises to the industrial users and the commercial users in each month is represented, index data of the total amount of the power consumption of the enterprises in the year is obtained, and row vectors of 1 x 12, namely 12-dimensional training data are formed;
(2) Total amount of electricity used by residents in the month: the total electricity consumption of the current area of each month is represented, index data of the total electricity consumption of the current year resident month is obtained, and row vectors of 1 x 12 are formed, namely 12-dimensional training data;
(3) Daytime load density: representing the ratio of average load to area in the area between 6 points in daytime and 6 points at night, reflecting the daytime economic index in the area, and forming row vectors of 1 x 365 in units of days, namely 365-dimensional training data;
(4) Night load density: the ratio of average load to area in the area between 6 points at night and 6 points at day after the next day is represented, the night economic index in the area is reflected, and row vectors of 1 x 365 are formed by taking day as a unit, namely 365-dimensional training data.
S2: a test set is established based on historical enterprise revenue data for the power enterprise.
And the test set of the convolutional neural network comprises two parts of data of an electric consumption elastic coefficient and a power generation value. Firstly, historical enterprise revenue data of a power enterprise stored in advance is obtained, wherein the historical enterprise revenue data comprises the electricity fee growth rate and the electricity generation amount of the power enterprise in a preset unit time. The total domestic production value (Gross Domestic Product, GDP) and the total domestic production value increase rate within a preset unit time are obtained, and in this embodiment, the total domestic production value increase rate and the GDP are known in advance. And calculating the ratio of the electricity fee increase rate to the domestic total production increase rate within a preset time period to obtain the power consumption elastic coefficient of the power enterprise. In this embodiment, the ratio of the rate of increase of the electric charge for a certain year to the rate of increase of the total domestic production value for the same year is calculated in units of years, and the elastic coefficient of electric power consumption reflects the relationship between electric power consumption and national economy development. And calculating the ratio of the generated energy to the total domestic production value to obtain the electricity generation value of the power enterprise, wherein the electricity generation value reflects GDP generated by electricity per degree. In the present embodiment, the annual power consumption elastic coefficient and the yearly power generation value are added to the test set as test data in units of years.
The data in the training set reflects the condition of power development, and the data in the testing set reflects the condition of economic benefit created by the power enterprises.
S3: inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, and training the convolutional neural network based on the training result and the test set.
In this embodiment, the convolutional neural network includes four parts of an input layer, a convolutional layer, a pooling layer, and a classifier.
The 12-dimensional training data in the training set are spliced into a first matrix of 4 x 12, the 365-dimensional training data are spliced into a second matrix of 2 x 365, the convolutional neural network is input through an input layer, namely 4 types of training data, namely the electricity selling growth rate of 12 months in one year, the number of newly increased users in each month, the total amount of electricity consumption of enterprises and the total amount of electricity consumption of residents in one month, and the night economic index and day economic index 2 types of training data of 365 days in one year are respectively input into the convolutional neural network.
And extracting the characteristics of the input training data through a convolution layer of the convolution neural network, and outputting a characteristic diagram. In this embodiment, the size of the convolution kernel used by the first matrix composed of 12-dimensional vectors is set to 3*3, the number is 3, and the sliding step s is 1; the size of the convolution kernel used by the second matrix composed of 365-dimensional vectors is set to 1*5, the number is 1, and the sliding step s is 12. The length H 'and width w' of the output feature map can be calculated by the following formulas:
For the first matrix, H is the length of the input matrix, 4 in this embodiment; w is the width of the input matrix, in this embodiment 12; k 1 is the length of the convolution kernel, in this embodiment 3, and k 2 is the width of the convolution kernel, in this embodiment 3; p is a filling value, in this embodiment 0; s is the sliding step length, which is 1 in this embodiment. Therefore, the output feature map size is 2×10×3.
For the second matrix, H is the length of the input matrix, in this embodiment 2; w is the width of the input matrix, 365 in this embodiment; k 1 is the length of the convolution kernel, in this embodiment 1, and k 2 is the width of the convolution kernel, in this embodiment 5; p is a filling value, in this embodiment 0; s is the sliding step length, which is 12 in this embodiment. Therefore, the output feature map size is 2×31×1.
And carrying out dimension reduction processing on the feature map through a pooling layer of the convolutional neural network, and pooling the first matrix and the second matrix by using maximum values respectively to obtain two training matrices after dimension reduction processing. In this embodiment, the size of the pooling window of the first matrix is 2×2, the step size is 1, and the output size can be calculated to be 1×9×3 by using a similar manner to convolution; the pooling window size of the second matrix is 2×2, the step size is 1, and the output size is 1×30×1.
And finally, expanding the two training matrixes subjected to dimension reduction processing into vectors, then merging, inputting the merged matrixes into a preset classifier, and outputting a training result through the classifier. In this embodiment, the number of neurons in the input layer of the classifier is the sum of 1×9×3 of the first matrix pool and 1×30×1 of the second matrix pool, and the total number of neurons is 57, the hidden layer contains 8 neurons, and the output layer is one neuron. The classifier adjusts the dimension of the output result of the pooling layer through the output layer so that the dimension can be compared with the data in the test set.
Based on the training result and the test set, training the convolutional neural network, including:
Firstly, calculating the average value of the data in the test set, namely summing the power consumption elastic coefficient and the power output value and taking the average value.
Calculating an error of a training result of the convolutional neural network relative to the calculated average value, wherein the error is an error L between a training result f (x) of the convolutional neural network and the calculated average value y based on a formula I, and the training result f (x) is a result output by the classifier output layer:
L= (y-f (x)) 2 formula one;
wherein x is the data input into the convolutional neural network, and the value ranges of x, y and f (x) are positive numbers.
If the error L does not exceed the preset threshold, training of the convolutional neural network is finished, and the fact that a regression model between the historical electricity selling data, the number of the historical users and the profit of the historical enterprise of the power enterprise is established at the moment is explained, and the relationship between the electricity selling quantity of the current year, the number of the historical users and the profit of the enterprise can be predicted subsequently. If the error L exceeds a preset threshold, it is indicated that the training of the convolutional neural network at this time does not reach the expected effect, and the model parameters in the convolutional neural network need to be adjusted to optimize the convolutional neural network, where the optimizing process includes:
The gradient descent is one of the most commonly adopted methods when solving model parameters of a machine learning algorithm, namely, unconstrained optimization problems. The gradient descent algorithm includes the formula:
Wherein, θ j' is an optimized parameter, θ j is a parameter before optimization, j is a label of the parameter, α is a preset learning rate, and L (θ) is a preset objective function; the values of theta j'、θj and alpha are real numbers, and the value of j is a positive integer.
S4: the expected electricity selling data of the year is input into a trained convolutional neural network, the expected load is determined through the convolutional neural network, and the electricity generating capacity of the year is determined based on the expected load.
And determining the expected electricity selling data of the year according to the income situation of the enterprise of the last year.
Inputting the predicted electricity selling data and the preset load data into a trained convolutional neural network, and outputting annual enterprise benefits through the convolutional neural network;
If the predicted result does not reach the preset profit threshold, adjusting the input preset historical load until the predicted result reaches the preset profit threshold;
And if the predicted result reaches the preset income limit value, taking the adjusted preset load data as the expected load, distributing the expected load and putting the expected load into power generation equipment to obtain the predicted power generation amount of the year, namely, connecting the expected power generation amount of the year into the power generation equipment according to the input daytime load density from 6 points in the daytime to 6 points at night, connecting the expected power generation amount of the year into the power generation equipment according to the input nighttime load density from 6 points at night to 6 points in the daytime at the day.
The electric power selling quantity expected to be achieved by the electric power enterprises in the current year and the number of users are input into the trained convolutional neural network, enterprise income which can be achieved by the electric power enterprises is predicted through the convolutional neural network, and therefore the electric power enterprises can reasonably plan electric power resources according to prediction results, and production and management plans in the current year can be adjusted in time.
Example two
As shown in fig. 3, the present invention further provides a device 5 for predicting the power generation capacity of an electric power enterprise based on a convolutional neural network, including:
training set unit 51: the power sensor is used for acquiring historical electricity selling data and historical load of the power enterprise, and a training set is established based on the acquired data. The method is particularly used for:
The historical electricity selling data and the historical load are obtained through the electric power sensors such as the electric energy collecting device deployed in the electric power enterprise, in the embodiment, the historical electricity selling data comprise the electricity selling quantity of the electric power enterprise in a preset unit time, the number of users of each type, and the historical load comprises daily load density and night load density. Types of users include industrial, commercial, agricultural production, and residential class 4 users. And calculating the electricity sales volume increase rate in unit time according to the electricity sales volume, calculating the user increase rate in unit time according to the number of users of each type, and adding the electricity sales volume increase rate and the user increase rate as training data into a training set. In this embodiment, the increase rate of the sales power of each month relative to the sales power of the previous month, that is, the sales power increase rate= (nth sales power/nth-1 month sales power) ×100% and the user increase rate= (nth month user number/nth-1 month user number) ×100% is calculated in the unit time of month, and the calculated sales power increase rate is added to the training set in the form of a row vector of 1×12, that is, training data of 12 dimensions.
Besides the electricity sales quantity and the electricity sales quantity increase rate, other indexes reflecting the electricity development in the training set are specifically as follows:
(1) Total amount of monthly electricity for enterprises: the power supply sum of the power enterprises to the industrial users and the commercial users in each month is represented, index data of the total amount of the power consumption of the enterprises in the year is obtained, and row vectors of 1 x 12, namely 12-dimensional training data are formed;
(2) Total amount of electricity used by residents in the month: the total electricity consumption of the current area of each month is represented, index data of the total electricity consumption of the current year resident month is obtained, and row vectors of 1 x 12 are formed, namely 12-dimensional training data;
(3) Daytime load density: representing the ratio of average load to area in the area between 6 points in daytime and 6 points at night, reflecting the daytime economic index in the area, and forming row vectors of 1 x 365 in units of days, namely 365-dimensional training data;
(4) Night load density: the ratio of average load to area in the area between 6 points at night and 6 points at day after the next day is represented, the night economic index in the area is reflected, and row vectors of 1 x 365 are formed by taking day as a unit, namely 365-dimensional training data.
Test set unit 52: for building a test set based on historical enterprise revenue data for the power enterprise. The method is particularly used for:
And the test set of the convolutional neural network comprises two parts of data of an electric consumption elastic coefficient and a power generation value. Firstly, historical enterprise revenue data of a power enterprise stored in advance is obtained, wherein the historical enterprise revenue data comprises the electricity fee growth rate and the electricity generation amount of the power enterprise in a preset unit time. The total domestic production value (Gross Domestic Product, GDP) and the total domestic production value increase rate within a preset unit time are obtained, and in this embodiment, the total domestic production value increase rate and the GDP are known in advance. And calculating the ratio of the electricity fee increase rate to the domestic total production increase rate within a preset time period to obtain the power consumption elastic coefficient of the power enterprise. In this embodiment, the ratio of the rate of increase of the electric charge for a certain year to the rate of increase of the total domestic production value for the same year is calculated in units of years, and the elastic coefficient of electric power consumption reflects the relationship between electric power consumption and national economy development. And calculating the ratio of the generated energy to the total domestic production value to obtain the electricity generation value of the power enterprise, wherein the electricity generation value reflects GDP generated by electricity per degree. In the present embodiment, the annual power consumption elastic coefficient and the yearly power generation value are added to the test set as test data in units of years.
The data in the training set reflects the condition of power development, and the data in the testing set reflects the condition of economic benefit created by the power enterprises.
Training unit 53: the training set is used for inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, and training is carried out on the convolutional neural network based on the training result and the test set. The method is particularly used for:
In this embodiment, the convolutional neural network includes four parts of an input layer, a convolutional layer, a pooling layer, and a classifier.
The 12-dimensional training data in the training set are spliced into a first matrix of 4 x 12, the 365-dimensional training data are spliced into a second matrix of 2 x 365, the convolutional neural network is input through an input layer, namely 4 types of training data, namely the electricity selling growth rate of 12 months in one year, the number of newly increased users in each month, the total amount of electricity consumption of enterprises and the total amount of electricity consumption of residents in one month, and the night economic index and day economic index 2 types of training data of 365 days in one year are respectively input into the convolutional neural network.
And extracting the characteristics of the input training data through a convolution layer of the convolution neural network, and outputting a characteristic diagram. In this embodiment, the size of the convolution kernel used by the first matrix composed of 12-dimensional vectors is set to 3*3, the number is 3, and the sliding step s is 1; the size of the convolution kernel used by the second matrix composed of 365-dimensional vectors is set to 1*5, the number is 1, and the sliding step s is 12. The length H 'and width w' of the output feature map can be calculated by the following formulas:
For the first matrix, H is the length of the input matrix, 4 in this embodiment; w is the width of the input matrix, in this embodiment 12; k 1 is the length of the convolution kernel, in this embodiment 3, and k 2 is the width of the convolution kernel, in this embodiment 3; p is a filling value, in this embodiment 0; s is the sliding step length, which is 1 in this embodiment. Therefore, the output feature map size is 2×10×3.
For the second matrix, H is the length of the input matrix, in this embodiment 2; w is the width of the input matrix, 365 in this embodiment; k 1 is the length of the convolution kernel, in this embodiment 1, and k 2 is the width of the convolution kernel, in this embodiment 5; p is a filling value, in this embodiment 0; s is the sliding step length, which is 12 in this embodiment. Therefore, the output feature map size is 2×31×1.
And carrying out dimension reduction processing on the feature map through a pooling layer of the convolutional neural network, and pooling the first matrix and the second matrix by using maximum values respectively to obtain two training matrices after dimension reduction processing. In this embodiment, the size of the pooling window of the first matrix is 2×2, the step size is 1, and the output size can be calculated to be 1×9×3 by using a similar manner to convolution; the pooling window size of the second matrix is 2×2, the step size is 1, and the output size is 1×30×1.
And finally, expanding the two training matrixes subjected to dimension reduction processing into vectors, then merging, inputting the merged matrixes into a preset classifier, and outputting a training result through the classifier. In this embodiment, the number of neurons in the input layer of the classifier is the sum of 1×9×3 of the first matrix pool and 1×30×1 of the second matrix pool, and the total number of neurons is 57, the hidden layer contains 8 neurons, and the output layer is one neuron. The classifier adjusts the dimension of the output result of the pooling layer through the output layer so that the dimension can be compared with the data in the test set.
The training unit 53 is further configured to:
Firstly, calculating the average value of the data in the test set, namely summing the power consumption elastic coefficient and the power output value and taking the average value.
Calculating an error of a training result of the convolutional neural network relative to the calculated average value, wherein the error is an error L between a training result f (x) of the convolutional neural network and the calculated average value y based on a formula I, and the training result f (x) is a result output by the classifier output layer:
L= (y-f (x)) 2 formula one;
wherein x is the data input into the convolutional neural network, and the value ranges of x, y and f (x) are positive numbers.
If the error L does not exceed the preset threshold, training of the convolutional neural network is finished, and the fact that a regression model between the historical electricity selling data, the number of the historical users and the profit of the historical enterprise of the power enterprise is established at the moment is explained, and the relationship between the electricity selling quantity of the current year, the number of the historical users and the profit of the enterprise can be predicted subsequently. If the error L exceeds a preset threshold, it is indicated that the training of the convolutional neural network at this time does not reach the expected effect, and the model parameters in the convolutional neural network need to be adjusted to optimize the convolutional neural network, where the optimizing process includes:
The gradient descent is one of the most commonly adopted methods when solving model parameters of a machine learning algorithm, namely, unconstrained optimization problems. The gradient descent algorithm includes the formula:
Wherein, θ j' is an optimized parameter, θ j is a parameter before optimization, j is a label of the parameter, α is a preset learning rate, and L (θ) is a preset objective function; the values of theta j'、θj and alpha are real numbers, and the value of j is a positive integer.
Prediction unit 54: the method is used for inputting expected electricity selling data of the year into the trained convolutional neural network, determining expected load through the convolutional neural network, and determining the electricity generation amount of the year based on the expected load. The method is particularly used for:
and determining the expected electricity selling data of the year according to the income situation of the enterprise of the last year.
Inputting the predicted electricity selling data and the preset load data into a trained convolutional neural network, and outputting annual enterprise benefits through the convolutional neural network;
If the predicted result does not reach the preset profit threshold, adjusting the input preset historical load until the predicted result reaches the preset profit threshold;
And if the predicted result reaches the preset income limit value, taking the adjusted preset load data as the expected load, distributing the expected load and putting the expected load into power generation equipment to obtain the predicted power generation amount of the year, namely, connecting the expected power generation amount of the year into the power generation equipment according to the input daytime load density from 6 points in the daytime to 6 points at night, connecting the expected power generation amount of the year into the power generation equipment according to the input nighttime load density from 6 points at night to 6 points in the daytime at the day.
The electric power selling quantity expected to be achieved by the electric power enterprises in the current year and the number of users are input into the trained convolutional neural network, enterprise income which can be achieved by the electric power enterprises is predicted through the convolutional neural network, and therefore the electric power enterprises can reasonably plan electric power resources according to prediction results, and production and management plans in the current year can be adjusted in time.
The various numbers in the above embodiments are for illustration only and do not represent the order of assembly or use of the various components.
The foregoing is illustrative of the present invention and is not to be construed as limiting thereof, but rather, the present invention is to be construed as limited to the appended claims.
Claims (8)
1. The method for predicting the power generation capacity of the power enterprise based on the convolutional neural network is characterized by comprising the following steps:
Acquiring historical electricity selling data and historical load of an electric power enterprise through an electric power sensor, establishing a training set based on the acquired data, and establishing a test set based on historical enterprise income data of the electric power enterprise;
inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, and training the convolutional neural network based on the training result and the test set;
inputting expected electricity selling data of the year into a trained convolutional neural network, determining expected load through the convolutional neural network, and determining the electricity generation capacity of the year based on the expected load;
The method for establishing the test set based on the historical enterprise revenue data of the power enterprise comprises the following steps:
Acquiring prestored historical enterprise revenue data of the power enterprise, wherein the historical enterprise revenue data comprises the electricity fee growth rate and the electricity generation amount of the power enterprise in a preset unit time;
Acquiring a domestic production total value and a domestic production total value increase rate within a preset unit time;
calculating the ratio of the electricity charge increase rate to the domestic total production increase rate to obtain the power consumption elastic coefficient of the power enterprise;
calculating the ratio of the generated energy to the total domestic production value to obtain the electricity generation value of the power enterprise;
Taking the power consumption elastic coefficient and the power output value as test data, and adding the test data into a test set;
The method for determining the annual generating capacity based on the expected load comprises the steps of:
Determining expected electricity selling data of the year according to the income situation of enterprises in the last year;
inputting the predicted electricity selling data and the preset load data into a trained convolutional neural network, and outputting annual enterprise benefits through the convolutional neural network;
if the predicted result does not reach the preset profit threshold, the input preset load data is adjusted until the predicted result reaches the preset profit threshold;
And if the predicted result reaches the preset income limit value, taking the adjusted preset load data as the expected load, and distributing the expected load to input power generation equipment according to the expected load to obtain the annual predicted power generation amount.
2. The method for predicting the power generation capacity of an electric power enterprise based on a convolutional neural network according to claim 1, wherein the acquiring the historical electricity selling data and the historical load of the electric power enterprise through the electric power sensor, and establishing the training set based on the acquired data comprises:
Acquiring historical electricity selling data and historical load of an electric power enterprise through an electric power sensor, wherein the historical electricity selling data comprises historical electricity selling quantity of the electric power enterprise in preset unit time, and the historical load comprises daily load density and night load density;
Calculating the electricity sales quantity increase rate in unit time according to the historical electricity sales quantity;
and taking the electricity sales volume increase rate, the historical electricity sales data and the historical load as training data, and adding the training data into a training set.
3. The method for predicting the power generation capacity of an electric power enterprise based on the convolutional neural network according to claim 1, wherein the step of inputting the training set into a preset convolutional neural network to obtain the training result of the convolutional neural network comprises the following steps:
inputting training data in a training set into a convolutional neural network;
Carrying out feature extraction on input training data through a convolution layer of a convolution neural network, and outputting a feature map;
performing dimension reduction processing on the feature map through a pooling layer of the convolutional neural network to obtain a training matrix after the dimension reduction processing;
inputting the training matrix into a preset classifier, and outputting a training result through the classifier.
4. The method for predicting the power generation capacity of an electric power enterprise based on a convolutional neural network according to claim 1, wherein training the convolutional neural network based on the training result and the test set comprises:
Calculating an average value of data in the test set;
Calculating the error of the training result of the convolutional neural network relative to the calculated average value, and ending the training of the convolutional neural network if the error does not exceed a preset threshold value.
5. The method for predicting the power generation capacity of an electric power enterprise based on the convolutional neural network according to claim 3, wherein calculating the error of the training result of the convolutional neural network relative to the calculated average value comprises:
calculating an error L between a training result f (x) of the convolutional neural network and a calculated average value y based on a formula I, wherein the formula I is as follows:
L= (y-f (x)) 2 formula one;
wherein x is the data input into the convolutional neural network, and the value ranges of x, y and f (x) are positive numbers.
6. The method for predicting power generation capacity of an electrical enterprise based on a convolutional neural network of claim 1, wherein the training the convolutional neural network further comprises optimizing the convolutional neural network, the optimizing comprising: adjusting various parameters in the convolutional neural network based on a gradient descent algorithm comprising the formula:
Wherein, θ j' is an optimized parameter, θ j is a parameter before optimization, j is a label of the parameter, α is a preset learning rate, and L (θ j) is an objective function of a preset parameter θ j; the values of theta j′、θj and alpha are real numbers, and the value of j is a positive integer.
7. The prediction device for generating capacity of an electric power enterprise based on a convolutional neural network is suitable for the prediction method as claimed in claim 1, and is characterized in that the prediction device comprises:
Training set unit: the power sensor is used for acquiring historical electricity selling data and historical load of an electric power enterprise, and a training set is established based on the acquired data;
Test set unit: the method comprises the steps of establishing a test set based on historical enterprise revenue data of a power enterprise;
Training unit: the training set is used for inputting the training set into a preset convolutional neural network to obtain a training result of the convolutional neural network, and training is carried out on the convolutional neural network based on the training result and the test set;
Prediction unit: the method is used for inputting expected electricity selling data of the year into the trained convolutional neural network, determining expected load through the convolutional neural network, and determining the electricity generation amount of the year based on the expected load.
8. The prediction device for generating capacity of an electric power enterprise based on a convolutional neural network according to claim 1, wherein the prediction unit is configured to:
Determining expected electricity selling data of the year according to the income situation of enterprises in the last year;
inputting the predicted electricity selling data and the preset load data into a trained convolutional neural network, and outputting annual enterprise benefits through the convolutional neural network;
if the predicted result does not reach the preset profit threshold, the input preset load data is adjusted until the predicted result reaches the preset profit threshold;
And if the predicted result reaches the preset income limit value, taking the adjusted preset load data as the expected load, and distributing the expected load to input power generation equipment according to the expected load to obtain the annual predicted power generation amount.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011363422.7A CN112508734B (en) | 2020-11-27 | 2020-11-27 | Method and device for predicting power generation capacity of power enterprise based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011363422.7A CN112508734B (en) | 2020-11-27 | 2020-11-27 | Method and device for predicting power generation capacity of power enterprise based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112508734A CN112508734A (en) | 2021-03-16 |
| CN112508734B true CN112508734B (en) | 2024-04-26 |
Family
ID=74967211
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011363422.7A Active CN112508734B (en) | 2020-11-27 | 2020-11-27 | Method and device for predicting power generation capacity of power enterprise based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112508734B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113822561A (en) * | 2021-09-14 | 2021-12-21 | 国网江苏省电力有限公司营销服务中心 | Resource configuration method based on power consumption data and radial basis function neural network |
| CN115169991B (en) * | 2022-09-02 | 2023-01-24 | 南京华盾电力信息安全测评有限公司 | Multi-type user energy supply and energy supply intelligent analysis system based on data mining |
| CN118246705B (en) * | 2024-05-28 | 2024-07-30 | 国网浙江省电力有限公司杭州供电公司 | Energy demand prediction method, system, equipment and medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107578124A (en) * | 2017-08-28 | 2018-01-12 | 国网山东省电力公司电力科学研究院 | The Short-Term Load Forecasting Method of GRU neutral nets is improved based on multilayer |
| CN108090563A (en) * | 2017-12-15 | 2018-05-29 | 烟台港股份有限公司 | A kind of electric flux Forecasting Methodology based on BP neural network |
| CN109543901A (en) * | 2018-11-20 | 2019-03-29 | 国网辽宁省电力有限公司经济技术研究院 | Short-Term Load Forecasting Method based on information fusion convolutional neural networks model |
| CN110651273A (en) * | 2017-11-17 | 2020-01-03 | 华为技术有限公司 | Data processing method and equipment |
| CN111080477A (en) * | 2019-12-06 | 2020-04-28 | 北京国电通网络技术有限公司 | Household power load prediction method and system |
| CN111695731A (en) * | 2020-06-09 | 2020-09-22 | 中国电力科学研究院有限公司 | Load prediction method, system and equipment based on multi-source data and hybrid neural network |
| CN111985701A (en) * | 2020-07-31 | 2020-11-24 | 国网上海市电力公司 | Power utilization prediction method based on power supply enterprise big data model base |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040215551A1 (en) * | 2001-11-28 | 2004-10-28 | Eder Jeff S. | Value and risk management system for multi-enterprise organization |
-
2020
- 2020-11-27 CN CN202011363422.7A patent/CN112508734B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107578124A (en) * | 2017-08-28 | 2018-01-12 | 国网山东省电力公司电力科学研究院 | The Short-Term Load Forecasting Method of GRU neutral nets is improved based on multilayer |
| CN110651273A (en) * | 2017-11-17 | 2020-01-03 | 华为技术有限公司 | Data processing method and equipment |
| CN108090563A (en) * | 2017-12-15 | 2018-05-29 | 烟台港股份有限公司 | A kind of electric flux Forecasting Methodology based on BP neural network |
| CN109543901A (en) * | 2018-11-20 | 2019-03-29 | 国网辽宁省电力有限公司经济技术研究院 | Short-Term Load Forecasting Method based on information fusion convolutional neural networks model |
| CN111080477A (en) * | 2019-12-06 | 2020-04-28 | 北京国电通网络技术有限公司 | Household power load prediction method and system |
| CN111695731A (en) * | 2020-06-09 | 2020-09-22 | 中国电力科学研究院有限公司 | Load prediction method, system and equipment based on multi-source data and hybrid neural network |
| CN111985701A (en) * | 2020-07-31 | 2020-11-24 | 国网上海市电力公司 | Power utilization prediction method based on power supply enterprise big data model base |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112508734A (en) | 2021-03-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112508734B (en) | Method and device for predicting power generation capacity of power enterprise based on convolutional neural network | |
| Agarwal et al. | Runoff modelling through back propagation artificial neural network with variable rainfall-runoff data | |
| CN107506868B (en) | Method and device for predicting short-time power load | |
| CN112381673B (en) | Park electricity utilization information analysis method and device based on digital twin | |
| CN110212524A (en) | A kind of region Methods of electric load forecasting | |
| CN110334879A (en) | Power grid bus reactive load forecasting method and device | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| CN108491982A (en) | A kind of short-term load forecasting method and system based on echo state network | |
| CN108596242A (en) | Power grid meteorology load forecasting method based on wavelet neural network and support vector machines | |
| CN114065653A (en) | Construction method of power load prediction model and power load prediction method | |
| CN111178585A (en) | Fault reporting amount prediction method based on multi-algorithm model fusion | |
| CN109034500A (en) | A kind of mid-term electric load forecasting method of multiple timings collaboration | |
| CN112288140A (en) | Keras-based short-term power load prediction method, storage medium and equipment | |
| CN112803398A (en) | Load prediction method and system based on empirical mode decomposition and deep neural network | |
| CN112613637A (en) | Method and device for processing charging load | |
| CN113722997A (en) | New well dynamic yield prediction method based on static oil and gas field data | |
| CN115660725A (en) | Method for depicting multi-dimensional energy user portrait | |
| CN114330934A (en) | Model parameter self-adaptive GRU new energy short-term power generation power prediction method | |
| CN116128116A (en) | Intelligent load prediction method for distribution network | |
| CN116562454A (en) | Manufacturing cost prediction method applied to BIM long-short-time attention mechanism network | |
| CN115238854A (en) | Short-term load prediction method based on TCN-LSTM-AM | |
| Dai et al. | Multimodal deep learning water level forecasting model for multiscale drought alert in Feiyun River basin | |
| CN104751253B (en) | Distribution power flow Forecasting Methodology based on B- spline Basis bottom developed curve cluster | |
| CN114091776A (en) | K-means-based multi-branch AGCNN short-term power load prediction method | |
| CN113361776A (en) | Power load probability prediction method based on user power consumption behavior clustering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |