CN-116094068-B - Power grid dispatching method, equipment and medium based on carbon emission prediction mechanism

CN116094068BCN 116094068 BCN116094068 BCN 116094068BCN-116094068-B

Abstract

The invention discloses a power grid dispatching method, equipment and medium based on a carbon emission prediction mechanism, which are used for acquiring carbon emission data related to a power plant, preprocessing the carbon emission data related to the power plant, taking the preprocessed data as sample data, dividing the sample data into a training set and a testing set, training an improved BP neural network by using the training set and the testing set to obtain the trained improved BP neural network, predicting the future carbon emission of the power plant by using the trained improved BP neural network, inputting the future carbon emission of the power plant into an objective function taking the carbon emission as the minimum optimization target, and outputting a power plant start-stop and output plan. According to the power grid dispatching method, equipment and medium based on the carbon emission prediction mechanism, which are provided by the invention, the future electricity carbon emission capacity of the thermal power plant unit is accurately mastered by improving the BP neural network, so that the low-carbon dispatching of a large power grid under a double-carbon background is realized, and data support and auxiliary decision making are provided for low-carbon operation work.

Inventors

ZHANG YINGLING
CHEN ZHENGPING
XU ZHENGQI
LI WENZHONG
HAN YE
WANG FANGDONG

Assignees

国网福建省电力有限公司

Dates

Publication Date: 20260505
Application Date: 20230321

Claims (3)

1. A power grid dispatching method based on a carbon emission prediction mechanism is characterized by comprising the following steps: step 1, acquiring carbon emission data related to a power plant; Step 2, preprocessing carbon emission data related to a power plant, and taking the preprocessed data as sample data; step 3, dividing sample data into a training set and a testing set; Training the improved BP neural network by using a training set and a testing set to obtain a trained improved BP neural network; Step 5, predicting the future carbon emission of the power plant by using the trained improved BP neural network; Step 6, inputting the future carbon emission amount of the power plant into an objective function taking the minimum carbon emission as an optimization target, and outputting a power plant start-stop and output plan; The plant-related carbon emission data includes, but is not limited to, a power plant unit, a raw coal conversion factor, a data month, a fossil combustion carbon emission amount, a coal consumption amount, a desulfurizing agent carbonate content, a limestone consumption amount, a desulfurizing machine emission factor, an electricity consumption rate, a desulfurizing process carbon emission amount, elemental carbon, a unit maintenance state, a volatile matter, a temperature, a low calorific value, an outsourcing electricity amount, a unit calorific value carbon amount, an outsourcing electricity carbon emission factor, a carbon oxidation rate, an outsourcing electricity carbon emission amount, an overall carbon emission amount, and an electricity carbon emission capability; the preprocessing method comprises the steps of checking and correcting error data, and replacing historical average value of empty data; The improved BP neural network comprises MobileNetV parts of BP network, a first convolution layer, MobileNetV 2a second convolution layer, an average pooling layer and a BP neural network are sequentially connected; the MobileNetV network structure is characterized in that the main body of the MobileNetV network structure consists of 17 anti-residual units; the BP neural network structure comprises a hierarchical neural network consisting of an input layer, an hidden layer and an output layer, wherein the hidden layer can be expanded into a plurality of adjacent layers, all neurons of the adjacent layers are connected, all neurons of each layer are not connected, the BP neural network learns in a teacher teaching mode, after a pair of learning modes are provided for the BP neural network, all neurons acquire input response of the BP neural network to generate a connection weight, and then all connection weights are corrected layer by layer from the output layer through the hidden layer according to the direction of reducing the expected output and actual output errors and returned to the input layer; The training of the improved BP neural network by using the training set and the testing set to obtain the trained improved BP neural network comprises the following steps: inputting the training set into an improved BP neural network for training until the improved BP neural network converges, and obtaining an initial improved BP neural network model; inputting the test set into an initial improved BP neural network model for testing, and obtaining a trained improved BP neural network model when an error function E corresponding to the test set reaches a threshold value; The input of the future carbon emission amount of the power plant takes the minimum carbon emission as an objective function of an optimization target, and the output of the start-stop and output plan of the power plant comprises the following steps: obtaining an objective function, wherein the calculation formula of the objective function is as follows: Wherein, delta co2 is an objective function, Is the corresponding carbon emission amount when the unit works normally, In order to correspond to the carbon emission when the unit is started, The method is characterized in that the method comprises the steps of (1) corresponding carbon emission when a unit is stopped, M is the number of power plant units, and T is the number of scheduling time periods; the predicted value lambda co2 of the future carbon emission of the power plant is obtained, and the calculation formulas of the constraint conditions st. and st. are constructed as follows: And under the precondition that constraint conditions st. are met, solving an objective function by taking the minimum carbon emission as an optimization target to obtain a power plant start-stop and output plan.
2. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor, implements a carbon emission prediction mechanism based power grid scheduling method as claimed in claim 1.
3. A computer apparatus, comprising: a memory for storing instructions; a processor for executing the instructions to cause the computer device to perform the operations of a grid dispatching method based on a carbon emission prediction mechanism as claimed in claim 1.

Description

Power grid dispatching method, equipment and medium based on carbon emission prediction mechanism Technical Field The invention relates to a power grid dispatching method, equipment and medium based on a carbon emission prediction mechanism, and belongs to the technical field of power grid dispatching automation. Background Carbon emissions reduction becomes a global hot spot problem. Therefore, the trading rules of the power grid dispatching main bodies are influenced, and in order to ensure that the power grid dispatching is safe and stable to operate, the carbon emission of each power grid dispatching main body needs to be accurately predicted so as to realize the accurate dispatching plan of the power grid dispatching main body based on different carbon emission. At present, most of carbon emission prediction technologies adopt linear or nonlinear theory prediction technologies, regional carbon emission factors issued by related departments are considered, the range is wider, the carbon emission prediction technologies are applicable to carbon emission data with wider range, the carbon emission data can only be used as trend prediction, and the prediction result error is larger. Furthermore, since no history data of a single sample is introduced, the uniqueness of the sample data is not considered. Therefore, it is necessary to study an accurate carbon emission prediction method and provide effective and highly reliable data support for low-carbon dispatching, so as to realize a more accurate power grid dispatching plan. Disclosure of Invention The invention aims to overcome the defects in the prior art and provides a power grid dispatching method, equipment and medium based on a carbon emission prediction mechanism, which provide data support and auxiliary decision for low-carbon dispatching. The technical scheme adopted by the invention is as follows: in a first aspect, a power grid dispatching method based on a carbon emission prediction mechanism includes the steps of: And step 1, acquiring carbon emission data related to the power plant. And 2, preprocessing the carbon emission data related to the power plant, and taking the preprocessed data as sample data. And 3, dividing the sample data into a training set and a testing set. And 4, training the improved BP neural network by using the training set and the testing set to obtain the trained improved BP neural network. And 5, predicting the future carbon emission of the power plant by using the trained improved BP neural network. And 6, inputting the future carbon emission amount of the power plant into an objective function taking the minimum carbon emission as an optimization target, and outputting a power plant start-stop and output plan. Further, the plant-related carbon emission data includes, but is not limited to, a power plant unit, a raw coal conversion factor, a data month, a fossil combustion carbon emission amount, a coal consumption amount, a desulfurizing agent carbonate content, a limestone consumption amount, a desulfurizing machine emission factor, an electricity consumption rate, a desulfurizing process carbon emission amount, elemental carbon, a unit maintenance state, a volatile matter, a temperature, a low calorific value, an outsourcing electricity amount, a unit calorific value carbon content, an outsourcing electricity carbon emission factor, a carbon oxidation rate, an outsourcing electricity carbon emission amount, an overall carbon emission amount, and an electricity carbon emission capability. Further, the preprocessing method comprises the steps of checking and correcting error data and replacing historical means of empty data. Further, the improved BP neural network comprises MobileNetV < 2 > and a BP network. The first convolution layer, mobileNetV, the second convolution layer, the averaging pooling layer and the BP neural network are sequentially connected. Further, the MobileNetV network structure, the main body is composed of 17 anti-residual units. Further, the BP neural network structure is a hierarchical neural network consisting of an input layer, an hidden layer and an output layer. The hidden layer may be extended to multiple adjacent layers. All the neurons of adjacent layers are connected, and all the neurons of each layer are not connected. Furthermore, the BP neural network learns in a teacher teaching mode, and when a pair of learning modes are provided for the BP neural network, each neuron obtains the input response of the BP neural network to generate a connection Weight (Weight). And then correcting each connection weight layer by layer from the output layer through the hidden layer according to the direction of reducing the error between the expected output and the actual output, and returning to the input layer. The process is repeatedly and alternately performed until the global error of the network tends to a given minimum value, namely, the learning process is completed, and all the connection w