Real-time electric power carbon emission factor measuring and calculating method for thermal power generating unit

Abstract

The invention discloses a thermal power generating unit real-time electric power carbon emission factor measuring and calculating method, and belongs to the technical field of electric power carbon emission monitoring. The method comprises the following steps: 1, data processing: collecting historical operation data of a thermal power generating unit; performing data cleaning and feature engineering processing on the collected historical operation data; step 2, model establishment: constructing a Transform model, and training the Transform model by using the processed historical operation data; 3, real-time measurement and calculation are conducted, specifically, according to the real-time fuel consumption output by the Transform model, the real-time total carbon emission of the thermal power generating unit is calculated in combination with the fuel carbon emission coefficient; the heat supply condition of the thermal power generating unit is considered, real-time total carbon emission is shared according to the proportion of power generation and heat supply, and a dynamic electric power carbon emission factor is obtained through calculation. The power generation carbon emission factor of the thermal power generating unit can be measured in real time, and support is provided for real-time scheduling and carbon emission management of a power system. Through combination of historical data and real-time operation data, the accuracy of carbon emission factor measurement and calculation is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power carbon emission monitoring, and more specifically, to a method for calculating a real-time power carbon emission factor of a thermal power unit. Background Art

[0002] With global climate change and environmental issues becoming increasingly severe, the monitoring and management of carbon emissions from power systems has become particularly important. As the primary source of carbon emissions in the power system, the accurate measurement of carbon emission factors for thermal power units is crucial for developing emission reduction strategies and optimizing power dispatch. However, traditional carbon emission factor calculation methods often rely on static data and empirical models. Furthermore, the actual data granularity of thermal power plants is limited, typically only providing daily fuel data and hourly power generation data. This makes it impossible to measure carbon emissions from thermal power units under different operating conditions in real time, resulting in significant errors in the measurement results. Furthermore, the calculation of power carbon emission factors is subject to lags.

[0003] In view of this, we propose a method for calculating the real-time carbon emission factor of thermal power units. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for calculating the real-time carbon emission factor of electricity of a thermal power unit, so as to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A method for calculating the real-time carbon emission factor of thermal power generation units comprises the following steps: Step 1: Data processing: Collect historical operating data of thermal power units, including daily power generation, daily fuel consumption, hourly power generation, and heat supply ratio; perform data cleaning and feature engineering on the collected historical operating data; Step 2: Model Building: Build a Transformer model and train it using processed historical operating data. The Transformer model is used to calculate the fuel consumption of the thermal power unit based on the unit's real-time operating power. The output of the Transformer model is then verified against actual gas data. Step 3: Real-time measurement: Based on the real-time fuel consumption output by the Transformer model and the fuel carbon emission coefficient, the real-time total carbon emissions of the thermal power unit are calculated; considering the heating situation of the thermal power unit, the real-time total carbon emissions are allocated according to the ratio of power generation and heating, thereby calculating the dynamic electricity carbon emission factor.

[0006] Preferably, step 1 comprises: Step 1.1: Data cleaning: clean the collected historical operating data of thermal power units to remove outliers, noise and interference, and normalize or standardize the historical operating data; Step 1.2: Feature engineering is used to extract the temporal features of the historical operating data of the thermal power units, organize them into a form that can be understood by the Transformer model, and input them into the Transformer model.

[0007] 3. A method for calculating the real-time carbon emission factor of thermal power generation units according to claim 1, characterized in that step 2 comprises: Step 2.1, position code calculation: The Transformer model needs to use position encoding to capture the order of elements in the sequence. The position encoding is calculated using the following formula: in, is the position index, is the dimension index, is the dimension of the model; Step 2.2, Transformer model construction: The Transformer model includes an input layer, an encoder, a decoder, and an output layer; The input layer takes the processed daily power generation, daily fuel consumption, hourly power generation data and time features as input and uses the embedding layer to convert the input data into a fixed-dimensional vector representation; The encoder is composed of multiple encoder layers. Each encoder layer contains a self-attention mechanism and a feed-forward network. The calculation formula of the self-attention mechanism is as follows: Where, , They are query, key and value. is the dimension of the key; The decoder consists of multiple decoder layers, each of which includes a self-attention mechanism, an encoder-decoder attention mechanism, and a feedforward neural network. The decoder's self-attention mechanism takes into account masks to prevent future information leakage: The mask is an upper triangular matrix that ensures that each element in the sequence can only see the position before it. The output layer uses a fully connected layer to convert the decoder output into a calculated value of hourly gas consumption; Step 2.3, Model training: Use daily and hourly electricity data and daily gas consumption data to train the Transformer model; Step 2.4, parameter update: use Adam optimizer to update parameters: in, is the current parameter, is the learning rate.

[0008] Preferably, in step 2.3, the mean square error (MSE) is used as the loss function: in, is the actual daily gas consumption, is the hourly gas consumption calculated by the model, is the daily gas consumption accumulated by calculation data, is the sample size.

[0009] Preferably, step 3 includes: Step 3.1: Calculate the gas consumption in real time based on the Transformer model and calculate the total carbon emissions of the thermal power unit in real time ; in, is the carbon emission coefficient of the fuel; Step 3.2: Based on the heat supply ratio of thermal power units , calculate the carbon emissions generated by power generation according to the principle of proportional allocation ; ; Step 3.3: Calculate the dynamic carbon emission factor of electricity generated by thermal power units : .

[0010] Compared with the prior art, the present invention has the following beneficial effects: This method can calculate the carbon emission factor of thermal power generation units in real time, supporting real-time scheduling and carbon emission management of power systems. By combining historical data with real-time operational data, the accuracy of carbon emission factor calculation is improved. The Transformer model considers parameters such as the heat supply ratio and can adapt to different operating conditions. This method is applicable to all types of thermal power units and has strong universality and promotional value. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 This is a flow chart for calculating the dynamic power carbon emission factor of a thermal power unit according to the present invention. DETAILED DESCRIPTION

[0012] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0013] Example: See also Figure 1 A method for calculating the real-time carbon emission factor of thermal power generation units comprises the following steps: Step 1: Data processing: Collect historical operating data of thermal power units, including daily power generation, daily fuel consumption, hourly power generation, and heat supply ratio; perform data cleaning and feature engineering on the collected historical operating data.

[0014] In this step, historical operating data is collected in real time through the DCS (distributed control system) of the thermal power unit, including: daily data: cumulative power generation from 0:00 to 24:00 daily (unit: MWh), fuel consumption (unit: tons / cubic meter, based on fuel type); hourly data: instantaneous power generation per hour (unit: MW), real-time operating power (as a percentage of rated power); heating ratio: the ratio of daily heating energy to total fuel energy (calculated through the heat supply sensor and the unit heat balance).

[0015] Specifically, step 1 includes: Step 1.1: Data cleaning: clean the collected historical operating data of thermal power units to remove outliers, noise, and interference, and normalize or standardize the historical operating data to improve the training effect of the model; outliers, noise, and interference are determined by setting thresholds and statistical analysis.

[0016] Step 1.2: Feature engineering is used to extract the time features (such as hours, days of the week, and months) of the historical operating data of the thermal power units. These features are organized into a form that can be understood by the Transformer model and then input into the Transformer model.

[0017] Step 2: Model establishment: Construct a Transformer model and train it using processed historical operating data. The Transformer model is used to calculate the fuel consumption of the thermal power unit based on the real-time operating power of the thermal power unit. The output of the Transformer model is then verified against actual gas data.

[0018] Specifically, step 2 includes: Step 2.1, position code calculation: The Transformer model needs to use position encoding to capture the order of elements in the sequence. The position encoding is calculated using the following formula: in, is a position index, indicating the position of the element in the sequence; It is a dimension index used to represent the dimension of the vector. The sine function is used for even dimensions and the cosine function is used for odd dimensions to avoid periodic overlap of position encoding. is the dimension of the model, that is, the dimension of the vector in the model.

[0019] Step 2.2, Transformer model construction: The Transformer model consists of an input layer, an encoder, a decoder, and an output layer. The encoder processes the input sequence, i.e., daily power generation and daily fuel consumption, and the decoder generates the output sequence, i.e., hourly gas consumption. The input layer takes the processed daily power generation, daily fuel consumption, hourly power generation data and time features as input, and uses the embedding layer to convert the input data into a fixed-dimensional vector representation; the input data includes: daily power generation / fuel consumption (converted to a 24-hour mean vector), hourly power generation (real-time value), and time features (One-Hot encoding); the embedding layer maps data of different scales into fixed-dimensional vectors (such as d model=512) and superimposed position encoding.

[0020] The encoder is composed of multiple encoder layers. Each encoder layer contains a self-attention mechanism and a feed-forward network. The calculation formula of the self-attention mechanism is as follows: Where, , They are query, key and value. is the dimension of the key, used to scale the attention score; specifically Q is the query matrix, which is used to represent the query information of the current element; K It is the key matrix, which is used to represent the key information of other elements in the sequence; V It is a value matrix, which is used to represent the specific information of other elements in the sequence.

[0021] The decoder consists of multiple decoder layers, each of which includes a self-attention mechanism, an encoder-decoder attention mechanism, and a feedforward neural network. The decoder's self-attention mechanism takes into account masks to prevent future information leakage: The mask is an upper triangular matrix that ensures that each element in the sequence can only see its previous position; a causal mask (upper triangular matrix) is used to ensure that the decoding t Hourly fuel consumption, only using the front t −1 hour prediction results to avoid future information leakage; encoder-decoder attention mechanism: combines the encoder output with the current decoder input to generate hourly fuel consumption predictions.

[0022] The output layer uses a fully connected layer to convert the decoder output into a calculated value of hourly gas consumption.

[0023] Step 2.3, Model training: Use daily and hourly electricity data and daily gas consumption data to train the Transformer model; The training objective of the model is to minimize the difference between the predicted gas consumption and the actual gas consumption, using the mean square error (MSE) as the loss function: in, is the actual daily gas consumption, is the hourly gas consumption calculated by the model, is the daily gas consumption accumulated by calculation data, The loss function is based on the following logic: The hourly fuel consumption output by the model over 24 hours is accumulated and compared with the actual daily fuel consumption to calculate the mean square error (MSE). This forces the model to learn the daily total consumption constraint and reduce the cumulative error of the hourly prediction.

[0024] Step 2.4, parameter update: use Adam optimizer to update parameters: in, is the current parameter, i.e. the model is t The parameter value at the iteration; is the learning rate, which is used to control the step size of parameter updates.

[0025] Step 3: Real-time measurement: Based on the real-time fuel consumption output by the Transformer model and the fuel carbon emission coefficient, the real-time total carbon emissions of the thermal power unit are calculated; considering the heating situation of the thermal power unit, the real-time total carbon emissions are allocated according to the ratio of power generation and heating, thereby calculating the dynamic electricity carbon emission factor.

[0026] Specifically, step 3 includes: Step 3.1: Calculate the gas consumption in real time based on the Transformer model and calculate the total carbon emissions of the thermal power unit in real time ; in, is the carbon emission coefficient of the fuel; Step 3.2: Based on the heat supply ratio of thermal power units , calculate the carbon emissions generated by power generation according to the principle of proportional allocation ; ; Step 3.3: Calculate the dynamic carbon emission factor of electricity generated by thermal power units : .

[0027] The method for calculating the carbon emission factor of thermal power units in this application has many advantages: in the time dimension, it can calculate the carbon emission factor of power generation in real time, assisting in real-time scheduling and carbon emission management of the power system; in terms of precision, it integrates historical and real-time operating data to improve the accuracy of calculation; in terms of adaptability, it can adapt to different operating conditions by considering parameters such as the heat supply ratio; in terms of scope of application, it is applicable to all types of thermal power units and has strong universality and promotion value.

[0028] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for calculating the real-time carbon emission factor of thermal power generation units, characterized in that: The steps include: Step 1: Data processing: Collect historical operating data of thermal power units, including daily power generation, daily fuel consumption, hourly power generation, and heat supply ratio; perform data cleaning and feature engineering on the collected historical operating data; Step 2: Model establishment: Construct a Transformer model and train it using processed historical operating data. The Transformer model is used to calculate the fuel consumption of the thermal power unit based on the real-time operating power of the thermal power unit. The output of the Transformer model is then verified against actual gas data. Step 3, real-time measurement: Calculate the real-time total carbon emissions of the thermal power unit based on the real-time fuel consumption output by the Transformer model and the fuel carbon emission coefficient; consider the heating situation of the thermal power unit and allocate the real-time total carbon emissions according to the ratio of power generation and heating to calculate the dynamic electricity carbon emission factor.

2. The method for calculating the real-time carbon emission factor of thermal power generation units according to claim 1, characterized in that: The step 1 comprises: Step 1.1: Data cleaning: clean the collected historical operating data of thermal power units to remove outliers, noise and interference, and normalize or standardize the historical operating data; Step 1.2: Feature engineering is used to extract the temporal features of the historical operating data of the thermal power units, organize them into a form that can be understood by the Transformer model, and input them into the Transformer model.

3. The method for calculating the real-time carbon emission factor of thermal power generation units according to claim 1, characterized in that: The step 2 includes: Step 2.1, position code calculation: The Transformer model needs to use position encoding to capture the order of elements in the sequence. The position encoding is calculated using the following formula: in, is the position index, is the dimension index, is the dimension of the model; Step 2.2, Transformer model construction: The Transformer model includes an input layer, an encoder, a decoder, and an output layer; The input layer takes the processed daily power generation, daily fuel consumption, hourly power generation data and time features as input and uses the embedding layer to convert the input data into a fixed-dimensional vector representation; The encoder is composed of multiple encoder layers. Each encoder layer contains a self-attention mechanism and a feed-forward network. The calculation formula of the self-attention mechanism is as follows: Where, , They are query, key and value. is the dimension of the key; The decoder consists of multiple decoder layers, each of which includes a self-attention mechanism, an encoder-decoder attention mechanism, and a feedforward neural network. The decoder's self-attention mechanism takes into account masks to prevent future information leakage: The mask is an upper triangular matrix that ensures that each element in the sequence can only see the position before it. The output layer uses a fully connected layer to convert the decoder output into a calculated value of hourly gas consumption; Step 2.3, Model training: Use daily and hourly electricity data and daily gas consumption data to train the Transformer model; Step 2.4, parameter update: use Adam optimizer to update parameters: in, is the current parameter, is the learning rate.

4. The method for calculating the real-time carbon emission factor of thermal power generation units according to claim 3, characterized in that: In step 2.3, the mean square error (MSE) is used as the loss function: in, is the actual daily gas consumption, is the hourly gas consumption calculated by the model, is the daily gas consumption accumulated by calculation data, is the sample size.

5. The method for calculating the real-time carbon emission factor of thermal power generation units according to claim 1, characterized in that: Step 3 includes: Step 3.1: Calculate the gas consumption in real time based on the Transformer model and calculate the total carbon emissions of the thermal power unit in real time ; in, is the carbon emission coefficient of the fuel; Step 3.2: Based on the heat supply ratio of thermal power units , calculate the carbon emissions generated by power generation according to the principle of proportional allocation ; ； Step 3.3: Calculate the dynamic carbon emission factor of electricity generated by thermal power units : 。

Images