Unit load advanced control method based on battery energy storage state

Abstract

The present invention relates to the technical field of automatic control for thermal power energy storage, and relates to a unit load advanced control method based on a battery energy storage state, comprising: determining the state of a thermal power unit on the basis of an AGC command and current load of the thermal power unit; determining whether an energy storage output of an energy storage system meets the AGC command; if not, establishing an initial thermal power unit heat storage coefficient and a prediction thermal power unit heat storage data set on the basis of the state of the thermal power unit; constructing a thermal power unit heat storage coefficient prediction model; and inputting operation data of the thermal power unit into a trained thermal power unit heat storage coefficient prediction model, generating a predicted thermal power unit heat storage coefficient, constructing a correspondence between the predicted thermal power unit heat storage coefficient and a load command advance amount, and generating a load command and executing same by the thermal power unit. The present invention solves the problem in the prior art of excessively high unit load caused by allocating frequency regulation tasks to energy storage systems and conventional thermal power units according to a fixed proportional coefficient.

Description

Battery-based load advance control method Technical Field

[0001] This invention belongs to the field of thermal power energy storage automation control technology, specifically relating to a method for unit load advance control based on battery energy storage status. Background Technology

[0002] With the expansion of wind and solar power grid connection, the role of thermal power units in the power grid is gradually shifting towards providing more flexible and efficient ancillary services. However, when participating in grid frequency regulation, thermal power units may suffer severe wear and tear due to equipment characteristics, leading to a decrease in short-term power throughput capacity, affecting load regulation potential, and consequently prolonging system response time and reducing frequency regulation capability, making it difficult to achieve the expected frequency regulation goals. The development of battery energy storage technology provides important technical support for new energy power systems. Its rapid response characteristics can effectively reduce the frequency regulation burden of conventional thermal power units and improve the quality of frequency regulation services.

[0003] Currently, when receiving Automatic Generation Control (AGC) commands, generating units equipped with battery energy storage typically allocate frequency regulation tasks to the energy storage system and conventional thermal power units according to a fixed proportional coefficient. However, this method has some limitations: for thermal power units, it cannot flexibly determine whether their output can meet the load allocation commands based on the actual operating conditions of the unit, nor can it fully utilize the unit's heat storage capacity, thus failing to fully realize its frequency regulation potential. For battery energy storage systems, their response capability is limited when overcharging or over-discharging occurs, making it unable to effectively execute the allocated commands. Furthermore, the number of charge-discharge cycles of the energy storage system varies with fluctuations in the frequency regulation signal, which may adversely affect the lifespan of the energy storage system. Summary of the Invention

[0004] The purpose of this invention is to provide a method for advance control of unit load based on battery energy storage status, which solves the problem of excessive unit load caused by allocating frequency regulation tasks to energy storage systems and conventional thermal power units according to a fixed proportional coefficient in the prior art.

[0005] The technical solution adopted in this invention is a method for advance control of unit load based on battery energy storage status, which is implemented according to the following steps:

[0006] Step 1: Determine the status of the thermal power unit based on the AGC command and the current load of the thermal power unit;

[0007] Step 2: Determine whether the energy storage output of the energy storage system meets the AGC command. If the energy storage output does not meet the AGC command, proceed to steps 3 to 6; if the energy storage output meets the AGC command, do not proceed.

[0008] Step 3: Obtain the initial thermal power unit heat storage coefficient based on the thermal power unit status, and then construct a predicted thermal power unit heat storage dataset.

[0009] Step 4: Construct a prediction model for the thermal power unit's heat storage coefficient and train the model based on the predicted thermal power unit heat storage dataset from Step 3.

[0010] Step 5: Input the data of the thermal power unit during operation into the thermal power unit heat storage coefficient prediction model trained in Step 4 to generate the predicted thermal power unit heat storage coefficient and construct the correspondence between the predicted thermal power unit heat storage coefficient and the load command advance amount.

[0011] Step 6: Based on the status of the thermal power unit and the energy storage charge status of the energy storage system, generate the load command to be executed by the thermal power unit; and in the correspondence between the predicted thermal power unit heat storage coefficient and the load command advance, determine the corresponding load command advance based on the predicted thermal power unit heat storage coefficient, and dynamically adjust the load command to be executed by the thermal power unit based on the load command advance.

[0012] Preferably, the thermal power unit status in step 1 includes load increase, load decrease, and stable state; the load increase is when the AGC command is greater than the current load of the unit, the load decrease is when the AGC command is less than the current load of the unit, and the stable state is when the AGC command is equal to the current load of the unit.

[0013] Preferably, in step 2, determining whether the energy storage output of the energy storage system meets the AGC command means that when the energy storage state of charge of the energy storage system is not within the adjustment range, the energy storage output of the energy storage system cannot meet the AGC command.

[0014] Preferably, the initial thermal power unit heat storage coefficient C in step 3 is calculated using the formula shown in equation (1):

[0015] (1);

[0016] In the formula, C is the thermal storage coefficient of the thermal power unit; m represents the main steam flow rate; p is the steam drum pressure; and t is the time.

[0017] Preferably, the predicted thermal power unit heat storage dataset in step 3 includes the input set {X} i} and output set {Y i}, the input set {X i The input set {X} represents the unit parameters for the 300 seconds prior to time t0, and the drum pressure and main steam flow rate at time t1. These unit parameters include load, main steam flow rate, feedwater flow rate, main steam pressure setpoint, actual main steam pressure, total air volume, fuel quantity, main steam temperature, furnace negative pressure, integrated valve position, load change rate, and drum pressure. Based on different times, i input samples are constructed to form the input set {X}.i};

[0018] The output set {Y i Let} be the heat storage coefficient of the thermal power unit at time t0. Based on different times, construct i output samples to form the output set {Y}. i}

[0019] Preferably, the thermal power unit heat storage coefficient prediction model includes an input layer, a convolutional neural network layer, a bidirectional gated recurrent unit layer, a multi-head self-attention mechanism layer, and an output layer;

[0020] The convolutional neural network layer includes two convolutional layers and one pooling layer; the bidirectional gated recurrent unit layer includes two GRU networks; the multi-head self-attention mechanism layer and the output layer are connected through a fully connected layer.

[0021] Preferably, in step 4, the thermal power unit thermal storage coefficient prediction model is trained based on the predicted thermal power unit thermal storage dataset from step 3, specifically implemented according to the following steps:

[0022] Step 4.1, input set {X i The input layer contains a 13×16 vector of input features.

[0023] Step 4.2: Extract the feature vectors of the dataset through convolutional layers, and then reduce the extracted feature vectors to one-dimensional data through pooling layers;

[0024] Step 4.3: The bidirectional gated recurrent unit layer performs bidirectional training on the one-dimensional data output in step 4.2 to learn the deeper temporal characteristics between the unit operation data and the thermal power unit heat storage coefficient. Then, all the one-dimensional data output after training is used as the input data of the multi-head self-attention mechanism layer.

[0025] Step 4.4, the hidden state h output by the multi-head self-attention mechanism layer to the bidirectional gated recurrent unit layer. t Assign different weight values;

[0026] Step 4.5, the output layer assigns different weight values ​​to the hidden states h from step 4.4. t After using Tanh as the activation function for mapping, the predicted value of the thermal power unit's heat storage coefficient is output. The predicted value of the thermal power unit's heat storage coefficient is then mapped to the interval (-1,1) to obtain the predicted thermal power unit's heat storage coefficient.

[0027] Step 4.6, compare the predicted thermal power unit heat storage coefficient with the output set {Y} i The comparison is performed, and steps 4.1 to 4.5 are repeated for multiple iterations of training until the thermal power unit heat storage coefficient prediction model converges.

[0028] Preferably, in step 4.4, the state h is hidden.t Different weight values ​​are assigned, as shown in formulas (2) and (3):

[0029] (2);

[0030] (3);

[0031] In the formula, Let be a GRU network, and q be the prediction vector of the GRU network. Let t be the GRU hidden state. Weights for different time steps t; To calculate the correlation score between the hidden state and the prediction vector at time step t; The normalization factor for time step t;

[0032] Step 4.5 is specifically shown in formula (4):

[0033] (4);

[0034] In the formula, x is the input value and e is the base of the natural logarithm.

[0035] Preferably, the correspondence between the predicted thermal power unit heat storage coefficient and the load command advance is as follows: when the predicted thermal power unit heat storage coefficient is -1, the load command advance is -10; when the predicted thermal power unit heat storage coefficient is -0.6, the load command advance is -5; when the predicted thermal power unit heat storage coefficient is -0.4, the load command advance is -2; when the predicted thermal power unit heat storage coefficient is -0.1, the load command advance is -0; when the heat storage coefficient is 0, the load command advance is 0; when the predicted thermal power unit heat storage coefficient is 0.1, the load command advance is 0; when the predicted thermal power unit heat storage coefficient is 0.4, the load command advance is 2; when the predicted thermal power unit heat storage coefficient is 0.6, the load command advance is 5; and when the predicted thermal power unit heat storage coefficient is 1, the load command advance is 10.

[0036] Preferably, in step 6, a load command to be executed by the thermal power unit is generated based on the state of the thermal power unit and the energy storage charge state of the energy storage system; and in the correspondence between the predicted thermal power unit heat storage coefficient and the load command advance, the corresponding load command advance is determined based on the predicted thermal power unit heat storage coefficient, and the load command executed by the thermal power unit is dynamically adjusted based on the load command advance. Specifically,

[0037] When the thermal power unit is under increased load and the energy storage system's state of charge is less than the minimum output limit, the load command advance corresponding to the thermal power unit's heat storage coefficient is increased in the generated load command to be executed by the thermal power unit; when the thermal power unit is under decreased load and the energy storage system's state of charge is greater than the energy storage charging limit, the load command advance corresponding to the thermal power unit's heat storage coefficient is reduced in the generated load command to be executed by the thermal power unit.

[0038] The beneficial effects of this invention are:

[0039] This invention relates to a load advance control method for power units based on battery energy storage status. It reads relevant parameters of the energy storage battery and the thermal power unit in real time, and determines the load increase, decrease, and stable state of the thermal power unit based on the unit's AGC commands. Based on the energy storage battery's state of charge (SOC) and output parameters, it determines whether the battery can meet the load change requirements of the thermal power unit. If the energy storage battery cannot meet the requirements, it automatically adjusts the load advance control setting by combining the thermal power unit's real-time operating conditions with a predicted heat storage coefficient. This method fully utilizes the thermal power unit's heat storage while avoiding overcharging or over-discharging of the energy storage system, thus extending the system's lifespan and comprehensively considering both the energy storage status and the unit's condition. It utilizes the thermal power unit's heat storage under different operating conditions, improving the unit's load regulation performance. Attached Figure Description

[0040] Figure 1 is a flowchart of the unit load advance control method based on battery energy storage status according to the present invention.

[0041] Figure 2 is a schematic diagram of the thermal power unit heat storage coefficient prediction model in the unit load advance control method based on battery energy storage status of the present invention.

[0042] Figure 3 is a schematic diagram of the unit's response to AGC command load changes in Embodiment 3 of the present invention. Embodiments of the present invention

[0043] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0044] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0045] Example 1

[0046] The present invention provides a unit load advance control method based on battery energy storage status, as shown in Figure 1, which is implemented according to the following steps:

[0047] Step 1: Determine the status of the thermal power unit based on the Automatic Generation Control (AGC) command and the current load of the thermal power unit;

[0048] In step 1, the status of the thermal power unit includes load increase, load decrease, and stable state. The load increase is when the AGC command is greater than the current load of the thermal power unit, the load decrease is when the AGC command is less than the current load of the thermal power unit, and the stable state is when the AGC command is equal to the current load of the thermal power unit.

[0049] Step 2: Determine whether the energy storage output of the energy storage system meets the AGC command.

[0050] When the grid AGC command for thermal power units changes, the energy storage system works in conjunction with the thermal power units to output power. Considering the lifespan of the energy storage system, the state of charge (SOC) of the energy storage is generally maintained at 0.2~0.9, which is the regulation range of the energy storage system;

[0051] When the SOC of the energy storage system is outside the regulation range, the energy storage system will not respond to the AGC command;

[0052] When the AGC command exceeds the dead zone of energy storage regulation, and the SOC is within the allowable range, the energy storage responds to the AGC command with the rated maximum power.

[0053] Dead zone refers to the set range within which the energy storage system will not perform adjustment actions when the changes in energy storage regulation are within a certain range;

[0054] Step 3: When the energy storage state of charge (SOC) is not within the adjustment range, obtain the initial thermal power unit heat storage coefficient based on the thermal power unit status, and then construct a predicted thermal power unit heat storage dataset.

[0055] When the state of charge (SOC) of the energy storage is outside the regulation range, the time it takes for the thermal power unit to reach the AGC command will be prolonged. In order to ensure that the response capability of the thermal power unit does not decrease and to make full use of the output of the thermal power unit, the computer unit anti-load is adjusted.

[0056] The amount of advance load of a generating unit mainly depends on the thermal power unit's heat storage status when AGC commands change. When the thermal power unit has sufficient heat storage, it can provide a larger amount of advance load, and the combined output of energy storage can respond to AGC commands faster and more accurately, better meeting the grid regulation needs. When the thermal power unit has insufficient heat storage, its actual output in a short period of time is far less than that of a thermal power unit under normal conditions, and the output of the energy storage system is also small and cannot quickly compensate for the insufficient actual output of the thermal power unit, resulting in a deficiency in the output of the thermal power energy storage system and the inability to meet the overall AGC indicators.

[0057] The initial thermal power unit heat storage coefficient is defined as the ratio of the change in steam flow rate stored in the water-cooled walls, steam drum, and superheater to the rate of change in steam drum pressure, as shown in equation (1):

[0058] (1);

[0059] In the formula, C is the initial thermal power unit heat storage coefficient; m represents the main steam flow rate; p is the steam drum pressure; and t is the time.

[0060] Furthermore, the thermal power unit's heat storage coefficient C0 at the time of the AGC command change t0 is determined by the thermal power unit's combustion status. Due to the inertia of boiler combustion, the heat storage coefficient can be predicted based on the thermal power unit's operating data.

[0061] Step 3 predicts the thermal power unit's heat storage dataset, which includes the input set {X}. i} and output set {Y i}, the input set {X i The input set {X} represents the parameters of the thermal power unit 300 seconds before time t0 and the steam drum pressure and main steam flow rate of the unit at time t1. These parameters include load, main steam flow rate, feedwater flow rate, main steam pressure setpoint, actual main steam pressure, total air volume, fuel quantity, main steam temperature, furnace negative pressure, integrated valve position, load change rate, and steam drum pressure. Based on different times, i input samples are constructed to form the input set {X}. i};

[0062] The output set {Y i Let} be the heat storage coefficient of the thermal power unit at time t0. Based on different times, construct i output samples to form the output set {Y}. i}

[0063] Step 4: Construct a prediction model for the thermal power unit's heat storage coefficient and train the model based on the predicted thermal power unit heat storage dataset from Step 3.

[0064] Step 5: Input the data from the operation of the thermal power unit into the thermal power unit heat storage coefficient prediction model trained in Step 4 to generate the predicted thermal power unit heat storage coefficient. Construct the correspondence between the predicted thermal power unit heat storage coefficient and the load command advance amount. The specific correspondence is shown in the table below:

[0065]

[0066] Step 6: Based on the status of the thermal power unit and the energy storage charge status of the energy storage system, generate the load command to be executed by the thermal power unit; and in the correspondence between the predicted thermal power unit heat storage coefficient and the load command advance, determine the corresponding load command advance based on the predicted thermal power unit heat storage coefficient, and dynamically adjust the load command to be executed by the thermal power unit based on the load command advance.

[0067] Specifically, when the thermal power unit is under increased load and the energy storage system's state of charge is less than the minimum output limit, the load command advance corresponding to the thermal power unit's heat storage coefficient is increased in the generated load command to be executed by the thermal power unit; when the thermal power unit is under decreased load and the energy storage system's state of charge is greater than the energy storage charging limit, the load command advance corresponding to the thermal power unit's heat storage coefficient is reduced in the generated load command to be executed by the thermal power unit.

[0068] Example 2

[0069] Based on Example 1, as shown in Figure 2, the thermal power unit heat storage coefficient prediction model in the unit load advance control method based on battery energy storage status of the present invention includes a convolutional neural network (CNN) layer, a bi-directional gated recurrent unit (BiGRU) layer, a multihead self-attention (MSA) layer, and an output layer.

[0070] The convolutional neural network layer includes two convolutional layers and one pooling layer; the multi-head self-attention mechanism layer and the output layer are connected through a fully connected layer.

[0071] In step 4, the thermal power unit heat storage coefficient prediction model is trained based on the predicted thermal power unit heat storage dataset from step 3. This is specifically implemented according to the following steps:

[0072] Step 4.1, input set {X i The input layer contains a 13×16 vector of input features.

[0073] Step 4.2: Extract the feature vectors of the dataset through convolutional layers, and then reduce the extracted feature vectors to one-dimensional data through pooling layers;

[0074] Step 4.3: The bidirectional gated recurrent unit layer consists of two GRU networks. One processes time series data from front to back, and the other processes time series data from back to front. The one-dimensional data output from Step 4.2 is trained bidirectionally to learn the deeper time series characteristics between the unit operation data and the thermal power unit heat storage coefficient. Then, all the one-dimensional data output after training is used as the input data of the multi-head self-attention mechanism layer.

[0075] Step 4.4, the hidden state h output by the multi-head self-attention mechanism layer to the bidirectional gated recurrent unit layer. t Assign different weight values; specifically:

[0076] The first step is to process the hidden state h at each time step.t Using functions Calculate the relevance score ;

[0077] The second step is to process all time steps. Perform softmax normalization to obtain normalized weights at different time steps. ;

[0078] The third step is to use... As weights, for all hidden states h t By performing a weighted sum, we can finally obtain the hidden state h. t Assign different weight values;

[0079] Specifically, as shown in formulas (2) and (3):

[0080] (2);

[0081] (3);

[0082] In the formula, Let be a GRU network, and q be the prediction vector of the GRU network. Let t be the GRU hidden state. The weights for different time steps t, The normalization factor for the time step; To calculate the correlation score between the hidden state and the prediction vector at time step t, so as to highlight the impact of key features on the prediction results, and enable the prediction model to more effectively focus on the time points that have a greater impact on the prediction results.

[0083] Step 4.5, the output layer assigns different weight values ​​to the hidden states h from step 4.4. t After using Tanh as the activation function for mapping, the predicted value of the thermal power unit's heat storage coefficient is output. The predicted value of the thermal power unit's heat storage coefficient is then mapped to the interval (-1,1) to obtain the predicted thermal power unit's heat storage coefficient, as shown in formula (4):

[0084] (4);

[0085] In the formula, x is the input value and e is the base of the natural logarithm.

[0086] Step 4.6, compare the predicted thermal power unit heat storage coefficient with the output set {Y} i The comparison is performed, and steps 4.1 to 4.5 are repeated for multiple iterations of training until the thermal power unit heat storage coefficient prediction model converges.

[0087] Example 3

[0088] This embodiment of the unit load advance control method based on battery energy storage status is implemented according to the following steps:

[0089] Step 1: Determine the status of the thermal power unit based on the AGC command and the current load of the thermal power unit;

[0090] The generator set is determined to be in a state of increasing load, decreasing load, or steady state based on the Automatic Generation Control (AGC) command and the current load of the thermal power unit. An AGC command greater than the current load indicates an increasing load, an AGC command less than the current load indicates a decreasing load, and an AGC command equal to the current load indicates a steady state.

[0091] Step 2: Analyze whether the energy storage output of the thermal power unit and the energy storage system meets the AGC command;

[0092] As shown in Figure 3, when the unit's AGC command changes at time t0, the AGC command is converted into a load command after passing through a set rate limit, and the unit adjusts the load according to the load command. When the energy storage SOC is within the allowable range, the energy storage output P... e Then the load of the thermal power unit at this time is the load P1 of the superimposed energy storage system.

[0093] Step 3: Establish the thermal power unit heat storage coefficient and predict the thermal power unit heat storage dataset;

[0094] As shown in Figure 3, when the unit responds to the AGC command, the heat storage is maintained for a short period of time. That is, the unit receives the AGC command change at time t0, and the heat storage of the thermal power unit is exhausted at time t1. The unit then relies on the energy of the fuel system to respond to the load. Therefore, the heat storage coefficient C0 of the unit at time t0 when it receives the AGC command change is as shown in formula (5):

[0095] (5);

[0096] In the formula, C0 is the heat storage coefficient when the unit's AGC command changes to t0. The main steam flow rate at time t0 when the unit's AGC command changes. The main steam flow rate when the unit's AGC command changes by t1;

[0097] The time interval between t0 and t1 is selected based on the characteristics of the thermal power unit. When the unit is selected as 120s, the heat storage coefficient C0 at t0, when the unit's AGC command changes, is determined by the unit's combustion status. Due to the inertia of boiler combustion, the heat storage coefficient can be predicted based on the unit's operating data. From the distributed control system (DCS) of the thermal power unit, the unit parameters for the 300 seconds prior to t0 are selected, including load, main steam flow rate, feedwater flow rate, main steam pressure setpoint, actual main steam pressure, total air volume, fuel quantity, main steam temperature, furnace negative pressure, integrated valve position, load change rate, and drum pressure, as well as the drum pressure and main steam flow rate at t1. These unit operating data are combined into an input sample. Based on the model training requirements, i input samples are constructed, and these constructed samples form the input set {X}. i}. Based on formula (2), the heat storage coefficient of the thermal power unit at time t0 is calculated as the output sample. Similarly, i output samples are constructed, and the constructed samples are combined into the output set {Y}. i}

[0098] Step 4: Construct a prediction model for the thermal power unit's heat storage coefficient and train the model based on the predicted thermal power unit heat storage dataset from Step 3.

[0099] Step 5: Input the data from the unit's operation into the thermal power unit heat storage coefficient prediction model trained in Step 4 to generate the predicted thermal power unit heat storage coefficient. Then, analyze the correlation between the predicted thermal power unit heat storage coefficient and the load command lead amount.

[0100] Step 6: Based on the status of the thermal power unit and the energy storage charge status of the energy storage system, generate the load command to be executed by the thermal power unit; and in the correspondence between the predicted thermal power unit heat storage coefficient and the load command advance, determine the corresponding load command advance based on the predicted thermal power unit heat storage coefficient, and dynamically adjust the load command to be executed by the thermal power unit based on the load command advance.

[0101] Specifically, when the thermal power unit is under increased load and the energy storage SOC of the energy storage system is less than the minimum output limit, the load command of the thermal power unit is increased by a calculated load command lead, as shown in Figure 3. The superimposed load lead enables the thermal power unit to respond to AGC load requirements more quickly and accurately. When the thermal power unit is under decreased load and the energy storage SOC of the energy storage system is greater than the energy storage charging limit, the thermal power unit reduces the corresponding load according to the calculated load command lead to respond to AGC commands more quickly.

[0102] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0103] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0104] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Those skilled in the art can readily implement the present invention based on the accompanying drawings and the above description. However, any modifications, alterations, or variations made by those skilled in the art without departing from the scope of the present invention, utilizing the disclosed technical content, are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A method for unit load advance control based on battery energy storage state, characterized in that, The method is implemented according to the following steps: Step 1, judging the state of the thermal power unit according to the AGC instruction and the current load of the thermal power unit; Step 2, judging whether the energy storage output of the energy storage system meets the AGC instruction, if the energy storage output does not meet the AGC instruction, executing steps 3 to 6, if the energy storage output meets the AGC instruction, not executing; Step 3, obtaining the initial thermal storage coefficient of the thermal power unit according to the state of the thermal power unit, and then constructing a predicted thermal storage data set of the thermal power unit; Step 4, constructing a thermal storage coefficient prediction model of the thermal power unit and training the thermal storage coefficient prediction model of the thermal power unit based on the predicted thermal storage data set of the thermal power unit in step 3; Step 5, inputting the data of the running thermal power unit into the thermal storage coefficient prediction model of the thermal power unit trained in step 4, generating a predicted thermal storage coefficient of the thermal power unit, and constructing a corresponding relationship between the predicted thermal storage coefficient of the thermal power unit and the load instruction lead; Step 6, generating a load instruction executed by the thermal power unit according to the state of the thermal power unit and the state of charge of the energy storage system, and in the corresponding relationship between the predicted thermal storage coefficient of the thermal power unit and the load instruction lead, determining the corresponding load instruction lead according to the predicted thermal storage coefficient of the thermal power unit, and dynamically adjusting the load instruction executed by the thermal power unit according to the load instruction lead.

2. The battery state of energy storage based unit load look-ahead control method of claim 1, wherein, The state of the thermal power unit in step 1 includes load increase, load decrease and stable state, the load increase is that the AGC instruction is greater than the current load of the thermal power unit, the load decrease is that the AGC instruction is less than the current load of the thermal power unit, and the stable state is that the AGC instruction is equal to the current load of the thermal power unit.

3. The battery state of energy storage based unit load look-ahead control method of claim 1, wherein, In step 2, whether the energy storage output of the energy storage system meets the AGC instruction is specifically that when the state of charge of the energy storage system is not in the adjustment range, the energy storage output of the energy storage system cannot meet the AGC instruction.

4. The battery state of energy storage based unit load look-ahead control method of claim 1, wherein, In step 3, the calculation formula of the initial thermal storage coefficient C of the thermal power unit is: （1）； In the formulae m represents the main steam flow, p is the drum pressure, and t is the time.

5. The battery state of energy storage based unit load look-ahead control method of claim 1, wherein, The step 3 of predicting the thermal storage data set of the thermal power generating unit comprises an input set {X i} and an output set {Y i}. The input set {X i} is the unit parameters 300 seconds before the time t0 and the drum pressure and main steam flow of the unit at the time t1, the unit parameters including load, main steam flow, feed water flow, main steam pressure set value, main steam pressure actual value, total air volume, fuel quantity, main steam temperature, furnace negative pressure, comprehensive valve position, load change rate, drum pressure; according to different time, i input samples are constructed to form the input set {X i}. The output set {Y i} is the thermal storage coefficient of the thermal power unit at time t0, and the output set {Y i} is formed by constructing i output samples according to different time.

6. The battery state of energy storage based unit load look-ahead control method of claim 5, wherein, The thermal storage coefficient prediction model of the thermal power unit includes an input layer, a convolutional neural network layer, a bidirectional gated recurrent unit layer, a multi-head self-attention mechanism layer and an output layer; The convolutional neural network layer includes two convolutional layers and a pooling layer, the bidirectional gated recurrent unit layer includes two GRU networks, and the multi-head self-attention mechanism layer and the output layer are connected through a fully connected layer.

7. The battery state of energy storage based unit load look-ahead control method of claim 6, wherein, In step 4, the thermal storage coefficient prediction model of the thermal power unit is trained based on the predicted thermal storage data set of the thermal power unit in step 3, which is implemented according to the following steps: Step 4.1, the input set {X i} is input into the input layer, and the vector of input features is 13x16; Step 4.2, extracting the feature vectors of the data set through the convolutional layer, and reducing the extracted feature vectors to one-dimensional data through the pooling layer; Step 4.3, the bidirectional gated recurrent unit layer performs bidirectional training on the one-dimensional data output by step 4.2, learns the deeper time sequence features between the thermal power unit operation data and the thermal storage coefficient of the thermal power unit, and then takes all the trained one-dimensional data as the input data of the multi-head self-attention mechanism layer; Step 4.4, the multi-head self-attention mechanism layer takes the hidden state h output by the bidirectional gated recurrent unit layer as input t assigning different weight values; Step 4.5, the output layer assigns different weight values to the hidden state h of step 4.4 t After mapping by using Tanh as the activation function, the output is the predicted value of the thermal storage coefficient of the thermal power unit, the predicted value of the thermal storage coefficient of the thermal power unit is mapped to the interval (-1, 1), and the predicted thermal storage coefficient of the thermal power unit is obtained. Step 4.6, compare the predicted heat storage coefficient of the thermal power unit with the output set {Y i} Repeat steps 4.1-4.5 for multiple iterations until the thermal power unit heat storage coefficient prediction model converges.

8. The battery state of energy storage based unit load look-ahead control method of claim 7, wherein, The step 4.4 is a hidden state h t Different weight values are assigned, as shown in equation (2) and equation (3): （2）； （3）； In the formulae, for the GRU network, q is a prediction vector of the GRU network, GRU hidden state for time step t, weights for different time steps t; to compute a relevance score of the hidden state at time step t and the prediction vector; is a normalization factor for time step t; Step 4.5 is specifically shown in formula (4): （4）； In the formula, x is the input value, and e is the natural logarithm base.

9. The battery state of energy storage based unit load look-ahead control method of claim 1, wherein, The corresponding relationship between the predicted heat storage coefficient of the thermal power unit and the load instruction advance amount is: When the predicted heat storage coefficient of the thermal power unit is -1, the load instruction advance amount is -10; When the predicted heat storage coefficient of the thermal power unit is -0.6, the load instruction advance amount is -5; When the predicted heat storage coefficient of the thermal power unit is -0.4, the load instruction advance amount is -2; When the predicted heat storage coefficient of the thermal power unit is -0.1, the load instruction advance amount is 0; When the predicted heat storage coefficient of the thermal power unit is 0, the load instruction advance amount is 0; When the predicted heat storage coefficient of the thermal power unit is 0.1, the load instruction advance amount is 0; When the predicted heat storage coefficient of the thermal power unit is 0.4, the load instruction advance amount is 2; When the predicted heat storage coefficient of the thermal power unit is 0.6, the load instruction advance amount is 5; When the predicted heat storage coefficient of the thermal power unit is 1, the load instruction advance amount is 10.

10. The battery state of energy storage based unit load look-ahead control method of claim 1, wherein, In step 6, the load instruction executed by the thermal power unit is generated according to the state of the thermal power unit and the energy storage state of the energy storage system; and in the corresponding relationship between the predicted heat storage coefficient of the thermal power unit and the load instruction advance amount, the corresponding load instruction advance amount is determined according to the predicted heat storage coefficient of the thermal power unit, and the load instruction executed by the thermal power unit is dynamically adjusted according to the load instruction advance amount, specifically, When the thermal power unit is in the ascending load state and the energy storage state of the energy storage system is less than the minimum output limit, the load instruction advance amount corresponding to the heat storage coefficient of the thermal power unit is increased in the generated load instruction executed by the thermal power unit; when the thermal power unit is in the descending load state and the energy storage state of the energy storage system is greater than the energy storage charging limit, the load instruction advance amount corresponding to the heat storage coefficient of the thermal power unit is reduced in the generated load instruction executed by the thermal power unit.

Images